article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in this paper we consider triangular systems of the form @xmath5 where @xmath6 and @xmath7 are real sequences , and @xmath8 , @xmath9 and @xmath3 are continuous functions .
observe that system ( [ problema1 ] ) preserves the fibration of the plane given by @xmath10 , that is , it sends fibers of @xmath11 to other ones .
we will assume that there exists a point @xmath12 which is a local stable attractor of the subsystem @xmath13 .
in this case , we say that system ( [ problema1 ] ) has a _ local attractive fiber _ @xmath14 .
our objective is to know if the asymptotic dynamics of the orbits corresponding to points in the _ basin of attraction _ of the _ limit fiber _
@xmath2 is characterized by the dynamics on this fiber , the _ limit dynamics_. in all the cases , we will assume that the limit dynamics is very simple , that is : either ( a ) the fiber @xmath2 contains a global attractor , see proposition [ propoprob1attractor ] ; ( b ) the fiber is filled by fixed points , theorem [ propoprob1identitatv2](a ) ; or ( c ) it is filled by @xmath4-periodic orbits , theorem [ propoprob1identitatv2](b ) .
observe that there is no need to consider the case in which there is a global repellor on the fiber @xmath2 , since in this situation it is clear that any orbit with initial conditions in the basin of attraction of this fiber is unbounded ( otherwise there should be accumulation points in the fiber ) .
the paper is structured as follows . in section
[ main ] we present the main results ( proposition [ propoprob1attractor ] , theorem [ propoprob1identitatv2 ] and corollary [ corolaridiff ] ) together with some motivating examples .
the proofs are given in section [ proofs ] .
section [ aplications1 ] is devoted to show some applications of the main results .
for instance , in section [ quasi1 ] we study the limit dynamics of some linear quasi - homogeneous maps .
the example considered there shows that , for this class of maps , the shape of the basins of attraction of the origin can have a certain level of complexity . in section [ secondordermultiplicative ]
we apply our results to study the global dynamics of difference equations of multiplicative type @xmath15 , with @xmath16 being a @xmath17 function . in particular , we reobtain the results in @xcite concerning the difference equations @xmath18 the more general recurrences @xmath19 are studied in section [ higher ] .
we also apply the results to study the additive difference equations @xmath20 in section [ aplications3 ] . finally , we also present several different type of recurrences for which the theory developed in this work applies in order to obtain a full characterization of their dynamics .
we started motivated by the following clarifying example , inspired in the ones developed for continuous systems in @xcite ( a particular case is also considered in @xcite , see also @xcite and @xcite ) , which shows that there are systems of type @xmath21 with @xmath22 , such that @xmath23 is a global limit fiber , having a global attractor for the restricted dynamics on @xmath23 , but having unbounded orbits @xmath24 , with @xmath25 , in their the global dynamics .
_ example a : hyperbolic globally attracting fiber with a restricted global attractor but having unbounded solutions . _
consider the system @xmath26 with @xmath22 , @xmath27 and @xmath28 .
of course , this system preserves the fibration @xmath29 , and it has a _ global _ attracting limit fiber @xmath23 . on this fiber
the dynamics is the following : the origin is attractive if @xmath30 ; the fiber is a continuum of fixed points if @xmath31 ; and the fiber is a continuum of @xmath4periodic orbits if @xmath32 .
however , as shown in proposition [ propocascurios ] , if @xmath33 then there are initial conditions giving rise to unbounded solutions , and therefore the global dynamics is not characterized by the dynamics on the global limit fiber . [ propocascurios ] the curve @xmath34 is invariant for system ( [ cascuriosgeneral ] ) .
moreover , for any initial condition @xmath35 the associated orbit is given by @xmath36 , @xmath37 .
if @xmath33 , then any initial condition @xmath35 gives rise to unbounded solutions , with @xmath25 .
imposing that system ( [ cascuriosgeneral ] ) has solutions of the form @xmath38 , @xmath37 , we easily get @xmath39 observe that if @xmath40 ( that is when @xmath41 ) we obtain that @xmath42 is an invariant curve . moreover , if @xmath33 , then @xmath43 , and the orbits on @xmath42 are unbounded solutions .
example a shows that , in general , for systems of type ( [ triang ] ) with terms of degree greater or equal than @xmath4 in @xmath44 , we can not expect that the global dynamics in the basin of attraction of the limit fiber @xmath23 is characterized by the dynamics on the fiber .
however , it remains to explore the case of systems of type ( [ problema1 ] ) with affine terms in @xmath44 .
for these systems , the possible limit dynamics that we consider are the cases ( a ) , ( b ) and ( c ) mentioned in the introduction .
the first result concerns the case ( a ) , when the the limit fiber contains a global attractor .
this case is characterized by the fact that @xmath45 , and the attractor is @xmath46 ; [ propoprob1attractor ] consider the system ( [ problema1 ] ) with @xmath8 and @xmath9 continuous and @xmath45 .
suppose that @xmath12 is an attractive point of @xmath13 .
then , for all initial condition @xmath47 with @xmath48 in the basin of attraction of @xmath12 , we have @xmath49 .
we want to point out that the above one is not a local result .
in fact , the convergence is guaranteed for all @xmath47 such that @xmath48 is in the basin of attraction of @xmath50 it is not difficult to see that in the case ( b ) , when the limit fiber @xmath2 is filled by fixed points , there are systems such that the orbits corresponding to initial conditions in the basin of attraction of the limit fiber tend to some of the fixed points on the limit fiber , which depend on the initial conditions . indeed : _ example b : hyperbolic attracting fiber and fast enough convergence to the limit dynamics .
_ consider the system given by @xmath51 with @xmath22 , and @xmath28 .
then @xmath52 observe that the infinite product @xmath53 is convergent since , as @xmath22 , we have @xmath54 so , for each initial condition @xmath47 the associated orbit converges to the fixed point @xmath55 .
the next example shows , however , that the above situation does not occur when the convergence of @xmath56 to @xmath57 is too slow . _
example c : hyperbolic attracting fiber but slow convergence to the limit dynamics . _
consider the system @xmath58 with @xmath22 , and @xmath59 again , an straightforward computation gives @xmath60 using that @xmath61 , which is a divergent series , we get that the infinite product @xmath62 is also divergent .
hence each initial condition @xmath47 gives rise to an unbounded sequence @xmath6 and , therefore , the dynamics on the basin of attraction of the limit fiber @xmath23 is not characterized by the dynamics on the fiber .
the above example leads us to introduce the following definition .
as we will see in example e below , it gives an optimal characterization of the speed of convergence of the terms @xmath63 and @xmath56 to guarantee the convergence of the orbits in the basin of attraction of the limit fiber .
[ fecld ] if @xmath64 is a stable attractor of @xmath13 , we say that the attracting fiber @xmath2 of system ( [ problema1 ] ) with @xmath65 , has _ the fast enough convergence to the limit dynamics property
_ if there exists @xmath66 such that if @xmath7 is a solution of @xmath13 satisfying @xmath67 for some @xmath68 , then there exists @xmath69 such that @xmath70 for all @xmath71 , @xmath72 , and also there exist two functions @xmath73 such that for all @xmath74 it is satisfied : 1 .
2 . @xmath76 .
@xmath77 and @xmath78 are non decreasing for @xmath79 ; 4 .
@xmath80 and @xmath81 .
let @xmath12 be a stable attractive fixed point of @xmath82 .
we say that @xmath82 is _ locally contractive _ at @xmath12 if there exists an open neighborhood @xmath83 of @xmath64 such that any @xmath84 @xmath85 let @xmath86 be a continuous function and let @xmath64 be a stable attractive fixed point of @xmath82 . 1 .
if @xmath3 is an orientation preserving function , then @xmath82 is locally contractive at @xmath12 .
2 . there exist orientation reversing functions @xmath82 which are not locally contractive at @xmath12 .
if @xmath87 and @xmath12 is an hyperbolic fixed point of @xmath82 then @xmath82 is locally contractive at @xmath12 .
now we are ready to present our result about the cases ( b ) and ( c ) .
observe that if @xmath88 and @xmath89 , then the limit fiber is filled by fixed points ( if @xmath90 then there are unbounded solutions ) .
when @xmath91 , the fiber is filled by @xmath4-periodic orbits and there is a fixed point given by @xmath92 .
[ propoprob1identitatv2 ] consider system ( [ problema1 ] ) where @xmath8 and @xmath9 are continuous functions , and @xmath65 .
suppose that @xmath2 is an attracting fiber satisfying the fast enough convergence to the limit dynamics property , and consider initial conditions @xmath47 with @xmath48 in the basin of attraction of @xmath12 for the recurrence @xmath93 then 1 . if @xmath88 and @xmath89 , then there exists @xmath94 such that @xmath95 2 . if @xmath91 and additionally @xmath3 is locally contractive at @xmath12 , then @xmath96 and @xmath97 notice that the above result must not be interpreted in the sense that the limit @xmath98 is different for each initial condition @xmath47 .
_ example d : hyperbolic attracting fiber and fast enough convergence to the limit dynamics via theorem [ propoprob1identitatv2 ] .
_ consider the systems given by @xmath99 with @xmath22 , @xmath100 , @xmath101 and @xmath102 .
fixing @xmath103 and taking @xmath104 ; setting @xmath105 , @xmath106 , @xmath107 , @xmath108 with @xmath109 , @xmath110 we have that hypothesis @xmath111 and @xmath112 are trivially fulfilled .
with respect hypothesis @xmath113 we have @xmath114 on the other hand @xmath115 and @xmath116 so @xmath117 is also fulfilled . from theorem [ propoprob1identitatv2 ] , for each initial condition @xmath47 there exists @xmath68 such that @xmath118 , and also there exists @xmath94 such that : if @xmath119 then @xmath120 , and if @xmath121 , then @xmath122 and @xmath123 .
the next example shows that theorem [ propoprob1identitatv2 ] is optimal when @xmath12 is not a hyperbolic attractor of @xmath13 .
_ example e : non - hyperbolic attracting fiber .
optimal characterization of fast enough and slow convergence to the limit dynamics .
_ consider the systems given by @xmath124 with @xmath125 , @xmath126 , and @xmath127 .
take @xmath128 small enough and such that @xmath48 is in the basin of attraction of @xmath129 .
now , we claim that if @xmath130 then there exists @xmath94 such that @xmath120 ; and if @xmath131 then @xmath6 is a divergent sequence if @xmath132 , and @xmath133 if @xmath134 .
to prove this claim , we will use the following result that we learned from s. stevi @xcite .
its proof is attributed to e. jacobsthal , see also ( * ? ? ?
* problem 174 , page 217 ) : [ stev ] let @xmath135 with @xmath136 , be a continuous function such that @xmath137 for every @xmath138 and @xmath139 , where @xmath127 , @xmath140 and @xmath141 are positive .
let @xmath142 and @xmath143 @xmath144
. then @xmath145 taking logarithms in equation ( [ case ] ) we have @xmath146 . by theorem [ stev ]
we have , @xmath147 hence if @xmath148 , then @xmath149 is a convergent sequence , and if @xmath150 then @xmath151 , and the claim is proved .
at this point we will see that the criterium given by theorem [ propoprob1identitatv2 ] is optimal . by fixing @xmath66 small enough , taking @xmath104 , and using theorem [ stev ]
, we have that for the sequence @xmath7 defined by ( [ case ] ) and for all @xmath152 , there exists @xmath68 and @xmath153 such that if @xmath154 then @xmath155 setting @xmath156 @xmath106 , @xmath107 , @xmath157 , and @xmath158 we have that for any @xmath152 , the hypothesis @xmath111@xmath112 are trivially fulfilled , and with respect @xmath117 we have that @xmath115 and @xmath159 observe that @xmath160 is convergent if and only if @xmath161 for all @xmath152 or , in other words , if and only if @xmath162 hence , theorem [ propoprob1identitatv2 ] guarantees convergence of the sequence @xmath163 if @xmath148 for all @xmath126 , which is the optimal value .
the next result shows that if system ( [ problema1 ] ) is a differentiable one , and @xmath12 is a hyperbolic attractor of @xmath13 , then the hypotheses of theorem [ propoprob1identitatv2 ] are fulfilled .
furthermore , each point in the attracting fiber is the limit of an orbit of the basin of attraction of the fiber .
[ corolaridiff ] consider system ( [ problema1 ] ) where @xmath164 where @xmath165 is a neighborhood of @xmath12 , and @xmath65 .
suppose that @xmath12 is a hyperbolic attractor of @xmath3 .
then , for any @xmath48 in the basin of attraction of @xmath12 there exists @xmath166 such that : 1 . if @xmath89 and @xmath88 , then @xmath167 .
2 . if @xmath91 , then @xmath168 and @xmath169 .
furthermore , for any point @xmath170 there exists an initial condition @xmath47 in the basin of attraction of the limit fiber such that @xmath171 .
in this section we prove the main results of the paper , proposition [ propoprob1attractor ] , theorem [ propoprob1identitatv2 ] and corollary [ corolaridiff ] .
first observe that , from the continuity of @xmath9 , there exists @xmath103 such that for all @xmath172 such that @xmath74 , we have @xmath173 with @xmath174 .
consider @xmath47 such that @xmath48 is in the basin of attraction of @xmath12 . from the hypotheses
, we can assume that there exists @xmath68 such that for all @xmath175 , we have @xmath176 hence , since @xmath8 is continuous , @xmath177 .
thus we have @xmath178 .
applying this last inequality we obtain @xmath179 and therefore the sequence is bounded . as a consequence @xmath180 and @xmath181 .
hence , there are subsequences @xmath182 and @xmath183 .
for these subsequences , from equation ( [ problema1 ] ) , and using the continuity of @xmath8 and @xmath9 , we have @xmath184 hence we have that @xmath185 . to prove @xmath186 , since @xmath12 is a stable attractor
, we can take @xmath66 such that for all @xmath172 with @xmath187 @xmath188 in particular , @xmath64 is the only fixed point in the neighborhood @xmath189 observe that since @xmath3 preserves orientation , if @xmath190 then @xmath191 assume that statement @xmath186 is not true . given a sequence @xmath192 , then for each @xmath193 with @xmath194 we can find a point @xmath195 such that @xmath196 for all @xmath193 ( the opposite assumption can be treated in a similar manner ) , and also that the subsequence is monotone decreasing .
notice that , since @xmath197 and @xmath3 preserves orientation , also @xmath198 and the second inequality in ( [ estrella ] ) reads as @xmath199 on the other hand , we can consider the sequence of iterates by @xmath3 of one point @xmath200 such that @xmath201 , obtaining @xmath202 for all @xmath203 and that @xmath204 that is , we can choose a subsequence @xmath205 with the property that @xmath206 and , as before , being monotone decreasing .
considering the continuous function @xmath207 we have that for all @xmath208 @xmath209 hence , for all @xmath203 it exists @xmath210 such that @xmath211 it implies that @xmath212 and @xmath64 is not isolated as a fixed point of @xmath3 . a contradiction with our assumptions . in order to see @xmath213 consider the function @xmath214 this map has two fixed points @xmath215 and @xmath216 since @xmath217 the point @xmath218 is a repellor of the function @xmath219
the point @xmath215 is not an hyperbolic fixed point because @xmath220 but it is easy to see that the interval @xmath221 $ ] is invariant under the action of @xmath3 and that for all @xmath222 and @xmath223 then , @xmath224 is an attracting fixed point of @xmath219 on the other hand , if @xmath225 then @xmath82 satisfies @xmath226 hence condition ( [ locatr1 ] ) is not satisfied .
statement ( c ) is a simple consequence of the mean value theorem . in order to prove statement @xmath186
consider @xmath47 such that @xmath48 is in the basin of attraction of @xmath12 .
first we start proving that the sequence @xmath44 is bounded . from the hypothesis ,
if @xmath66 is small enough , there exists @xmath68 such that for all @xmath175 , @xmath176 .
hence , to simplify the notation , in the following we will assume that the point @xmath47 is such that @xmath227 .
furthermore , 1 . from the fact that @xmath78 is non decreasing for @xmath79 , if @xmath193 is large enough @xmath228 2 .
analogously @xmath229 . in summary
, we have @xmath230 . applying this last inequality we obtain @xmath231+\prod\limits_{j=0}^{n } \left(1+v(p_j)\right)|x_{0}|
. \end{array}\ ] ] observe that from hypothesis @xmath113 we have @xmath232 , and by hypothesis @xmath112 , for all @xmath233 we have that @xmath234 .
so @xmath235 observe that @xmath236 , because @xmath237 hence , regarding that if @xmath193 is large enough then @xmath238 , we have @xmath239 so the sequence @xmath6 is bounded , and therefore also the sequence @xmath163 is bounded .
now we are going to see that @xmath6 is a cauchy sequence and therefore it has a limit @xmath98 . first observe that if @xmath240 is a bound of @xmath241 then @xmath242 therefore , since the series @xmath243 and @xmath160 are convergent , then for all @xmath66 there exists @xmath244 such that for all @xmath245 @xmath246 and therefore @xmath6 is a cauchy sequence , which completes the proof of statement ( a ) . in order to see @xmath213 ,
observe that @xmath247 where @xmath248 and @xmath249 . after renaming @xmath250 , @xmath251 @xmath252 and @xmath253
we get the system @xmath254 which is a system of type ( [ problema1 ] ) .
notice that when we take @xmath47 ( resp .
@xmath255 ) as initial condition and we apply ( [ dospassosy ] ) repeatedly we get @xmath256 ( resp .
@xmath257 ) . we are going to see that system ( [ dospassosy ] ) satisfies the hypothesis @xmath111@xmath117 , and since @xmath258 and @xmath259 , the result will follow from the convergence of the sequence @xmath260 guaranteed by statement ( a ) . adding and subtracting both @xmath261 and @xmath262 to @xmath263 we get , @xmath264 since @xmath64 is a stable attractor of @xmath265 and @xmath82 is locally contractive at @xmath266 if @xmath267 then @xmath268 . using that @xmath269 and @xmath83 satisfy @xmath112 , we get @xmath270 where @xmath271 and @xmath272 .
now we obtain a new function @xmath273 , satisfying @xmath111 for system ( [ dospassosy ] ) . on the other hand , using the inequality @xmath274 where @xmath275 , obtaining the new function @xmath276 , satisfying @xmath113 for system ( [ dospassosy ] ) .
now , it is straightforward to prove that @xmath277 and @xmath278 satisfy hypothesis @xmath112 and @xmath117 , hence each sequence @xmath279 is convergent , and therefore the sequences @xmath280 and @xmath281 are convergent . finally , observe that setting @xmath282 and @xmath283 , and using the continuity of @xmath8 and @xmath9 and equation ( [ problema1 ] ) we obtain that @xmath284 . if @xmath12 is a hyperbolic attractor of @xmath3 , then for @xmath103 small enough , we have that @xmath285 . setting @xmath286 , @xmath287 , and @xmath288 , where @xmath289 and @xmath290 are the supremum in @xmath74 of @xmath291 and @xmath292 respectively , and using the mean value theorem , one gets straightforwardly , that the hypothesis @xmath111@xmath117 are satisfied .
so statements ( a ) and ( b ) follow from theorem [ propoprob1identitatv2 ] .
observe that for each point @xmath170 , the differential matrix of the map associated to system ( [ problema1 ] ) , @xmath293 , is given by @xmath294 where @xmath295 .
hence from the stable manifold theorem ( see @xcite or @xcite for instance ) there is an invariant @xmath17 curve , transversal to @xmath2 at @xmath296 , such that any initial condition on this curve gives rise to a solution with limit @xmath296 . from the above proof , we notice that the statements ( a ) and ( b ) in corollary [ corolaridiff ] are also satisfied in the case that @xmath3 is not differentiable but @xmath297 with @xmath298 , for those values of @xmath172 such that @xmath74 , for @xmath66 small enough .
we say that @xmath299 is a _ quasi - homogeneous function _ with weights @xmath300 and quasi - degree @xmath301 if @xmath302 notice that if @xmath303 is a quasi - homogeneous function with weights @xmath300 then , for all constant @xmath304 , @xmath303 is also a quasi - homogeneous function with weights @xmath305 hence we will consider @xmath306 with gcd@xmath307 we say that @xmath308 is a _ quasi - homogeneous map _ with weights @xmath300 and quasi - degree @xmath301 if @xmath309 and @xmath310 are quasi - homogeneous functions with weights @xmath300 and quasi - degrees @xmath311 and @xmath312 , respectively .
this section deals with quasi - homogeneous maps of quasi - degree @xmath57 ( which are called _ linear quasi - homogeneous maps _ ) with weights of different sign . assuming that @xmath100 and @xmath313 from the definition it follows that in the class of analytic maps , any linear quasi - homogeneous map takes the form @xmath314 where @xmath315 and @xmath316 are analytic functions .
these maps preserve the fibration given by @xmath317 @xmath318 , since they send the curves @xmath319 to the curves @xmath320 the dynamical system associated to @xmath321 is : @xmath322 applying the transformation @xmath323 and calling @xmath324the map @xmath325 is transformed into @xmath326 which is a triangular map , whose corresponding system is : @xmath327 notice that such a system is of the form ( [ problema1 ] ) and the fiber @xmath215 is invariant . in order to apply proposition [ propoprob1attractor ] to study the basin of attraction of the origin , we need that @xmath328 and that @xmath215 is an attractive point of the subsystem @xmath329 which is guaranteed by the condition @xmath330 where @xmath331 and @xmath332 so , in this case we have @xmath333 for all @xmath334 where @xmath335 is the basin of attraction of @xmath336 and any arbitrary value of @xmath337 . as before we can apply proposition [ propoprob1attractor ] .
thus , if @xmath341 and @xmath330 then @xmath342 for all @xmath334 where @xmath335 is the basin of attraction of @xmath336 for the subsystem @xmath343 and @xmath344 is arbitrary .
hence , we get : [ hom ] assume that @xmath328 and @xmath341 and let @xmath335 be the basin of attraction of @xmath215 for the one - dimensional system @xmath345 then for all @xmath346 such that @xmath347 the sequence @xmath348 tends to @xmath349 as @xmath350 _ example f. _ as a particular easy case we can take @xmath351 and @xmath352 which is transformed in : @xmath353 with @xmath354 and @xmath355 to determine the basin of attraction of @xmath215 for arbitrary values of @xmath356 is not an easy task . in fact , the dynamics of the one dimensional system @xmath13 with @xmath357 can be very complicated .
for instance , if such map has a @xmath358-periodic point , then it has periodic points for all the periods .
it means that the _ intrafibration _ dynamics of the hyperbolas @xmath359 for system ( [ qh3 ] ) can be complicated .
one case for which the dynamics of @xmath13 is simple is just when we have three fixed points with alternating stability .
for instance , if we consider @xmath360 and @xmath361 then we get three fixed points @xmath362 and @xmath363 moreover @xmath364 so @xmath129 and @xmath4 are attractive points and @xmath365 and @xmath57 is a repelling point .
there are three preimages of the fixed point @xmath366 that are @xmath366 @xmath367 and @xmath368 .
it easily follows that @xmath369 is contained in @xmath370 the basin of attraction of @xmath336 while @xmath371 is contained in @xmath372 the basin of attraction of @xmath363 setting @xmath373 and @xmath374 for @xmath375 ; and @xmath376 , @xmath377 we have that these intervals are interlaced as @xmath378 where the right extreme of each interval @xmath379 equals the left one of @xmath380 and the right extreme of each interval @xmath381 equals the left one of @xmath382 these boundary points are exactly the preimages of the fixed point @xmath383 furthermore , these intervals have decreasing length , tending to the two - periodic points @xmath384 [ l10 ] consider the one dimensional discrete system generated by @xmath386 let @xmath387(resp .
@xmath388 or @xmath389 ) be the basin of attraction of @xmath215 ( resp .
@xmath390 or infinity ) and let @xmath391 be , the set of preimages of the repulsive fixed point @xmath383 then @xmath392 where @xmath393 is the unique orbit of @xmath82 with minimal period @xmath4 . moreover @xmath394 to prove ( b ) , assume that @xmath405 then , again by lemma , the sequence @xmath7 , where @xmath406 , has four possibilities when @xmath193 goes to infinity : either it tends to @xmath4 , or after some iterates it is constant equal to 1 , or it takes the values @xmath407 , or it tends to infinity .
using @xmath408 , we see that in the first case , when @xmath193 is big enough @xmath409 .
since @xmath398 , it follows that @xmath410 tends to @xmath411 when @xmath193 goes to infinity .
the other cases follow similarly .
the proof of ( ii ) is a consequence of lemma [ l10 ] and the dynamics of on the invariant sets @xmath412 @xmath413 and @xmath414 for instance , on the third one , @xmath415 and clearly the orbits go towards infinity on it .
notice that system presents interesting bifurcation when @xmath416 in particular , as a consequence of the shape of @xmath387 explained above , when @xmath398 , the basin of attraction of @xmath349 of this system is formed by the union of infinitely many disjoint hyperbolic - shaped bands which accumulate to the hyperbolas @xmath417 ; see figure 1 .
we consider the next family of second order multiplicative difference equations @xmath419 where @xmath420 is a @xmath17 function defined in an open set of @xmath421 . multiplying both sides of ( [ g1 ] ) by @xmath422 , and setting @xmath423
we get that equation ( [ g1 ] ) can be written as @xmath424 which has the associated map @xmath425
if we consider the map @xmath426 we obtain @xmath427 hence , by calling @xmath428 and @xmath429 ( respectively @xmath430 and @xmath431 ) we get the system @xmath432 which is of the form ( [ problema1 ] ) . hence we can apply proposition [ propoprob1attractor ] and corollary [ corolaridiff ] to system ( [ g4 ] ) and study the behavior of @xmath256 ( respectively @xmath257 ) .
notice that when we consider the initial conditions @xmath47 ( resp .
@xmath255 and we apply system ( [ g4 ] ) iteratively we get @xmath256 ( resp .
@xmath257 ) . _
example g. _ among the recurrences of type ( [ g1 ] ) , we are going to consider the one given by @xmath433 the global behavior of equation ( [ bliz ] ) is now completely understood after the work done by bajo and liz @xcite .
their approach is based on the computation of the explicit solutions of ( [ bliz ] ) .
previously , particular cases of this equation have been studied by a large numbers of experts , see @xcite and @xcite , the reader is also referred to @xcite for a summary of these previous references concerning this equation .
equation ( [ bliz ] ) has the form ( [ g1 ] ) , where @xmath434 .
furthermore , it corresponds with the systems of type ( [ g3 ] ) : @xmath435 this system has two invariant fibers given by @xmath23 and @xmath436 whose stability is determined by the parameter @xmath437 , since setting @xmath438 we have @xmath439 and @xmath440 .
the global dynamics of equation ( [ bliz ] ) in its _ good set _ ( that is , the set of all initial conditions for which the dynamical system is well defined , @xcite ) @xmath441 $ ] , and then the good set would be @xmath442 .
] , is determined by the graph of @xmath3 .
first it is necessary to know the dynamics of the recurrence @xmath13 which , since @xmath3 is a mbius transformation , is well known .
for instance , using ( * ? ? ?
* corollary 7 ) we get : 1 . if @xmath446 , then for any initial condition @xmath447 , @xmath448 .
2 . if @xmath354 , then for any initial condition @xmath449 , @xmath450 .
3 . if @xmath121 , then for any initial condition in @xmath451 , the sequence @xmath7 is @xmath4-periodic .
4 . if @xmath119 , then for any initial condition in @xmath451 , @xmath448 . 1 . if @xmath452 , then the sequence @xmath453 for all @xmath203 .
2 . if @xmath446 , then for any initial condition @xmath454 such that @xmath455 , @xmath133 .
if @xmath456 then @xmath6 is @xmath4-periodic .
3 . if @xmath354 , then for any initial condition @xmath454 such that @xmath457 we have : if @xmath458 and @xmath459 then @xmath6 tends to a @xmath4-periodic orbit @xmath460 such that @xmath461 ; if @xmath456 then @xmath6 is @xmath4-periodic ; and if @xmath462 then @xmath463 .
4 . if @xmath121 , then for any initial condition @xmath454 such that @xmath457 we have : if @xmath458 and @xmath464 then the solution @xmath6 is unbounded ; if @xmath464 then @xmath6 is @xmath4-periodic ; and if @xmath462 then @xmath6 is @xmath465-periodic .
5 . if @xmath119 , then for any initial condition @xmath454 such that @xmath466 , @xmath133 .
if @xmath462 then @xmath6 is @xmath4-periodic . in order to prove ( b ) we consider @xmath467 then , by lemma [ moeb ] ( a ) @xmath215 is an attractor of the recurrence ( [ moeb2 ] ) in @xmath468 .
we now note that for equation ( [ bliz ] ) , system ( [ g4 ] ) writes as @xmath469 applying proposition [ propoprob1attractor ] to system ( [ dup ] ) we deduce that @xmath470 it implies that @xmath471 and @xmath472 too .
hence , for each initial condition such that @xmath455 @xmath133 . on the other hand , substituting @xmath456 in equation ( [ bliz ] ) we get @xmath473 , obtaining a @xmath4-periodic orbit .
\(c ) if @xmath354 , then by lemma [ moeb ] ( b ) , for all @xmath474 the sequence @xmath475 since @xmath88 and @xmath64 is a hyperbolic attractor of ( [ moeb2 ] ) , we can use corollary [ corolaridiff ] to assert that the sequence @xmath476 converges to a point which depend on the initial condition . then , if we take the initial condition @xmath477 ( resp .
@xmath478 ) we have that @xmath479 ( resp .
@xmath480 ) . then ,
since @xmath481 the condition @xmath482 must be satisfied .
the other assertions of statement @xmath483 are easily deduced from the equation ( [ bliz ] ) .
\(d ) if @xmath121 and @xmath484 then @xmath485 hence @xmath44 is unbounded . if @xmath486 , then @xmath487 and @xmath6 is 2-periodic .
if @xmath488 , then @xmath489 and @xmath6 is 4-periodic . if @xmath490 , then @xmath491 and @xmath6 is 4-periodic .
an straightforward computation shows that the second component of system ( [ nousistema30bis ] ) has the following explicit solution : @xmath495 hence the first equation of system ( [ nousistema30bis ] ) is @xmath496 and we can construct the explicit solution @xmath497 so if @xmath154 for a suitable @xmath68 , we can take logarithms in the above equations , obtaining : @xmath498 observe that , for any value of @xmath141 and @xmath499 we have @xmath500 hence @xmath501 , and the result follows .
the next examples show that the approach in this section can be applied to other families of difference equations not in the class ( [ g1 ] ) .
we will not develop the analysis in this paper , our main purpose is only to illustrate a range of applications of the results in section [ main ] .
_ example h : a third order equation . _ the exact solutions of the third order equation @xmath502 have been obtained in @xcite .
a complete analysis of the dynamics associated to this equation can be done using a similar approach as above .
setting @xmath503 , we have that the subsequences @xmath504 and @xmath505 can be studied using the system @xmath506 where @xmath507 , and @xmath508 for @xmath509 .
hence , as in the case of example g , the sequences @xmath510 can be straightforwardly characterized using the behavior of the real mebius recurrence @xmath511 .
example i. _ consider the equation @xmath512 with @xmath513 , @xmath514 and being @xmath16 a continuous positive function . by multiplying both sides of the equation by @xmath422 ; setting @xmath515 , and taking @xmath516
we get the system of type ( [ problema1 ] ) @xmath517 that can be studied using the results in section [ main ] , if @xmath518 and there is a limit fiber @xmath519 with the fast enough convergence property . the above approach to study equation ( [ g1 ] ) can be applied to higher order multiplicative - type of difference equations .
consider the order @xmath520 difference equation : @xmath521 some straightforward computations using the associated map @xmath522 show that the sets @xmath523 and @xmath524 for @xmath525 such that @xmath526 ( if it exists ) are invariant , and also lead to the following result : 1 . if @xmath529 for all @xmath530 then @xmath453 for all @xmath203
. 2 . suppose that @xmath523 and @xmath531 . if additionally @xmath532 then @xmath6 is @xmath533-periodic ; if @xmath534 then @xmath6 is @xmath520-periodic ; if @xmath535 then @xmath536 ; and if @xmath537 then @xmath538 .
3 . if there exists @xmath525 in @xmath539 such that @xmath526 , and @xmath540 then @xmath6 is @xmath520-periodic ( non - minimal ) .
[ corolgk0 ] consider the equation ( [ gk ] ) being @xmath16 a continuous function defined in an open set @xmath421 .
assume that @xmath527 and @xmath535 , then @xmath215 is an attractor for the system @xmath541 , and for all initial condition @xmath542 such that @xmath543 is in the basin of attraction of @xmath23 we have @xmath133 .
multiplying both sides of ( [ gk ] ) by @xmath544 and setting @xmath545 , we obtain that equation ( [ gk ] ) can be written as @xmath546 hence , we have @xmath547 and @xmath548 . dividing these expressions
we have @xmath549 from the above equation , and after renaming @xmath550 and @xmath551 we get @xmath520 systems of type ( [ problema1 ] ) associated to each initial condition @xmath552 for @xmath553 .
@xmath554 where @xmath555 and @xmath556 .
the result follows because each of these systems are under the hypothesis of proposition [ propoprob1attractor ] .
[ corolgk02 ] consider the equation ( [ gk ] ) being @xmath16 a @xmath17 function defined in an open set @xmath421 .
let @xmath557 be a hyperbolic attractor of @xmath558 .
then , for all initial condition @xmath559 such that @xmath560 is in the basin of attraction of @xmath561 we have that : we consider again the systems ( [ nousistemak30 ] ) which , under the current hypotheses , also satisfy the ones of corollary [ corolaridiff ]
. hence if @xmath562 , there exists @xmath564 for @xmath553 such that each sequence @xmath565 satisfies @xmath566 . if @xmath532 , then there exist @xmath567 and @xmath568 such that @xmath569 and @xmath570 .
it is easy to see that for @xmath571 each subsequence @xmath572 corresponds with the subsequence @xmath573 , obtaining that there are @xmath533 convergent subsequences of @xmath6 . in this section
we consider the second order difference equations @xmath574 we call these equations _ additive _ ones .
equation ( [ s1 ] ) has the associated map @xmath575 which preserves the fibration @xmath576 .
so if @xmath577 is a fixed point of @xmath16 , the map preserves the fiber @xmath578 .
it is easy to observe that if @xmath584 , or @xmath585 and @xmath525 then there are iterates of map @xmath321 on the invariant fiber @xmath578 which are unbounded , and therefore these cases are out of our scope .
in fact it is straightforward to obtain the following result .
consider the equation ( [ s1 ] ) where @xmath16 is a continuous function defined in an open set @xmath421 .
let @xmath586 be a fixed attracting point @xmath587 .
then , for each initial condition @xmath454 such that @xmath588 , we have @xmath589 if @xmath590 ; the orbits are @xmath4-periodic if @xmath591 ; are fixed points if @xmath585 and @xmath592 ; and there are unbounded orbits if @xmath585 and @xmath525 , or @xmath584 .
consider the equation ( [ s1 ] ) being @xmath593 function defined in an open set @xmath421 .
let @xmath586 be a hyperbolic attracting point of @xmath587 .
then , for all initial condition @xmath454 such that @xmath594 is in the basin of attraction of @xmath577 , we have : 1 . if @xmath590 , then @xmath589 .
if @xmath591 and system ( [ s13 ] ) has fast enough convergence to the limit fiber @xmath2 , then @xmath6 tends to a @xmath4-periodic orbit @xmath595 .
if @xmath585 , @xmath592 and system ( [ s13 ] ) has fast enough convergence to the limit fiber @xmath23 , then there exists @xmath596 such that @xmath597 . to prove statements ( b ) and ( c ) , observe that by applying corollary [ corolaridiff ] to system ( [ s13 ] ) , we obtain that if @xmath591 , then for each @xmath509 there exists @xmath564 such that @xmath599 . and
if @xmath585 and @xmath592 , there exists @xmath596 and @xmath600 such that @xmath597 .
_ example j. _ consider the equation @xmath601 adding the term @xmath602 in both sides ; setting @xmath603 , and after renaming @xmath550 and @xmath551 we get @xmath520 systems of type ( [ problema1 ] ) associated to each initial condition @xmath552 for @xmath553 .
@xmath604 where @xmath555 and @xmath605 .
if @xmath64 is an attractor of @xmath606 and the system has the fast enough convergence property to the limit fiber @xmath2 .
the dynamics on the basin of attraction of this fiber can be studied using theorem [ propoprob1identitatv2 ] . _
example k. _ in an analogous way as in all the previous examples , by adding @xmath607 in both sides of equation @xmath608 and setting @xmath609 , we get that the above equation can be studied using via the system of type ( [ problema1 ] ) @xmath610 a. cima , a. gasull , f. maosas . _ some remarks on continuous and discrete markus - yamabe problems_. a conference on polynomial maps and the jacobian conjecture , 1997 .
special issue in honour of the mathematical work of gary meisters .
. _ existence of unbounded solutions of time varying systems and failure of global asymptotic stability in discrete time casacade systems_. i m a j. math . control and information 22 ( 2005 ) , 8087 . | we consider planar triangular maps @xmath0 .
these maps preserve the fibration of the plane given by @xmath1 .
we assume that there exists an invariant attracting fiber @xmath2 for the dynamical system generated by @xmath3 and we study the limit dynamics of those points in the basin of attraction of this invariant fiber , assuming that either it contains a global attractor , or it is filled by fixed or @xmath4-periodic points .
finally , we apply our results to a variety of examples , from particular cases of triangular systems to some planar quasi - homogeneous maps , and some multiplicative and additive difference equations , as well . _
2000 mathematics subject classification : _ ` 39a10 , 39a11 , 39a20 ` _ keywords _ : attractors ; difference equations ; discrete dynamical systems ; triangular maps ; periodic solutions ; quasi - homogeneous maps . |
in recent years it has been discovered that bound states for one - electron systems with three / four protons exist in nature in presence of strong magnetic fields @xcite . in particular , for the molecular ion @xmath7 it has been shown that the geometrical configuration changes depending on the field strength , being an equilateral triangle for @xmath8 g@xmath9 g and linear for @xmath10 g @xmath11 g. systems with more than three protons also exist in presence of such strong fields , but the only linear configurations parallel to the field seems to be optimal . other other geometrical configuration might exist in such circumstances but have not been studied , so far . in a very naive way of thinking
a clue to find other configurations is to assume the proton to have a non - integer charge . although the physics behind this two phenomena are totally different , it is possible that for a given field strength there will be a preferred direction for which a geometrical configuration , different from linear , might be realized .
for example , in ref . @xcite it is presented the case of one - electron systems with charged centers @xmath12 , @xmath13 , turned out that there are bound states for positive charges less that a certain critical one , @xmath14 , with non - linear optimal geometrical configuration . a restriction to integer values of charges
do not appear in classical electrodynamics and not even in more advanced theories such as the theory of atomic - molecular physics , it is only in elementary particle physics @xcite where a justification for the existence of integer charge appears . in classical electrostatics where
stable configurations of point charges are known to be absent ( earnshaw s theorem ) , zero charge is a singular point where the nature of interaction changes from repulsion to attraction . usually , at a singular charge the whole or some part of the potential vanishes .
this behavior is not exclusive of classical physics , it also can be found in non - relativistic quantum electrodynamics where these singular charges exist as well .
however , a new phenomenon occurs - there are some critical charges which separate the domain of the existence of the bound states from the domain of non - existence , although the nature of the potential remains unchanged . in some cases
a system gets bound at a critical charge with polynomialy - decaying eigenfunctions at large distances , unlike standard exponentially - decaying eigenfunctions @xcite .
the well - known examples where such a transition , of existence to non - existence , occurs are the square - well of finite depth , the pschl - teller potential and the yukawa potential . about coulomb molecular systems , there are some works related to molecular ion @xmath15 with non - integer charge @xcite . for more than two charged centers and one electron ,
the first study was presented in @xcite for @xmath13 : @xmath16 , @xmath17 and @xmath18 , where critical charges were calculated .
the optimal geometric configurations were found to correspond to platonic solids ; a line for @xmath16 , an equilateral triangle for @xmath17 and a regular tetrahedron for @xmath18 .
for all studied systems , both atomic and molecular , the total energy has a square - root branch point with exponent @xmath6 at @xmath5 .
this work can be considered as a continuation of work @xcite addressing five and six infinitely massive centers of the same charge @xmath0 and one electron .
one of the goals is to find the possible optimal geometrical configuration for which the system is realized .
another goal is to identify the domain(s ) in @xmath0 where the system is bound , focusing on the critical charges @xmath19 which separate the domains of existence / non - existence of bound state .
this study is performed in the framework of non - relativistic quantum mechanics .
atomic units @xmath20 are used throughout , but energies are given in rydbergs .
let us consider a coulomb molecular system which consists of @xmath21 fixed charges @xmath0 and one electron , @xmath12 .
the hamiltonian which describes this system as follows @xmath22 where @xmath23 is the distance between charge centers @xmath24 and @xmath25 , and @xmath26 is the distance from the electron to @xmath24th charge center . from a physical point of view
, it is clear that such a system is not bound , both for large positive and for negative @xmath0 values .
however , it has to be bound at small @xmath27 .
it is clear that there must be a critical charge which separates the domains of existence and non - existence of a bound state ; one such a critical charge is @xmath28 , the other one is at some finite @xmath0 , @xmath5 , which does not represent any type of singularity of operator .
it is worth noting that it is possible to describe the same problem by considering that the charged centers have positive charge equal to the unity and the electron has negative charge greater than one . in order to calculate the total energy @xmath29 we use the variational method .
we employ a physics - inspired trial functions .
the choice of the trial function is based on arguments of physical relevance , i.e. the trial function should support the symmetries of the system , has to reproduce the coulomb singularities and the asymptotic behavior at large distances adequately ( for more details see @xcite ) .
let us consider five fixed charges @xmath0 and one electron .
there is an infinite number of configurations for this system , but only one that is optimal .
it is impossible to study all configurations but it is a good guess to consider the most symmetric ones .
three particular configurations were studied : pentagon , square pyramid and triangular dipyramid . for the pentagon configuration , protons were located on the the edges of the figure , defining the @xmath30 plane with origin in the centroid , and @xmath31 the distance between two adjacent protons . in the square pyramid configuration four protons were placed on the corners of a square of side @xmath31 , defining the @xmath30 plane with its sides parallel to the axes ; the fifth proton is placed on the @xmath32axis perpendicular to the square .
finally the triangular dipyramid configuration is given by three centers forming an equilateral triangle of side @xmath33 and two others which are symmetrically placed perpendicular to the triangle coming through its center with distance @xmath34 between them .
the variational method is used to obtain all numerical results .
the trial function is taken in a form of a symmetrized product of five @xmath35-coulomb orbitals ( slater functions ) @xmath36 where @xmath37 is the symmetrizer with respect a permutation of the charged centers .
after doing permutations the trial function contains 120 terms .
the function ( [ trial5ze ] ) depends on five @xmath38 variational parameters .
it is expected that this type of function gives results with an accuracy of @xmath39 in the total energy . in order to find these critical charges , curves of the total energy as function of the charge
can be obtained for the three different configurations , see fig [ 5zeconfigs ] .
calculations show that critical charges are @xmath40 , @xmath41 and @xmath42 for pentagon , square pyramid and triangle dipyramid , respectively .
the curves also show that the most stable geometrical configuration is given by the triangular dipyramid , see fig .
[ 5ze ] . for the optimal geometrical configuration , the triangular dipyramid , it is possible to find the domains of metastability ; i.e. points where systems begin to have decay channels .
these domains are given by crossing points between one - electron and @xmath21 @xmath0-charged systems , @xmath12 .
[ e : nz+e ] shows curves of the total energy as function of the charge , @xmath29 , for systems @xmath43 ( as reference curve ) , @xmath17 , @xmath18 ( see @xcite ) , @xmath1 and @xmath2 , all taken in the optimal configuration .
thus , for charges @xmath44 the system is metastable with two decay channels @xmath45 for @xmath46 the system is metastable with one decay channel @xmath47 and finally for @xmath48 the system is stable . beyond determining the critical charge , it is important to understand the behavior of the energy , as function of the charge , close to critically . in this way it turns out that this behavior is described by means of a puiseux expansion @xmath49 with the condition that @xmath50 .
the goal is to find parameters @xmath51 and @xmath52 of this expansion .
restricting the expansion ( [ pui ] ) to a finite number of terms a fit of the total energy calculated numerically . by making the fit it is found , in particular , that exponents @xmath52 are very close to @xmath53 ( @xmath54 ) ; then
, it is convenient to assume @xmath55 and only determine the coefficients @xmath51 .
the fit is based on data in the domain @xmath56 ( 10 points ) .
this behavior indicates that critical point might be a square - root branch point . @xmath57 where the critical point is @xmath58 for the critical point @xmath28 ( which is the singular point of the schrdinger equation ) , we study the behavior of the energy as function of the charge .
the behavior on the energy ( @xmath59 with 15 points ) is given by a taylor expansion : @xmath60 such a behavior does not provide an indication of a singular nature at @xmath28 .
however , the total energy can not be analytically continued to @xmath61 .
table [ fitcorr5ze ] shows a comparison between the results of the energy fit and fit near critical charges @xmath5 , @xmath28 , correspondingly , and data . .
[ cols="^,^,^",options="header " , ] finally , fig .
[ 6zedistances ] presents curves for the distances ( height @xmath62 , and radius of circumscribed circle @xmath63 ) as functions of the charge @xmath0 for the geometrical configuration of the system ( octahedron ) . from fig . [
6zedistances ] we see that the optimal geometrical configuration is an octahedron , where the height @xmath62 is always smaller than the radius of circumscribed circle @xmath63 .
this tell us that the charges are placed on a oblate spheroid with semi - axes @xmath63 , see fig.[elipse6ze ] .
in this paper we calculated for the first time the critical charges for two molecular systems : @xmath1 and @xmath64 .
for all those systems the total energy and equilibrium distances vs. @xmath0 are smooth curves without any indication of charge quantization .
moreover the optimal geometric configuration seems to be always the maximally symmetric : a triangular dipyramid for @xmath1 and an octahedron for @xmath2 .
it is important to mention that the behavior near the critical charge of the total energy as a function of the charge , @xmath29 , indicates a square - root branch point with exponent 3/2 .
this agrees with the previous results found for various atomic and molecular systems @xcite .
the author is grateful to a. v. turbiner , a. alijah and w. bietenholz for their valuable suggestions and useful comments , and for reading the manuscript .
he further thanks j. c. lpez vieyra for his assistance with computer calculations and interest in the present work .
this work was supported in part by conacyt grants * 58962-f * , * 171494 * and * 202139 * ( mexico ) , and by the computer center of the universit de reims champagne - ardenne ( france ) . | we consider a coulomb system of one electron and five or six infinitely massive centers of charge @xmath0 : @xmath1 and @xmath2 .
critical charges and the possible optimal geometrical configurations are found .
it is shown that the domain of stability for @xmath1 is @xmath3 with the optimal geometrical configuration given by a dipyramid ( equilateral triangle base ) circumscribed in a prolate spheroid .
for @xmath2 the stability is @xmath4 with the optimal geometrical configuration given by an octahedron ( square base ) , circumscribed in an oblate spheroid . for both systems
we obtain an indication that total energy at @xmath5 has a square - root branch point singularity with exponent @xmath6 . |
ugca 86 is a magellanic spiral ( sm ) galaxy originally catalogued from optical plates by @xcite .
it is also known as a0355 + 66 @xcite , and was discovered independently through its hiemission by rots ( 1979 ) , who referred to it as a0355 .
ugca 86 was proposed as a member of the local group based on an analysis of a colour - magnitude diagram @xcite , but photometry of the brightest stars place it at a distance of 2.6mpc , in the ic 342/maffei group @xcite .
ugca 86 is located @xmath7 southeast of the center of the scd galaxy ic 342 , implying a minimum separation of 71 kpc for the assumed distance of @xmath8 .
the adopted distance is somewhat less than the 3.3 mpc distance of ic 342 @xcite , but in view of the uncertainties , ugca 86 is likely a satellite of ic 342 .
its present distance from ic 342 is at least 50% larger than the distance between the milky way and the magellanic clouds .
optical studies are complicated by the low galactic latitude of this galaxy ( @xmath9 ) .
@xcite found reddening @xmath10 for hii regions in ugca 86 , with no evidence for significant extinction inside ugca 86 .
surface photometry in @xmath11 and @xmath12 by @xcite shows a nearly exponential surface brightness profile with central surface brightness @xmath13 and integrated magnitudes @xmath14 , @xmath15 . applying a standard galactic extinction curve ,
the absolute magnitude in the @xmath11 band corrected for extinction is @xmath16 with an uncertainty of at least one magnitude .
the extinction - corrected central surface brightness is @xmath17 .
the shape of the optical isophotes is elliptical with mean axial ratio @xmath18 and major axis in p.a .
the optical axial ratio implies an inclination @xmath20 for an intrinsic axial ratio @xmath21 .
@xcite detected numerous hii regions in ugca 86 adding up to a total star formation rate of @xmath22 .
most of the star formation is concentrated in three hii region complexes southeast and southwest of the center of ugca 86 .
the most prominent star formation complex is the `` cometary '' component to the south east ( see figure 20 of buta & mccall 1999 ) .
that component is catalogued as viizw 009 , and is well resolved into stars .
it was suggested to be a separate , disturbed galaxy by @xcite , but has since been shown to be a part of ugca 86 @xcite .
@xcite likened the cometary component to the 30 doradus complex in the large magellanic cloud .
the metallicity of hii regions 81 and 101 in @xcite is @xmath23 , comparable to the large magellanic cloud @xcite .
both these hii regions are associated with the cometary knot southeast of the center of ugca 86 .
@xcite argued that the cometary component is a more recent burst of star formation than the central part of the galaxy , and speculated that it was triggered by infall of gas .
infrared work on this galaxy has concentrated on the prominent hii regions in the south - west regions of the galaxy , with similar conclusions @xcite .
an x - ray source is located near the core ( rosat source 1rxp j035951 + 6708.6 ) , and has been suggested as being a low - mass x - ray binary @xcite .
@xcite showed that ugca 86 is rotating , with a perturbed velocity field , and identified it as a magellanic dwarf irregular galaxy .
his hi mass reduced to a distance of 2.6 mpc is @xmath24 .
@xcite reported a systemic velocity @xmath25 .
other authors found a significantly smaller value @xmath26 @xcite and @xmath27 @xcite .
the galactic coordinates of ugca 86 are @xmath28 . at this longitude
, positive velocities are forbidden by galactic rotation .
confusion with foreground galactic hi is not a problem at velocities exceeding @xmath29 .
the systemic velocity of ugca 86close to the systemic velocity of ic 342 ( @xmath30 ) @xcite , which adds to the suspicion that ugca 86 is a satellite of ic 342 .
@xcite further suggested that ugca 86 is tidally interacting with ic 342 as a means of explaining the perturbations in the velocity fields of both galaxies . a suggestion by @xcite that ugc 2826 might be the perturber for ic 342 has since been ruled out by the high radial velocity of that galaxy @xcite .
rots low - resolution ( @xmath31 , @xmath32 ) hi results are the only published hi maps of ugca 86 to date . to address this lack and investigate the possible link between ugca 86 and ic 342 ,
we have mapped the decimetre radio continuum and neutral hydrogen distribution of both galaxies with @xmath33 spatial and @xmath34 velocity resolution , using the synthesis telescope at the dominion radio astrophysical observatory ( drao ) .
data from a pointing centered on the optical disk of ugca 86 were acquired simultaneously in the hi - line and in continuum bands at 1420 mhz and 408 mhz ( wavelengths of 21 and 74 cm , respectively ) .
data were also collected for ic 342 in a separate but overlapping pointing ( the 1420mhz primary beam diameter of the drao array is @xmath35 to half power ; at 408mhz it is @xmath36 ) .
analysis of the ic 342 field is deferred to another paper .
twelve sets of spacings were used , giving baselines from 12.9 m to 604.3 m at 4.3 m intervals , and yielding sensitivity to spatial structures from @xmath33 to @xmath37 at 1.4ghz ( and about 3.5 times those values at 408mhz ) .
these data were collected during 2000 october december .
some brief technical details of the observations and reduction are given in the following sections ; refer to landecker et al .
( 2000 ) for full details of the synthesis telescope .
all images presented below have been corrected for primary beam attenuation .
hi line data were collected in 256 contiguous channels over a 4mhz band , spanning 844kms@xmath38 ( 3.3kms@xmath38 per channel ) with a velocity resolution of 5.2kms@xmath38 .
the observation was centered at @xmath39kms@xmath38 , the heliocentric velocity of ic 342 ; ugca 86 is nominally at + 67kms@xmath38 , which is well within the same band .
both right- and left - hand circular polarizations were recorded , with the two being averaged to make the final images , resulting in an rms noise per channel of 9.5mjybeam@xmath38 . to remove continuum sources ,
channels free from hydrogen emission and band - edge effects were selected from each end of data - cube and averaged together ( 120 such channels were used ) , then the result was subtracted from each channel image .
the spatial resolution achieved from uniformly - weighted visibilities was @xmath40 at p.a .
@xmath41 ; the data presented in this paper have had a radial taper applied to increase the brightness temperature sensitivity , and have a beam of @xmath42 at p.a .
a spectral line cube tapered to @xmath44 resolution was made to obtain a better signal to noise ratio for faint extended emission . the well - sampled aperture plane and inherent low signal - to - noise meant that it was not necessary to deconvolve the channel images ( a standard practice at drao ) . missing short spacings
cause a mild negative bowl .
the 1.4ghz continuum system records full polarization data in four 7.5mhz bands , pairs of which flank the spectrometer band .
data from the four bands were gridded separately onto a common aperture plane to make a single image in each of the four stokes parameters .
the synthesized beam at 1.4ghz was the same as for the radially tapered hi data described above , and the rms noise was 0.25mjybeam@xmath38 .
no significant linear or circular polarization was seen from ugca 86 . the 408mhz continuum system records data over a 3.5mhz bandwidth , in right - hand circular polarization only . in the absence of circularly polarized signal , which is generally the case ,
this may be interpreted as stokes @xmath12 . at 408mhz
the tapered synthesized beam was @xmath45 at p.a .
@xmath46 , and the confusion - limited rms noise achieved was 4mjybeam@xmath38 .
the data were processed in drao s ` madr ` package .
the raw stokes @xmath12 image at each frequency was deconvolved using clean , and self - calibrated to remove artefacts arising from variations of amplitude and phase on short time - scales during the observation .
figure [ chanmaps - fig ] shows the hi channel maps of ugca 86 at positive velocities .
emission of ugca 86 is blended with galactic emission at velocities below @xmath47 .
the location of the brightest emission associated with ugca 86 changes gradually with velocity in a way consistent with a rotating disk . at velocities beyond @xmath48
the structure in the channel maps is more complex .
in addition to the brighter component , a faint spur of emission extends towards the northwest .
figures [ gray : hicolumn ] and [ gray : velfield ] show the hicolumn density and velocity field , respectively , for ugca 86 . these images were constructed from a cube in which the emission from ugca 86was manually identified in each channel image and selectively masked to exclude areas with no signal . in this way we minimized the adverse effects of noise in empty regions , and the contaminating galactic hiemission at velocities below @xmath49 . in these contaminated channels
we also examined position - velocity slices though the data - cube to aid in the identification of emission from ugca 86 , which has higher velocity dispersion than the galactic emission .
the total hi mass of ugca 86 derived from this masked dataset is @xmath50 .
this is significantly smaller than the value found by @xcite .
this difference can not be attributed to resolution of the extended emission in rots map by the shortest baselines of the drao interferometer .
the present data lack the sensitivity to detect the extended low column density emission in rots map , and the modest negative bowl has a negative effect on the flux when integrating over a large area .
differences in the treatment of confusing galactic emission may also have contributed .
note that the extended emission in the first panel of figure [ chanmaps - fig ] is considered galactic here .
the distribution of neutral hydrogen has a maximum column density of @xmath51@xmath52 .
the distribution is approximately elliptical in outline for column densities higher than @xmath53@xmath52 ( figure [ gray : hicolumn ] ) , with an extent of @xmath54 at the @xmath55 level , oriented at p.a .
the axial ratio of hi column density contours in the range @xmath57 to @xmath58 , excluding the spur , is @xmath59 .
this part of ugca 86 is centered on the optical galaxy with similar axial ratio and orientation of the apparent major axis @xcite .
the velocity field in this area shows a well - defined rotational velocity gradient with the receding side of the galaxy in p.a .
both the hydrogen and optical disks are centered at ( @xmath61 , @xmath62 ) = ( @xmath63 , @xmath64 ) .
there are three prominent `` holes '' in the hi at ( @xmath65 , @xmath66 ) , ( @xmath67 , @xmath64 ) , and ( @xmath68 , @xmath69 ) , in which the column density drops below @xmath70@xmath52 . at low column densities
the narrow hi spur noticed in figure [ chanmaps - fig ] extends towards the northwest ( the approximate direction of ic 342 ) , up to @xmath71 ( 9 kpc ) from the center of ugca 86 .
this structure is also visible in the low - resolution column density map of @xcite .
a faint extension of the hi isophotes is seen on the opposite side of the disk near 04@xmath72 01@xmath73 , @xmath74 @xmath75 .
the suggestion of an elongated structure that partially overlaps with the hi disk on the sky will be confirmed later by position - velocity diagrams and hi line profiles .
the kinematics of the spur are ordered , with lower velocities occurring on the eastern side of the spur .
note that the velocity changes more rapidly along the minor axis of the spur than in the disk , and that the ( apparent ) velocity gradient has a different direction .
the velocity field ( figure [ gray : velfield ] ) clearly shows the regular rotation of the disk .
however , the spur affects the velocity field as it overlaps in position with the disk .
the approaching side of the disk is also affected by confusion with galactic hi .
a tilted ring fit @xcite to the velocity field of figure [ gray : velfield ] was attempted to obtain the rotation curve of the disk . however , the results of the fit to the velocity field are skewed towards smaller rotation velocities because of the contaminating emission of the spur .
this will be illustrated further in section [ spur - section ] .
the best representation of the rotation curve of the disk was obtained by fitting the shape of the position - velocity diagram though the kinematic major axis of the disk . the p.a . of the kinematic major axis and an initial rotation curve
were determined from the tilted ring fits .
the rotation velocities of the annuli and the systemic velocity were adjusted manually to force the rotation curve to follow the bright curved ridge of emission in figure [ gray : rotcurve ] .
the resulting rotation curve is listed in table [ rotcur - table ] .
the total mass of ugca 86 within the outermost radius of the rotation curve is @xmath76 .
the inclination of the hi disk is uncertain .
the adopted inclination @xmath77 is based on the hi axial ratio @xmath78 and an intrinsic axial ratio @xmath21 .
if the optical inclination @xmath79 is adopted , the velocities are 15% higher than listed in table [ rotcur - table ] .
the systemic velocity found here is consistent with the values of @xcite and @xcite , but significantly smaller than the @xmath80 found by @xcite .
this systematic difference may be the result of rots discarding channels affected by galactic emission , and inclusion of the spur . a significant misalignment between the p.a . of the apparent major axis and the kinematic major axis
is observed in ugca 86 .
a small but significant misalignment between the kinematic and morphological major axis is not exceptional in dwarf galaxies .
@xcite interpreted a similar misalignment in the dwarf irregular galaxy sex a as evidence that the gas in the disk describes non - circular orbits .
figure [ holeslice - fig ] shows a position - velocity slice in p.a .
@xmath81 , approximately along the apparent major axis of the hi disk . a small area @xmath82 southwest of the center of ugca 86displays a significant departure from the regular rotation pattern of the disk . locally
the line of sight velocity deviates by @xmath83 from the value expected from the rotation curve model of table [ rotcur - table ] .
the area where the velocity deviates more than @xmath84 from the rotation model is indicated by the ellipse in the upper panel of figure [ holeslice - fig ] .
the hi mass of this area integrated over the velocity range @xmath3 to @xmath85 is @xmath86 .
accounting for 30% helium by mass , the kinetic energy associated with this region is @xmath87 .
this energy is equivalent with the mechanical energy released by 240 supernovae .
ccccc 60 & 35 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 120 & 65 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 180 & 87 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 240 & 103 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 300 & 114 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 360 & 119 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 420 & 122 @xmath88 5 & 45 & 287 @xmath88 2 & 72 @xmath88 5 + 480 & 122 @xmath88 5 & 45 & 287 @xmath88 2 &
72 @xmath88 5 + [ rotcur - table ] position - velocity slices parallel to the major axis ( p.a .
322@xmath89 ) and the minor axis ( p.a .
232@xmath89 ) of the spur are shown in figure [ gray : majorslice ] and figure [ gray : minorslice ] respectively .
these slices were taken from the @xmath44 resolution cube to show faint emission in the spur .
galactic foreground hi is seen at negative velocities and close to zero .
the spur is clearly visible in the velocity range @xmath90 to @xmath91 , far outside the velocity range permitted by galactic rotation assuming circular orbits and a flat rotation curve .
it is possible in principle that the spur is hi in the foreground that does not follow galactic rotation .
we note that the difference between heliocentric velocity and velocity with respect to the local standard of rest ( lsr ) is only @xmath92 for ugca 86 .
one possibility is that the spur is a high - velocity cloud ( hvc ) along the line of sight .
the hvc complexes nearest to the line of sight are complex a and complex h in the velocity range @xmath93 @xcite .
association of the spur with these complexes is dismissed because of the large velocity difference . instead , the spur could be considered a compact high - velocity cloud ( chvc ) , not associated with a large complex . however , the all - sky distribution of velocities of chvcs in the lsr reference frame is dominated by the orbital velocity of the lsr around the galactic center .
the velocity of the lsr has a large component in the direction of ugca 86 , and no chvcs with positive velocities in the lsr frame have been found in this part of the sky ( * ? ? ?
* and references therein ) .
the probability that an extraordinary chvc with a positive velocity is found in this location is very small .
for this reason , the interpretation of the spur as a chvc is rejected .
@xcite reported a population of hi clouds in the galactic halo with velocities up to @xmath48 beyond that allowed by galactic rotation
. some of these clouds would be detectable in the present data @xcite .
the spur is observed at much higher forbidden velocities .
also , the velocity gradient of the spur and the width of the hi line profile in the spur do not match the narrow line widths observed in these halo clouds @xcite .
a distinct velocity gradient is observed along both the major and the minor axes of the spur .
this indicates that the actual velocity gradient is not aligned with either the major axis or the minor axis .
the direction of the actual velocity gradient across the spur was determined by interactively changing the p.a .
of position - velocity slices until no velocity gradient was discernible .
this gives the p.a .
of the direction perpendicular to the velocity gradient .
the velocity gradient across the spur was found to be @xmath94 in p.a .
@xmath95 . evidence for a reversal of the velocity gradient is seen in slice d of figure [ gray : majorslice ] towards the extreme end of the spur .
figure [ gray : majorslice ] also shows that the spur does not begin at the northwest edge of the disk , but rather overlaps on the sky with the hi disk . the hi line profiles are double - peaked or blended where the spur and the disk overlap as illustrated in figure [ centerprofiles ] .
double - peaked hi line profiles with a velocity separation of @xmath96 are seen near the center of ugca 86 . southeast of the center of ugca 86
the line profiles are blended so much that a kinematic separation is not possible .
the asymmetric line profiles on the southeast side do suggest that the spur extends across the disk . on the extreme east side of ugca 86 , the disk itself is difficult to disentangle from galactic hi .
the difference @xmath97 in the line of sight velocity of the spur and the disk towards the center of ugca 86 is a direct measurement of the velocity component of spur gas in the direction of the center of mass of ugca 86 , as defined by the optical galaxy .
the radial component of the spur gas in this direction is a significant fraction , approximately one third , of the circular velocity at the last measured point of the rotation curve ( table [ rotcur - table ] ) . as the rotation curve appears to turn over around a radius of @xmath98 , the velocity of a circular orbit at larger radii
is not expected to exceed @xmath99 by a significant amount .
therefore , the radial component of the velocity of this part of the spur is at least one third of the local circular velocity , independent of the actual distance from the center of ugca 86 .
this significant radial component of the orbital velocity is a direct indication that the orbits of gas in the spur are significantly non - circular with respect to the center of ugca 86 .
a determination of the mass of the spur requires decomposition of the line profiles in a disk and a spur component in places where the spur and the disk overlap on the sky .
figure [ centerprofiles ] shows that a decomposition of the line profiles into gaussians will not provide a unique solution in many places .
this ambiguity was addressed by subtracting a sequence of models for the hi disk from the data .
these models were generated in three steps . in the first step ,
a decomposition into at most two gaussians was attempted at every location .
the strongest gaussian component was tentatively assigned to the disk .
if no clear decomposition could be made , the profile was skipped in this step . in the second step ,
the central velocity of the gaussian assigned to the disk was compared with the line of sight velocity expected from the rotation curve , the inclination of the disk , and the location in the disk .
if the velocity difference exceeded a certain threshold , the fit was rejected .
thresholds of 0 , 5 , 10 , 20 and 50 @xmath100 were applied to generate a sequence of disk models that assign less weight to the rotation curve and increasing weight to the gaussian decomposition . in the third step , rejected profiles and profiles that could not be decomposed into separate components in step 1 , were replaced by a model gaussian .
this model gaussian has a central velocity calculated from the rotation curve , amplitude equal to the observed intensity at the central velocity , and dispersion equal to the median velocity dispersion of the hi disk ( @xmath101 ) .
each model was subtracted from the data .
the flux of the residual emssion was measured to obtain the mass of the spur . at this time , the anomalous velocity region shown in figure [ holeslice - fig ] was masked out .
it is not included in the mass of the spur .
the hi mass of the spur was found to be in the range @xmath102 to @xmath103 .
the hi mass of the spur is approximately 10% of the himass of ugca 86 , but well within the range of hi masses of nearby dwarf galaxies @xcite .
the velocity dispersion of hi in the spur and the disk was determined by fitting a single gaussian to the hi line profiles .
the @xmath44 resolution data cube was regridded to 15 pixels to obtain a single spectrum per beam .
line profiles with multiple components in places where the disk and the spur overlap were excluded from this analysis .
measurements of the velocity dispersion in the spur include areas west of @xmath104 and north of @xmath105 , which are separated spatially from the disk .
only 12 spatially independent profiles with sufficient signal were available in the spur .
some of the profiles in the spur appear to be the superposition of two or more components , but the limited signal to noise ratio does not warrant a decomposition into multiple gaussians .
figure [ gray : histo ] shows histograms of the hi velocity dispersion in the disk and in the spur .
the median error for velocity dispersions is @xmath106 in the disk , and @xmath107 in the spur .
the larger errors for the velocity dispersion in the spur are because of the generally lower brightness and the larger velocity dispersion .
the distribution of velocity dispersions in the disk is similar to that of other dwarf galaxies .
the median velocity dispersion in the disk is @xmath108 , before corrections for spectral resolution and broadening by the rotational velocity gradient .
even though the number of measurements in the spur is small , figure [ gray : histo ] indicates that the distributions of velocity dispersion differ between the spur and the disk .
the median velocity dispersion in the spur is @xmath109 .
@xcite found similar velocity dispersions in the outer cloud complexes of ngc 4449 .
the line of sight velocity dispersions shown in figure [ gray : histo ] have not been corrected for spectral resolution or for a contribution of the rotational velocity gradient .
the correction for rotational broadening of the line profiles was done following @xcite .
the velocity gradient due to rotation of the disk derived from the rotation curve is @xmath110 .
the instrumental spectral resolution is @xmath34 .
the median velocity dispersion @xmath111 then results in a corrected line of sight velocity dispersion and spectral resolution @xmath112 .
the velocity gradient across the spur contributes to the observed line of sight velocity dispersion in a similar way . for the velocity gradient @xmath113 ,
observed velocity dispersions of @xmath114 , @xmath84 , and @xmath6 imply intrinsic velocity dispersions of 7 , 15 , and 27 @xmath100 respectively .
the velocity gradient over the beam contributes to the high velocity dispersion observed in the spur , but its effect is too small to explain the difference in the distributions of the velocity dispersion in figure [ gray : histo ] . adopting a single velocity gradient for the spur is a simplification .
the smallest velocity dispersion measured in the spur is smaller than the broadening implied by the mean gradient .
a more detailed analysis requires observations with higher spatial resolution . the 1.4 ghz radio continuum emission from ugca 86(figure [ gray : continuum ] )
shows a lop - sided distribution , with several emission peaks to the south and west of the optical center , embedded in an extended component that covers the entire optical disk .
the strongest 1.4ghz continuum emission coincides with the optical cometary component , and has a similar morphology .
emission peaks to the west and south - west of the optical center coincide with optical hii region complexes @xcite that are also seen in reprocessed iras data @xcite .
the hii region complexes appear on the periphery of depressions in hi column density .
two unresolved sources to the east have no optical or infrared counterparts , and are presumably background sources .
the northern one is catalogued as nvss j040022 + 670830 @xcite .
no radio counterpart to the rosat x - ray source 1rxp j035951 + 6708.6 is evident .
ugca 86 is also detected at 408 mhz .
three independent complete synthesis observations with ugca 86 near the pointing center were analysed .
one was centered on ugca 86 , one on ic 342 , and one unrelated observation was available in the drao archive .
ugca 86 was detected in each of these 408 mhz observations .
the 408 mhz image in figure [ gray : continuum ] is the linear mosaic of these three observations .
the total flux density at 1.4ghz is @xmath115mjy , and at 408mhz is @xmath116mjybeam@xmath38 .
the implied spectral index of ugca 86 between 408 mhz and 1.4 ghz is @xmath117 ( @xmath118 ) .
while this is consistent with a nearly flat spectrum , it is likely that not all of the emission is thermal .
the association of the compact emission with hii regions suggests that the the extended emission seen in the 21-cm continuum image may be significantly non - thermal .
this is reminiscent of the extended synchrotron halo of the post - starburst dwarf galaxy ngc 1569 , which has the same global spectral index @xmath119 @xcite .
extended synchrotron haloes are a rare phenomenon in dwarf galaxies .
ugca 86 is a prime candidate dwarf galaxy with a synchrotron halo .
ugca 86 is a member of a heterogeneous set of dwarf galaxies with very significant departures from ordered rotation in their extended hi envelopes .
these galaxies are of interest for our understanding of the formation and evolution of galaxies because of the origin of the chaotic kinematics .
we distinguish three broad categories that arguably share some common characteristics : 1 .
tidal interaction with another galaxy ; 2 . collision with or capture of an intergalactic hi cloud ; 3 .
assembly of the hi disk on timescales longer than a hubble time ( `` continuing galaxy formation '' @xcite ) . in case of tidal interaction ,
the hi displaying the peculiar kinematics is in fact the disturbed outer hi disk .
the most extreme case of tidal interaction is a merger with another galaxy . in the case of a collision or capture of an intergalactic gas cloud or a smaller dwarf galaxy ,
the gravitational force on ugca 86 may be negligible , but the infalling material may create large holes in the gas distribution and compress the interstellar medium to high densities in some places , triggering star formation @xcite . the case of ongoing galaxy formation
assumes that the hi was always bound to the dwarf galaxy , but has not yet been assimilated into the disk .
an interesting question is why most dwarf galaxies do appear to have formed an hi disk , or even converted all of the available gas into stars .
the main difference between a collision and ongoing assimilation is the time scale of the process . both may be characterized as infall .
the relative importance of these processes is not well known .
the number of ( dwarf ) galaxies with chaotic kinematics in their outer parts and the details of their kinematics provide important clues regarding the relative importance of these processes .
this may in turn provide insight in similar processes occurring in galaxies at high redshift .
although the orbits of the gas are not uniquely constrained by the observations , some constraints may be given if a geometry is specified .
an important observation is the steep velocity gradient along the minor axis of the spur observed in figure [ gray : minorslice ] .
this velocity gradient is difficult to explain if the the spur is a filament , with its dimension along the line of sight equal to its minor axis on the sky .
it would require a special short - lived geometry , or a mass concentration in the spur to balance the velocity shear and maintain its filamentary structure .
it is more likely that the velocity gradient across the spur displays the line of sight component of the orbital velocity of gas in a structure that is extended along the line of sight .
this argument is equally valid whether the orbits of the spur gas are open or closed .
the large range in the hi velocity dispersion of the spur ( figure [ gray : histo ] ) is also most easily explained in terms of a superposition of multiple velocity components along the line of sight .
much of the following discussion is based on this argument .
the hi velocity field around the optical galaxy shows the structure expected for a disk with well - ordered rotation .
the first question that comes to mind is whether the spur is an extension of the hidisk of ugca 86 .
the outer hi disk may not show well - ordered rotation as a result of tidal interaction with ic 342 , or because it has not had the time to settle into a disk .
however , if the spur is to be considered an extension of the disk , the orbits of hi in the spur should be reasonably circular around the optical galaxy and approximately in the plane of the observed hi disk .
it should be noted that some dwarf galaxies appear to have significant warps in their outer hi disks , suggested by their twisted hi isophotes and a change in p.a . of the velocity field .
examples are ddo 168 @xcite , um 439 @xcite , and ngc 4449 @xcite
. three lines of argument indicate that the gas in the spur does not satisfy these requirements .
first , the spur partially overlaps in position on the sky with the hi disk , but not in velocity .
this implies that at least this part of the spur is significantly out of the plane of the disk , because the orbits of gas in the spur can not cross the orbits of gas in the disk .
the second argument follows from the difference in the line of sight velocity between the spur and the disk in the direction of the optical center of ugca 86 .
although the rotation curve analysis of the velocity field did not uniquely define the position of the center of mass of ugca 86 , it is reasonable to assume that the center of mass is close to the optical center , as it is for larger disk galaxies .
the correspondence between the morphology of the optical and the hi isophotes supports this assumption .
the velocity difference @xmath97 is therefore the radial component of the orbital velocity of gas in the spur , with respect to the dynamical center of ugca 86 . for ugca 86 ,
@xmath120 is about one third of the circular velocity of the outer disk at radius @xmath121 ( 6 kpc ) .
this large radial velocity component relative to the local circular velocity implies that the orbit is significantly non - circular .
the first and second arguments combined imply that the part of the spur that is observed in the direction of the optical center is probably further from the dynamic center than the observed radius of the hi disk ( @xmath122 ) .
the distance of this part of the spur from the plane of the disk is then at least @xmath123 for the adopted inclination @xmath124 .
the third argument that the spur is not an extension of the hi disk is that differential rotation would wind - up the spur on an orbital time scale @xmath125 .
the linear appearance of the spur would be highly coincidental because it would exist only for a short time .
the conclusion is that the hi spur is not an extension of the hidisk .
this also excludes the possibility that the spur is an hispiral arm similar to those observed in the extended outer hi disks of some dwarf galaxies , such as ngc 2915 @xcite and ddo 47 @xcite .
at least some hi in the spur is found to follow non - circular orbits at a high inclination with respect to the plane of the hi disk .
ugca 86 is not unique in this respect .
highly structured outer hi with kinematics different from a disk that coincides with the optical galaxy has been reported for ngc 4449 @xcite , izw 18 @xcite , iizw 40 @xcite and ngc 1569 @xcite .
iizw 40 is a clear example of a merger in progress .
the hi filaments in ngc 4449 may be the result of tidal interaction .
however , @xcite suggested that the extended hi in izw 18 is a remnant of the nascent cloud from which the galaxy formed .
there is no known companion of ngc 1569 that could have caused the disturbed kinematics in its outer regions through tidal interaction .
these examples show that there may not be a single origin of the disturbed hi kinematics in these systems .
it is not possible in general to decide unambiguously which process occurs from hi observations of a particular galaxy .
the proximity of ic 342 suggests that the spur may be the result of tidal interaction with this much larger neighbor .
if the spur is a tidal tail induced by ic 342 , some inferences can be made based on the kinematics and morphology of the spur .
first , the velocity of the spur in the direction of the center of ugca 86 is blue shifted relative to the hi disk .
a tidal tail is expected to have a radial velocity component away from the center .
this implies that the blue shifted spur gas should be located in front of the hi disk .
second , the 3-dimensional space velocity of the gas in a tidal tail is expected to vary smoothly with position along the tidal tail because adjacent test particles in the tidal tail always experienced similar accelerations during the interaction .
this continuity of the spacial velocity should be seen in the data as a smooth gradient of the line of sight velocity along the tidal tail .
the observed variation of the line of sight velocity along the _ minor _ axis of the spur is not necessarily in conflict with this expectation .
the minor - axis velocity gradient may be the result of a projection effect where different parts of the tidal tail are seen close together on the plane of the sky .
for this to be the case , the tidal tail must be in a plane that is observed nearly edge - on .
the orientation of the spur along the minor axis of the optical and hi isophotes of the disk of ugca 86 further means that the plane of the tidal tail makes a large angle ( close to @xmath126 ) with the plane of the disk , no matter what the inclination of the disk might be .
third , the fact that the spur extends on the sky in the direction of ic 342 , and the necessity that the spur is in a plane observed nearly edge - on , mean that ic 342 must be close to the plane of the spur .
fourth , if different parts of the tidal tail are to be seen in projection on the sky along the entire spur , the tip of the spur that appears most distant from ugca 86 is a point where the line of sight is approximately a tangent to the tidal tail .
if this is the case , one would expect that the spur is brighter at this point and that the velocity gradient along the minor axis vanishes .
this expectation is confirmed by figure [ gray : minorslice ] . the inference that ic 342 is in the same plane as the tidal tail is an interesting argument in favor of an interpretation of the spur as a tidal tail induced by ic 342 .
however , significant problems exist with this interpretation .
if the spur is a tidal tail in a nearly polar plane , significant warping of the outer hi disk of ugca 86 is also expected to occur .
this warp should manifest itself by a change in p.a . of the hi isophotes and the velocity field of the disk .
no such variation in p.a . is observed in figure [ gray : hicolumn ] and figure [ gray : velfield ] .
@xcite similarly argued that the component hi - sx in izw 18 is not a tidal tail because of lack of continuity in the velocity field between the components hi - a and hi - sx .
the necessity of a tidal tail in a plane that is perpendicular to the plane of the disk is by itself a problem because low - inclination encounters seem to be most effective in the formation of tidal tails @xcite .
finally , in the inferred geometry the tidal tail appears in the direction of ic 342 , with no evidence for a counter tail .
this is contrary to simulations of tidally interacting disk galaxies , where the counter tail is the most prominent tidal feature @xcite .
detailed modeling specific to this system is required to make these arguments conclusive .
ugca 86 currently experiences a period of enhanced star formation .
the detection of ugca 86 in the radio continuum , especially at 408 mhz , indicate exceptional star forming activity for a dwarf galaxy .
further evidence for significant recent star formation is seen in the numerous hii regions and the cometary knot . it is tempting to associate the spur with the current star formation in ugca 86 .
gas falling into the disk as suggested by @xcite and @xcite may have triggered the recent episode of star formation .
indirect evidence that this may have occurred is found in the region with anomalous velocity shown in figure [ holeslice - fig ] .
this region occurs on the side of ugca 86that displays the most intense star formation .
if the kinetic energy of this region is supplied by stellar winds and supernovae , the equivalent of more than 200 supernova explosions is required .
the integrated star formation rate of ugca 86 of @xmath127 implies a supernova rate of @xmath128 supernovae per year for the entire galaxy , assuming a salpeter imf with low mass cut - off @xmath129 and high - mass cut - off @xmath130 .
the main hii regions are found on the periphery of the anomalous velocity region , not near the center .
the power and the location of the star formation regions suggest that the observable population of high - mass stars is not the source of the anomalous velocity region .
the impact of high - velocity clouds in a galactic disk has been suggested as an alternative energy source for super shells with very high energy requirements @xcite .
the presence of the spur and a kinematically disturbed region in the disk that requires more energy input than the observable stars can provide , lends some credibility to the suggestion of gas falling into ugca 86 .
we present hi , and 1.4 ghz and 408 mhz continuum observations of the magellanic dwarf irregular galaxy ugca 86 with the drao interferometer .
the hi morphology of ugca 86 is characterized by a rotating disk with inclination @xmath131 , but a significant misalignment exists between the optical / hi major axis and the kinematic major axis .
in addition to this , a narrow spur of hiextends to the northwest , up to @xmath71 ( 9 kpc ) from the center of ugca 86 .
the spur is oriented along the apparent minor axis of the disk of ugca 86 , and its kinematics are significantly different from the disk .
the spur partially overlaps with the disk on the sky .
the velocity difference between the spur and the disk towards the center of ugca 86 is @xmath120 . in this direction
we measure directly the radial component of the orbital velocity of gas in the spur .
this large radial component , one third of the circular velocity at the edge of the hi disk , is evidence that the orbits of gas in the spur are significantly non - circular . also
, the overlap of the spur and the disk on the sky , and the modest inclination of the disk imply that part of the spur is located well outside the plane of the disk , but no evidence for a warp of the outer hi disk is found . a well - defined velocity gradient is observed along both the major and the minor axes of the spur .
the mean hi velocity dispersion of the disk @xmath132 is similar to what is observed in other ( dwarf ) galaxies .
however , the velocity dispersion of hi in the spur shows a large range , from @xmath29 to @xmath6 .
ugca 86 is also detected in the continuum at 1.4 ghz and 408 mhz .
the 1.4 ghz continuum image shows the three largest hii region complexes , and an extended component that covers the entire optical disk of ugca 86 . at 408 mhz ,
the galaxy is unresolved .
the global spectral index between 408 mhz and 1.4 ghz is @xmath117 ( @xmath118 ) .
the interpretation of the spur as a tidal tail induced by ic 342 , following an earlier suggestion by @xcite is considered . in order to explain the observed hi kinematics of the spur
, the tidal tail must be in a plane that is nearly perpendicular to the disk , although no warping of the outer hi disk is observed .
also , the spur is oriented in the direction of ic 342 , which means that the prominent tidal tail is located in between the two interacting galaxies .
detailed modeling of the interaction between ic 342 and ugca 86 is necessary to address these objections against a tidal tail interpretation of the spur .
if the spur is not a tidal tail , it is likely gas in an extended , non - circular orbit around ugca 86 .
this gas may then be a remnant of the from the formation of ugca 86 that has not yet settled into the disk , or an unlucky small dwarf galaxy that was tidally disrupted by ugca 86 .
the dominion radio astrophysical observatory is operated as a national facility by the national research council of canada .
this research made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory at caltech , under contract with the national aeronautics and space administration .
jms is grateful to dr .
a. r. taylor for financial support during this work .
the authors thank dr .
h. lee for his comments on the manuscript .
landecker , t. l. , dewdney , p. e. , burgess , t. a. , gray , a. d. , higgs , l. a. , hoffmann , a. p. , hovey , g. j. , karpa , d. r. , lacey , j. d. , prowse , n. , purton , c. r. , roger , r. s. , willis , a. g. , wyslouzil , w. , routledge , d. , vaneldik , j. f. 2000 , a&as , 145 , 509 | we present 21-cm hi line and 408 mhz and 1.4 ghz continuum observations of the magellanic dwarf galaxy ugca 86 , made with the drao synthesis telescope .
ugca 86 is detected in the continuum at both frequencies , with 408 mhz flux density @xmath0 and 1.4 ghz flux density @xmath1 .
the hi structure of ugca 86 is complex , with two separate components : a rotating disk and a highly elongated spur that is kinematically disjunct from the disk .
the hi disk is centered on the optical galaxy with similar axial ratio and orientation of the major axis .
an area of the disk with a peculiar velocity of @xmath2 relative to the regular rotation of the disk is found on the southern side , where most of the star formation is concentrated .
the spur is seen along the minor axis of ugca 86 and overlaps in part with the disk . towards the optical center of ugca 86 , the velocity difference between the spur and the disk is @xmath3 , about one third of the rotation velocity of the hi disk at 6 kpc from the center .
this implies a large radial component of the orbital velocity of the spur , and therefore a significantly non - circular orbit .
the median hi velocity dispersion of the disk is @xmath4 , similar to other ( dwarf ) galaxies .
the hi velocity dispersion of the spur varies from @xmath5 to @xmath6 .
a possible tidal origin of the spur is considered in view of the proximity of the large scd galaxy ic 342 .
however , the orientation of the spur along the minor axis and its spatial overlap with the disk suggest that the spur is located far outside the plane of the hi disk .
no evidence is found that the outer hi disk is warped , which poses a problem for the interpretation of the spur as a tidal tail induced by ic 342 .
detailed modelling of the ic 342/ugca 86system will be required before a tidal origin of the spur can be dismissed conclusively .
the possibility that the spur is part of the nascent cloud of ugca 86 or the remains of a small dwarf galaxy is presented as an alternative interpretation . |
we consider the hyperbolic conservation laws @xmath0 with the state vector u @xmath1 u(t , x ) : @xmath2 , for a time interval @xmath3 $ ] for some @xmath4 and an open bounded computational domain @xmath5 .
@xmath6 is the flux function .
an initial condition @xmath7 is given along with appropriate boundary conditions . despite the smoothness of @xmath8 , the solution to ( [ eq1 ] ) may develop a discontinuity within a finite time .
high order numerical approximations of the developed discontinuity suffer from the gibbs phenomenon yielding spurious oscillations near the discontinuity .
these oscillations not only degrade the accuracy of the approximation but may also cause instability of the scheme . since the publications by harten et al .
@xcite and by jiang and shu @xcite , the essentially non - oscillatory ( eno ) and weighted essentially non - oscillatory ( weno ) methods have been one of the most powerful numerical methods that can successfully deal with the gibbs oscillations and widely used in various applications .
numerous modifications of the original eno and weno methods have been also developed .
these include recent works such as weno - m @xcite , weno - z @xcite , power - eno @xcite , weno - p @xcite , modification of the eno basis @xcite and weno-@xmath9 methods @xcite , to name a few .
there is no _ best _ eno / weno variation because all variations have their own strengths and weaknesses .
however , most variations have the common element for their construction : the polynomial reconstruction . in recent reviews of the weno methods by shu @xcite , the eno / weno reconstruction based on the non - polynomial functions such as the fourier functions
is briefly mentioned such as the one used in @xcite . however , most eno / weno variations are based on the polynomial reconstruction . in this paper , let us call the eno / weno method based on the polynomial reconstruction as the regular eno / weno method . in this paper , we present a simple new type of the eno / weno method based on the non - polynomial interpolation . as an example of the non - polynomial bases , we first use radial basis functions ( rbfs ) . as we will show later in this paper ,
the choice of rbfs as a non - polynomial basis function is not necessary but the perturbation form works in more general sense .
furthermore , the presented method is not a hybrid method that combines the eno / weno method with the rbf method or other high order methods .
the presented method is basically the eno / weno method but with the interpolation coefficients slightly modified . in the seminal work of @xcite ,
the ader method was developed based on the polyharmonic spline , which belongs to the family of piecewise smooth rbfs .
the motivation of the method presented in @xcite was to adopt the weno method efficiently for the arbitrary geometry and unstructured mesh by using the meshless feature of rbfs .
our main motivation in this paper , however , is not in using the meshless feature of rbfs but is to enhance the original eno / weno accuracy by modifying the interpolation coefficients . for this
, we need free parameters and need to optimize them . in @xcite
the polyharmonic spline was used and there was no undetermined shape parameter or the shape parameter is fixed as @xmath10 and the order of convergence is overall fixed once the size of each stencil @xmath11 is fixed .
but in our proposed method , the shape parameter is essential .
thus the presented method in this paper is different from the one in @xcite .
.commonly used radial basis functions @xmath12 with @xmath13 known as the shape parameter .
[ cols="^,^,^ , < " , ] [ table0_2d ] we consider the same linear advection equation with the following discontinuous initial condition and the boundary condition @xmath14 \\
-1 & \text{if } & y \in ( 0.5,1 ] \end{array } \right . , \\
v(x,0,t ) & = & v(x,1,t ) = 1 , \quad t>0 . \label{example_2d_2 } \end{aligned}\ ] ] with this example , we check how the rbf - eno solution behaves near the discontinuity . to handle the discontinuous solution
, we apply the same monotone polynomial method as in 1d . in figure
[ figure_2d_1 ] , the solutions by the regular eno ( blue ) and rbf - eno ( red ) methods are given with time at @xmath15 .
the figure clearly shows that the rbf - eno yields non - oscillatory solutions for all time .
the figure also shows that the rbf - eno method yields much sharper solution profile than the regular eno method , which is clearly shown in the zoomed profile in the bottom right figure . with the discontinuous initial condition , for the eno ( blue ) , rbf - eno ( red ) methods with @xmath16 and @xmath17.,scaledwidth=70.0% ] we consider the 2d burgers equation for @xmath18\times [ 0,1]$ ] @xmath19 with the initial condition @xmath20 figure [ 2d_figure3 ] shows the rbf - eno solutions at various times at @xmath15 ( left figure ) and the pointwise errors by the eno ( blue ) and rbf - eno ( red ) methods with @xmath16 and @xmath17 at @xmath21 .
the left figure clearly shows that the rbf - eno solution is not oscillatory yet yielding a sharp shock profile near the boundaries .
the right figure shows that the rbf - eno method yields more accurate results than the regular eno method in the smooth region . by the eno ( blue ) and rbf - eno ( red ) on logarithmic scale with @xmath16 and @xmath17 .
, title="fig:",scaledwidth=50.0% ] by the eno ( blue ) and rbf - eno ( red ) on logarithmic scale with @xmath16 and @xmath17 .
, title="fig:",scaledwidth=40.0% ]
in this paper , we developed a non - polynomial eno method for solving hyperbolic equations .
as an example of non - polynomial bases , we used rbfs . the formulation based on the non - polynomial basis yields the flexibility of improving the original eno accuracy .
the key idea of the developed method lies in the adaptation of the shape parameters in the expansion with a non - polynomial basis that can make the leading error term vanish or at least become small in the local interpolation .
the new non - polynomial eno method improves local accuracy and convergence if the underlying solution is smooth . for the non - smooth solution such as a shock
, we adopted the monotone interpolation method so that the non - polynomial eno reconstruction is reduced into the regular eno reconstruction resulting in the suppression of the gibbs oscillations .
the numerical results show that the non - polynomial eno method is superior to the regular eno method and even better than the weno - js method for @xmath16 .
the non - polynomial eno method yields @xmath22th order accuracy while the regular eno solution is only @xmath23rd order accurate for @xmath24 in the smooth region .
the numerical results show that the developed method yields highly accurate results for the scalar problems for both @xmath16 and @xmath24 . for the system problems ,
the non - polynomial eno solutions are similar to or better than the regular eno solutions .
the non - polynomial eno scheme is slightly more costly than the regular eno scheme because it has a procedure of computing the optimal shape parameter values .
but it is less expensive than the weno scheme for the given value of @xmath11 .
for some cases , the non - polynomial eno method achieves the same level of accuracy as the weno method or even better accuracy than the weno method while its computational cost is less demanding than the weno method .
the weno method based on the non - polynomial eno reconstruction is also better then the regular weno method .
the 2d non - polynomial finite volume interpolation is more beneficial than the 2d polynomial interpolation .
we showed that the non - polynomial interpolation can raise the order which can not be obtained with the polynomial interpolation even though all the cell averages are used .
we provided the table of the reconstruction coefficients for @xmath16 and @xmath24 for the non - polynomial eno method .
the non - polynomial eno formulation for higher values of @xmath11 will be considered in our future work .
the current work considered the uniform mesh only . in our future work , we will investigate the non - polynomial eno method with the nonuniform and unstructured mesh .
as mentioned in introduction , the meshless feature of rbfs was combined with the weno method in @xcite where the shape parameter was globally fixed for the reconstruction .
it will be interesting to investigate how the optimization of the shape parameter can be realized with the meshless properties of rbfs on the unstructured mesh .
.1 in * acknowledgments : *
the authors thank w .- s .
don for his useful comments on the construction of the rbf - eno / weno method .
the second author thanks grady wright for his useful comments on the rbf interpolation . ,
essentially non - oscillatory and weighted essentially non - oscillatory schemes for hyperbolic conservation laws , advanced numerical approximation of nonlinear hyperbolic equations " ( lecture notes in mathematics 1697 ) , a. quarteroni ( ed . ) , springer - verlag , 1998 . | the essentially non - oscillatory ( eno ) method is an efficient high order numerical method for solving hyperbolic conservation laws designed to reduce the gibbs oscillations , if existent , by adaptively choosing the local stencil for the interpolation .
the original eno method is constructed based on the polynomial interpolation and the overall rate of convergence provided by the method is uniquely determined by the total number of interpolation points involved for the approximation . in this paper , we propose simple non - polynomial eno and weighted eno ( weno ) finite volume methods in order to enhance the local accuracy and convergence .
we first adopt the infinitely smooth radial basis functions ( rbfs ) for a non - polynomial interpolation .
particularly we use the multi - quadric and gaussian rbfs .
the non - polynomial interpolation such as the rbf interpolation offers the flexibility to control the local error by optimizing the free parameter .
then we show that the non - polynomial interpolation can be represented as a perturbation of the polynomial interpolation .
that is , it is not necessary to know the exact form of the non - polynomial basis for the interpolation . in this paper , we formulate the eno and weno methods based on the non - polynomial interpolation and derive the optimization condition of the perturbation .
to guarantee the essentially non - oscillatory property , we switch the non - polynomial reconstruction to the polynomial reconstruction adaptively near the non - smooth area by using the monotone polynomial interpolation method .
the numerical results show that the developed non - polynomial eno and weno methods enhance the local accuracy .
* keywords * essentially non - oscillatory method , weighted essentially non - oscillatory method , radial basis function interpolation , finite volume method , hyperbolic conservation laws . |
with the commissioning of the relativistic heavy ion collider ( rhic ) , nuclear collision physics entered the collider era .
new phenomena have been presented in @xmath2 collisions at the center - of - mass energy @xmath3 130 gev / nucleon@xcite .
negative hadron multiplicities are significantly increased compared to proton - antiproton collisions@xcite .
some of the data , such as the antiproton - to - proton ratios@xcite , are reported to contradict perturbative quantum chromodynamics ( pqcd ) calculations .
initial efforts to understand hadron spectra from nuclear reactions at rhic using pqcd have been published recently@xcite .
however , there is still a strong debate on how to apply pqcd to hadron production even in proton - proton ( @xmath4 ) collisions . with rhic running at the full energy , and experiments at the large hadron collider ( lhc ) at several
tev / nucleon soon to follow , `` clean '' processes , where pqcd works well in @xmath4 collisions , are needed to test pqcd in nuclear reactions at collider energies .
it is particularly important to distinguish between expected nuclear effects and `` new physics '' .
the production of @xmath0 and @xmath1 bosons has been extensively studied at tevatron energy ( @xmath5 tev ) .
perturbative qcd proved to be very successful in explaining the cdf@xcite and d0@xcite data .
due to the large mass of @xmath0 and @xmath1 , and no final state rescattering in their production process , power corrections are expected to be small .
therefore , @xmath0 and @xmath1 production could provide a bench mark test for pqcd at rhic and lhc in both @xmath4 and nuclear collisions .
luminosity remains a concern at rhic , in particular in @xmath4 collisions .
the bulk of the data for @xmath0 and @xmath1 production is concentrated in the small transverse momentum ( @xmath6 ) region , where @xmath6 is much smaller than the corresponding heavy boson mass ( @xmath7 ) . when @xmath8 , the @xmath6 distributions calculated order - by - order in @xmath9 in conventional fixed - order perturbation theory receive a large logarithm , @xmath10 , at every power of @xmath9 , even in the leading order in @xmath9 .
therefore , at sufficiently small @xmath6 , the convergence of the conventional perturbative expansion in powers of @xmath9 is impaired , and the logarithms must be resummed .
resummation of the large logarithms in qcd can be carried out either in @xmath6-space directly , or in the so - called `` impact parameter '' , @xmath11-space , which is a fourier conjugate of the @xmath6-space .
( since we are also interested in _ nuclear _ collisions , we reserve the notation @xmath12 for the usual impact parameter associated with the geometry of colliding heavy nuclei , and denote the mathematical tool used in the resummation procedure by the symbol @xmath11 . ) using the renormalization group equation technique , collins and soper improved the @xmath11-space resummation to resum all logarithms as singular as , @xmath13 , as @xmath14 @xcite . in the frame work of this renormalization group improved @xmath11-space resummation , collins , soper , and sterman ( css ) derived a formalism for the transverse momentum distribution of vector boson production in hadronic collisions@xcite . in the css formalism ,
non - perturbative input is needed for the large @xmath11 region .
the dependence of the pqcd results on the non - perturbative input is not weak if the original extrapolation proposed by css is used .
recently , a new extrapolation scheme was proposed , based on solving the renormalization group equations , including power corrections@xcite . using the new extrapolation formula ,
the dependence of the pqcd result on the non - perturbative input was significantly reduced .
this was achieved without matching the fixed - order calculations .
the results agree with the d0 nd cdf data very well in the entire @xmath6 interval from @xmath15 gev to @xmath6 as large as the the vector mass .
in addition , in this new extrapolation formalism , power corrections can be studied systematically for @xmath4 to proton - nucleus ( @xmath16 ) to nucleus - nucleus ( @xmath17 ) collisions .
factorization plays a key role in pqcd predictions of observables . at rhic and lhc
, nuclear collisions reach an unprecedented energy scale .
factorization can not be extrapolated easily to @xmath17 collisions from hadron - hadron collisions@xcite , and power corrections may become important @xcite .
there are two types of power corrections : ( 1 ) power corrections directly to the physical observables ( type - i ) and ( 2 ) power corrections to the evolution of renormalization group equations ( type - ii ) .
type - i power correction are proportional to powers of @xmath18 , q being the physical large scale ( momentum ) .
these are small for @xmath0 and @xmath1 production as a result of the large mass of the particles in question .
type - ii power corrections are proportional to powers of @xmath19 , with evolution scale @xmath20 .
therefore , physical observables carry the effect of type - ii power corrections for all @xmath21 $ ] , with the boundary condition at the scale @xmath22 .
even with large mass , @xmath0 and @xmath1 can still carry a large effect of type - ii power corrections .
however , ref .
@xcite provides a way to study the effect of type - ii power corrections numerically .
based on the numerical results presented in ref .
@xcite , the effect of type - ii power corrections is also expected to be small for heavy vector boson production in @xmath4 collisions . in nuclear collisions , type - ii power corrections will be enhanced by the nuclear size . here
we study the effects of enhanced power correction in @xmath16 and @xmath17 collisions .
the rapidity distribution of @xmath0 and @xmath1 bosons at lhc was calculated recently and the effect of shadowing was discussed in this context@xcite .
the decay spectrum of @xmath1 was suggested as an alternative reference process for @xmath23 suppression at lhc , since the drell - yan lepton pairs at low mass are expected to be dominated by @xmath24 decay . as most of the @xmath0 and @xmath1 bosons observed in the experiment
will be in the small transverse momentum region , it is of great interest to study the nuclear effects on massive gauge boson transverse momentum distributions .
these effects include both the nuclear modification of the parton distribution function ( shadowing ) , and the initial state power corrections . the second part of the present paper is devoted to these subjects .
the rest of the paper is organized as follows . in section [ sec2 ] , we review the css @xmath11-space resummation formalism and the new procedure@xcite to extrapolate the pqcd calculation to the large @xmath11 region is outlined .
we give predictions for @xmath0 and @xmath1 transverse momentum distributions from @xmath4 collision at both rhic and lhc energies .
the sensitivity of the prediction to the non - perturbative parameters is discussed .
nuclear effects on @xmath0 and @xmath1 transverse momentum distributions in @xmath16 and @xmath17 collisions are studied in sections [ sec3 ] and [ sec4 ] . in section [ sec5 ] , we present our conclusions . to make the paper reasonably self - contained , the appendix summarizes relevant details of the css formalism .
we use @xmath25 units .
in this section , we briefly review the css @xmath11-space resummation formalism ( section [ subsec21 ] ) . in section [ subsec22 ]
we introduce the recently - proposed new formalism to extrapolate the perturbative calculation to the large @xmath11 region@xcite , and discuss the transverse momentum distribution of heavy vector bosons in @xmath4 collisions at rhic and lhc energies . in the production of heavy vector bosons ( of mass @xmath7 ) , when @xmath8 , the @xmath6 distribution calculated order - by - order in @xmath9 in conventional fixed - order perturbation theory receives a large logarithm , @xmath10 , at every power of @xmath9 . even at the leading order in @xmath9 , the cross section
@xmath26 contains a term proportional to @xmath27 coming from the partonic subprocess @xmath28 , where @xmath29 or @xmath1 . beyond the leading order ,
we actually get two powers of the logarithm for every power of @xmath9 , due to soft and collinear gluons emitted by the incoming partons . therefore , at sufficiently small @xmath6 , the convergence of the conventional perturbative expansion in powers of @xmath9 is impaired , and the logarithms must be resummed . for vector boson production in hadronic collisions between hadrons @xmath30 and @xmath31 , @xmath32 with @xmath29 , or @xmath1 ,
the css resummation formalism yields the following generic form @xcite : @xmath33 where @xmath34 and @xmath35 , with rapidity @xmath36 and collision energy @xmath37 . in eq .
( [ css - gen ] ) , the @xmath38 term dominates the @xmath6 distributions when @xmath8 , and the @xmath39 term gives corrections that are negligible for small @xmath6 , but become important when @xmath40 . in the css formalism , the function @xmath38 appears as a superposition , @xmath41 where @xmath42 is the lowest order cross section for a quark - antiquark pair of invariant mass @xmath7 to annihilate into a vector boson @xmath43 , and the @xmath44 runs over all possible quark and antiquark flavors that can annihilate into the vector boson at the born level@xcite . in eq .
( [ css - w - ij ] ) , @xmath45 is an effective flux to have partons of flavor @xmath46 and @xmath47 from the respective hadrons @xmath30 and @xmath31 , and it has the form @xmath48 where @xmath49 is given in the appendix , and @xmath50 is a constant of order unity@xcite .
the functions @xmath51 in eq .
( [ css - w - sol ] ) depend on only one momentum scale , @xmath52 , and are perturbatively calculable as long as @xmath52 is large enough .
all large logarithms from @xmath53 to @xmath54 in @xmath45 are completely resummed into the exponential factor @xmath55 $ ] . since the perturbatively resummed @xmath45 in eq .
( [ css - w - sol ] ) is only reliable for the small @xmath11 region , an extrapolation to the large @xmath11 region is necessary in order to complete the fourier transform in eq .
( [ css - gen ] ) .
this is the point where the dependence of the result on the nonperturbative parameters enters into the formalism , possibly limiting the predicative power of pqcd , especially for small @xmath6 .
we will next discuss this in detail .
the css formalism in its original form introduces a modification to the output of the perturbative calculation ( see appendix , eq .
( [ css - w - b ] ) ) .
the size of the modification depends on the non - perturbative input parameters , and can be as large as 50% of the perturbative result .
therefore the predictive power of the css formalism was questioned and new efforts have been devoted to resum the large logarithms directly in @xmath6 space @xcite . in oder to improve the predictive power of the css @xmath11-space resummation formalism , the following new form was proposed@xcite by solving the renormalization group equation including power corrections : @xmath56 where the nonperturbative function @xmath57 in eq .
( [ qz - w - sol - m ] ) is given by @xmath58 \nonumber \\ & \ & { \hskip 0.8 in } -\bar{g}_2 \left(\tb^2 - \tb_{max}^2\right ) \bigg\ } \,\,\ , .
\label{qz - fnp - m}\end{aligned}\ ] ] in eq . (
[ qz - w - sol - m ] ) @xmath59 is a parameter to separate the perturbatively calculated part from the non - perturbative input . unlike in the original css formalism , when @xmath60 , the perturbatively calculated @xmath61 is not altered and is independent of the nonperturbative parameters .
in addition , the @xmath11-dependence in eq .
( [ qz - fnp - m ] ) is separated according to different physics origins .
the @xmath62-dependence mimics the summation of the perturbatively calculable leading power contributions to the kernels @xmath63 and @xmath64 in eq.s ( [ css - k - rg ] ) and ( [ css - g - rg ] ) of the appendix , respectively , to all orders in the running coupling constant @xmath65 with the scale @xmath20 running into the nonperturbative region @xcite . the @xmath66-dependence of the @xmath67 term is a direct consequence of dynamical power corrections to the renormalization group equations of the kernels @xmath63 and @xmath64 , and has an explicit dependence on @xmath7 .
the @xmath68 term represents the effect of the non - vanishing intrinsic parton transverse momentum .
as we discuss later , these two terms behave differently in nuclear collisions .
a remarkable feature of the @xmath11-space resummation formalism is that the resummed exponential factor @xmath55 $ ] ( see eq . ( [ css - w - sol ] ) ) suppresses the @xmath11-integral when @xmath11 is larger than @xmath69 .
therefore , when @xmath70 , ( where @xmath71 represents the scale at which pqcd starts to be applicable ) , it is possible that the fourier transform in eq .
( [ css - gen ] ) is dominated by a region of @xmath11 much smaller than @xmath72 , and the calculated @xmath6 distributions are insensitive to the nonperturbative information at large @xmath11 .
in fact , it was shown using the saddle point method that , for a large enough @xmath7 , a qcd perturbation calculation is valid even at @xmath73@xcite . as discussed in ref .
@xcite , the value of the saddle point strongly depends on the collision energy @xmath37 , in addition to its well - known @xmath74 dependence . because of the steep evolution of parton distributions at small @xmath75 , the @xmath37 dependence of the @xmath38 in eq .
( [ css - gen ] ) significantly decreases the value of the saddle point and improves the predictive power of the @xmath11-space resummation formalism at collider energies . to display the saddle point clearly ,
let us rewrite the first term on the right - hand side of eq .
( [ css - gen ] ) , taking into account that there is no preferred transverse direction and that @xmath38 in eq .
( [ css - gen ] ) is a function of @xmath76 only : @xmath77 where @xmath38 contains the exponential factor @xmath55 $ ] , and @xmath78 with @xmath79 represents the bessel function of order zero .
when @xmath80 , the bessel function suppresses the large @xmath11 region of the integration .
because the argument of the bessel function is proportional to @xmath6 , the large @xmath11 region is more suppressed if @xmath6 is larger . in the following , we focus on the saddle point at @xmath73 .
1 is a graph ( on an arbitrary scale ) of the integrand of the right hand side of eq .
( [ css - w - f ] ) with @xmath73 for the case of @xmath1 production at the collision energies where we intend to carry out our calculations : @xmath81 350 gev ( dotted line ) , @xmath81 500 gev ( short dashes ) , and at @xmath81 5.5 tev ( solid ) .
the position of the saddle point decreases as the collision energy increases .
we will show that the predictive power of the formalism is very good at rhic . as a consequence of the behavior of the saddle point position
, we expect excellent precision at @xmath81 5.5 tev . at the lhc @xmath4 energy of @xmath81 14 tev ,
the saddle point moves to even smaller values of @xmath11 , improving the accuracy of the calculation even further .
the parameters @xmath82 and @xmath83 of eq .
( [ qz - fnp - m ] ) are fixed by the requirement of continuity of the function @xmath84 and its derivative at @xmath85 .
( the results are insensitive to changes of @xmath59 in the interval 0.3 gev@xmath86 0.7 gev@xmath87 .
we use @xmath88 0.5 gev@xmath87 in ref .
@xcite and here . ) the value of @xmath67 and @xmath89 can be obtained by fitting the low - energy drell - yan data , and are taken to be @xmath90 gev@xmath91 , and @xmath92 gev@xmath91 following ref.s @xcite and @xcite . as the @xmath11 dependence of the @xmath67 and @xmath89 terms in eq .
( [ qz - fnp - m ] ) is identical , it is convenient to combine these terms and define @xmath93 using the values of the parameters listed above , we get @xmath94 gev@xmath91 for @xmath0 and @xmath1 production in @xmath4 collisions .
the parameter @xmath95 can be considered the only free parameter in the non - perturbative input in eq .
( [ qz - fnp - m ] ) , arising from the power corrections in the renormalization group equation .
our results are not sensitive to small variations of @xmath95 around its estimated value .
we use @xmath96 gev@xmath91 in the following .
an impression about the importance of power corrections can be obtained by comparing results with @xmath96 gev@xmath91 to those with power corrections turned off ( @xmath97 ) .
we therefore define the ratio of these quantities : @xmath98 the cross sections in the above equation and in the results presented in this paper have been integrated over rapidity and invariant mass squared from eq .
( [ css - gen ] ) . for the integration over @xmath99
, we use the `` narrow width approximation''@xcite . with respect to rapidity
, we integrate over the interval @xmath100 $ ] for rhic , and over [ -2.4,2.4 ] for lhc , representing the central acceptances of the appropriate detectors . for the parton distribution functions
, we use the cteq5 m set@xcite in the present work .
the following figures display differential cross sections and the corresponding @xmath101 ratios for @xmath1 and @xmath0 production as functions of @xmath6 , in order of increasing energy .
( in what follows , by @xmath0 production cross section we mean the cross section for @xmath102 . ) in fig .
2(a ) we show the differential cross section for @xmath1 production in @xmath4 collisions at @xmath103 gev , where a rhic @xmath4 run is expected . the accompanying fig .
2(b ) displays @xmath101 of eq . ( [ sigma - g2 ] ) .
it can be seen that the @xmath101 ratio converges to unity ( i.e. very small power corrections ) with increasing @xmath6 , and the effect of power corrections is about five percent for @xmath73 . however , when @xmath104 gev , the deviation of @xmath101 from unity is under two percent .
a very similar conclusion can be drawn from fig . 3 , concerning @xmath0 production in @xmath4 collisions at the same energy : the dependence on the nonperturbative input is weak even at the lowest transverse momenta .
fig .- s 4 and 5 contain similar information for @xmath1 and @xmath0 production in @xmath4 collisions at the planned lhc nucleus - nucleus collision energy of @xmath105 tev , while the cross section and @xmath101 are given for the lhc @xmath4 energy of @xmath106 tev in fig .-
s 6 and 7 . at all energies ,
the magnitude of the @xmath0 production cross section is about 3 to 5 times that of the @xmath1 production cross section . otherwise , the results for @xmath1 and @xmath0 production are very similar , because the saddle points are almost at the same place in @xmath11 space in the two cases .
as it has been pointed out in ref@xcite , the position of the saddle point is determined by two terms : the first term is inversely proportional to the mass of the vector boson , and the second term is proportional to the derivative of the parton distribution function with respect to the scale at @xmath107 or @xmath108 . at the collider energies considered here , the second term is negative . while @xmath109 is a little smaller than @xmath110 , and thus the first term is larger for @xmath111 than for @xmath1 , the contribution from the second term will effect the saddle point for @xmath0 more than that of the @xmath1 with a small @xmath107 and @xmath108 . as expected from fig . 1
, the dependence on the nonperturbative input decreases with increasing collision energy .
the @xmath101 ratio is smaller than one percent at lhc for both @xmath112 tev and @xmath113 tev , even when @xmath73 .
these results confirm that the predictive power of the formalism is very good at rhic and excellent at lhc .
they also show that the effect of power corrections is very small at lhc for almost the whole @xmath6 region and it is also negligible for rhic when @xmath6 is larger than @xmath114 gev .
in lack of nuclear effects on the hard collision , the production of heavy vector bosons in nucleus - nucleus ( @xmath17 ) collisions should scale , compared to the production in @xmath4 collisions , as the number of hard collisions , @xmath17 . however , there are several nuclear effects on the hard collision in a heavy - ion reaction .
we distinguish three categories of these effects : isospin effects , the modification of the parton distribution function in the nucleus , and the enhancement of power corrections by the nuclear size . in this section
we discuss the first two of these nuclear effects .
power corrections will be treated separately in the next section .
as the parton distribution of neutrons is different from that of the protons , the production cross section of heavy vector bosons in proton - neutron and neutron - neutron interactions differs from the corresponding production cross section in @xmath4 collisions .
this difference is the source of the so - called isospin effects . to study the isospin effects ,
let us introduce @xmath115 where @xmath116 and @xmath117 are the atomic numbers and @xmath118 and @xmath119 are the mass numbers of the colliding nuclei , and the cross section @xmath120 has been averaged over @xmath17 , while @xmath121 is the @xmath4 cross section ( with @xmath122 ) . since the @xmath123 coupling in @xmath1 production is replaced by @xmath124 coupling to produce @xmath0 bosons , we expect @xmath125 to be complementary for @xmath0 and @xmath1 production , i.e. if one of these ratios is larger than one , the other ratio should be smaller than unity .
next we turn to the phenomenon of shadowing .
we use the term shadowing in the general sense , referring to all modifications of the parton distribution function in the environment of the nucleus@xcite .
this includes , in different regions of @xmath75 , the phenomena of shadowing ( in the strict sense ) , anti - shadowing , and the emc effect .
it is also important to note that we define shadowing effects as leading twist effects , not including power corrections@xcite .
thus , shadowing effects and power correction are clearly separated in the present work . in general ,
shadowing is expected to be a function of @xmath75 , the scale @xmath126 , and of the position in the nucleus .
the latter dependence means that in heavy - ion collisions , shadowing should be impact parameter ( @xmath12 ) dependent .
the parameterizations of shadowing in the literature take into account some of these effects , but no complete parameterization exists to date to our knowledge .
for example , the hijing parameterization includes impact parameter dependence , but does not deal with the scale dependence@xcite . on the other hand ,
the eks98@xcite and hkm@xcite parameterizations have a scale dependence , but do not consider impact parameter dependence .
( the latter parameterizations have been compared recently@xcite . ) since in this paper we concentrate on impact - parameter integrated results , where the effect of the @xmath12-dependence of shadowing is relatively unimportant@xcite , and we focus more attention on scale dependence , we find the eks98 shadowing parameterization most appropriate for our present purposes .
therefore we use eks98 shadowing@xcite in this work . similarly to eq .
( [ sigma_iso ] ) for isospin effects , we define a shadowing ratio , @xmath127 thus , @xmath128 incorporates the effects of both shadowing and the isospin composition of the nuclei .
furthermore , in this section we take @xmath97 throughout , switching off power corrections .
the enhanced power corrections in @xmath16 and @xmath17 collisions will be discussed separately in the next section .
displays @xmath125 ( dashed lines ) and @xmath128 ( full lines ) as functions of @xmath6 for @xmath1 production ( fig .
8(a ) ) and @xmath0 production ( fig .
8(b ) ) , respectively , at @xmath3 350 mev .
the same information is presented at @xmath3 5.5 tev in fig .
we find that @xmath129 for @xmath1 production and @xmath130 for @xmath0 production at both energies , complementing each other , as expected .
how close @xmath125 is to unity depends on the region of @xmath75 that has the dominant contribution to the cross section . at lhc , @xmath131 , and
the magnitude of the isospin effects is about 2% .
this is because when @xmath75 is in this range , the difference of the distributions of @xmath132 and @xmath133 quarks in the nucleus is very small . at rhic ,
where the dominant @xmath134 , the isospin effects are about 20% in the small @xmath6 region .
we have also checked that the scale dependence of isospin effects is not important . in fig .
8 , we also show a dotted line , which is the ratio of @xmath128 to @xmath125 , and thus can be considered to display the effect of `` pure shadowing '' .
pure shadowing appears as an approximately constant reduction of 3 - 5% at @xmath134 .
the appearance of @xmath128 ( full lines ) is more surprising at @xmath3 5.5 tev ( fig .
9 ) . since at lhc , even at @xmath135 gev , @xmath136
, we are still in the `` strict shadowing '' region .
therefore , the fact that @xmath137 for 20 gev @xmath138 70 gev is not `` antishadowing '' ; rather , it is a consequence of the change of the shape of the cross section from @xmath4 to @xmath17 reactions . in @xmath4 collisions ,
the maximum of the cross section is at @xmath139 3.7 gev . in @xmath17 collisions ,
the peak is shifted by about 0.4 gev in the direction of large transverse momenta . on the other hand
, the contributions from the scale @xmath140 to @xmath7 have been resummed .
resummation plays a more important role for small @xmath6 than for large @xmath6 . in eks98 ,
the strict shadowing effect is more important for small scales , such as @xmath141 gev , than for @xmath142 .
thus , a relatively small shift in the peak position brings about the rise of @xmath128 above unity .
the shadowing ratio @xmath128 dips below one again at a certain @xmath6 , to approach the fixed - order pqcd result when @xmath142 .
the situation in fig .
9 can be understood by noting that shadowing in the resummed transverse momentum distributions is sensitive to the scale used to describe the nuclear effects on the parton distribution function .
after this discussion one may want to revisit fig . 8 and ask why is @xmath128 flat at rhic energies .
the answer lies in the observation that the emc region of @xmath75 dominates the cross section at @xmath3 350 mev .
there is no sensitivity to the scale in that region in eks98 . in summary ,
the shadowing effects in the @xmath6 distribution of heavy bosons at rhic and lhc are sensitive to the scale of the nuclear parton distribution .
the available data on the scale dependence of nuclear parton distribution functions are very limited@xcite .
theoretical studies ( such as eks98 ) are based on the the assumption that the nuclear parton distribution functions differ from the parton distributions in the free proton , but obey the same dglap evolution@xcite .
therefore , the tranverse momentum distribution of heavy bosons at lhc in @xmath143 collisions can provide a further test of our understanding of the nuclear parton distributions .
we have seen in section [ sec2 ] that the effect of power corrections on the resummed cross section is very weak in @xmath4 collisions at both rhic and lhc . in @xmath16 and @xmath17
collisions , power corrections will be enhanced due to rescattering in the nucleus . in this section
we discuss the enhancement of power corrections in nuclear collisions .
as @xmath89 represents the effect of the partons non - vanishing intrinsic transverse momentum , it should not have a strong nuclear dependence . on the other hand , @xmath67 arises as a result of dynamical power corrections , and should be enhanced by the nuclear size , i.e. proportional to @xmath144 . considering the enhancement , we get @xmath145 gev@xmath91 for @xmath146 and @xmath147 gev@xmath91 for @xmath143 reactions . in order to study the effects of power corrections , we now define @xmath148 in nuclear collision as follows : @xmath149 where the numerator represents the result of the full calculation , with isospin effects , shadowing , and power corrections taken into account , and the denominator was introduced in connection with eq .
( [ sigma - sh ] ) and contains no power corrections ( @xmath122 ) .
fig.10(a ) and fig.11(a ) present the @xmath150 for @xmath0 production at @xmath151 gev in @xmath146 reactions and @xmath112 tev in @xmath143 collisions , respectively . in fig .
10(b ) the solid line is @xmath148 in @xmath146 collisions and the dashed line is @xmath101 in @xmath4 collisions at the same energy . at @xmath152 gev , @xmath148 is about 0.88 , in contrast to about 0.95 in @xmath4 collisions .
the effects of power corrections are enhanced about three times in @xmath153 collision relative to @xmath4 collisions . however , even for @xmath154 , the effect of power corrections is only about 10% , and for @xmath155 gev , the effect remains under 5% .
thus the effect of power corrections is weak in @xmath146 collisions at rhic . as for the situation at lhc in @xmath143 collisions , displayed in fig .
11 , the effects of power corrections appear to be enhanced by a similar factor ( about three ) from @xmath4 to @xmath143 collisions at lhc .
the reason for the similar enhancement is the interplay of the higher energy and the larger value of @xmath95 mentioned above . in any case , the enhanced power corrections remain under 3% in the [ 0,80 ] gev @xmath6 interval for @xmath143 at lhc .
the results for @xmath1 production are similar to those for @xmath0 production .
we have demonstrated that the predictive power of a resummed pqcd calculation for heavy boson production is very good at rhic and excellent at lhc , because the dependence of the cross sections on the non - perturbative input is weak .
this makes @xmath0 and @xmath1 production reliably calculable at the energies considered . on the experimental side
, @xmath0 and @xmath1 production should be an early observable at lhc ( with sufficiently high production rates ) in both @xmath4 and nuclear collisions . at rhic
, production rates may remain on the low side , in particular in @xmath4 collisions . in all cases , the cross sections peak at relatively low transverse momenta , and we calculate the peak position for all systems considered .
we have examined the effects of the isospin composition of nuclei and of shadowing in nuclear collisions .
we find that isospin effects are modest in @xmath0 and @xmath1 production at the energies considered , while shadowing is relatively unimportant at rhic , but plays an important role at lhc in shaping the transverse momentum distributions .
this comes about because of the sensitivity of shadowing effects to the scale of the parton distribution functions .
we also studied the magnitude of power corrections in @xmath4 and nuclear collisions .
power corrections are small in @xmath4 collisions , and , even though part of these corrections is enhanced in nuclear collisions by a factor of @xmath144 , they remain below 10% ( even at very low transverse momenta ) in @xmath17 reactions . as indicated by the above observations , the resummed pqcd formalism yields accurate results for @xmath0 and @xmath1 transverse momentum distributions at rhic and lhc .
the production of heavy bosons can therefore serve as a further test of the nuclear parton distributions . in more general terms ,
@xmath0 and @xmath1 production at rhic and lhc provides a bench mark test of resummed pqcd for @xmath4 and nuclear collisions at these energies .
we hope that our predictions will soon be compared to data .
we thank t. hallman , d. keane , and r. vogt for stimulating discussions and j.w .
qiu for a careful reading of the manuscript .
this work was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 86er-40251 and by the nsf under grant int-0000211 .
here we summarize relevant details of the css formalism concerning the structure of the terms in eq .
( [ css - gen ] ) .
we address , in particular , the construction of the exponential factor @xmath55 $ ] in eq .
( [ css - w - sol ] ) .
it was shown in ref .
@xcite that for @xmath156 , the function @xmath61 of eq .
( [ css - gen ] ) is directly related to the singular parts of the @xmath6 distribution as @xmath157 .
more precisely , @xmath61 includes all singular terms like @xmath158 and @xmath159_{\rm reg}$ ] with @xmath160 .
less singular terms as @xmath14 are included in the @xmath39 term of eq .
( [ css - gen ] ) .
the qcd resummation of the large logarithms in the css formalism is achieved by solving the evolution equation for the @xmath161 of eq .
( [ css - w - ij ] ) , @xmath162 \tilde{w}_{ij}(\tb , m , x_a , x_b ) \,\,\ , , \label{css - w - evo}\ ] ] and the corresponding renormalization group equations for the kernels @xmath63 and @xmath64 , @xmath163 the anomalous dimension @xmath164 in eqs .
( [ css - k - rg ] ) and ( [ css - g - rg ] ) is perturbatively calculable @xcite .
the renormalization group equations ( [ css - k - rg ] ) and ( [ css - g - rg ] ) for @xmath63 and @xmath64 ensure the correct renormalization scale dependence , @xmath165 the solution given in eq . ( [ css - w - sol ] ) corresponds to simultaneously solving the evolution equation ( [ css - w - evo ] ) from @xmath166 to @xmath54 , and the renormalization group equations ( [ css - k - rg ] ) and ( [ css - g - rg ] ) from @xmath166 to @xmath167 and from @xmath54 to @xmath167 , respectively , where @xmath50 is a constant of order unity . integrating eq .
( [ css - k - rg ] ) from @xmath166 to @xmath167 , and eq .
( [ css - g - rg ] ) from @xmath54 to @xmath167 , one derives @xmath168 where @xmath118 is a function of @xmath169 and @xmath170 while @xmath119 depends on both @xmath171 and @xmath172 .
the functions @xmath118 and @xmath119 do not have large logarithms and have perturbative expansions @xmath173 and @xmath174 , respectively .
the first two coefficients in the perturbative expansions are known@xcite : @xmath175\ , , \nonumber\\ b^{(1 ) } & = & -\frac{3}{2}c_f\ , , \nonumber\\ b^{(2 ) } & = & \left(\frac{c_f}{2}\right)^2 \left[\pi^2 -\frac{3}{4}-12\zeta(3)\right ] + \frac{c_f}{2}\ , n \left[\frac{11}{18}\pi^2-\frac{193}{24 } + 3\zeta(3 ) \right ] \nonumber\\ & + & \frac{c_f}{2}t_r\ , n_f \left[\frac{17}{6}-\frac{2}{9}\pi^2\right]\ , , \label{css - ab}\end{aligned}\ ] ] where @xmath176 for su(3 ) color , the color factor @xmath177 for quarks , @xmath178 , and @xmath179 is the number of active quark flavors .
the coefficients @xmath118 and @xmath119 given in eq .
( [ css - ab ] ) are derived from the general expressions in ref .
@xcite with renormalization constant @xmath180 , where @xmath181 is euler s constant@xcite .
substituting eq .
( [ css - kg ] ) into eq .
( [ css - w - evo ] ) , and integrating from @xmath166 to @xmath54 , one obtains the @xmath161 given in eq .
( [ css - w - sol ] ) with @xmath182\ , .
\label{css - s}\ ] ] in eq .
( [ css - w - sol ] ) , all large logarithms from @xmath166 to @xmath54 in @xmath45 are completely resummed into the exponential factor @xmath55 $ ] , leaving the @xmath51 with only one momentum scale @xmath52 .
the @xmath51 in eq .
( [ css - w - sol ] ) is then perturbatively calculable when the momentum scale @xmath52 is large enough , and is given by @xcite @xmath183 the functions @xmath184 and @xmath185 are the modified parton distributions @xcite , @xmath186 where @xmath187 runs over all parton flavors . in eq .
( [ mod - pdf ] ) , @xmath188 is the normal parton distribution for finding a parton of flavor @xmath189 in hadron @xmath118 , and @xmath190 are perturbatively calculable coefficient functions for finding a parton @xmath46 in the modified parton distribution @xmath191 from a parton @xmath189 in the normal parton distribution @xmath188 . the first two coefficients of the @xmath192 are available@xcite : @xmath193 , \nonumber \\ c^{(1)}_{i / g}(z,\mu = c/\tb ) & = & t_r\ , z\ , ( 1-z ) \,\,\ , , \label{css - coef}\end{aligned}\ ] ] where @xmath46 and @xmath47 represent quark or antiquark flavors and @xmath194 represents a gluon .
the coefficient functions given in eq .
( [ css - coef ] ) are derived from the general functional forms in ref .
@xcite by setting the renormalization constants and the factorization scale as , @xmath195 , @xmath196 , and @xmath197 .
the @xmath42 in eq .
( [ css - w - ij ] ) is the lowest order cross section for a quark - antiquark pair to annihilate into the vector boson ( @xmath198 , @xmath199 or @xmath1 ) . for @xmath200 ,
we have@xcite @xmath201 where @xmath202 is the weak charge , @xmath203 ^ 2 + 1 \over 16\sin^2\theta_w\cos^2\theta_w } \,\,\ , , \ ] ] with @xmath204 denoting the weinberg angle and @xmath205 denoting the charge of the quark . here
, @xmath46 and @xmath47 must represent a quark - antiquark pair of the same flavor . for the weinberg angle we use @xmath206 .
the @xmath42 for @xmath207 and @xmath208 can be found in refs .
@xcite . in the css resummation formalism , the @xmath39 term in eq .
( [ css - gen ] ) represents a small correction to the @xmath6 distribution when @xmath8 .
however , the @xmath39 term dominates the @xmath6-distributions when @xmath142 .
the @xmath39 term has the perturbative expansion @xmath209 , and the coefficients @xmath210 have the factorized form@xcite @xmath211 where @xmath212 runs over all possible parton flavors and @xmath20 represents both the factorization and renormalization scales .
the @xmath213 in eq .
( [ css - w - yn ] ) are perturbatively calculable and have the same normalization as those introduced in ref .
@xcite . since the perturbatively resummed @xmath45 in eq .
( [ css - w - sol ] ) is only reliable for the small @xmath11 region , an extrapolation to the large @xmath11 region is necessary in order to complete the fourier transform in eq .
( [ css - gen ] ) . in the original css formalism , a variable @xmath214 and a nonperturbative function @xmath215 were introduced@xcite in such a way that @xmath38 was modified for all @xmath11 ( except @xmath216 ) as @xmath217 where @xmath218 , with @xmath219 gev@xmath87 .
this construction ensures that @xmath220 for all values of @xmath11 .
the function @xmath221 is a gaussian in @xmath11 , @xmath222 with @xmath223 having a certain @xmath74 , @xmath107 , and @xmath108 dependence@xcite .
as it has been pointed out in this paper and earlier@xcite , the original css extrapolation introduces a modification to the perturbative calculation , and the size of the modifications depends on the non - perturbative parameters in @xmath215 .
this motivated the development of the alternative extrapolation formula@xcite used in the present work . | we predict @xmath0 and @xmath1 transverse momentum distributions from proton - proton and nuclear collisions at rhic and lhc . a resummation formalism with power corrections to the renormalization group equations is used .
the dependence of the resummed qcd results on the non - perturbative input is very weak for the systems considered .
shadowing effects are discussed and found to be unimportant at rhic , but important for lhc .
we study the enhancement of power corrections due to multiple scattering in nuclear collisions and numerically illustrate the weak effects of the dependence on the nuclear mass .
ksucnr-201 - 02 |
let @xmath2 be a bounded and simply - connected polyhedral domain in @xmath3 with boundary @xmath4 and @xmath5 , and let @xmath6 be the outward unit vector normal to the boundary
. denote by @xmath7 the electric field , we consider the following @xmath0 model problem originated from a second order hyperbolic equation by eliminating the magnetic field in maxwell s equations : @xmath8 { \bm{u}}{\!\times\!}{\bm{n}}&=&{\bm{g}}_{_d } , & \ ; \text { on } \ , { \gamma}_d , \\[2 mm ] ( \mu^{-1 } { \nabla { \!\times\!}}{\bm{u } } ) { \!\times\!}{\bm{n } } & = & { \bm{g}}_{_n } , & \ ; \text { on } \ , { \gamma}_n , \end{array } \right.\ ] ] where @xmath9 is the curl operator ; the @xmath10 , @xmath11 , and @xmath12 are given vector fields which are assumed to be well - defined on @xmath2 , @xmath13 , and @xmath14 , respectively ; the @xmath15 is the magnetic permeability ; and the @xmath16 depends on the electrical conductivity , the dielectric constant , and the time step size .
assume that the coefficients @xmath17 and @xmath18 are bounded below @xmath19 for almost all @xmath20 .
the _ a posteriori _ error estimation for the conforming finite element approximation to the @xmath0 problem in has been studied recently by several researchers .
several types of _ a posteriori _ error estimators have been introduced and analyzed .
these include residual - based estimators and the corresponding convergence analysis ( explicit @xcite , and implicit @xcite ) , equilibrated estimators @xcite , and recovery - based estimators @xcite .
there are four types of errors in the explicit residual - based estimator ( see @xcite ) .
two of them are standard , i.e. , the element residual , and the interelement face jump induced by the discrepancy induced by integration by parts associated with the original equation in .
the other two are also the element residual and the interelement face jump , but associated with the divergence of the original equation : @xmath21 , where @xmath22 is the divergence operator .
these two quantities measure how good the approximation is in the kernel space of the curl operator .
recently , the idea of the robust recovery estimator explored in @xcite for the diffusion interface problem has been extended to the @xmath0 interface problem in @xcite .
instead of recovering two quantities in the continuous polynomial spaces like the extension of the popular zienkiewicz - zhu ( zz ) error estimator in @xcite , two quantities related to @xmath23 and @xmath24 are recovered in the respective @xmath0- and @xmath25-conforming finite element spaces .
the resulting estimator consists of four terms similar to the residual estimator in the pioneering work @xcite on this topic by beck , hiptmair , hoppe , and wohlmuth : two of them measure the face jumps of the tangential components and the normal component of the numerical approximations to @xmath23 and @xmath24 , respectively , and the other two are element residuals of the recovery type .
all existing a posteriori error estimators for the @xmath0 problem assume that the right - hand side @xmath10 is in @xmath25 or divergence free .
this assumption does not hold in many applications ( e.g. the implicit marching scheme mentioned in @xcite ) .
moreover , two terms of the estimators are associated with the divergence of the original equation . in the proof , these two terms come to existence up after performing the integration by parts for the irrotational gradient part of the error , which lies in the kernel of the curl operator .
one of the key technical tools , a helmholtz decomposition , used in this proving mechanism , relies on @xmath10 being in @xmath25 , and fails if @xmath26 . in @xcite , the assumption that @xmath27 is weakened to @xmath10 being in the piecewise @xmath25 space with respect to the triangulation , at the same time
, the divergence residual and norm jump are modified to incorporate this relaxation .
another drawback of using helmholtz decomposition on the error is that it introduces the assumption of the coefficients quasi - monotonicity into the proof pipeline .
an interpolant with a coefficient independent stability bound is impossible to construct in a `` checkerboard '' scenario ( see @xcite for diffusion case , and @xcite for @xmath0 case ) . to gain certain robustness for the error estimator in the proof
, one has to assume the coefficients distribution is quasi - monotone .
however , in an earlier work of chen , xu , and zou ( @xcite ) , it is shown that numerically this quasi - monotonicy assumption is more of an artifact introduced by the proof pipeline , at least for the irrotational vector fields . as a result
, we conjecture that the divergence related terms should not be part of an estimator if it is appropriately constructed . in section [ sec : numex ] , some numerical justifications are presented to show the unnecessity of including the divergence related terms . the pioneering work in using the dual problems for a posteriori error estimation dates back to @xcite . in @xcite ,
oden , demkowicz , rachowicz , and westermann studied the a posteriori error estimation through duality for the diffusion - reaction problem .
the finite element approximation to a dual problem is used to estimate the error for the original primal problem ( diffusion - reaction ) .
the result shares the same form to the prague - synge identity ( @xcite ) for diffusion - reaction problem .
the method presented in this paper may be viewed as an extension of the duality method in @xcite to the @xmath0 interface problem . the auxiliary magnetizing field introduced in section [ sec : aux
] is the dual variable resembling the flux variable in @xcite .
the connection is illustrated in details in section [ sec : dual ] .
later , repin ( @xcite ) proposes a functional type a posteriori error estimator of @xmath0 problem , which can be viewed as an extension of the general approach in @xcite .
repin et al ( @xcite ) improve the estimate by assuming that the data @xmath10 is divergence free and the finite element approximation is in @xmath25 . in @xcite ,
the upper bound is established through integration by parts by introducing an auxiliary variable in an integral identity for @xmath0 .
an auxiliary variable is recovered by globally solving an @xmath0 finite element approximation problem and is used in the error estimator . for the global lower bound
, the error equation is solved globally in an @xmath0 conforming finite element space .
then the solution is inserted into the functional as the error estimator of which the maximizer corresponds to the solution to the error equation .
the purpose of this paper is to develop a novel a posteriori error estimator for the conforming finite element approximation to the @xmath0 problem in ( [ eq : pb - ef - weak ] ) that overcomes the above drawbacks of the existing estimators , e.g. the helmholtz decomposition proof mechanism , restricted by the assumption that @xmath28 or divergence free , which brings in the divergence related terms .
specifically , the estimator studied in this paper is of the recovery type , requires the right - hand side merely having a regularity of @xmath1 , and has only two terms that measure the element residual and the tangential face jump of the original equation .
based on the current approximation to the primary variable @xmath7 ( the electric field ) , an auxiliary variable @xmath29 ( the magnetizing field ) is recovered by approximating a similar auxiliary @xmath0 problem . to this end
, a multigrid smoother is used to approximate this auxiliary problem , which is independent of the primary equation and is performed in parallel with the primary problem .
the cost is the same order of complexity with computing the residual - based estimator , which is much less than solving the original @xmath0 problem .
an alternate route is illustrated as well in section [ sec : loc ] by approximating a localized auxiliary problem . while embracing the locality , the parallel nature using the multigrid smoother
the recovery through approximating localized problem requires the user to provide element residual and tangential face jump of the numerical magnetizing field based on the finite element solution of the primary equation .
the estimator is then defined as the sum of the modified element residual and the residual of the auxiliary constitutive equation .
it is proved that the estimator is equal to the true error in the energy norm globally . moreover , in contrast to the mechanism of the proof using helmholtz decomposition mentioned previously , the decomposition is avoided by using the joint energy norm . as a result , the new estimator s reliability does not rely on the coefficients distribution ( theorem [ th : rel ] ) .
meanwhile , in this paper , the method and analysis extend the functional - type error estimator in @xcite to a more pragmatic context by including the mixed boundary conditions , and furthermore , the auxiliary variable @xmath29 is approximated by a fast multigrid smoother , or by solving a localized @xmath0 problem on vertex patches , to avoid solving a global finite element approximation problem .
lastly , in order to compare the new estimator introduced in this paper with existing estimators , we present numerical results for @xmath0 intersecting interface problems .
when @xmath26 , the mesh generated by our indicator is much more efficient than those by existing indicators ( section [ sec : numex ] ) .
denote by @xmath30 the space of the square integrable vector fields in @xmath3 equipped with the standard @xmath1 norm : @xmath31 , where @xmath32 denotes the standard @xmath1 inner product over an open subset @xmath33 , when @xmath34 , the subscript is dropped for @xmath35 and @xmath36 .
let @xmath37 which is a hilbert space equipped with the norm @xmath38 denote its subspaces by @xmath39 \mbox{and } \quad { { { \mathop{{}{\bm{h}}}\limits^{\vbox to -.1ex{\kern -0.5ex\hbox{$\scriptstyle \circ$}\vss}}}}_{b}({\mathbf{curl}\hspace{0.7pt}};\omega ) } : = \{{\bm{v}}\in { { \bm{h}}_{b}({\mathbf{curl}\hspace{0.7pt}};\omega ) } : { \bm{g}}_{_b}={{\bf 0}}\ } \end{array}\ ] ] for @xmath40 or @xmath41 .
for any @xmath42 , multiplying the first equation in by a suitable test function @xmath43 with vanishing tangential part on @xmath13 , integrating over the domain @xmath44 , and using integration by parts formula for @xmath0-regular vector fields ( e.g. see @xcite ) , we have @xmath45 & = & ( \mu^{-1}{\nabla { \!\times\!}}{\bm{u}},\,{\nabla { \!\times\!}}{\bm{v}})+ ( { \beta}\,{\bm{u}},\,{\bm{v } } ) - \int_{\gamma_{n}}{\bm{g}}_{_n } \cdot { \bm{v}}\,ds.\end{aligned}\ ] ] then the weak form associated to problem is to find @xmath46 such that @xmath47 where the bilinear and linear forms are given by @xmath48 respectively . here , @xmath49 denotes the duality pair over @xmath14 .
denote by @xmath50 the `` energy '' norm induced by the bilinear form @xmath51 .
[ th : pb - ef - weak ] assume that @xmath52 , @xmath53 , and @xmath54 .
then the weak formulation of has a unique solution @xmath55 satisfying the following a priori estimate @xmath56 for the notations and proof , see the appendix [ appendix ] . for simplicity of the presentation , only the tetrahedral elements are considered .
let @xmath57 be a finite element partition of the domain @xmath44 .
denote by @xmath58 the diameter of the element @xmath59 .
assume that the triangulation @xmath60 is regular and quasi - uniform .
let @xmath61 where @xmath62 is the space of polynomials of degree less than or equal to @xmath63 .
let @xmath64 and @xmath65 be the spaces of homogeneous polynomials of scalar functions and vector fields
. denote by the first or second kind ndlec elements ( e.g. see @xcite ) @xmath66 for @xmath67 , respectively , where the local ndlec elements are given by @xmath68 \mbox{and}\quad & { \bm{{{\mathcal{n}}}\!{{\mathcal{d}}}}}^{k,2}(k ) = \{{\bm{p}}+ \nabla s : { \bm{p}}\in { \bm{{{\mathcal{n}}}\!{{\mathcal{d}}}}}^{k,1}(k ) , s\in { \widetilde}{p}_{k+2}(k ) \}.
\end{aligned}\ ] ] for simplicity of the presentation , we assume that both boundary data @xmath11 and @xmath12 are piecewise polynomials , and the polynomial extension ( see @xcite ) of the dirichlet boundary data as the tangential trace is in @xmath69 .
now , the conforming finite element approximation to ( [ eq : pb - ef ] ) is to find @xmath70 such that @xmath71 assume that @xmath7 and @xmath72 are the solutions of the problems in ( [ eq : pb - ef ] ) and ( [ eq : pb - ef - fem ] ) , respectively , and that @xmath73 , @xmath74 ( when the regularity assumption is not met , one can construct a curl - preserving mollification , see @xcite ) , by the interpolation result from @xcite chapter 5 and ca s lemma , one has the following a priori error estimation : @xmath75 where @xmath76 is a positive constant independent of the mesh size @xmath77 .
introducing the magnetizing field @xmath78 then the first equation in ( [ eq : pb - ef ] ) becomes @xmath79 the boundary condition on @xmath80 may be rewritten as follows @xmath81 for any @xmath82 , multiplying equation ( [ eq : pb - mf - aux ] ) by @xmath83 , integrating over the domain @xmath2 , and using integration by parts and ( [ eq : pb - mf ] ) , we have @xmath84 & = & ( { \beta}^{-1}{\nabla { \!\times\!}}{\bm{\sigma}},\,{\nabla { \!\times\!}}{\bm{\tau}})+ ( { \nabla { \!\times\!}}{\bm{u}},\,{\bm{\tau } } ) \\ & & \ ; + \int_{\gamma_d } ( { \bm{u}}{\!\times\!}{\bm{n } } ) \cdot { \bm{\tau}}\,ds -\int_{\gamma_n } { \bm{u}}\cdot ( { \bm{\tau}}{\!\times\!}{\bm{n}})\,ds \\[1 mm ] & = & ( { \beta}^{-1}{\nabla { \!\times\!}}{\bm{\sigma}},\,{\nabla { \!\times\!}}{\bm{\tau}})+ ( \mu\,{\bm{\sigma}},\,{\bm{\tau } } ) + \int_{\gamma_{d}}{\bm{g}}_{_d } \cdot { \bm{\tau}}\,ds.\end{aligned}\ ] ] hence , the variational formulation for the magnetizing field is to find @xmath85 such that @xmath86 where the bilinear and linear forms are given by @xmath87 respectively .
the natural boundary condition for the primary problem becomes the essential boundary condition for the auxiliary problem , while the essential boundary condition for the primary problem is now incorporated into the right - hand side and becomes the natural boundary condition .
denote the `` energy '' norm induced by @xmath88 by @xmath89 [ th : pb - mf - weak ] assume that @xmath52 , @xmath90 , and @xmath91 .
then problem _ _ has a unique solution @xmath92 satisfying the following a priori estimate @xmath93 the theorem may be proved in a similar fashion as theorem [ th : pb - ef - weak ] .
similarly to that for the essential boundary condition , it is assumed that the polynomial extension of the neumman boundary data as the tangential trace is in @xmath69 as well .
now , the conforming finite element approximation to ( [ eq : pb - mf - weak ] ) is to find @xmath94 such that @xmath95 assume that @xmath29 and @xmath96 are the solutions of the problems in ( [ eq : pb - mf ] ) and ( [ eq : pb - mf - fem ] ) , respectively , and that @xmath97 , @xmath98 , one has the following a priori error estimation similar to @xmath99 the a priori estimate shows that heuristically , for the auxiliary magnetizing field @xmath29 , using the same order @xmath0-conforming finite element approximation spaces with the primary variable @xmath7 may be served as the building blocks for the a posteriori error estimation .
the localization of the recovery of @xmath96 for this new @xmath0 recovery shares similar methodology with the one used in the equilibrated flux recovery ( see @xcite ) .
however , due to the presence of the @xmath1-term , exact equilibration is impossible due to several discrepancies : @xmath100 if @xmath96 and @xmath72 are in ndlec spaces of the same order ; if @xmath101 is used for @xmath96 and @xmath69 for @xmath72 , the inter - element continuity conditions come into the context in that @xmath102 , which has different inter - element continuity requirement than @xmath69 . due to these two concerns ,
the local problem is approximated using a constraint @xmath0-minimization .
let @xmath103 be the correction from @xmath104 to the true magnetizing field : @xmath105 .
now @xmath103 can be decomposed using a partition of unity : let @xmath106 be the linear lagrange nodal basis function associated with a vertex @xmath107 , which is the collection of all the vertices , @xmath108 denote @xmath109 .
let the vertex patch @xmath110 , where @xmath111 is the collection of vertices of element @xmath59 .
then the following local problem is what the localized magnetizing field correction satisfies : @xmath112 with the following jump condition on each interior face @xmath113 , and boundary face @xmath114 : @xmath115 the element residual is @xmath116 , and the tangential jump is @xmath117 . to find the correction , following piecewise polynomial spaces
are defined : @xmath118 & \bm{{\mathcal{w}}}^k({{\mathcal{f}}}_{{\bm{z}}})= \ { { \bm{\tau}}\in { \bm{l}}^2({{\mathcal{f}}}_{{\bm{z}}}):\ , { \bm{\tau}}{\big\vert_{\raisebox{-0.5pt}{\scriptsize$f$ } } } \in { \bm{{\mathcal{r}}\!{{\mathcal{t}}}}}^k(f),\ ; \forall f\in{{\mathcal{f}}}_{{\bm{z } } } ; \\ & \hspace{2em } { \bm{\tau}}{\big\vert_{\raisebox{-0.5pt}{\scriptsize$f_i$ } } } \cdot ( { \bm{t}}_{ij}{\!\times\!}{\bm{n}}_{i } ) = { \bm{\tau}}{\big\vert_{\raisebox{-0.5pt}{\scriptsize$f_j$ } } } \cdot ( { \bm{t}}_{ij}{\!\times\!}{\bm{n}}_{j } ) , \forall f_i , f_j\in { { \mathcal{f}}}_{{\bm{z } } } , { \partial}f_i \cap { \partial}f_j = e_{ij } \ } , \\[3pt ] & \bm{{\mathcal{h}}}_{{\bm{z } } } = \{{\bm{\tau}}\in { \bm{{{\mathcal{n}}}\!{{\mathcal{d}}}}}^{k}_{-1}({\omega_{\bm{z } } } ) : \ ; { { \lbrack\hspace{-1.5pt}\lbrack}{{\bm{\tau}}{\!\times\!}{\bm{n}}_f } { \rbrack\hspace{-1.5pt}\rbrack}_{\raisebox{-2pt}{\scriptsize$f$ } } } = - { \overline}{{\bm{j}}}_{f,{\bm{z } } } \ ; \forall\ , f \in { { \mathcal{f}}}_{{\bm{z}}}\ } , \\[3pt ] \text{and } \ ; & \bm{{\mathcal{h}}}_{{{\bf 0 } } , { \bm{z } } } = \{{\bm{\tau}}\in \bm{{\mathcal{h}}}_{{\bm{z } } } : \ , { \bm{\tau}}{\!\times\!}{\bm{n}}_f{\big\vert_{\raisebox{-0.5pt}{\scriptsize$f$ } } } = { { \bf 0 } } , \ ; \forall\ ; f\ , \subset { \omega_{\bm{z}}}\}. \end{aligned}\ ] ] here @xmath119 is the planar raviart - thomas space on a given face @xmath120 , of which the degrees of freedom can be defined using conormal of an edge with respect to the face normal @xmath121 .
for example , @xmath122 is the unit tangential vector of edge @xmath123 joining face @xmath124 and @xmath125 , then the conormal vector of @xmath123 with respect to face @xmath124 is @xmath126 .
@xmath127 can be viewed as the trace space of the broken ndlec space @xmath128 . for detail
please refer to section 4 and 5 in @xcite . to approximate the local correction for magnetizing field , @xmath129 and @xmath130
are projected onto proper piecewise polynomial spaces . to this end , let @xmath131 where @xmath132 is the @xmath1 projection onto the space @xmath133 , and @xmath134 is the @xmath1 projection onto the space @xmath119 . dropping the uncomputable terms in , and using as a constraint , the following local @xmath0-minimization problem is to be approximated : @xmath135 the hybridized problem associated with above minimization is obtained by taking variation with respect to @xmath136 of the functional by the tangential face jump as a lagrange multiplier : @xmath137 for any @xmath138 , using the fact that @xmath139 , and @xmath140 on @xmath141 @xmath142 as a result , the local approximation problem is : @xmath143 & a_{{\beta},\mu;{{\bm{z}}}}\big({{\bm{\sigma}}^{{\delta}}_{{\bm{z}},{\raisebox{-0.5pt}{$\scriptscriptstyle{{{\mathcal{t}}}}$}}}},{{\bm{\tau}}}\big ) + b_{{\bm{z } } } \big({\bm{\tau } } , { \bm{\theta}}_{{\bm{z}}}\big ) = { \bigl ( { { \beta}^{-1 } { \overline}{{\bm{r}}}_{k,{\bm{z } } } } , \,{{\nabla { \!\times\!}}{\bm{\tau } } } \bigr)}_{{\omega_{\bm{z } } } } , \quad \forall\ , { \bm{\tau}}\in { \bm{{{\mathcal{n}}}\!{{\mathcal{d}}}}}^k_{-1}({\omega_{\bm{z } } } ) , \\[0.5em ] & \qquad \qquad \qquad\quad b_{{\bm{z } } } \big({{\bm{\sigma}}^{{\delta}}_{{\bm{z}},{\raisebox{-0.5pt}{$\scriptscriptstyle{{{\mathcal{t}}}}$ } } } } , { \bm{\gamma}}\big ) = -\sum_{f\in{{\mathcal{f}}}_{{\bm{z } } } } { \bigl ( { { \overline}{{\bm{j}}}_{f,{\bm{z}}}},\,{{\bm{\gamma } } } \bigr)}_f , \quad \forall\ , { \bm{\gamma}}\in \bm{{\mathcal{w}}}^k({{\mathcal{f}}}_{{\bm{z } } } ) , \end{aligned } \right.\ ] ] wherein the local bilinear forms are defined as follows : @xmath144 \text { and } \ ; b_{{\bm{z } } } ( { \bm{\tau}},{\bm{\gamma } } ) & : = \sum_{f\in{{\mathcal{f}}}_{{\bm{z } } } } { \bigl ( { { { \lbrack\hspace{-1.5pt}\lbrack}{{\bm{\tau}}{\!\times\!}{\bm{n}}_f } { \rbrack\hspace{-1.5pt}\rbrack}_{\raisebox{-2pt}{\scriptsize$f$ } } } } , \,{{\bm{\gamma } } } \bigr)}_f .
\end{aligned}\ ] ] problem has a unique solution . for a finite dimensional problem ,
uniqueness implies existence .
it suffices to show that letting both the right hand sides be zeros results trivial solution .
first by @xmath145 for any @xmath138 ( direct implication of proposition 4.3 and theorem 4.4 in @xcite ) , setting @xmath146 in the second equation of immediately implies that @xmath147 . as a result , @xmath148 .
now let @xmath149 in the first equation of , since @xmath150 induces a norm in @xmath151 , @xmath152 .
for @xmath153 , it suffices to show that @xmath154 on each @xmath120 if @xmath155 using theorem 4.4 in @xcite , if @xmath156 is non - trivial and satisfies above equation , there always exists a @xmath157 such that @xmath158 . as a result , @xmath159 , which is a contradiction .
thus , the local problem is uniquely solvable . with the local correction to the magnetizing field , @xmath136 for all @xmath160 , computed above ,
let @xmath161 then the recovered magnetizing field is @xmath162
in this section , we study the following a posteriori error estimator : @xmath163 where the local indicator @xmath164 is defined by @xmath165 it is easy to see that @xmath166 the @xmath72 and @xmath96 are the finite element approximations in problems and respectively . with the locally recovered @xmath167 , the local error indicator @xmath168 and the global error estimator @xmath169 are defined in the same way as and : @xmath170 and @xmath171 in practice , @xmath96 does not have to be the finite element solution of a global problem . in the numerical computation ,
the hiptmair - xu multigrid preconditioner in @xcite is used for discrete problem with two @xmath172 multigrid v - cycles for each component of the vector laplacian , and two @xmath173 multigrid v - cycles for the kernel part of the curl operator .
the @xmath96 used to evaluate the estimator is the pcg iterate .
the computational cost is the same order with computing the explicit residual based estimator in @xcite .
generally speaking , to approximate the auxiliary problem , the same black - box solver for the original @xmath0 problem can be applied requiring minimum modifications .
for example , if the boomeramg in _ hypre _
( @xcite ) is used for the discretizations of the primary problem , then the user has to provide exactly the same discrete gradient matrix and vertex coordinates of the mesh , and in constructing the the hx preconditioner , the assembling routines for the vector laplacian and scalar laplacian matrices can be called twice with only the coefficients input switched .
[ th : rel ] locally , the indicator @xmath174 and @xmath175 both have the following efficiency bound @xmath176 for all @xmath177 .
the estimator @xmath178 and @xmath169 satisfy the following global upper bound @xmath179 denote the true errors in the electric and magnetizing fields by @xmath180 respectively .
it follows from ( [ eq : pb - mf ] ) , ( [ eq : pb - mf - aux ] ) , and the triangle inequality that @xmath181\nonumber & \leq & \left({\left\|{\mu^{1/2 } { \bm{e}}}\right\|}_{k } ^2 + { \left\|{\mu^{-1/2}{\nabla { \!\times\!}}{\bm{e}}}\right\|}_{k } ^2 + { \left\|{{\beta}^{-1/2}{\nabla { \!\times\!}}{\bm{e}}}\right\|}^2_k + { \left\|{{\beta}^{1/2}{\bm{e}}}\right\|}_{k}^2\right ) \\[2mm]\nonumber & = & \left({\@ifstar{\mathopen{|\mkern-2.5mu|\mkern-2.5mu| } s \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { \mathopen{|\mkern-2.5mu|\mkern-2.5mu| } } \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { { \bm{e}}}_{\mu,{\beta},k}^2 + { \@ifstar{\mathopen{|\mkern-2.5mu|\mkern-2.5mu| } s \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { \mathopen{|\mkern-2.5mu|\mkern-2.5mu| } } \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { { \bm{e}}}_{\beta,\mu , k}^2\right),\end{aligned}\ ] ] which implies the validity of for @xmath174 . for @xmath175 , the exact same argument follows except by switching @xmath182 by locally recovered @xmath183 . to prove the global identity in , summing over all @xmath177 gives @xmath184 & = & { \@ifstar{\mathopen{|\mkern-2.5mu|\mkern-2.5mu| } s \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { \mathopen{|\mkern-2.5mu|\mkern-2.5mu| } } \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { { \bm{e}}}_{\mu,{\beta}}^2 + { \@ifstar{\mathopen{|\mkern-2.5mu|\mkern-2.5mu| } s \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { \mathopen{|\mkern-2.5mu|\mkern-2.5mu| } } \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { { \bm{e}}}_{\beta,\mu}^2 -2({\bm{e}},\,{\nabla { \!\times\!}}{\bm{e}})+ 2({\nabla { \!\times\!}}{\bm{e}},\,{\bm{e}}).\end{aligned}\ ] ] now , follows from the fact that @xmath185 lastly , the global upper bound for the locally recovered @xmath169 follows from the fact that @xmath72 and @xmath96 are the solutions to the following global problem : @xmath186 as a result , @xmath187 which is the global minimum achieved in the finite element spaces .
this completes the proof of the theorem .
in theorem [ th : rel ] it is assumed that the boundary data are admissible so that they can be represented as tangential traces of the finite element space @xmath69 .
if this assumption is not met , it can be still assumed that divergence - free extension @xmath188 of its tangential trace @xmath11 to each boundary tetrahedron @xmath59 on @xmath13 is at least @xmath189-regular ( @xmath190 ) , and @xmath191 as well ( @xmath192 ) , so that the conventional edge interpolant is well - defined ( e.g. see @xcite chapter 5 ) . when the same assumption is applied to @xmath29 and @xmath193 ,
the reliability bound derived by still holds ( for notations please refer to appendix [ appendix ] ) : @xmath194 using the fact that @xmath7 and @xmath29 are approximated by the conventional edge interpolants @xmath195 on @xmath13 and @xmath196 on @xmath14 respectively yields : @xmath197 by the interpolation estimates for boundary elements together with the weighted trace inequalities from appendix [ appendix ] , the reliability constant is not harmed if the interface singularity does not touch the boundary .
a posteriori error estimation by the duality method for the diffusion - reaction problem was studied by oden , demkowicz , rachowicz , and westermann in @xcite . in this section ,
we describe the duality method for the @xmath0 problem and its relation with the estimator @xmath178 defined in ( [ eq : eta ] ) . to this end , define the energy and complimentary functionals by @xmath198 \mbox{and}\quad { { \mathcal{j}}}^*({\bm{\tau } } ) & = & -\frac{1}{2}(\mu\,{\bm{\tau}},\,{\bm{\tau } } ) - \frac{1}{2 } { \bigl ( { { \beta}^{-1}({\bm{f}}- { \nabla { \!\times\!}}{\bm{\tau}})},\,{{\bm{f}}-{\nabla { \!\times\!}}{\bm{\tau } } } \bigr ) } -{{\left\langle { { \bm{g}}_{_d}},{{\bm{\tau } } } \right\rangle}}_{{\gamma}_d},\end{aligned}\ ] ] respectively
. then problems ( [ eq : pb - ef - weak ] ) and ( [ eq : pb - mf - weak ] ) are equivalent to the following minimization and maximization problems : @xmath199 respectively . by the duality theory for a lower semi - continuous convex functional ( see e.g. @xcite ) , we have @xmath200 a simple calculation gives that the true errors of the finite element approximations in the `` energy '' norm can be represented by the difference between the functional values as follows : @xmath201 hence , the `` energy '' error in the finite element approximation is bounded above by the estimator @xmath178 defined in ( [ eq : eta ] ) ( and the locally - recovered @xmath169 as well ) : @xmath202 & \leq & 2\big ( { { \mathcal{j}}}({\bm{u}}_{_{{\mathcal{t } } } } ) - { { \mathcal{j}}}^*({\bm{\sigma}}_{_{{\mathcal{t } } } } ) \big)= \eta^2,\end{aligned}\ ] ] where the last equality is obtained by evaluating @xmath203 through integration by parts .
note that the above calculation indicates @xmath204 which leads us back to the identity on the global reliability in .
in this section , we present numerical results for interface problems , i.e. , the problem parameters @xmath15 and @xmath16 in ( [ eq : pb - ef ] ) are piecewise constants with respect to a partition of the domain @xmath205 .
assume that interfaces @xmath206 do not cut through any element @xmath177 .
the @xmath72 is solved in @xmath207 , and the @xmath96 is recovered in @xmath207 as well .
the numerical experiments are prepared using ` delaunaytriangulation ` in matlab for generating meshes , l. chen s ifem ( @xcite ) for the adaptively refining procedure , the ` matlab2hypre ` interface in blopex ( @xcite ) for converting sparse matrices , and mfem ( @xcite ) to set up the serial version of auxiliary - space maxwell solver ( ams ) in _ hypre _ ( @xcite ) as preconditioners .
we compare numerical results generated by adaptive finite element method using following error estimators : * the new indicator @xmath208 defined in , and its locally - recovered sibling @xmath209 defined in . * the residual - based indicator @xmath210 introduced in @xcite with the appropriate weights for piecewise constant coefficients defined in @xcite : @xmath211 & + \sum_{f\in { { \mathcal{f}}}_h(k ) } \frac{h_f}{2 } \left ( { \beta}_f^{-1 } { \left\|{{{\lbrack\hspace{-1.5pt}\lbrack}{{\beta}{\bm{u}}_{_{{\mathcal{t}}}}\cdot { \bm{n}}_f } { \rbrack\hspace{-1.5pt}\rbrack}_{\raisebox{-2pt}{\scriptsize$f$ } } } } \right\|}_{l^2(f)}^2 + \mu_f{\left\|{{{\lbrack\hspace{-1.5pt}\lbrack}{(\mu^{-1 } { \nabla { \!\times\!}}{\bm{u}}_{_{{\mathcal{t}}}}){\!\times\!}{\bm{n } } } { \rbrack\hspace{-1.5pt}\rbrack}_{\raisebox{-2pt}{\scriptsize$f$ } } } } \right\|}_{{\bm{l}}^2(f)}^2 \right ) , \end{aligned}\ ] ] * the recovery - based indicator @xmath212 presented in @xcite : @xmath213 & \qquad + { \left\|{\mu^{1/2}{\bm{\sigma}}_{_{{\mathcal{t}}}}-\mu^{-1/2 } { \nabla { \!\times\!}}{\bm{u}}_{_{{\mathcal{t } } } } } \right\|}_{{\bm{l}}^2(k)}^2 , \end{aligned}\ ] ] where @xmath214 and @xmath215 are the @xmath216 recoveries of @xmath217 and @xmath218 , respectively . in our computation , the energy norms @xmath219 are used for the estimators @xmath220 and @xmath221 and the estimator @xmath222 , respectively . the respective relative errors and effectivity indices are computed at each iteration by @xmath223 for the estimators @xmath220 and @xmath221 and by @xmath224for the estimator @xmath222 and @xmath225 , where @xmath226 .
in all the experiements , the lowest order ndlec element space is used , and , hence , the optimal rate of convergence for the adaptive algorithm is @xmath227 .
* example 1 * : this is an example similar to that in @xcite with a few additions and tweaks , in which the kellogg intersecting interface problem is adapted to the @xmath0-problem . the computational domain is a slice along @xmath228-direction : @xmath229 with @xmath230 .
let @xmath231 be a piecewise constant given by @xmath232 the exact solution @xmath7 of ( [ eq : pb - ef ] ) is given in cylindrical coordinates @xmath233 : @xmath234 where @xmath235 is a continuous function defined by @xmath236 here we set parameters to be @xmath237 the initial mesh is depicted in figure [ fig : ex1-init ] which is aligned with four interfaces . 0.5 cm it is easy to see that the exact solution of the auxiliary problem in ( [ eq : pb - mf ] ) for this example is @xmath238 .
hence , the true error for the finite element approximation to ( [ eq : pb - mf ] ) is simply the energy norm of the finite element solution @xmath96 defined in ( [ eq : pb - mf - weak ] ) @xmath239 in the first experiment , we choose the coefficients @xmath240 and @xmath241 .
this choice enables that @xmath242 , i.e. , @xmath243 , and that @xmath7 satisfies the @xmath244-weighted normal continuity : @xmath245 for any surface @xmath120 in the domain @xmath44 .
this is the prerequisite for establishing efficiency and reliability bounds in @xcite and @xcite and the base for recovering @xmath246 in @xmath247 in @xcite .
the quasi - monotonicity assumption is not met in this situation ( for the analysis of the quasi - monotonicity affects the robustness of the estimator for @xmath0 problems , please refer to @xcite ) .
the meshes generated by @xmath208 , @xmath210 , and @xmath212 are almost the same ( see figure [ fig : ex1-hdiv - mesh ] ) . in terms of the convergence , we observe that the error estimator @xmath248 exhibits asymptotical exactness .
this is impossible for the error estimators in @xcite and @xcite because of the presence of the element residuals .
table 1 shows that the number of the dof for the @xmath222 is about @xmath249 less than those of the other two estimators while achieving a better accuracy . as the reliability of the estimator does not depend on the quasi - monotonicity of the coefficient , the rate of the convergence is not hampered by checkerboard pattern of the @xmath244 . .estimators comparison , example 1 , @xmath242 [ cols="^,^,^,^,^,^,^",options="header " , ] [ table : ex3 - 1 ] in the second experiment , the @xmath244 is chosen to be : @xmath250 we test the case where @xmath251 .
similar to example 1 , the necessary tangential jump conditions across the interfaces for the primary problem are satisfied . yet
the choice of @xmath244 implies that the right hand side @xmath252 .
using the residual - based or recovery - based estimator will again lead to unnecessary over - refinement along the interfaces ( see figure [ fig : ex3-mesh ] ) , and the order of convergence is sub - optimal than the optimal order for linear elements ( see table [ table : ex3 - 1 ] and figure [ fig : ex3-conv ] ) . 0.5 cm 0.5 cm the new estimator in this paper shows convergence in the optimal order no matter how we set up the jump of the coefficients .
the conclusion of comparison with the other two estimators remains almost the same with example 1 . in this example , the differences are more drastic : @xmath253 the degrees of freedom for the new estimator to get roughly the same level of approximation with the other two .
in this appendix , an a priori estimate for the mixed boundary value problem with weights is studied following the arguments and notations mainly from @xcite . in our study , it is found that , due to the duality pairing on the neumann boundary and the nature of the trace space of @xmath0 , a higher regularity is needed for the neumann boundary data @xmath12 than those for elliptic mixed boundary value problem . first we define the tangential trace operator and tangential component projection operator , and their range acting on the @xmath254 .
secondly we construct a weighted extension of the dirichlet boundary data to the interior of the domain .
lastly the a priori estimate for the solution of problem is established after a trace inequality is set up for the piecewise smooth vector field .
@xmath257 \bm{0 } & \mbox{on } \ , { \partial}{\omega}\backslash { \overline}{{\gamma}}_b , \end{array}\right .
\mbox{and}\,\,\ , \pi_{\top , b } : { \bm{v}}\mapsto \left\{\begin{array}{ll } { \bm{n}}{\!\times\!}({\bm{v}}{\!\times\!}{\bm{n } } ) & \mbox{on } \,{\gamma}_b , \\[2 mm ] \bm{0 } & \mbox{on } \,{\partial}{\omega}\backslash { \overline}{{\gamma}}_b , \end{array}\right . \ ] ] define the following spaces as the trace spaces of @xmath260 : @xmath261 for the @xmath262-regular vector fields , define the trace spaces @xmath263 and @xmath264 as : @xmath265 it is proved in @xcite that the tangential trace space and the tangential component space can be characterized by @xmath266 the @xmath267 supscripted spaces @xmath268 and @xmath269 are defined as the dual spaces of @xmath263 and @xmath264 . [
assumption : bd ] let the dirichlet or neumann boundary @xmath258 ( @xmath274 or @xmath41 ) be decomposed into simply - connected components : @xmath275 . for any @xmath276 , there exists a single @xmath277 , such that @xmath278 .
assumption [ assumption : bd ] is to say , each connected component on the dirichlet or neumann boundary only serves as the boundary of exactly one subdomain .
assumption [ assumption : bd ] is here solely for the a priori error estimate .
the robustness of the estimator in section 5 does not rely on this assumption if the boundary data are piecewise polynomials .
due to assumption [ assumption : bd ] , the tangential trace and tangential component of a @xmath279 vector field is the same space as those of a @xmath280 vector field on @xmath13 or @xmath14 respectively . with slightly abuse of notation , define @xmath281 now we define the weighted @xmath282-norm for the value of any @xmath283 on boundary as : @xmath284 now thanks to the embedding results from @xcite , @xmath285 is equivalent to the unweighted @xmath286 which can be defined as : @xmath287 the fact that @xmath295 implies the following problem is well - posed : @xmath296 & { \mathfrak{a}}_{{\beta},\mu}({\bm{w}},{\bm{v } } ) = { { \left\langle { { \bm{g}}_d},{{\bm{v } } } \right\rangle } } , \quad \forall\ , { \bm{v}}\in { { p\!{\bm{h}}}^{1}({\omega},{\mathscr{p}})}\cap { \bm{x}}_0({\omega},{\alpha},{\gamma}_n ) , \end{aligned } \right.\ ] ] where the bilinear form @xmath297 is given as @xmath298 on this weighted divergence free subspace @xmath299 : @xmath300 with slightly abuse of notation , the zero extension of @xmath11 to the neumann boundary is denoted as @xmath11 itself .
now for the trial function space and the test function space in problem are the same , letting @xmath301 leads to @xmath302 together with their tangential traces vanish on the neumann boundary , this implies @xmath303 the extension is now letting @xmath304 . to prove the estimate
, we first notice that the problem is a consistent variational formulation for the following pde : @xmath305 therefore , the energy norm of @xmath306 is @xmath307 for the second equality in the lemma , it is straightforward to verify that for any @xmath42 , with @xmath308 is from the above construction , the following identity holds @xmath309 the last equality follows from the fact that @xmath310 on @xmath311 and @xmath312 on @xmath259 . in this section we want to establish a trace inequality for the tangential component space of @xmath313 . for any @xmath42 , consider the tangential component space @xmath314 defined in that contains all the tangential components of @xmath42 on the neumann boundary and zero on the dirichlet boundary . first we notice that @xmath317 which is the dual space of @xmath318 . for any @xmath319
, there exists @xmath320 such that @xmath321 and @xmath322 by the integration by parts formula from @xcite and cauchy - schwarz inequality , we have @xmath323 hence by definition the lemma follows .
now , to show the validity of the theorem , it suffices to prove that problem has a unique solution @xmath327 satisfying the following a priori estimate @xmath330 to this end , for any @xmath42 , we have from trace lemma [ lemma : trace ] @xmath331 which , together with the cauchy - schwarz inequality , implies @xmath332 & \leq & \left({\left\|{{\beta}^{-1/2}{\bm{f}}}\right\|}+ { \left\|{{\bm{g}}_{_n}}\right\|}_{1/2,{\beta},\mu,{\gamma}_n}\right ) { \@ifstar{\mathopen{|\mkern-2.5mu|\mkern-2.5mu| } s \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { \mathopen{|\mkern-2.5mu|\mkern-2.5mu| } } \mathclose{|\mkern-2.5mu|\mkern-2.5mu| } } { { \bm{v}}}_{\mu,\beta } .\end{aligned}\ ] ] by the lax - milgram lemma , has a unique solution @xmath327 . taking @xmath333 in
, we have @xmath334 dividing @xmath335 on the both sides of the above inequality yields .
this completes the proof of the theorem . | in this paper , we introduce a novel a posteriori error estimator for the conforming finite element approximation to the @xmath0 problem with inhomogeneous media and with the right - hand side only in @xmath1 .
the estimator is of the recovery type .
independent with the current approximation to the primary variable ( the electric field ) , an auxiliary variable ( the magnetizing field ) is recovered in parallel by solving a similar @xmath0 problem .
an alternate way of recovery is presented as well by localizing the error flux .
the estimator is then defined as the sum of the modified element residual and the residual of the constitutive equation defining the auxiliary variable .
it is proved that the estimator is approximately equal to the true error in the energy norm without the quasi - monotonicity assumption . finally
, we present numerical results for several @xmath0 interface problems .
[ section ] [ theorem]assumption [ theorem]remark [ theorem]definition [ theorem]lemma [ theorem]proposition |
since the 1990 s , many authors have successfully applied the so - called monotone lhpital s rules to establish many new and useful inequalities . in this article
we make use of the connection of these rules to some well - known properties of concave / convex functions ( via a change of variable ) to extend the usability of the rules . we show how
the rules can be formulated under no differentiability requirements , and how to characterize all possible situations , which are normally not covered by the conventional form of the rules .
these include situations in which the functions involved have non - zero initial values and when the derived functions are not necessarily monotone .
this perspective is not new ; it can be dated back to hardy , littlewood and plya ( abbreviated as hlp ) .
the concepts of non - decreasing , increasing ( we use this term in the sense of what some authors prefer to call `` strictly increasing '' ) , non - increasing , and decreasing functions are defined as usual .
the term `` monotone '' can refer to any one of these senses . for convenience
, we use the symbols @xmath0 and @xmath1 to denote increasing and decreasing , respectively .
the rules have appeared in various formulations .
let us start with the most popularly known form .
let @xmath2 .
a dual form assumes @xmath3 , @xmath4 and @xmath5 .
the same conclusion holds .
note that even though the rule allows @xmath6 , and that @xmath7 or @xmath8 need not be defined at @xmath9 , as far as the proof is concerned we may assume without loss of generality that @xmath9 is finite , and @xmath7 and @xmath8 are defined and continuous up to @xmath9 , because we can first study the functions in a smaller subinterval @xmath10\subset(0,b ) $ ] and then let @xmath11 and @xmath12 .
when @xmath7 and @xmath8 are assumed to be differentiable , we take @xmath13 and @xmath14 to mean the righthand derivative at @xmath15 , and @xmath16 , @xmath17 to mean the lefthand derivative at @xmath18 .
moreover , we can take @xmath19 after a suitable translation .
these simplifications will be assumed in the rest of the paper .
some of the variations in other formulations are merely cosmetic .
for example , if @xmath20 , then use @xmath21 instead , or if @xmath7 is not defined at @xmath22 , use the righthand limit @xmath23 , if it exists
. we will not dwell further on these types .
some other variations are attempts to weaken the differentiability requirement on @xmath7 and @xmath8 .
a different type concerns stronger formulations in which strict monotonicity can be deduced from non - strict hypothesis .
we defer a discussion of such variations to section 2 . in the majority of concrete practical applications , however , the functions involved are fairly smooth , often infinitely differentiable and the stronger forms are seldom used .
propositions 147 and 148 ( page 106 ) in the famous classic by hlp @xcite ( first edition 1934 ) read : other than the phrases in red , proposition 148 is essentially the increasing part of the monotone rule , and proposition 147 is an application of the rule in the same spirit as in the more recent work . at first reading ,
proposition 148 appears to be weaker than the modern rule because it requires the additional conditions that @xmath7 is + and @xmath0 , and @xmath24 is + .
let us take a closer look at hlp s short and elegant proof which is reproduced below .
[ hardy , littlewood , plya ] to prove this , observe that @xmath25 and consider the possible intersections of the curves @xmath26 , @xmath27 . at one of these intersections the first curve has a horizontal and the second a rising tangent , and therefore there can be at most one intersection . if we take @xmath8 as the independent variable , write @xmath28 , and suppose , as in the last clause of the theorem , that @xmath29 or @xmath30 , then the theorem takes the form : _
if @xmath30 and @xmath31 increases for @xmath32 , then @xmath33 increases for @xmath32 .
_ this is a slight generalization of part of theorem 127 . in the proof
, the + @xmath34 property of @xmath8 is needed to guarantee that the denominators of the fractions @xmath35 and @xmath24 will not become 0 .
the + @xmath34 property of @xmath7 and the positivity of @xmath24 , however , is not needed anywhere .
once the extra conditions are disposed of , we see that the @xmath36 version of rule also holds , by considering @xmath37 instead of @xmath38 .
hlp did not attribute the result to anyone .
it could mean that it is one of their own , or it was widely known .
we did not attempt to track it down further in earlier literature . in one of the bourbaki books , @xcite ( 1958 ) , exercise 10 of chapter 1 , 2 ( page 38 )
is exactly hlp s proposition 148 , minus those extra conditions .
it seems that in the intervening years , someone must have figured out that those are superfluous .
neither a solution nor any attribution is given .
mitrinovic @xcite quoted bourbaki s exercise as 3.9.49 .
hlp s proposition , on the other hand , is broader than the modern rule .
it says something about the general situation when neither @xmath39 nor @xmath40 is assumed . in section 3 , we will fully characterize all such situations .
a classroom note by mott @xcite ( 1963 ) in the monthly presented a much weaker version of the monotone rule , in the integral formulation ( see section 2 ) .
in the ( more or less ) equivalent differential formulation , the hypotheses require @xmath41 and @xmath42 ( which implies @xmath24 is @xmath34 ) .
two follow - up papers by redheffer @xcite ( 1964 ) and boas @xcite ( 1965 ) contained alternative proofs and further comments .
the authors were unaware of the earlier results . then came the work of gromov @xcite , anderson , vamanamurthy , and vuorinen @xcite @xcite , pinelis @xcite , and many subsequent authors who have done a tremendous amount of good work to enrich the subject area .
the readers should have no difficulty finding them by searching for `` monotone lhpital rule '' on the internet , and by referring to the references cited in the papers listed .
analogous to the classical lhpital s rule for finding limits of indeterminant forms , one can also consider the situation when @xmath43 .
results , examples and counterexamples in this regard are presented in anderson , et al .
@xcite . in this article
, we confine ourselves to the case when @xmath44 and @xmath45 are finite .
if we let @xmath46 and @xmath47 , then the rule stated in section 1 becomes : this integral formulation has two advantages .
one is that now we do not have to separately assume that @xmath48 .
second is that the rule , stated as it is , can be applied to functions @xmath49 and @xmath50 with discontinuities .
for instance , @xmath51 and @xmath52 may be piecewise continuous , such as step functions , as long as their quotient is still monotone .
in such cases , the differential form would have failed because the requirement that @xmath7 and @xmath8 be differentiable at every point is not satisfied .
if one has proved the differential form of the rule previously based on this assumption , such as invoking the generalized cauchy mean value theorem , one has to seek a different proof .
alternatively , one can use an approximation technique , as suggested by redheffer @xcite , to deduce the integral form from the differential form . choose a sequence of continuous functions
@xmath53 that converge to @xmath52 in the uniform norm in the finite interval @xmath54 $ ] . next , choose another sequence of continuous functions @xmath55 that converge to @xmath56 , this time requiring that each @xmath55 preserves the same monotone property of the latter .
this can be done , for instance , by using the well - known method of mollifiers of s. sobolev and k.o .
friedrich in the theory of partial differential equations .
since the functions @xmath57 and @xmath58 are now differentiable everywhere , the differential form of the rule implies that @xmath59 is monotone . letting @xmath60 gives the desired conclusion .
there are other ways to relax the differentiability requirement in the differential formulation .
for example , one can use one - sided dini derivatives instead of regular derivatives . in the next section
, we will see how the rule can be formulated even without mentioning differentiability . as we have remarked before
, such extensions may be of theoretical value , but they are often not needed for practical applications . another direction of extension is to strengthen the conclusion of the rule .
pinelis @xcite showed that if @xmath24 is @xmath34 , then @xmath61 is in fact strictly + ( with strict monotonicity following as a corollary ) .
one can compare this with the regular and strong forms of the maximum principle in the theory of differential equations .
the weaker form states that the global maximum must be attained at the boundary of the region in consideration , while the strong form maintains that at the boundary point where the global maximum is attained , the directional derivative along the outward normal must be strictly + .
one is also able to deduce strict monotonicity in a certain sense from non - strict monotone assumptions .
one such case will be discussed in the next section .
in the particular case when @xmath62 , the increasing monotone rule reduces to : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if @xmath39 and @xmath63 , so is @xmath64 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the slightly weaker ( since @xmath7 is required to be twice differentiable ) version , @xmath65 , is what hlp referred to as proposition 127 .
both of these results are subsumed by the following well - known property of convex functions . @xmath66{5 in }
\par\vspace*{2mm}\par \em if $ f $ is a continuous , strictly convex function in $ [ 0,b ) $ , satisfying $ f(0)\leq 0 $ , then $ f / x{\,\,\nearrow}$ in $ [ 0,b ) $ .
\par\vspace*{1mm}\par \end{minipage } } \ ] ] a function is defined to be convex in @xmath54 $ ] if for all @xmath67 , @xmath68 it is said to be strictly convex if @xmath69 is replaced by @xmath70 .
concavity is defined by reversing the inequality signs .
no differentiability requirement is assumed .
it is well - known that for a continuous convex function , the following inequality holds : @xmath71[l]{\qquad\fbox{\rmcon } } \else\marginpar{\vspace{0.7\baselineskip}\hspace{-1.15\textwidth}\fbox{\rmcon } } \fi \fi}\ ] ] if the convexity is strict , replace @xmath69 by @xmath70 .
its geometric interpretation is that the arc of the graph of @xmath38 , @xmath72 $ ] , lies below the the chord joining the two points @xmath73 and @xmath74 . figure 1 .
convex functions with @xmath39 and @xmath75 .
@xmath76 @xmath76 @xmath77 @xmath77 @xmath78 @xmath78 figure 1 depicts two such functions ( green curves ) ; the first has @xmath39 and the second @xmath75 .
the quantity @xmath79 represents the slope of the straight line @xmath80 joining the origin @xmath76 and the point @xmath81 on the curve . as @xmath82 increases , the point @xmath83 slides along the curve towards the right and it is intuitively clear that the slope of @xmath80 increases . a vigorous proof can be given using ( [ con ] ) .
the proof is not new , but we include it here for easy reference . in the case
@xmath39 , by convexity , the arc of the curve between @xmath76 and @xmath78 lies below the straight line @xmath84 .
in particular , the point @xmath77 lies below @xmath84 and the desired conclusion follows .
for the case @xmath75 , we modify the function @xmath7 in @xmath85 $ ] by replacing the arc over @xmath85 $ ] with the chord @xmath86 .
the resulting new curve represents the function @xmath87 which is again convex .
then we are back to the first case @xmath39 .
* alternative proof*. another simple proof uses the fact that a straight line can not intersect a strict convex / concave curve at more than two points . as above , we only have to consider the case @xmath39 .
suppose @xmath64 is not monotone ; then there are @xmath88 such that @xmath89 . then the straight line @xmath90 intersects the graph of @xmath7 at the two points @xmath91 and @xmath92 . a third intersection point , however , is @xmath93 , giving a contradiction .
width4pt depth1pt but how do we make the quantum jump from the special case @xmath62 to the general case ?
the trick of change of variable , just as hlp pointed out in their proof , is the key . by the way
, this trick has been known to work for the classical lhpital rule for indeterminant limits as well .
see , for example , taylor @xcite .
let us rephrase hlp s argument to make it more transparent .
suppose that @xmath40 and @xmath94 is continuous and @xmath34 in @xmath54 $ ] .
the inverse function @xmath95 is well - defined .
substitute this into the definition of @xmath38 to get the composite function @xmath96 .
we have then the following generalized monotone rule . in practical applications , to check the convexity or concavity of @xmath97 ,
we often resort to showing that @xmath98 is monotone . by the chain rule
, @xmath98 is nothing but @xmath99 .
this new rule is more general than the popular one described on p. 1 , because it does not impose any differentiability on @xmath7 and @xmath8 , and it covers situations when @xmath20 .
if we omit `` strictly '' in the hypotheses , we can not guarantee that @xmath35 is strictly monotone , as the degenerate example @xmath100 shows .
however , this is pretty much the only exceptional situation , in the following sense . if there exist two points @xmath101 in @xmath102 such that @xmath103 , then @xmath104 in @xmath105 $ ] for some constant @xmath106 . to prove this , one only has to study the special case when @xmath62
then the assertion becomes geometrically obvious .
we omit the details .
convex function with @xmath107 .
@xmath76 @xmath83 @xmath108 next , let us look at the case @xmath107 when @xmath62 , and @xmath38 is strictly convex .
figure 2 depicts such a curve .
starting at @xmath109 , the point @xmath83 is where the curve intersects the vertical axis and the slope of @xmath80 is @xmath110 .
as @xmath82 increases , the slope of @xmath80 decreases until we reach the point @xmath108 such that @xmath111 is tangent to the curve . beyond that the slope of @xmath80 increases again .
this is an example of the general situation described in the conclusion of hlp s proposition .
the quintessential conclusion is that there exists at most one point @xmath112 such that @xmath64 is @xmath36 in @xmath113 and @xmath34 in @xmath114 .
as the example @xmath115 shows , the point @xmath116 may not exist at all .
after uplifting to the case of general @xmath94 , the pertinent part of hlp s proposition can be generalized as follows .
the symbol @xmath117 is a shorthand for `` there exists a unique '' .
of course , hlp s proof no longer works under the minimal assumption .
a rigorous proof can be given using only properties of convex functions .
the quotient @xmath64 is continuous in @xmath118 $ ] .
although it blows up at @xmath109 , it is easy to see that it attains a global minimum at some point @xmath119 $ ] . by its very construction , the line @xmath111 that joins the origin @xmath76 and the point @xmath120 lies below the curve of @xmath7 .
we can show that @xmath64 is decreasing in @xmath121 $ ] as follows .
let @xmath122 .
since the line joining @xmath76 and the point @xmath73 is above the line @xmath111 , the former must intersect the vertical line @xmath123 above @xmath108 , say at a point @xmath124 .
the arc of @xmath38 from @xmath77 to @xmath108 must be below the line @xmath125 , which lies below @xmath126 .
in particular , the point @xmath74 is below the line @xmath126 and hence @xmath127 . if it happens that @xmath128 , @xmath64 has no chance to bounce back from @xmath36 to @xmath34 .
if @xmath129 , then we can argue in a similar way as above that @xmath64 is @xmath34 in @xmath114 .
this completes the proof .
is there a practical way to delineate the two situations in the conclusion ? in the special case @xmath62 and @xmath38 is convex , it is easy to see that a necessary and sufficient condition for the existence of an interior @xmath112 is that in a left neighborhood of @xmath9 , the curve of @xmath7 lies below the chord joining @xmath76 and @xmath130 . in the general setting , this criteria can be expressed as @xmath131 one can also use limsup instead of liminf . for differentiable @xmath38 and @xmath94
, it simplifies to @xmath132 for concave @xmath97 , the inequality sign is reversed . if @xmath129 exists , it is determined by solving the equation @xmath133 .
if we allow the function @xmath97 to be non - strict convex in the hypotheses , then the curve of @xmath97 ( versus @xmath134 ) may contain flat portions ( line segments ) .
if it happens that the tangent line from the origin touches the curve and contains one of these flat portions , then the unique turning point @xmath116 in the above rule becomes an entire interval of turning points .
we have to modify the rule to say that now there exists a subinterval @xmath135\subset(a , b ) $ ] such that @xmath35 is @xmath36 in @xmath136 , constant in @xmath135 $ ] and @xmath34 in @xmath137 .
finally , let us consider the case when @xmath138 , which is also covered by hlp s proposition .
in the special case when @xmath139 , @xmath140 , and @xmath7 is convex , the corresponding problem is to investigate the @xmath34 and @xmath36 properties of @xmath141 , which is the slope of the line joining the point @xmath142 on the @xmath82-axis and the point @xmath143 on the graph of @xmath7 .
equivalently , we may apply a translation to shift the point @xmath142 to be the new origin . in this perspective
, we can exploit the same figures earlier in this section , only that now the curve of @xmath7 starts from @xmath144 instead of @xmath109 .
it is easy to check with simple examples that both possibilities discussed in the case @xmath40 , @xmath107 can occur . besides those , an additional possible third situation is that @xmath35 is @xmath34 in @xmath145 . in the special case @xmath139 ,
this happens when the line joining the points @xmath142 and @xmath146 lies below the curve of @xmath7 in a right neighborhood of @xmath109 . in the general
setting , if @xmath7 and @xmath8 are differentiable , this happens when @xmath147 however , the possibility of @xmath35 having the shape @xmath148 is not allowed .
we summarize all the findings : suppose that @xmath94 is a continuous , positive @xmath34 function in @xmath54 $ ] and @xmath96 is a continuous strictly convex ( concave ) function of @xmath149 in @xmath150 $ ] .
* @xmath40 and @xmath151 , then @xmath35 is @xmath34 @xmath152 . *
@xmath40 and @xmath153or @xmath138 . * * if @xmath154 and @xmath155 then @xmath35 is @xmath34 @xmath156 . * * if @xmath157 then @xmath35 is @xmath36 @xmath158 . * * otherwise , @xmath35 has the shape @xmath159 @xmath160 with a unique turning point in @xmath145 .
suppose that the convexity / concavity of @xmath97 is not assumed to be strict .
let @xmath161 $ ] be a maximal subinterval @xmath162which is possibly void@xmath163 in which @xmath104 for a constant @xmath106 .
then in @xmath164 $ ] , the same conclusions as above @xmath162with all monotonicity being strict@xmath163 hold .
case ( 1 ) can actually be combined with ( 2 ) i ) .
we prefer to separate it out since it is historically as well as application - wise the most prominent case .
two points in the result are worth noting .
the function @xmath35 can not change monotonicity more than once , and in the convex case , the shape @xmath148 is ruled out .
the usefulness of the criteria given in the cases ( 2 ) i ) and ii ) ( in addition to the @xmath165 condition ) lies in the fact that there is only one boundary condition at one of the endpoints to verify in order to deduce monotonicity of @xmath35 over the entire interval .
the last part of the theorem is a strong form of the rule .
even if strict convexity / concavity is not assumed , we can still obtain strict monotonicity , except possibly in an initial subinterval in a very special situation . if @xmath94 is a positive @xmath36 function in @xmath145 , an analogous result holds .
we only have to do a reflection @xmath166 to reduce it to the @xmath34 case .
the role of @xmath22 and @xmath9 are now exchanged .
one has to be careful in chasing the signs and inequalities in the conditions .
we state the result for ease of reference .
it is useful in handling functions such as @xmath167 .
suppose that @xmath94 is a continuous , positive @xmath36 function in @xmath145 and @xmath96 is a continuous strictly convex @xmath162concave@xmath163 function of @xmath149 in @xmath168 $ ] .
* @xmath169 and @xmath170 , then @xmath35 is @xmath36 @xmath171 .
* @xmath169 and @xmath172or @xmath173 . * * if @xmath174 then @xmath35 is @xmath34 @xmath156 . * * if @xmath175 and @xmath176 then @xmath35 is @xmath36 @xmath158 . * * otherwise , @xmath35 has the shape @xmath159 @xmath160 with a unique turning point in @xmath145 . in practical applications
when @xmath7 and @xmath8 are differentiable , the following simplified rule is easier to use .
corollary 1 deals with increasing @xmath8 and corollary 2 deals with decreasing @xmath8 .
suppose @xmath7 and @xmath8 are differentiable , @xmath177 , and @xmath178 in @xmath145 . * if @xmath40 and @xmath151 , then @xmath35 is @xmath34 @xmath152 . *
@xmath40 and @xmath153or @xmath138 . * * if @xmath154 and @xmath179 then @xmath35 is @xmath34 @xmath156 . * * if @xmath180 then @xmath35 is @xmath36 @xmath158 . *
* otherwise , @xmath35 has the shape @xmath159 @xmath160 with a unique turning point in @xmath145 .
it is instructive to visualize the three possibilities in the rule .
figure 3 shows the plots of @xmath35 ( red curves ) and @xmath24 ( green dashed curves ) in three typical examples .
all other examples exhibit the same features . by hypotheses ,
the dashed curves are @xmath34 .
cases ( 1 ) and ( 2 ) i ) ( 2 ) ii ) ( 2 ) iii ) figure 3 .
three possibilities of @xmath35 ( red curve ) when @xmath181 .
the first plot shows that if @xmath35 lies below @xmath24 , then @xmath35 is @xmath34 .
as the rule asserts , to guarantee this situation , you only have to check whether @xmath48 ( case ( 1 ) ) or in case ( 2 ) i ) , to check whether @xmath182 .
in other words , only the behaviors at the left endpoint matters .
the second plot shows that if @xmath35 lies above @xmath24 , then @xmath35 is @xmath36 .
again , to guarantee this situation , you only need to check whether @xmath183 at the right endpoint .
the third plot shows the @xmath159 possibility , which happens when the dashed curve intersects the red curve at some point @xmath112 .
this situation can be considered as a hybrid case : before the intersection , the plot looks like the second one , and after that it looks like the first one .
right at the intersection , the red curve has a horizontal tangent .
hlp have recorded the same observation in their proof of proposition 148 .
suppose @xmath7 and @xmath8 are differentiable , @xmath32 , @xmath184 , and @xmath185 in @xmath145 . * if @xmath169 and @xmath170 , then @xmath35 is @xmath36 @xmath171 . *
@xmath169 and @xmath172or @xmath173 . * * if @xmath186 then @xmath35 is @xmath34 @xmath156 . * * if @xmath175 and @xmath187 then @xmath35 is @xmath36 @xmath158 . * * otherwise , @xmath35 has the shape @xmath159 @xmath160 with a unique turning point in @xmath145 .
the analogous plots of @xmath35 and @xmath24 are shown in figure 4 .
cases ( 1 ) and ( 2 ) i ) ( 2 ) ii ) ( 2 ) iii ) figure 4 .
three possibilities of @xmath35 ( red curve ) when @xmath188 .
pinelis @xcite @xcite discovered an interesting extension of the monotone rule to the situation when @xmath24 is no longer monotone . in this section
, we look at this extension from the perspective of convex / concave functions .
figure 5 . a function that changes convexity three times .
@xmath76 @xmath189 @xmath190 @xmath191 suppose that @xmath145 can be divided into @xmath192 subintervals with points @xmath193 and @xmath24 is assumed to be @xmath34 in the odd subintervals @xmath194 , @xmath195 , @xmath196 and @xmath36 in the even ones .
the function changes its monotonicity @xmath197 times .
figure 5 depicts one such functions with @xmath198 .
it is convex in @xmath199 $ ] , concave in @xmath200 $ ] , etc .
the points @xmath201 are points of reflection of @xmath7 . for convenience , we can also say that @xmath24 `` oscillates '' @xmath197 times .
do not confuse this use of the term with the more conventional meaning , as in the `` oscillation '' of the pendulum .
when a pendulum oscillates 2 times , it has `` oscillated in our sense '' ( changed directions ) 3 times .
the case when @xmath24 is @xmath36 in the odd subintervals and @xmath34 in the even ones can be studied in a similar way .
: _ what can we say about the oscillatory property of @xmath35 ?
_ again , we first appeal to the special case when @xmath62 , and @xmath7 is assumed to be alternatively convex and concave in the subintervals . by theorem 1
( 1 ) , @xmath64 is @xmath34 in @xmath199 $ ] . in @xmath200
$ ] , @xmath7 is concave . using the concave version of theorem 1 ( 2 )
, we see that @xmath64 can have three possible behavior .
since before @xmath189 , it is @xmath34 , it will continue to @xmath34 at least for a little while after @xmath189 .
this rules out the case ( 2 ) i ) , that it is @xmath36 in the entire subinterval . in the remaining two possibilities , there may or may not exist a turning point @xmath202 , depending on whether a tangent line through the origin can be drawn touching the curve inside @xmath203 .
if such a @xmath204 exists , then @xmath64 will change monotonicity once in @xmath205 $ ] .
otherwise , @xmath64 remains @xmath34 in @xmath205 $ ] . to summarize , in @xmath205 $ ] , @xmath64 can not oscillate more times than @xmath206 .
furthermore , the turning point of @xmath35 , @xmath204 ( if it exists ) lags behind that of the latter , namely , @xmath189 . in other words , @xmath207 . we can continue with similar arguments in subsequent intervals and it is easy to see that the first statement in the above summary remains true throughout the entire interval @xmath102 .
the second statement has to be interpreted in the following way . in each subinterval @xmath208
, there is at most one turning point @xmath209 of @xmath210 .
the sense of the change of monotonicity ( i.e. from @xmath34 to @xmath36 , or from @xmath36 to @xmath211 is the same as that of @xmath206 at @xmath201 . in the list of all turning points of @xmath79
, some @xmath209 can be missing .
the case when @xmath107 is just a little more complicated . in view of theorem 1
( 2 ) , we have to append the possibility of a first turning point @xmath212 at which @xmath64 switches from @xmath36 to @xmath34 ; @xmath64 can change monotonicity at most @xmath192 times .
if there is no such @xmath213 , then either @xmath64 is @xmath34 in @xmath214 and the behavior is exactly the same as the case when @xmath198 , or @xmath64 is @xmath36 in @xmath205 $ ] and in the remaining subintervals the behavior mimics that of the case @xmath198 . in the first situation , @xmath64 changes monotonicity at most @xmath197 times . in the second situation ,
it changes at most @xmath215 times .
the case with @xmath139 , @xmath140 can be analyzed in a similar way . after translating to the more general setting
, we derive the following generalization of pinelis oscillation rules .
suppose that @xmath32 and @xmath34 in @xmath54 $ ] .
suppose that @xmath96 , as defined before , is a continuous function of @xmath149 and is alternatively strictly convex @xmath162concave@xmath163 and strictly concave @xmath162convex@xmath163 in the @xmath216 subintervals corresponding to the decomposition of @xmath102 described above .
if @xmath40 and @xmath217 , then @xmath35 is initially @xmath34 @xmath152 in @xmath214 , while in each subsequent subinterval @xmath208 , @xmath35 can change monotonicity at most once @xmath162 in the same sense as the change of monotonicity of @xmath24 at @xmath218 .
hence , @xmath35 can oscillate at most @xmath197 times . if @xmath40 and @xmath219 , or if @xmath138 , then @xmath35 may or may not have one additional change of monotonicity in @xmath214 .
the behavior in subsequent subintervals is the same as in the previous case .
hence , @xmath35 can change monotonicity at most @xmath192 times .
theorem 3 has an obvious analog for + @xmath36 @xmath8 .
two corollaries have found applications ( to be described in the next section ) in some recent work of the author .
suppose @xmath7 and @xmath8 are continuous in @xmath220 $ ] , differentiable , with @xmath221 in @xmath222 . if @xmath24 is initially @xmath223 and changes monotonicity only once in @xmath220 $ ] , then @xmath35 has a unique global maximum @xmath162minimum@xmath163 in @xmath220 $ ] .
let @xmath7 and @xmath8 be continuous and differentiable with @xmath221 in @xmath222 .
suppose that @xmath24 is initially @xmath34 and changes monotonicity once in @xmath220 $ ] .
if in addition @xmath224 and @xmath225 , then @xmath35 is @xmath34 in @xmath220 $ ] suppose that @xmath24 is @xmath34 in @xmath226 $ ] and @xmath36 in @xmath227 $ ] . by corollary 1 ( 2 ) i ) , the boundary condition at @xmath15 implies that @xmath35 is @xmath34 in @xmath226 $ ] . in @xmath227
$ ] , we use the concave version of corollary 1 ( 2 ) ii ) , applied to the boundary condition at @xmath9 , to conclude that @xmath35 is also @xmath34 there .
we can push corollary 4 a little further to still get @xmath35 @xmath34 when @xmath24 changes monotonicity two times .
let @xmath7 and @xmath8 be continuous and differentiable with @xmath221 in @xmath222 .
suppose that @xmath24 is initially @xmath34 and changes monotonicity exactly twice in @xmath220 $ ] .
let @xmath228 be the second turning point of @xmath24 .
if @xmath224 and @xmath229 , then @xmath35 is increasing in @xmath220 $ ] corollaries 4 and 5 have analogs for proving @xmath230 .
in some recent work with h. alzer ( on studying some properties of the error function ) we need to compute the global maximum value of the function @xmath231 in @xmath232 , where @xmath233 is a multiple of the error function . from a theoretical viewpoint ,
no matter how accurate the maximization algorithm is , these values can not be simply taken to be the correct ones .
that is because most algorithms can only guarantee to return a local maximum , which is not necessarily the global maximum . before any further justification ,
the best we can conclude is that the computed value represents a local maximum out of possibly multiple local maxima .
hence , it can only be taken as a lower bound of the true value sought . to put any doubt to rest
, we need to affirm that @xmath236 changes monotonicity only once .
letting @xmath238 , we compute the `` hpital derivative '' ( for lack of a better name ) of @xmath236 @xmath239 the derivative of which is @xmath240 it is easy to verify that @xmath241 has a unique positive root @xmath242 , and that @xmath243 is @xmath34 in @xmath194 and @xmath36 in @xmath244 . by corollary 2
, we conclude that @xmath245 also changes monotonicity only once in the same sense , just as desired .
the case of @xmath249 is a bit more complicated . with @xmath238 and @xmath250 , its hpital
derivative is @xmath251 it suffices to show that @xmath252 is initially increasing and then changes monotonicity only once in @xmath253 .
one is tempted to apply the oscillation rule one more time by computing @xmath254 however , the conditions of the rule are not satisfied because the denominator of @xmath252 is not a monotone function of @xmath82 .
in fact , it increases in @xmath255 and decreases in @xmath256 .
we have to investigate the behaviors of @xmath252 in these two subintervals separately .
first we study the monotonicity of @xmath257 in @xmath258 and @xmath256 .
the derivative of @xmath257 , after some simplification , is @xmath259 since the fraction on the righthand side is nonnegative , the monotonicity of @xmath257 depends on the sign of the polynomial in parentheses .
letting @xmath260 , the polynomial can be written as @xmath261 since @xmath262 and @xmath263 , @xmath264 has a root @xmath265 in @xmath255 .
we claim that this is the only positive root , by showing that @xmath264 is strictly decreasing for @xmath266 . to this end
, we note that @xmath267 attains its global maximum in @xmath253 when @xmath268 .
it is easy to verify that this global maximum is negative .
we are now ready to study @xmath252 in @xmath271 .
since its denominator is decreasing in @xmath271 , we invoke corollary 2 instead of corollary 1 .
since @xmath257 is decreasing , the function @xmath97 in the hypotheses is concave and we have to use the concave version of corollary 2 .
it is easy to verify that @xmath272 hence , for @xmath9 very large , @xmath273 . by corollary 2
( 2 ) ii ) , we conclude that @xmath274 in @xmath275 $ ] for large @xmath9 . in @xmath276 $ ] , @xmath277 changes monotonicity once , implying that @xmath278 changes monotonicity at most once .
the only way that this is compatible with @xmath274 in @xmath256 is that @xmath278 changes monotonicity exactly once in @xmath232 , as desired . .
the function @xmath279 occurs in the same study .
we want to show that it is @xmath34 in the interval @xmath280 $ ] .
note that we can not extend the claim to @xmath276 $ ] because @xmath281 is not @xmath34 at @xmath282 . letting @xmath283 and @xmath284
, we find that @xmath285 however , @xmath24 is not monotone in @xmath286 ; it changes monotonicity at @xmath287 .
hence , the regular monotone rule fails .
we can easily verify that the hypotheses of corollary 4 are satisfied and thus conclude that @xmath281 is @xmath34 in the interval .
figure 6 shows the graphs of @xmath288 and its hpital derivative .
it should be compared with figures 3 and 4 .
the green dashed curve is not monotone , but as long as it stays above the red curve , the latter is @xmath34 .
the hpital derivative is @xmath290 note that at @xmath109 , the denominator becomes 0 while the numerator is @xmath291 .
therefore , the usual monotone rule can not be used . instead of using the extended rule established in this article ,
an easier way is to note that @xmath292 the first two terms combined is @xmath34 .
thus , it suffices to show that the last term is @xmath34 . yet
the usual rule still can not be used directly because the denominator of the last term is not monotone in @xmath232 .
we can overcome that obstacle by showing that its reciprocal @xmath293 is @xmath36 .
then the usual rule can be applied .
99 = 1 g.d .
anderson , m.k .
vamanamurthy , and m. vuorinen , conformal invariants , inequalities , and quasiconformal maps , john wiley and sons , new york , 1997 .
anderson , m.k .
vamanamurthy , and m. vuorinen , monotonicity rules in calculus , american math .
monthly 113 ( 2006 ) , 806 - 816 = 1 r.p .
boas jr . ,
more about quotients of monotone functions , amer .
monthly 72 ( 1965 ) , 5960 .
= 1 n. bourbaki , fonctions dune variable relle , paris 1958 .
= 1 j. cheeger , m. gromov , and m. taylor , finite propagation speed , kernel estimates for functions of the laplace operator , and the geometry of complete riemannian manifolds , j. differential geom .
17 ( 1982 ) , 15 - 53 .
hardy , j.e .
littlewood , and g. plya , inequalities , second edition , cambridge ( first edition 1934 , second edition 1952 ) .
mitrinovic , analytic inequalities , springer - verlag , 1970 .
mott , on the quotient of monotone functions , amer . math .
monthly 70 ( 1963 ) , 195 - 196 .
= 1 i. pinelis , lhospital type rules for monotonicity , with applications j. ineq .
pure & appl .
, 3 ( 2002 ) , issue 1 , article 5 .
= 1 i. pinelis , lhospital type rules for oscillation , with applications , j. ineq . pure & appl . math . , 3 ( 2002 ) ,
issue 2 , article 33 .
= 1 i. pinelis , lhospital type rules for oscillation : an application to probability inequalities for sums of bounded random variables , j. ineq .
pure & appl .
, 3 ( 2002 ) , issue 1 , article 7 .
redheffer , remarks about quotients , amer .
monthly 71 ( 1964 ) , 69 - 71 .
taylor , lhpital s rule , amer . math .
monthly 59 ( 1952 ) , 20 - 24 . | we point out the connection of the so - called hpital - style rules for monotonicity and oscillation to some well - known properties of concave / convex functions . from this standpoint
, we are able to generalize the rules under no differentiability requirements and greatly extend their usability .
the improved rules can handle situations in which the functions involved have non - zero initial values and when the derived functions are not necessarily monotone .
this perspective is not new ; it can be dated back to hardy , littlewood and plya . # 1 # 2 plus 2pt # 1 # 1 * # 1 * 1ex # 1 # 1 2ex # 1 # 1 # 2 man kam kwong _ department of applied mathematics + the hong kong polytechnic university , + hunghom , hong kong _ + 26.70 , 26a48 , 26a51 monotonicity , hpital s rule , oscillation , inequalities . |
the born infeld electromagnetic theory @xcite was introduced in order to avoid the infinite self energy of a charged point particle , which constitutes a well known problem within maxwell electrodynamics , and it is the only non linear theory without birefringence .
maxwell and born infeld theories have electric magnetic duality invariance @xcite , a property not shared with other electromagnetic theories .
recently , the born infeld electrodynamics has received considerable attention as it can be obtained as a low energy limit of string theory , which at present constitutes the main candidate for a unified theory .
the spherically symmetric solution of einstein gravity coupled to born infeld electrodynamics corresponds to a charged black hole @xcite .
the solution in @xmath1 dimensions @xcite also represents a black hole which is singular at the origin , where the ricci scalar and the ricci square diverge .
the metric is regular everywhere , and is given by ds^2=-f(r)dt^2+f^-1(r)dr^2+r^2d^2,[1 ] where the function @xmath2 has the form f(r)= -m++ r--2q^2.[1f ] the dimensionless constants @xmath3 and @xmath4 are identified as the adm mass and the charge , @xmath5 is the cosmological constant with dimensions @xmath6^{-2}$ ] , and @xmath7 is the born infeld parameter ; @xmath7 has dimensions of length , and therefore those of the inverse of the field strength .
the limit @xmath8 corresponds to maxwell linear electrodynamics ; in this limit the well known static btz solution with charge @xcite is recovered , and we simply have @xmath9 .
the spacetime geometry of the btz black hole has constant negative curvature and it is , locally , that of anti de sitter ( ads ) space ; it only differs from anti de sitter in its global properties .
this feature makes the btz solution of great interest within the framework of the ads / cft correspondence . in the geometry defined by eqs .
( [ 1 ] ) and ( [ 1f ] ) , the radii @xmath10 and @xmath11 are related by @xmath12 , and are associated to the zero of the electrostatic potential @xmath13 in each theory .
we will assume that the cosmological constant @xmath14 is negative ( i.e. @xmath15 ) , so that in the case of a vanishing charge the geometry is asymptotically anti - desitter .
for suitable values of the other parameters , this choice makes possible a standard horizon structure @xcite , in which the metric function is always positive beyond a certain radius ( different for each theory ) . thin layers of matter naturally appear in the context of general relativity and cosmology .
for instance , they can be used to model the gravitational collapse to a black hole , and also for the study of the evolution of bubbles and domain walls in a cosmological setting .
spherically symmetric shells around vacuum ( bubbles ) , around black holes and stars , and also being the throat of traversable wormholes , have been extensively studied @xcite ; these shells have also been considered in more than four dimensions @xcite .
the dynamics of collapsing shells in three spacetime dimensions has been presented and applied to several examples in refs .
@xcite . besides , shells in a three dimensional background within einstein
maxwell theory have been associated to thin - shell wormholes @xcite . on the other hand ,
we have recently analyzed the problem in the four dimensional case for einstein gravity coupled to born infeld electrodynamics in relation with thin - shell wormholes @xcite and shells around vacuum or around black holes @xcite .
the current interest in @xmath0-dimensional einstein maxwell and einstein born infeld black holes , leads to address the related problem of the behavior of thin layers associated with them . in the present article , we study the characterization and the linearized stability under perturbations preserving the symmetry of circular charged shells around black holes ; as a particular case , we also analyze shells around vacuum .
we mathematically build the shells starting from two black hole geometries with a negative cosmological constant , given by eqs .
( [ 1 ] ) and ( [ 1f ] ) .
we apply the well known cut and paste procedure ; we work under the condition that the resulting shell is constituted by normal matter , that is , by matter satisfying the energy conditions .
we set the units so that @xmath16 .
to define a shell ( more precisely a ring ) of radius @xmath17 we take two geometries of the form ( [ 1 ] ) with different metric functions @xmath18 and different coordinates @xmath19 and @xmath20 , and we remove the outer region @xmath21 of one of them and the inner region @xmath22 of the other one .
then , we join the resulting manifolds at @xmath23 to form a new one . for a shell surrounding a black hole , the radius @xmath17 is chosen beyond the largest horizon radius of the inner geometry and also @xmath17 should be large enough to remove the horizons ( if any ) of the outer geometry . in the case of a shell around vacuum ,
the first restriction is no longer needed .
the new complete spacetime in general includes a @xmath24 dimensional matter shell at @xmath25 ; this implies that the metric of the complete geometry , though continuous everywhere , must have discontinuous derivatives .
more precisely , at the shell radius @xmath23 the line element satisfies @xmath26 , while the derivatives of the metric at both sides of the @xmath24 dimensional surface @xmath25 are related with the surface energy - momentum tensor @xmath27 by the lanczos equations @xcite 8s_i^j=-[k_i^j]+_i^j[k ] where @xmath28 is the extrinsic curvature tensor , @xmath29 stands for its trace , and the brackets denote the jump of a given quantity across the circumference @xmath25 .
we let the radius @xmath17 to be a function of the proper time @xmath30 on the ring .
then the components of the extrinsic curvature tensor are given by ^1,2=- , ^1,2= , where a prime denotes a derivative with respect to @xmath31 and a dot stands for @xmath32 .
the surface energy density @xmath33 and the pressure @xmath34 are given by = -+ , p=- .
the energy and pressure for a static ( @xmath35 ) ring read _ 0=-+ , p_0=- .
a static shell of normal ( i.e. non exotic ) matter satisfies the weak energy condition , i.e. @xmath36 and @xmath37 ; this implies @xmath38 . with some changes ,
the cut and paste procedure can be used to construct a wormhole geometry ; in this case , both terms in the expressions for @xmath39 and @xmath40 will have the same sign . then , the weak energy condition can not be fulfilled because the joining of two exterior regions @xmath41 with metrics given by @xmath42 and @xmath43 would give @xmath44 . in our study of bubbles or shells around black holes
, we will work under the assumption of normal matter .
from the lanczos equations , we can obtain an equation for the evolution of the shell radius in the form analogous to the energy conservation for a point particle restricted to a motion in only one spatial dimension . squaring twice the expression for the energy density and after some algebra we obtain ^2+v(a)=0[mal ]
, where the potential @xmath45 has the form v(a)=-^2 - 4a(a ) ^2.[pot ] the energy and pressure satisfy the conservation equation ( a)+p=0 , which leads to =-(+p ) . if an equation of state @xmath46 is given , this can be integrated to obtain the energy density as a function of the ring radius by inverting the resulting relation a=-+c.[sol1 ] then @xmath47 can be substituted in @xmath45 and this would allow , in principle , to obtain the time evolution of the ring radius .
we are interested in a perturbative treatment of the dynamics , more precisely , the stability of static solutions . because of the analogy with the problem of a point particle in a one dimensional potential ,
the analysis is straightforwardly carried out in terms of the sign of the second derivative of the potential @xmath45 at an equilibrium configuration given by @xmath35 .
we introduce the definitions s(a)= , r(a)= , m(a)=2a .
this allows to write the potential as v(a)=s--4 m^2 .
a static configuration @xmath35 implies @xmath48 , equilibrium requires @xmath49 , and the condition for stability is that @xmath50 .
the first and second derivatives of the potential read v(a_0)=s-^2--8mm , @xmath51 in the perturbative treatment , only the first derivative of the pressure at the equilibrium position is involved in the dynamics .
second derivatives of @xmath52 and of @xmath53 can be expressed in terms of first derivatives by recalling the conservation equation @xmath54 and introducing the parameter which relates the derivatives of the pressure and energy density at the equilibrium radius @xmath55 ; in the case @xmath56 this parameter can be understood as the speed of sound along the ring . with this definition
we have m=-m- ,
= + m- . then taking into account that @xmath49 and defining the auxiliary functions x(a_0)=s--
, y(a_0)=s-^2+rr--^2 , z(a_0)=-m++m- , the stability condition for a static equilibrium configuration takes the concise form z(a_0)<y(a_0)-8x^2(a_0).[sta ] because we are interested in the study of bubbles and rings around black holes ( not in thin - shell wormholes ) , the additional condition of normal matter must be imposed . the natural way to understand the results
is then to present the stability regions by drawing the intersection of the relation ( [ sta ] ) and the inequalities @xmath57 for different values of the parameters . , @xmath58 , @xmath59 , @xmath60 ; the critical values of the charge are @xmath61 and @xmath62.,width=604 ] , @xmath63 , @xmath59 , @xmath64 ; the critical value of the charge is @xmath65.,width=453 ] , @xmath66 , @xmath59 , @xmath67.,width=604 ]
. the values of the parameters are @xmath68 , @xmath58 , @xmath59 , @xmath60 ; the critical values of the charge are @xmath69 and @xmath70.,width=604 ] .
the values of the parameters are @xmath68 , @xmath71 , @xmath59 , @xmath72 ; the critical value of the charge is @xmath73.,width=453 ] .
the values of the parameters are @xmath68 , @xmath66 , @xmath59 , @xmath67.,width=604 ] in the case of shells around black holes in einstein born infeld theory ,
we fix the scale with the choice of the cosmological constant ; the inner metric corresponds to non charged black holes ( @xmath59 ) with different masses , thus the ring charge is simply the charge associated to the exterior metric .
we first consider the limit @xmath8 , which corresponds to maxwell electrodynamics .
the values of the masses and charges of the exterior metrics are shown in figs .
1 to 3 . for low values of the mass @xmath74 , two critical values of the charge @xmath75 exist : @xmath76 and @xmath77 are such that for a charge under @xmath76 and for a charge beyond @xmath77 there is an event horizon in the original outer metric , while for @xmath78 it presents a naked singularity . for a certain value of the mass , the critical values of the charge fuse into one , and
beyond that an event horizon always exists in the original exterior metric for any value of the charge . in the complete manifold , the features of the outer original metric , i. e. its horizon structure , determine the behavior ( range , location and shape ) of the stability regions in parameter space as the charge increases .
in particular , only in the case in which the horizon of the original metric is lost , stability could be possible for vanishing or negative @xmath79 . in all cases
a vanishing charge makes stability compatible only with @xmath80 . figs .
1 to 3 illustrate the behavior of the stability regions with an increase of the charge , for a fixed value of the cosmological constant and three different values of the mass .
the main difference with the results for the four dimensional case @xcite is that in three dimensions the evolution of the regions is not monotonous : as the considered charge becomes larger , so that the original outer geometry has an event horizon , the stability regions recover the same form of those corresponding to low values of the charge . in the case of einstein gravity
coupled to born infeld non linear electrodynamics , the departure from standard maxwell theory is determined by the parameter @xmath7 , which can be taken non negative without losing generality .
the stability study is performed in terms of the dimensionless quantity @xmath81 .
as before , we consider charged shells around @xmath0-dimensional non charged black holes ; the results are shown in figs .
the behavior of the original outer metric with the charge is analogous to that of the maxwell case ; however , for all other parameters fixed , the difference between the values of the two critical charges is smaller in born infeld electrodynamics .
more precisely : for a given value of @xmath81 , the difference between @xmath77 and @xmath82 decreases as the mass @xmath74 is enlarged , while for a given value of @xmath74 this difference becomes smaller as a larger @xmath81 is chosen .
this is reflected in the evolution of the shape and size of the stability regions as functions of the parameters .
we display only the results corresponding to @xmath83 .
the plots show that the region such that smaller positive and also negative values of @xmath79 are compatible with stability is reduced and finally disappears as the mass @xmath74 is taken larger .
another point to be noted is that , for charges near the critical ones , the stability regions for the same values of @xmath84 are in general slightly larger in the case of born infeld electrodynamics , making stability compatible with lower values of the parameter @xmath79 . as a particular case
, we can study bubbles rings around vacuum by taking the inner black hole mass @xmath85 equal to zero .
this can be done for both maxwell and born infeld theories . within both theoretical frameworks
the results are very similar to those corresponding to rings around black holes , so figures are not included .
however , the numerical analysis shows that , in general , for the same values of the ring charge , rings around black holes admit slightly larger stability regions , associated to smaller possible values of the parameter @xmath79 , than rings around vacuum .
we have constructed charged rings around @xmath0-dimensional non charged black holes within the framework of einstein born infeld theory , and we have studied their stability under linearized radial perturbations . the einstein
maxwell case corresponds to the limit in which the born infeld parameter @xmath7 is zero .
we have found that , for fixed @xmath86 and taking a given @xmath81 , for low values of the mass @xmath74 of the outer metric , there are two critical values of the charge such that the stability regions are larger for charges between them ; these critical charges are no longer present for large values of the mass @xmath74 , and the stability regions are smaller in this case .
there are ranges of the parameters for which the stability regions include the physically more interesting case @xmath87 , so @xmath88 can be understood as the velocity of sound on the shell . for fixed @xmath74 ,
the critical values of charge get closer to each other as @xmath81 grows , reducing the range of charge corresponding to larger stability regions .
we have also considered the particular case of bubbles , that is , charged rings around vacuum . though not displayed , the numerical results show that , for fixed values of the parameters , the stability regions of rings around vacuum are slightly smaller than those of rings around black holes , in the sense that stiffer matter is needed .
this work has been supported by universidad de buenos aires and conicet .
p. r. brady , j. louko and e. poisson , phys .
d * 44 * , 1891 ( 1991 ) ; m. ishak and k. lake , phys .
d * 65 * , 044011 ( 2002 ) ; s. m. c. v. gonalves , phys . rev .
d * 66 * , 084021 ( 2002 ) ; f. s. n. lobo and p. crawford , class .
quantum grav .
* 22 * , 4869 ( 2005 ) . m. visser and d.l .
wiltshire , class .
quantum grav .
* 21 * , 1135 ( 2004 ) ; n. bili , g.b .
tupper , and r. d. viollier , jcap * 02 * , 013 ( 2006 ) ; f. s. n. lobo and a. v. b. arellano , class .
quantum grav .
* 24 * , 1069 ( 2007 ) .
e. poisson and m. visser , phys .
d * 52 * , 7318 ( 1995 ) ; e. f. eiroa and g. e. romero , gen .
. grav . * 36 * , 651 ( 2004 ) ; f.s.n .
lobo and p. crawford , class .
quantum grav .
* 21 * , 391 ( 2004 ) ; e. f. eiroa and c. simeone , phys .
d * 71 * 127501 ( 2005 ) ; m. g. richarte and c. simeone , int . j. mod .
d17 * , 1179 ( 2008 ) ; j. p. s. lemos and f. s.n .
lobo , phys .
d * 78 * , 044030 ( 2008 ) ; e. f. eiroa , phys .
d * 78 * , 024018 ( 2008 ) ; g. a. s. dias and j. p. s. lemos , phys .
d * 82 * , 084023 ( 2010 ) .
m. thibeault , c. simeone and e. f. eiroa , gen .
. grav . * 38 * , 1593 ( 2006 ) ; m. g. richarte and c. simeone , phys .
d * 76 * , 087502 ( 2007 ) ; * 77 * , 089903(e ) ( 2008 ) ; s. habib mazharimousavi , m. halilsoy and z. amirabi , phys . rev .
d * 81 * , 104002 ( 2010 ) ; c. simeone , phys .
d * 83 * , 087503 ( 2011 ) .
j. crisstomo and r. olea , phys .
d * 69 * , 104023 ( 2004 ) ; r. olea , mod .
a20 * , 2649 ( 2005 ) ; r.b .
mann and j.j .
oh , phys .
d * 74 * , 124016 ( 2006 ) ; * 77 * , 129902(e ) ( 2008 ) ; r.b .
mann , j.j .
oh , and m .- i .
park , phys .
d * 79 * , 064005 ( 2009 ) .
n. sen , ann .
( berlin ) * 378 * , 365 ( 1924 ) ; k. lanczos , ann .
( berlin ) * 379 * , 518 ( 1924 ) ; g. darmois , mmorial des sciences mathmatiques , fascicule xxv ( gauthier - villars , paris , 1927 ) , chap . 5 ; w. israel , nuovo cimento b * 44 * , 1 ( 1966 ) ; * 48 * , 463(e ) ( 1967 ) . | we study circular shells in a @xmath0-dimensional background within the framework of einstein born infeld theory . for shells around black holes
we analyze the mechanical stability under perturbations preserving the symmetry .
shells around vacuum are also discussed .
we find a large range in the values of the parameters compatible with stable configurations .
+ pacs number(s ) : 04.20.gz ; 04.40.-b ; 04.70.bw + keywords : thin - shell ; spacetimes with charge ; nonlinear electrodynamics . |
orthogonal transform such as discrete wavelet transform is an important tool in statistical signal processing and analysis .
especially , wavelet denoising is a popular application of discrete wavelet transform . in wavelet denoising
, noisy signal is transformed into wavelet domain in which wavelet coefficients are obtained . by applying a thresholding method ,
noise - related parts of coefficients are removed in a sense ; e.g. some of coefficients are set to zero .
the inverse wavelet transform of the modified coefficients yields a denoised signal .
the most popular and simple methods of thresholding is hard and soft - thresholding in @xcite .
both thresholding methods have a parameter . in hard - thresholding method , the parameter works purely as a threshold level ; i.e. coefficients less than the parameter value are removed and un - removed coefficients are harmless . on the other hand , in soft - thresholding ,
the parameter works as a threshold level as in hard - thresholding and simultaneously as an amount of shrinkage for un - removed components .
coefficients less than the parameter value are removed and un - removed coefficients are shrunk toward zero by the parameter . for a better denoising performance , we need to determine an optimal value of the parameter .
for example , in hard - thresholding , if the parameter value is too large then most of coefficients are removed even when those are significant .
this results in an excess smoothing that yields a large bias between estimated output and target function output .
on the other hand , if the parameter value is too small then most of coefficients are un - removed even when those are not significant .
this results in a large variance of output estimate and , thus useless for denoising .
a problem of choice of an optimal parameter value is often referred as a model selection problem .
there are several model selection methods under thresholding .
@xcite has proposed universal hard and soft - thresholding in which a theoretically significant constant value is employed as a parameter value .
also , @xcite has derived a criterion for determining an optimal parameter value of soft - thresholding by applying stein s lemma@xcite . the soft - thresholding method with this criterion
is called as sure ( stein s unbiased risk estimator ) shrink in @xcite .
unfortunately , there is no such a theoretically supported criterion for hard - thresholding while modified cross validation approaches have been proposed @xcite .
we focus on a soft - thresholding method in this paper . as previously mentioned , soft - thresholding is a combination of hard - thresholding and shrinkage in which both of threshold level and amount of shrinkage are simultaneously controlled by a single parameter .
the parameter is a threshold level for removing un - necessary components and is also an amount of shift by which estimators of coefficients of un - removed components are shrunk toward to zero .
if the parameter value is large then threshold level is large .
therefore , the number of un - removed components is small .
however , at the same time , the amount of shrinkage is automatically large .
this can be an excess shrinkage amount which may yields a large bias of output estimate in representing a target function .
this may cause a high prediction error at a relatively small model even when it can represent a target function ; i.e. even when it can obtain a sparse representation .
therefore , the number of un - removed components in soft - thresholding tends to be large if we choose the parameter value based on a substitution of prediction error such as sure and cross - validation error .
this is an inevitable problem of soft - thresholding , which is brought about by an introduction of a single parameter for controlling both of threshold level and amount of shrinkage simultaneously .
note that , in the implementation of thresholding methods for wavelet denoising in @xcite , thresholding is recommended to apply only to detail coefficients .
this heuristics may be actually valid to avoid the problem mentioned here .
on the other hand , in machine learning and statistics , there are several model selection methods by using regularization , in which coefficient estimators are obtained by minimizing a regularized cost that consists of error term plus regularization term .
a regularization method has a parameter that is multiplied by regularizer in the regularization term and determines a balance between error and regularization .
lasso ( least absolute shrinkage and selection operator ) is a very popular regularization method for variable selection@xcite .
it employs sum of absolute values of coefficients as a regularizer ; i.e. @xmath0 norm of a coefficient vector .
lasso is known to be useful for obtaining a sparse representation of a target function ; i.e. the number of components for representing a target function is very small . in lasso ,
extra components are automatically removed by setting their coefficients to zero .
this property is clearly understood when it applied to orthogonal regression problems . in this case
, lasso reduces to a soft - thresholding method in which a parameter of soft - thresholding is a regularization parameter divided by 2 .
hence , a sparseness obtained by lasso comes from a sof - thresholding property . and , thus , lasso encounters the above mentioned problem of soft - thresholding .
this dilemma between sparsity and prediction of lasso has already been discussed in @xcite and @xcite .
@xcite has proposed scad ( smoothly clipped absolute deviation ) penalty which is a nonlinear modification of @xmath0 penalty .
@xcite has proposed adaptive lasso that employs weighted @xmath0 penalties .
an @xmath0 penalty term is modified by different ways ( functions ) in scad and adaptive lasso while shrinkage is suppressed for large values of estimators in both methods .
this may reduce an excess shrinkage at a relatively small model .
especially , in case of orthogonal regression , weights of adaptive lasso are effective for directly and adaptively reducing a shrinkage amount that is represented as a shift in soft - thresholding . in these methods ,
cross validation is used as a model selection method for choosing parameter values such as a regularization parameter .
unfortunately , usual cross validation can not be used in orthogonal regression unless it is heuristically modified as in @xcite . in this paper
, we introduce a scaling of soft - thresholding estimators ; i.e. a soft - thresholding estimator is multiplied by a scaling parameter . unlike adaptive lasso , introduction of scaling is intended to control threshold level and amount of shrinkage independently .
it is thus a direct solution for a problem of parametrization of soft - thresholding .
if the scaling parameter value is less than one then it works as shrinkage of soft - thresholding estimator . for an orthogonal regression problem
, this is equivalent to elastic net@xcite in machine learning .
however , the scaling parameter can be larger than one by which the above mentioned excess shrinkage in soft - thresholding is expected to be relaxed ; i.e. scaling expands a shrinkage estimator obtained by soft - thresholding . especially in this paper , we propose a component - wise and data - dependent scaling method ; i.e. scaling parameter value can be different for each coefficient and is calculated from data .
we refer the proposed scaling as adaptive scaling . in this paper
, we derive a risk under adaptive scaling and construct a model selection criterion as an unbiased risk estimate .
therefore , our work establishes a denoising method in which a drawback of a naive soft - thresholding is improved by the introduction of adaptive scaling and an optimal model is automatically selected according to a derived criterion under the adaptive scaling .
in section 2 , we state a setting of orthogonal non - parametric regression that includes a problem of wavelet denoising . in this section
, we also give a naive soft - thresholding method and several related methods . in this paper ,
especially , we employ a soft - thresholding method based on lars ( least angle regression)@xcite in these methods . in lars - based soft - thresholding , a model selection problem reduces to the determination of the number of un - removed components . in section 3
, we define an adaptive scaling and derive a risk under lars - based soft - thresholding with the adaptive scaling .
we then give a model selection criterion as an unbiased estimate of the risk .
we here also consider the properties of risk curve and reveals the model selection property .
the proofs of theorems in this section are included in appendix with some lemmas . in section 4 , the proposed adaptive scaling method
is examined for toy artificial problems including applications to wavelet denoising .
section 5 is devoted to conclusions and future works .
let @xmath1 and @xmath2 be input variables and an output variable , for which we have @xmath3 i.i.d .
samples : @xmath4 , where @xmath5 .
we assume that @xmath6 , @xmath7 , where @xmath8 are i.i.d additive noise sequence according to @xmath9 ; i.e. normal distribution with mean @xmath10 and variance @xmath11 .
@xmath12 is a target function .
we assume that @xmath13 are fixed below .
we define @xmath14 , @xmath15 and @xmath16 , where @xmath17 denotes a matrix transpose .
we then have @xmath18 and @xmath19=\h$ ] , where @xmath20 denotes the expectation with respect to the joint probability distribution of @xmath21 .
let @xmath22 be a series of functions on @xmath23 .
we consider to estimate a target function by a linear combination of @xmath3 functions in this series : @xmath24 where @xmath25 is a coefficient vector .
this is a non - parametric regression problem .
we call @xmath26 a component or basis function .
we assume that there exists @xmath27 and @xmath28 such that @xmath29 for any @xmath30 when @xmath31 .
@xmath32 can be zero for some @xmath33 .
we define @xmath34 and denote the complement of @xmath35 by @xmath36 .
we call @xmath26 with @xmath37 true component or non - zero component .
we also define @xmath38 which is the number of true components or non - zero components .
we assume that @xmath39 ; i.e. true components are always included in a model .
we also assume that @xmath40 is very small compared to @xmath3 .
these two assumptions say that there exists a sparse representation of a target function in terms of a set of @xmath3 components .
let @xmath41 be an @xmath42 matrix whose @xmath43 element is @xmath44 .
we assume that the orthogonality condition : @xmath45 where @xmath46 denotes an @xmath42 identity matrix .
we thus consider a non - parametric orthogonal regression problem ; e.g. discrete fourier transform and discrete wavelet transform for typical examples .
the least squares estimator under the orthogonality condition is given by @xmath47 note that we have @xmath48 here .
since there exists a @xmath49 such that @xmath50 , @xmath51 holds by the assumption on additive noise ; i.e. multivariate normal distribution with a mean vector @xmath49 and a unit covariance matrix multiplied by @xmath52 . in other words ,
@xmath53 , @xmath54 and @xmath55 are independent .
we define @xmath56 , @xmath54 , where @xmath57 is a sign function .
we define @xmath58 as an index sequence for which @xmath59 holds .
note that we can exclude the case of ties in our probabilistic evaluations in this paper since this is guaranteed with probability one by ( [ eq : dist - of - ecv ] ) .
let @xmath60 with a parameter @xmath61 be a soft - thresholding estimator , in which @xmath62 where @xmath63 .
@xmath64 determines both of a threshold level and amount of shrinkage . under the orthogonality condition
, several sparse modeling methods can be reduced to soft - thresholding estimator . for a fixed @xmath65 , cost function of lasso
is given by @xmath66 where @xmath67 is the euclidean norm and @xmath68 ; i.e. lasso introduces an @xmath0 regularizer .
@xmath69 is a regularization parameter .
a minimizer of ( [ eq : cost - lasso ] ) under the orthogonality condition is known to be a soft - thresholding estimator with @xmath70 ; i.e. it is @xmath71 . on the other hand , for fixed @xmath65 and @xmath72 , cost function of elastic net
is given by @xmath73 thus , elastic net introduces both of an @xmath0 regularizer and @xmath74 regularizer .
as shown in @xcite , a minimizer of ( [ eq : cost - elnet ] ) under the orthonormality condition is given by @xmath75 , @xmath76 . since @xmath72 ,
the solution of elastic net is obtained by shrinking lasso estimator which is a soft - thresholding estimator .
on the other hand , lars ( least angle regression ) @xcite is a greedy iterative algorithm in which a component is appended to a model at each step .
this can be viewed as a sparse modeling method if we can find an optimal step .
for this purpose , a @xmath77 type criterion is derived under a mild condition in @xcite . as shown in @xcite and lemma 1 in @xcite , under the orthonormality condition , lars
is also reduced to soft - thresholding estimator in which the parameter value is given by @xmath78 at the @xmath79th step ; i.e. it is the @xmath80th largest absolute value among the least squares estimators .
therefore , a set of candidates of parameter values is @xmath81 in lars . by this choice of threshold level , the number of un - removed components at the @xmath79th step is equal to @xmath79 .
therefore , a model selection problem of lars - based soft - thresholding is the determination of the number of un - removed components .
we refer to lars - based soft - thresholding as lst . as a modification of lasso , adaptive
lasso@xcite introduces a weighted @xmath0 regularizer , in which a weight for the @xmath33th component is @xmath82 and a choice of @xmath83 with @xmath84 is especially considered in @xcite .
the solution of adaptive lasso under the orthonormality condition is given by @xmath85 it is regarded as a soft - thresholding estimator with a component - wise and data - dependent parameter . if @xmath86 is large then @xmath82 is small . in this case ,
threshold level and amount of shrinkage for the corresponding estimator is small .
this reduces a bias , or equivalently , an excess shrinkage of estimator especially when the estimator is actually valid ; i.e. the corresponding component is needed . in other words ,
adaptive lasso avoids an excess shrinkage on estimators of un - removed components by an adaptive manner ; i.e. by controlling a component - wise and estimator - dependent `` shift '' in soft - thresholding estimator .
this relaxes the problem of employing a single parameter value for both of threshold level and amount of shrinkage in soft - thresholding .
we can choose a small parameter value for valid components and a large value for non - essential components ; i.e. the parameters mainly work as threshold levels for removing non - essential components . in this paper , by introducing scaling for soft - thresholding estimator
, we consider to control threshold level and amount of shrinkage independently .
our approach is different from adaptive lasso while they serves the same purpose .
as seen in later sections , the advantage of employing scaling is that we can construct a model selection criterion that is required in applications .
let @xmath87 be a vector of the above mentioned lst estimators that are defined by @xmath88 where @xmath89 .
we define @xmath90 for @xmath91 . in this paper
, we consider to employ @xmath92 in which @xmath93 we call @xmath94 , @xmath54 component - wise scaling parameters .
let @xmath95 be an @xmath42 diagonal matrix whose @xmath96 element is @xmath94 .
we can write @xmath97 .
we define @xmath98 .
note that , in a matrix formulation , @xmath99 and @xmath100 are used as vertical vectors .
as in the previous discussion , if we restrict @xmath101 then the method is elastic net which yields shrinkage of soft - thresholding estimator .
therefore , introduction of scaling parameter can be viewed as an extension of elastic net .
however , we expect that scaling is used for expanding soft - thresholding estimator ; i.e. @xmath102 is desirable .
note that @xmath103 is a two stage estimate in which lst is firstly applied and then scaling is applied . scaling re - adjusts only amount of shrinkage . a risk for lst with component - wise scaling
is defined by @xmath104 where the latter definition is due to the orthogonality condition ( [ eq : ot - cond ] ) .
a naive lst is a case of @xmath105 , where @xmath106 is an @xmath3-dimensional vector of one s . for this case , we have @xmath107 as a special case of @xcite . more generally , in case of introducing a single common scaling parameter @xmath108 on all components
, @xcite has shown that @xmath109 therefore , an unbiased risk estimate is given by @xmath110 which can be used as a model selection criterion for choosing an optimal @xmath79 if we replace @xmath11 with its estimate @xmath111 . for this case ,
an optimal scaling value that minimizes the risk is given by @xmath112+\sigma^2k / n } { \e\left[\sum_{j\in\ek_k}\eb_{k , j}^2\right]}.\ ] ] in practical application , for example , @xmath113 can be an estimate of the optimal value .
we state some definitions used below .
we define @xmath114 and @xmath115 .
we define @xmath116 and @xmath117 , by which @xmath118 ; i.e. @xmath119 and @xmath120 are mutually independent .
we define @xmath121 in applying lst . correspondingly , by ( [ eq : def - eb - k - j ] ) , we define @xmath122 and @xmath123 .
we also define @xmath124 , by which @xmath125 are i.i.d . according to @xmath126 by the definition of @xmath120 . for an event @xmath127
, we denote the complement of @xmath127 by @xmath128 and indicator function of @xmath127 by @xmath129 .
we define @xmath130 and @xmath131 .
we also define @xmath132 .
we denote @xmath133 distribution with one degree of freedom by @xmath134 .
the purpose of scaling is to avoid excess shrinkage of coefficients of un - removed components .
then , it is reasonable to choose @xmath94 so as to satisfy @xmath135 .
this yields @xmath136 when @xmath137 is small .
this approximation is valid since an un - removed component may have a coefficient estimate that is sufficiently larger than an appropriate threshold level . in this paper
, we hence employ @xmath138 as empirical values , where @xmath108 is a finite constant that is defined to avoid @xmath139 when @xmath140 .
we define @xmath141 .
( [ eq : component - wise - scaling ] ) gives data - dependent and component - wise scaling value .
we refer this scaling method as adaptive scaling . by ( [ eq : component - wise - scaling ] )
, the adaptive scaling value is always larger than one .
note also that @xmath142 is valid only to @xmath143 since @xmath144 for @xmath145 .
let @xmath146 be an @xmath42 diagonal matrix whose @xmath96 element is @xmath142 .
we define a risk for our adaptive scaling estimator by @xmath147 we state three theorems whose proofs are given in appendix with some lemmas .
[ theorem : rnk - evalpha ] for @xmath148 defined in ( [ eq : component - wise - scaling ] ) , @xmath149\ ] ] holds .
[ theorem : ealpha_j - convergence ] we define @xmath150 with @xmath151 . for @xmath37 , @xmath152=0\ ] ] holds .
this implies that , for @xmath37 , @xmath153=0\ ] ] holds for any @xmath154 . on the other hand , we assume that @xmath155 . then , for @xmath156 , @xmath157=0\ ] ] holds for any @xmath154 .
[ theorem : r(1)-r(evalpha ) ] @xmath158 holds for a sufficiently large @xmath3 .
we give some remarks .
* by theorem [ theorem : rnk - evalpha ] , @xmath159 is an unbiased estimator of risk under adaptive scaling with @xmath148 defined by ( [ eq : component - wise - scaling ] ) .
therefore , this can be a model selection criterion for choosing an optimal @xmath79 if we can set an appropriate estimate of noise varinace @xmath11 in ( [ eq : ernk_evalpha ] ) . * by lemma [ lemma : p - oen*-bound ] , the probability that all of true components are un - removed is high when @xmath160 and @xmath3 is sufficiently large ; i.e. lst has a kind of consistency in selecting true components if those exist .
note that since our adaptive scaling is applied to lst estimator , this consistency result applies to adaptive scaling estimators .
* theorem [ theorem : ealpha_j - convergence ] says that , in a large sample situation , scaling values are larger than @xmath161 for components that are not true .
some of non - true components are selected when @xmath162 .
this excess expansion of coefficient estimators for non - true components may cause a high risk for @xmath162 .
therefore , @xmath163 may hold for @xmath162 even though @xmath164 holds by theorem [ theorem : r(1)-r(evalpha ) ] .
this fact seems to be disadvantage of introducing our adaptive scaling .
however , it may not be so from the viewpoint of model selection since this property allows us to identify the minimum of risk curve ; i.e. risk is small at @xmath165 while it is large when @xmath166 .
hence , nearly optimal @xmath79 is expected to be found according to a model selection criterion given by ( [ eq : ernk_evalpha ] ) . and , at such an optimal @xmath79 , a consistent choice of a set of true components and a low risk value are guaranteed by lemma [ lemma : p - oen*-bound ] and theorem [ theorem : r(1)-r(evalpha ) ] respectively .
this speculation is verified in numerical experiments in the next section .
* since the least squares estimators of coefficients of true components tend to be large , approximation in ( [ eq : approx - in - adaptive - scaling ] ) is valid for them . therefore , lst estimators for true components are nearly the least squares estimators . and , as mentioned above , true components may be consistently selected according to ( [ eq : ernk_evalpha ] ) if variance estimate is suitable .
therefore , a model estimated by our adaptive scaling scheme may be close to one estimated by a hard thresholding for which it is difficult to establish a model selection procedure .
consider a set of @xmath3 functions @xmath167 in which @xmath168 the design matrix constructed by @xmath169 satisfies the orthogonality condition of ( [ eq : ot - cond ] ) if @xmath170 for @xmath7 and @xmath3 is even .
these two conditions are satisfied in our experiment here .
we set @xmath171 and @xmath172 , by which @xmath173 ; i.e. @xmath174 , @xmath175 are true components .
we set @xmath176 for gaussian noise variance .
we also set @xmath177 and the maximum number of components that is included in a model is @xmath178 . for an artificially generated data , we apply lst , lst - ssp(lst with single scaling parameter ) and lst - as(lst with an adaptive scaling ) . we employ ( [ eq : lst - ssp - ealpha ] ) as an empirical scaling value for lst - ssp .
adaptive scaling values for lst - as are given by ( [ eq : component - wise - scaling ] ) . for each method , we calculate the approximated risk that is the mean - squared error between a true function output and estimated output on data points .
we also calculate the risk estimate ( unbiased estimate of risk ) .
it is given by ( [ eq : risk - for - lst ] ) with @xmath179 for lst , ( [ eq : risk - for - lst - ssp ] ) with @xmath180 in ( [ eq : lst - ssp - ealpha ] ) for lst - ssp and ( [ eq : ernk_evalpha ] ) with @xmath142 in ( [ eq : component - wise - scaling ] ) for lst - as .
we need to estimate noise variance in calculating a risk estimate that is employed as a model selection criterion in applications .
we here estimate it by the unbiased estimate of noise variance under a linear regression with a set of @xmath181 components that includes true components .
we repeat this procedure for @xmath182 times .
we show averages of ( approximated ) risks and risk estimates for lst - as in figure [ fig : risk - and - risk - estimate - lst - as ] .
we also show averages of risks for lst , lst - ssp and lst - as in figure [ fig : risk - for - lst - ssp - as ] . in figure
[ fig : risk - and - risk - estimate - lst - as ] , we can see that ( [ eq : ernk_evalpha ] ) is actually valid as an unbiased risk estimate under lst - as even when noise variance is replaced with its estimator . in figure
[ fig : risk - for - lst - ssp - as ] , at around a small number of components , risk of lst - as is minimized and is smaller than those of lst and lst - ssp . however , risk of lst - as tends to be larger than those of lst and lst - ssp as the number of components increases .
this is consistent with the remark on theorem [ theorem : r(1)-r(evalpha ) ] and theorem [ theorem : ealpha_j - convergence ] . in other words , an optimal number of components
can clearly be identified in risk curve of lst - as while risk curves of lst and lst - ssp are nearly flat around the minimum value in figure [ fig : risk - for - lst - ssp - as ] .
we emphasize two important points in this result .
the first one is that , as guaranteed by theorem [ theorem : r(1)-r(evalpha ) ] , risk value of lst - as is smaller than that of lst at around an optimal number of components .
the second one is that it can be found via a model selection based on risk estimate . in table
[ tbl : risk - and - un - removed - components ] , we show the averaged risk value and the average numbers of un - removed components for models that are selected by risk estimates . from table [ tbl : risk - and - un - removed - components ] , we can say that lst - as gives low risk at a sparse representation .
.average of risk and the number of un - removed components [ tbl : risk - and - un - removed - components ] selected according to risk estimate .
the standard deviation is showed in the bracket . [ cols="^,^,^",options="header " , ] discrete wavelet transform is a popular tool for analysis , de - noising and compression of signals and images ; e.g. see @xcite .
we here consider an application of lst with adaptive scaling to a problem of wavelet denoising@xcite .
let @xmath183 , @xmath184 $ ] be a signal .
@xmath3 samples of @xmath183 is denoted by @xmath185 , @xmath186 , @xmath7 .
we define @xmath187 .
we assume that @xmath188 for a natural number @xmath189 .
let @xmath190 and @xmath191 be approximation and detail coefficients at a level @xmath189 in discrete wavelet transform , where @xmath192 .
we define @xmath193 in which we set @xmath194 for @xmath195 . by setting @xmath196 ,
the decomposition algorithm with pre - determined wavelets calculates @xmath197 from @xmath198 by decreasing @xmath199 , where @xmath200 is a fixed level determined by user .
this procedure can be written by @xmath201 where @xmath202 is an @xmath42 orthonormal matrix that is determined by coefficients of scaling and wavelet function ; e.g. see @xcite . on the other hand ,
the reconstruction algorithm calculates @xmath203 from @xmath204 by increasing @xmath205 .
this can be written by @xmath206 since @xmath202 is an orthonormal matrix .
let @xmath207 be an operator on @xmath208 into @xmath208 such as a thresholding operator . in wavelet denoising ,
@xmath209 is processed by using @xmath207 and obtain @xmath210 .
we then obtain a denoised signal by @xmath211 .
note that , in applications , a simple and fast decomposition / reconstruction algorithm is used instead of the above matrix calculation ; e.g. see @xcite .
we here compare the prediction accuracy and sparseness of lst - as to those of lst , lst - ssp and also universal soft - thresholding ( ust ) in @xcite .
note that sure shrink of @xcite is almost equivalent to lst here . in an application of ust , a threshold level on the absolute values of coefficients at the @xmath200th level
is given by @xmath212 where @xmath111 is an estimate of noise variance .
in wavelet denoising , the median absolute deviation ( mad ) is a standard robust estimate of noise variance .
it is given by @xmath213 where @xmath214 , @xmath215 is the smallest scale wavelet coefficients that are heuristically known to be noise dominated components . for lst , lst - ssp and lst - as
, we also employ this estimator in a model selection criterion that is an unbiased risk estimate .
we choose `` heavisine '' and `` blocks '' given in @xcite as test signals .
the former is almost smooth and the latter has many discontinuous points .
additive noise has a normal distribution with mean @xmath10 and variance @xmath176 . as in @xcite
, signals are rescaled so that signal - to - noise ratio is @xmath216 .
the number of samples is @xmath217 .
we set @xmath218 . in @xcite , in practical application , it is employed a heuristic method which applies soft - thresholding only for detail coefficients at a determined level .
we do not obey this heuristics and apply soft - thresholding to all coefficients in orthogonal transformation for a fair comparison .
this is because the choice of a level at which thresholding applies largely depends on the performance as in @xcite and there is no systematic choice of such level .
we employ the orthogonal daubechies wavelet with @xmath219 wavelet / scaling coefficients . for given samples ,
we apply lst , lst - ssp , lst - as and ust , in which the maximum number of un - removed components is set to @xmath220 ; i.e. the maximum value of @xmath79 to be examined .
we then calculate the mean squared error between true signal outputs and estimated outputs on the sampling points as an approximation of risk . for lst , lst - ssp and lst - as , the mean squared error and risk estimate are obtained at each @xmath79 .
for ust , the number of un - removed components and risk value at a selected size are obtained .
we repeat this procedure @xmath221 times .
we show averages of ( approximated ) risk and risk estimate of lst , lst - ssp and lst - as in figure [ fig : wl - risk - curve - heavisine ] for `` heavisine '' and figure [ fig : wl - risk - curve - blocks ] for `` blocks '' respectively .
we also show box plots of risk values at the selected number of components and those of the number of un - removed components in figure [ fig : wl - box - plot - heavisine ] for `` heavisine '' and figure [ fig : wl - box - plot - blocks ] for `` blocks '' respectively . by figure
[ fig : wl - risk - curve - heavisine ] ( b ) and figure [ fig : wl - risk - curve - blocks ] ( b ) , risk estimate approximates risk well for both signals even when noise variance is estimated by mad . by figure [ fig :
wl - risk - curve - heavisine ] and figure [ fig : wl - risk - curve - blocks ] , we can expect that a model estimated by lst - as shows a low risk and high sparsity compared to lst and lst - ssp ; i.e. this result leads to the same conclusions as in the previous numerical example . by comparing figure [ fig : wl - risk - curve - heavisine ] to figure [ fig : wl - risk - curve - blocks ] , the optimal number of components for `` blocks '' is larger than for `` heavisine '' , which is due to a degree of smoothness of signals . by figure
[ fig : wl - box - plot - heavisine ] and figure [ fig : wl - box - plot - blocks ] , for both signals , lst - as outperforms the other methods in terms of prediction accuracy and sparsity , in which especially it shows a nice sparseness property .
note that the worse results of lst and ust may be improved by applying a heuristics that thresholding methods are applied only to detail coefficients at a determined level while there is no systematic choice of the appropriate level .
\(a ) averaged risk curves of lst , lst - ssp and lst - as .
\(b ) averaged risk and risk estimate of lst - as .
\(a ) risk value at the selected number of components .
\(b ) the number of un - removed components .
\(a ) averaged risk curve of lst , lst - ssp and lst - as .
\(b ) averaged risk curve and risk estimate of lst - as .
\(a ) risk value at the selected number of components .
\(b ) the number of un - removed components .
soft - thresholding is a key modeling tool in statistical signal processing such as wavelet denoising .
it has a parameter that simultaneously controls threshold level and amount of shrinkage .
this parametrization is possible to suffer from an excess shrinkage for un - removed valid components at a sparse representation ; i.e. there is a dilemma between prediction accuracy and sparsity . in this paper ,
to overcome this problem , we introduced a component - wise and data - dependent scaling method for soft - thresholding estimators in a context of non - parametric orthogonal regression including discrete wavelet transform .
we refer this method as an adaptive scaling method . here , we employed a lars - based soft - thresholding method ; i.e. a soft - thresholding method that is implemented by lars under an orthogonality condition . in lars - based soft - thresholding , a parameter value is selected by a data - dependent manner by which a model selection problem reduces to the determination of the number of un - removed components .
we firstly derived a risk given by lar - based soft - thresholding estimate with our adaptive scaling .
for determining an optimal number of un - removed components , we then gave a model selection criterion as an unbiased estimate of the risk .
we also analyzed some properties of the risk curve and found that the model selection criterion is possible to select a model with low risk and high sparsity compared to a naive soft - thresholding .
this was verified by a simple numerical experiment and an application to wavelet denoising . as a future work ,
we need more application results . in doing this ,
estimate of noise variance should be established in general applications while mad was found to be a good choice for a wavelet denoising application .
although we gave scaling values in a top down manner in this paper , we may need to test the other forms of adaptive scaling values ; e.g. scaling values which are estimates of optimal values in some senses .
moreover , development of adaptive scaling for non - orthogonal case may be expected for more general applications .
00 abramovich , f. , b. yoav , 1996 .
adaptive thresholding of wavelet coefficients .
computational statistics & data analysis 22 , 351 - 361 .
burrus , c.s . ,
gopinath , r.a . ,
guo , h. , 1998 .
introduction to wavelets and wavelet transform .
prentice hall .
donoho , d.l . , johnstone , i.m . , 1994 .
ideal spatial adaptation via wavelet shrinkage .
biometrika 81 , 425 - 455 . donoho , d.l . ,
johnstone , i.m . ,
adapting to unknown smoothness via wavelet shrinkage .
90 , 1200 - 1224 .
efron , b. , hastie , t. , johnstone , i. , tibshirani , r. , 2004 .
least angle regression .
32 , 407 - 499 .
fan , j. and li , r. , 2001 .
variable selection via nonconcave penalized likelihood and its oracle properties .
96 , 1348 - 1360 .
hagiwara , k. , 2006 .
on the expected prediction error of orthogonal regression with variable components .
ieice trans .
fundamentals e89-a , 3699 - 3709 .
hagiwara , k. , 2014 .
least angle regression in orthogonal case , in : proceedings of iconip 2014 , part ii , lncs 8835 , springer , 540 - 547 .
hagiwara , k. , 2015 . on scaling of soft - thresholding estimator ,
submitted to neurocomputing .
hurvich c.m . and
tsai c. 1998 .
a crossvalidatory aic for hard wavelet thresholding in spatially adaptive function estimation .
biometrika 85 , 701 - 710 .
knight , k. , fu , w. , 2000 .
asymptotics for lasso - type estimators .
28 , 1356 - 1378 .
leadbetter , m.r . ,
lindgren , g. , rootzn , h. , 1983 .
extremes , and related properties of random sequences and processes .
springer - verlag .
nason , g.p . , 1996 .
wavelet shrinkage using cross - validation .
j. r. statist .
b 58 , 463 - 79 .
resnick , s.i . , 1987 .
extreme values , regular variation , and point processes .
springer - verlag .
stein , c. , 1981 .
estimation of the mean of a multivariate normal distribution .
ann . stat . 9 , 1135 - 1151 .
tibshirani , r. , 1996 .
regression shrinkage and selection via the lasso .
j. r. stat .
. methodol .
58 , 267 - 288 .
zhao , p. , yu , b. , 2006 . on model selection consistency of lasso .
res . 7 , 2541 - 2563 .
zou , h. , 2006 .
the adaptive lasso and its oracle properties .
assoc . 101 , 1418 - 1492 .
zou , h. , hastie , t. , tibshirani , r. , 2007 . on the degrees of freedom of lasso .
35 , 2173 - 2192 .
zou , h. , hastie , t. , 2005 .
regularization and variable selection via the elastic net .
j. r. stat .
67 , 301 - 320 .
we here give some lemmas that is used for proving the main theorems .
let @xmath222 be random variables .
we define the @xmath223th largest value among @xmath222 by @xmath224 .
[ lemma : max - chi2-e ] let @xmath222 be i.i.d .
random variables from @xmath134 .
we define @xmath225 . then , at each fixed @xmath226 , @xmath227 hold , where @xmath228 is the @xmath79th derivative of the gamma function at @xmath229 .
( [ eq : lemma - max - chi2-e ] ) implies that @xmath230=1.\end{aligned}\ ] ] by slightly modifying example 3 , pp.72 - 73 in @xcite , we can show that @xmath231 converges to the double exponential distribution .
then , ( [ eq : lemma - max - chi2-e ] ) is a direct conclusion of proposition 2.1 ( iii ) in @xcite .
[ lemma : prob - bound - mth - largest - chi2 ] let @xmath222 be i.i.d .
random variables from @xmath134 . at each fixed @xmath223 , @xmath232&=0\\ \label{eq : prob - bound - mth - largest - chi2-upper } \lim_{n\to\infty}\p\left[x_{(m)}>2\log n)\right]&=0\end{aligned}\ ] ] hold , where @xmath233 is an arbitrary positive constant .
we denote the probability distribution function of @xmath134 by @xmath234 .
the probability density function of @xmath134 is given by @xmath235 .
we have @xmath236 .
thus , we have @xmath237 as @xmath238 by applying @xmath239 and lhospital s rule . therefore , for a @xmath240 random variable @xmath241 , @xmath242\sim 2f_1(x)\ ] ] holds for a sufficiently large @xmath243 . by ( [ eq : lemma - chi2-p - bound-1 ] )
, we obtain @xmath244 & \le\sum_{i=1}^n\p\left[x_i>2\log n\right]\notag\\ & \sim 2nf_1\left(2\log n\right)\notag\\ & = \frac{1}{\pi}\frac{1}{\sqrt{\log n}}\to 0~(n\to\infty).\end{aligned}\ ] ] since @xmath245 for any @xmath223 , we have ( [ eq : prob - bound - mth - largest - chi2-upper ] ) . on the other hand , by ( [ eq : lemma - chi2-p - bound-1 ] ) , we have @xmath246 for a sufficiently large @xmath3 . since this goes to @xmath247 , we obtain ( [ eq : prob - bound - mth - largest - chi2-lower ] ) by theorem 2.2.1 in @xcite .
[ lemma : cj > maxci ] for any @xmath37 and any @xmath248 , @xmath249\le 2\pi^{-1/2}\rho^{-1/2}n^{-\rho}\ ] ] holds for a sufficiently large @xmath3 .
we define @xmath250 .
we obtain @xmath251 & \ge\p\left[\left[\tc_j^2>\tau_{n,\rho}\right ] \bigcap\left[\tau_{n,\rho}>\max_{i\in \ok^*}\tc_i^2\right]\right]\notag\\ & = 1-\p\left[\left[\tc_j^2\le\tau_{n,\rho}\right ] \bigcup\left[\tau_{n,\rho}\le\max_{i\in \ok^*}\tc_i^2\right]\right]\notag\\ & \ge 1-\p\left[\tc_j^2\le\tau_{n,\rho}\right]- \p\left[\max_{i\in \ok^*}\tc_i^2\ge \tau_{n,\rho}\right].\end{aligned}\ ] ] by the definition of @xmath252 , we have @xmath253&=\p\left[|\tc_j|\le\sqrt{\tau_{n,\rho}}\right]\notag\\ & = \p\left[|\sqrt{n}\beta_j/\sigma+\oc_j|\le\sqrt{\tau_{n,\rho}}\right]\notag\\ & \le\p\left[\sqrt{n}|\beta_j|/\sigma-|\oc_j|\le \sqrt{\tau_{n,\rho}}\right]\notag\\ & = \p\left[|\oc_j|\ge\sqrt{n}|\beta_j|/\sigma-\sqrt{\tau_{n,\rho}}\right]\notag\\ & \le\p\left[|\oc_j|\ge\sqrt{\tau_{n,\rho}}\right]\notag\\ & = \p\left[\oc_j^2\ge\tau_{n,\rho}\right]\end{aligned}\ ] ] for a sufficiently large @xmath3 .
note that this evaluation is not tight but is enough in this paper . since @xmath254 by the definition of @xmath252 , by ( [ eq : lemma - chi2-p - bound-1 ] ) and ( [ eq : cj > maxci-1 ] )
, we have @xmath255\le \pi^{-1/2}\rho^{-1/2}n^{-\rho}\ ] ] for a sufficiently large @xmath3 . on the other hand
, @xmath256 holds for @xmath257 since @xmath258 holds for @xmath257 . by ( [ eq : lemma - chi2-p - bound-1 ] )
, we thus have @xmath259 & \le\sum_{j\in\ok^*}\p[\tc_i^2\ge\tau_{n,\rho}]\notag\\ & \sim ( n - k^*)\pi^{-1/2}(\rho+1)^{-1/2}n^{-(\rho+1)}\notag\\ & \le\pi^{-1/2}\rho^{-1/2}n^{-\rho}\end{aligned}\ ] ] for a sufficiently large @xmath3 . by ( [ eq : cj > maxci-0 ] ) , ( [ eq : cj > maxci-2 ] ) and ( [ eq : cj > maxci-3 ] ) , we obtain ( [ eq : cj > maxci ] ) as desired .
[ lemma : p - oen*-bound ] @xmath260\le k^*\pi^{-1/2}\rho^{-1/2}n^{-\rho}\ ] ] holds for any @xmath248 and a sufficiently large @xmath3 .
if @xmath261 does not occur then there exist @xmath262 such that @xmath263 .
this implies that there exist @xmath37 and @xmath264 that satisfy @xmath265 .
therefore , we have @xmath266 . by lemma
[ lemma : cj > maxci ] , we then obtain ( [ eq : p - oen*-bound ] ) .
[ lemma : e - cp1 ^ 2-ioe * ] @xmath267 holds for a fixed @xmath268 .
we define @xmath269 and @xmath270 .
we also define an event @xmath271 . by the cauchy - schwarz inequality
, we have @xmath272\notag\\ & \le\e[(\oc_{p_1}+(\beta_{p_1}/\sigma)\sqrt{n})^{2m}i_{\oe_n^*}]\notag\\ & \le\e[(\oc+(\obeta/\sigma)\sqrt{n})^{2m}i_{\oe_n^*}]\notag\\ & \le\e[(\oc+(\obeta/\sigma)\sqrt{n})^{2m}i_fi_{\oe_n^ * } ] + \e[(\oc+(\obeta/\sigma)\sqrt{n})^{2m}i_{\of}i_{\oe_n^*}]\notag\\ & \le 2^{2m}\e[\oc^{2m}i_fi_{\oe_n^ * } ] + 2^{2m}(\obeta/\sigma)^{2m}n^m\e[i_{\of}i_{\oe_n^*}]\notag\\ & \le 2^{2m}\e[(\oc^2)^{m}i_{\oe_n^ * } ] + 2^{2m}(\obeta/\sigma)^{2m}n^m\e[i_{\oe_n^*}]\notag\\ & \le 2^{2m}\sqrt{\e[(\oc^2)^{2m}]}\sqrt{\p[\oe_n^ * ] } + 2^{2m}(\obeta/\sigma)^{2m}n^m\p[\oe_n^*].\end{aligned}\ ] ] by lemma [ lemma : p - oen*-bound ] with @xmath273 , the second term of ( [ eq : e - cp1 ^ 2-ioe*-2 ] ) goes to zero as @xmath274 . since @xmath275 is the largest value among i.i.d . @xmath134 sequence with size @xmath3 , the first term of ( [ eq : e - cp1 ^ 2-ioe*-2 ] ) goes to zero as @xmath274 by lemma [ lemma : max - chi2-e ] and lemma [ lemma : p - oen*-bound ] with the above choice of @xmath276 .
[ lemma : e - ttheta_k^2-bound ] if @xmath160 then @xmath277\le ( 2\log n)^m\ ] ] holds for a fixed @xmath268 and sufficiently large @xmath3 . we can write @xmath278=\e\left[\ttheta_k^{2m}i_{\oe_n^*}\right ] + \e\left[\ttheta_k^{2m}i_{e_n^*}\right].\ ] ] by lemma [ lemma : e - cp1 ^
2-ioe * ] and the definition of @xmath279 , @xmath280\le\e\left[\tc_{p_1}^{2m}i_{\oe_n^*}\right]\to 0~(n\to\infty).\ ] ] we define @xmath281 .
if @xmath261 occurs then @xmath282 and @xmath283 is the largest value among i.i.d . @xmath240 random sequence with length @xmath284 .
therefore , by lemma [ lemma : max - chi2-e ] , @xmath285}{(2\log n)^m } \le\frac{\e\left[\tc^{2m}\right]}{(2\log n)^m}\to 1~(n\to\infty).\ ] ]
we give the proofs of the main theorems below . for an @xmath148
, the risk is reformulated as @xmath286 where we used ( [ eq : dist - of - ecv ] ) at the third line and the orthogonality condition at the last line .
the last term is often called the degree of freedom ; see e.g. @xcite .
let @xmath287 be an @xmath288-dimensional vector that is constructed by removing @xmath289 from @xmath290 .
we define @xmath291 .
although @xmath292 is a function of @xmath290 , we regard this as a function @xmath289 under a fixed @xmath287 and denote it by @xmath293 .
let @xmath294 be the @xmath79th largest value in @xmath295 . by ( [ eq : component - wise - scaling ] ) , we have @xmath296 note here that @xmath293 is well - defined even when @xmath297 under the definition of @xmath142 in ( [ eq : component - wise - scaling ] ) .
this is lipschitz continuous as a function of @xmath289 when @xmath287 is fixed .
it is thus absolutely continuous . on the other hand , we denote expectation with respect to @xmath290 by @xmath298 .
we have @xmath299 and @xmath300 , where @xmath301 denotes the determinant of a matrix .
therefore , @xmath20 is always replaced with @xmath298 by change of variables .
we also denote a conditional expectation with respect to @xmath289 given @xmath287 by @xmath302 .
we define @xmath303}(\tc_j|\tcv_{-j})$ ] by @xmath303}(\tc_j|\tcv_{-j})=1 $ ] when @xmath304 $ ] and @xmath10 otherwise under a fixed @xmath287 .
then , by applying this change of variables and stein s lemma@xcite under the above absolutely continuity , we obtain @xmath305\notag\\ & = \sum_{j=1}^n\e_{\tcv}\left[d_j(\tcv)(\tc_j - c_j)\right]\notag\\ & = \sum_{j=1}^n\e_{\tcv_{-j}}\e_{\tc_j|\tcv_{-j } } \left[d_j(\tc_j|\tcv_{-j})(\tc_j - c_j)\right]\notag\\ & = \sum_{j=1}^n\e_{\tcv_{-j}}\e_{\tc_j|\tcv_{-j } } \left[\frac{\partial d_j(\tc_j|\tcv_{-j})}{\partial \tc_j}\right]\notag\\ & = \e_{\tcv}\left[\sum_{j=1}^n ( \theta_j^2/\tc_j^2)(1-i_{[-\theta_j,\theta_j]}(\tc_j|\tcv_{-j}))\right ] -\e_{\tcv}\left [ \sum_{j=1}^ni_{[-\theta_j,\theta_j]}(\tc_j|\tcv_{-j})\right]\notag\\ & = \e\left[\sum_{j\in\ek_k}(\ealpha_j-1)^2\right]-(n - k),\end{aligned}\ ] ] where the last line is obtained by the definition of @xmath294 and @xmath142 .
( [ eq : risk ] ) and ( [ eq : a - tbv - tcv - tcv - c ] ) yield ( [ eq : theorem - rnk - evalpha ] ) .
we show that @xmath306\to 0~(n\to\infty)\end{aligned}\ ] ] for @xmath37 .
we define @xmath307 for which @xmath308=0 $ ] . by the definition of @xmath142 in ( [ eq : component - wise - scaling ] )
, we then have @xmath309\notag\\ & = \p\left[\left\{\ealpha_j>1+\epsilon_{j , n}\right\}\bigcap \oe_0\right]\notag\\ & = \p\left[\left\{\frac{\ttheta_k}{|\tc_j|}>\epsilon_{j , n}\right\}\bigcap \oe_0\right]\notag\\ & \le\p\left[\frac{\ttheta_k}{\sqrt{2\log n } } -\frac{|\tc_j|}{(|\beta_j|+\delta)\sqrt{n}}>0\right]\notag\\ & \le \p\left[\ttheta_k>\sqrt{2\log n}\right ] + \p\left[|\tc_j|<(|\beta_j|+\delta)\sqrt{n}\right].\end{aligned}\ ] ] for the first term of ( [ eq : ealpha_j - bound - in*-2 ] ) , we have @xmath310\notag\\ & = \p\left[\ttheta_k^2>2\log n\right]\notag\\ & = \p\left[\ttheta_k^2>2\log n|e_n^*\right]\p\left[e_n^*\right]+ \p\left[\ttheta_k^2>2\log n|\oe_n^*\right]\p\left[\oe_n^*\right]\notag\\ & \le\p\left[\ttheta_k^2>2\log n|e_n^*\right]+ \p\left[\oe_n^*\right].\end{aligned}\ ] ] the second term of ( [ eq : ealpha_j - bound - in*-3 ] ) goes to zero as @xmath274 by lemma [ lemma : p - oen*-bound ] . if @xmath261 occurs then @xmath311 is the @xmath312th largest value among i.i.d . @xmath240
random sequence with size @xmath313 .
therefore , by ( [ eq : prob - bound - mth - largest - chi2-upper ] ) in lemma [ lemma : prob - bound - mth - largest - chi2 ] , the first term of ( [ eq : ealpha_j - bound - in*-3 ] ) goes to zero as @xmath274 .
thus , the first term of ( [ eq : ealpha_j - bound - in*-2 ] ) goes to zero as @xmath274 . recall that @xmath314 for @xmath37 , where @xmath315 .
then , for the second term of ( [ eq : ealpha_j - bound - in*-2 ] ) , we obtain @xmath316 & = \p\left[|\sqrt{n}\beta_j+\oc_j|<(|\beta_j|+\delta)\sqrt{n}\right]\notag\\ & \le\p\left[\sqrt{n}|\beta_j|-|\oc_j|<(|\beta_j|+\delta)\sqrt{n}\right]\notag\\ & = \p\left[|\oc_j|>\delta\sqrt{n}\right]\to 0~(n\to\infty).\end{aligned}\ ] ] since @xmath317 holds , we obtain ( [ eq : ealpha_j - lower - bound - k*-0 ] ) as desired .
on the other hand , we consider ( [ eq : ealpha_j - lower - bound - ok * ] ) . for any @xmath318
, we have @xmath319 & = \p\left[\{\ealpha_j\le 2-\epsilon\}\bigcap\oe_0\right]\notag\\ & \le\p\left[\ttheta_k\le ( 1-\epsilon)|\tc_j|\right]\notag\\ & \le\p\left[\ttheta_k\le\delta_n\right ] + \p\left[(1-\epsilon)|\tc_j|>\delta_n\right].\end{aligned}\ ] ] for the first term of ( [ eq : ealpha_j - bound - notin*-1 ] ) , we have @xmath320 & = \p\left[\ttheta_k^2\le\delta_n^2\right]\notag\\ & = \p\left[\ttheta_k^2\le\delta_n^2|e_n^*\right]\p[e_n^*]+ \p\left[\ttheta_k\le\delta_n^2|\oe_n^*\right]\p[\oe_n^*]\notag\\ & \le\p\left[\ttheta_k^2\le\delta_n^2|e_n^*\right]+\p[\oe_n^*].\end{aligned}\ ] ] by lemma [ lemma : p - oen*-bound ] , the second term of ( [ eq : ealpha_j - bound - notin*-2 ] ) goes to zero as @xmath274 .
we set @xmath321 .
if @xmath261 occurs then @xmath311 is the @xmath312th largest value among i.i.d . @xmath240
random sequence with size @xmath313 .
therefore , by ( [ eq : prob - bound - mth - largest - chi2-lower ] ) in lemma [ lemma : prob - bound - mth - largest - chi2 ] and the choice of @xmath318 , the first term of ( [ eq : ealpha_j - bound - notin*-2 ] ) goes to zero as @xmath274 .
we define @xmath281 . for the second term of ( [ eq : ealpha_j - bound - notin*-1 ] ) , we have @xmath322\le\p\left[\tc^2>2\log n\right]\end{aligned}\ ] ] since @xmath37 .
here , @xmath283 is the largest value among i.i.d . @xmath240
random sequence with size @xmath313 by the definitions of @xmath323 and @xmath35 .
hence , by ( [ eq : prob - bound - mth - largest - chi2-upper ] ) in lemma [ lemma : prob - bound - mth - largest - chi2 ] , ( [ eq : ealpha_j - bound - notin*-3 ] ) goes to zero as @xmath274 . by
( [ eq : risk - for - lst ] ) and ( [ eq : theorem - rnk - evalpha ] ) , we have @xmath324 -\e\left[(\ealpha_{p_j}-1)^2\ttheta_{k^*}^2\right ] -2\e\left[(\ealpha_{p_j}-1)^2\right]\right)\notag\\ & = \frac{\sigma^22\log n}{n } \sum_{j=1}^{k^*}\left(\frac{\e\left[\ttheta_{k^*}^2\right]}{2\log n } -\frac{\e\left[(\ealpha_{p_j}-1)^2\ttheta_{k^*}^2\right]}{2\log n } -2\frac{\e\left[(\ealpha_{p_j}-1)^2\right]}{2\log n}\right)\end{aligned}\ ] ] through a simple calculation .
we evaluate the three terms in the sum of ( [ eq : theorem - r(1)-r(evalpha)-1 ] ) .
we first have @xmath325/(2\log n)=1\ ] ] by lemma [ lemma : e - ttheta_k^2-bound ] .
hence , the proof is completed by showing that the second and third terms of ( [ eq : theorem - r(1)-r(evalpha)-1 ] ) goes to zero as @xmath274 .
we define @xmath326 and @xmath327 , where @xmath328 is defined in ( [ eq : def - epsilon_n ] ) .
we have @xmath329 for @xmath330 by the definition of @xmath331 . and , if @xmath261 occurs then @xmath332 for any @xmath333 .
we then obtain @xmath334 & = \e\left[(\ealpha_{p_j}-1)^2i_{g_j\bigcap e_n^*}\right ] + \e\left[(\ealpha_{p_j}-1)^2i_{\og_j\bigcup\oe_n^*}\right]\notag\\ & \le\e\left[i_{g_j\bigcap\oe_n^*}\right]+\epsilon_n^2+\p[\oe_n^*]\notag\\ & \le\sum_{l\in k^*}\p\left[(\ealpha_l-1)^2>\epsilon_n^2\right]+\epsilon_n^2+\p[\oe_n^*].\end{aligned}\ ] ] ( [ eq : theorem - r(1)-r(evalpha)-3 ] ) goes to zero as @xmath274 by ( [ eq : ealpha_j - lower - bound - k*-0 ] ) in theorem [ theorem : ealpha_j - convergence ] , the definition of @xmath335 and lemma [ lemma : p - oen*-bound ] .
we also have @xmath336\notag\\ & = \e\left[(\ealpha_{p_j}-1)^2\ttheta_{k^*}^2i_{g_j\bigcap e_n^*}\right ] + \e\left[(\ealpha_{p_j}-1)^2\ttheta_{k^*}^2i_{\og_j\bigcup\oe_n^*}\right]\notag\\ & \le\e\left[\ttheta_{k^*}^2i_{g_j}i_{e_n^*}\right ] + \epsilon_n^2\e[\ttheta_{k^*}^2]+\e[\ttheta_{k^*}^2i_{\oe_n^*}].\end{aligned}\ ] ] for the first term of ( [ eq : theorem - r(1)-r(evalpha)-4 ] ) , by the cauchy - schwarz inequality , we have @xmath337}{2\log n } & \le\frac{\sqrt{\e\left[\ttheta_{k^*}^4\right]}}{2\log n } \sqrt{\e\left[i_{g_j}i_{e_n^*}\right]}\notag\\ & \le\frac{\sqrt{\e\left[\ttheta_{k^*}^4\right]}}{2\log n } \sqrt{\sum_{l\in k^*}\p\left[(\ealpha_l-1)^2>\epsilon_n^2\right]}.\end{aligned}\ ] ] ( [ eq : theorem - r(1)-r(evalpha)-5 ] ) goes to zero as @xmath274 by lemma [ lemma : e - ttheta_k^2-bound ] and ( [ eq : ealpha_j - lower - bound - k*-0 ] ) in theorem [ theorem : ealpha_j - convergence ] .
the second term of ( [ eq : theorem - r(1)-r(evalpha)-4 ] ) goes to zero as @xmath274 by lemma [ lemma : e - ttheta_k^2-bound ] and the definition of @xmath335 . by the cauchy - schwarz inequality
, the third term of ( [ eq : theorem - r(1)-r(evalpha)-4 ] ) is bounded above by @xmath338}\sqrt{\p[\oe_n^*]}$ ] .
this goes to zero as @xmath274 by lemma [ lemma : e - ttheta_k^2-bound ] and lemma [ lemma : p - oen*-bound ] .
we thus obtain ( [ eq : theorem - r(1)-r(evalpha ) ] ) as desired . | soft - thresholding is a sparse modeling method that is typically applied to wavelet denoising in statistical signal processing and analysis .
it has a single parameter that controls a threshold level on wavelet coefficients and , simultaneously , amount of shrinkage for coefficients of un - removed components .
this parametrization is possible to cause excess shrinkage , thus , estimation bias at a sparse representation ; i.e. there is a dilemma between sparsity and prediction accuracy . to relax this problem
, we considered to introduce positive scaling on soft - thresholding estimator , by which threshold level and amount of shrinkage are independently controlled .
especially , in this paper , we proposed component - wise and data - dependent scaling in a setting of non - parametric orthogonal regression problem including discrete wavelet transform .
we call our scaling method adaptive scaling .
we here employed soft - thresholding method based on lars(least angle regression ) , by which the model selection problem reduces to the determination of the number of un - removed components .
we derived a risk under lars - based soft - thresholding with the proposed adaptive scaling and established a model selection criterion as an unbiased estimate of the risk .
we also analyzed some properties of the risk curve and found that the model selection criterion is possible to select a model with low risk and high sparsity compared to a naive soft - thresholding method .
this theoretical speculation was verified by a simple numerical experiment and an application to wavelet denoising .
non - parametric orthogonal regression , soft - thresholding , shrinkage , adaptive scaling , wavelet denoising |
the luminosity distribution of ellipticals and bulges of s0 galaxies has been modelled for long using the @xmath0 formula ( de vaucouleurs 1948 ) , which is characterized by just two scale parameters , the effective radius @xmath1 and the effective surface brightness @xmath2 of the isophote encircling half of the total light .
more recently showed that a generalized de vaucouleurs formula , @xmath3 , obtained by adding a free exponent @xmath4 @xcite , not only gives a better fit to the light profiles ( which is expected in view of the larger number of parameters ) , but it provides also an interesting correlation between shape ( @xmath4 ) and total luminosity ( @xmath5 ) , which in turn suggests that the family of elliptical galaxies is not homologous .
the main criticism moved to this result is that the observed trend of @xmath4 with @xmath6 may be due to the presence of hidden disks embedded in elliptical galaxies : the larger is this disk , the smaller comes the value of @xmath4 . caon et al .
( 1993 ) had obtained their result by fitting the main - axes light profiles using just the @xmath3 law . in view of the above criticism
, we decided to pinpoint the question by adopting a 2d fitting procedure allowing the presence of the disk ; 2d is needed in order to strongly couple and exploit the different projection properties of components such as disk and bulge which are expected to have quite different intrinsic thickness .
a non - linear least square fit minimizing the @xmath7 has been applied to the same volume - limited sample of caon et al .
( 1993 ) : it contains @xmath8 early - type galaxies belonging to either the virgo or the fornax cluster .
this sample is @xmath9 complete down to @xmath10 mag , equivalent to @xmath11 if we assume a distance modulus @xmath12 for both clusters . in the following we outline the adopted methodology and summarize the results of this work ( donofrio et al .
, in preparation ) .
our 2d fitting models are based on the superposition of up to two components , each one characterized by concentric and coaxial elliptical isophotes with constant flattening .
one such component may be thought as the projection of a spheroid with finite intrinsic thichness , thus mimicking a bulge . for this component
only we allow a moderate degree of boxiness / diskiness .
the other component is thought to be a disk .
* model 1 * is the 2d extension of the one used in their work by caon et al .
it consists of a spheroidal component only whose projected light distribution follows the generalized de vaucouleurs law : @xmath13\ ] ] with @xmath14 and @xmath15^{1/c}$ ] ; @xmath16 is the parameter accounting for some boxiness / diskiness ( @xmath17 for boxy isophotes and @xmath18 for disky isophotes ) .
is made by the superposition of a bulge - like component with a @xmath0 light distribution ( same as eq .
1 , with @xmath19 ) to an exponential disk represented as : @xmath20 with @xmath21^{1/2}$ ] .
is a generalization of model 2 ; the light distribution of the bulge follows a @xmath3 law , but the disk remains exponential . in summary ,
the structural parameters involved in these models are : the effective surface brightness @xmath2 and the effective radius @xmath1 of the spheroid , the free exponent @xmath4 when considered , the central surface brightness @xmath22 and the scale length of the disk @xmath23 if included , the apparent axial ratios @xmath24 for the bulge and the disk components , and the boxiness / diskiness parameter @xmath16 .
a number of simulations on artificial galaxies have been performed in order to test the tool and to unveil the systematic effects that influence the model parameters ; in particular the seeing , which produces strongly biased results even after proper convolution of the model , and can be taken care only by masking out the central area of the galaxies .
the simulations also provided a more correct estimate of the errors of the model parameters .
the three adopted models do not provide the same accuracy in reproducing the light distribution of early - type galaxies . by examining the reduced @xmath7 , the 2d residuals , and by comparing the geometry of the model ( ellipticity and @xmath25 profiles ) with the real galaxies , we concluded that the @xmath26 model has to be preferred to either the simple @xmath3 and the @xmath27 models .
model 1 is comparable to model 3 in reproducing dwarf elliptical galaxies ( @xmath28 ) .
model 2 often produces extended luminous halos not observed in the real objects .
it is evident that the reduced @xmath7 by itself does not guarantee the physical significance of the best - fit solution for models with different numbers of parameters .
we performed also an f - test of the best - fit @xmath7 distributions , in order to verify whether the variances are different even from a statistical point of view .
actually , however , we may try to attach a physical meaning to the most complex and best - fitting model , that of @xmath29 , only after analyzing and falsifying the correlations among the different parameters . in general the best fitting model
does not work in the inner seeing - dominated region ( @xmath30 ) and in the outer parts , where an extended luminous ` halo ' is often found , but it is good in the intermediate regions for a large range of @xmath31 .
we confirm the correlation between the effective radius @xmath1 and the effective surface brightness @xmath2 .
the @xmath32 relation ( fig .
1 ) shows a family of large bulges with @xmath33 kpc and with low surface brightness @xmath2 , as found by .
the separation between _ ordinary _ and _ bright _ galaxies is well visible , even if the gap between the two families is less clear .
the bulges of spirals all belong to the ordinary family ( @xmath34 kpc ) .
2 shows the @xmath35 relation .
the open circles represent the old distribution , found by for the same sample of galaxies by fitting the equivalent light profiles .
the dashed line marks the value @xmath19 of the de vaucouleurs law . by comparing the old relation with the new one , we observe that there is not a variation within the errors of the parameters .
the @xmath4 parameter spans a wide range of values ( @xmath36 ) : low values are typical of exponential bulges ( already known to exist in spiral galaxies ) , while high values are found in bright and large galaxies .
this result confirms that the different values of @xmath4 are not due to the presence of a disk , but that there is a clear trend in the values of @xmath4 with the galaxy size ( and therefore with the total luminosity ) for the spheroidal components .
it follows that the non - homology is likely a characteristic of hot stellar systems .
our best fit solutions tell us that in quite a few ( @xmath37 ) elliptical galaxies there is a central exponential component . this is not surprising for the disky ellipticals , in which an ` embedded disk ' , generally smaller and brighter than normal disks in s0 s and spirals ,
gives a characteristic disky shape to the isophotes ( , ) .
however , we found an inner exponential component ( with bright @xmath38 and small @xmath23 ) also in the bright , boxy ellipticals .
we verified _ a posteriori _ that such small and bright ` disks ' are a first order approximation of the inner cores of elliptical galaxies , well described by the nuker law ( , and ) .
fig.3 shows the central surface brightness @xmath39 ( corrected for inclination assuming a zero - thickness disk ) , plotted versus the scale length @xmath40 of the detected ` disks ' .
the distribution of values is large : it goes from small and bright ` disks ' ( @xmath41 , @xmath42 kpc ) , many of them hosted by luminous e galaxies , to less luminous and great disks , preferably found in s0 and low luminosity ellipticals .
the disk parameters of normal spirals ( from ) are also included in the diagram for comparison .
it seems that the whole distribution follows approximately the lines of constant disk luminosities ( dashed lines ) .
the spiral disks of de jong s sample are brighter than the early - type disks of our sample and extend towards fainter @xmath39 and longer scale length @xmath23 .
the distribution of @xmath39 does not peak around the freeman value ( solid line ) ( , ) . in fig .
4 we plotted the effective radius @xmath43 of the bulge component versus the scale length @xmath40 .
it is apparent there is a correlation between these parameters , but with two separated trends , one for the bulge dominated systems ( upper part of the diagram ) , and another for the disk dominated galaxies ( bottom part of the diagram ) .
it seems also that galaxies with very small bulges and disks ( @xmath44 kpc ) as well as systems with large @xmath1 and @xmath23 ( @xmath45kpc ) do not exist .
we note that the lack of objects with large size can not be a bias , since our sample is complete in luminosity .
by applying a full 2d fit of the light distribution of a volume - limited sample of early - type galaxies we obtained the following results : 1 . the luminosity distribution of bright e galaxies is well represented by the superposition of an inner exponential component and a large outer bulge .
this inner structure is characterized by an hight central surface brightness and a small scale length .
it is likely that the detected component is the core of the galaxy .
it seems important , however , that this component follows the same photometric properties of normal disks , although shifted in zero point .
2 . the distribution of @xmath39 for all detected ` disks ' is a function of the scale length @xmath23 .
the slope of the relation is that of constant disk luminosities .
the central surface brightness @xmath22 spans a wide range of values , in contrast with the claimed constant value .
the effective radii of the bulges and the scale lengths of the ` disks ' are correlated : this suggests a possible coupling during the galaxy formation between the two components , and that the angular momentum should be an important parameter for the hubble sequence .
the effective radius @xmath1 correlates with @xmath2 and with the exponent @xmath4 of the generalized de vaucouleurs law .
the existence of two families of spheroids is confirmed as well as the increasing of @xmath4 for large luminous bulges .
this suggests that elliptical galaxies are not homologous systems . | we outline the results of a two - dimensional ( 2d ) fit of the light distribution of early - type galaxies belonging to a complete volume - limited sample , and briefly discuss the significant correlations among the structural parameters . in particular we reconfirm that the lack of structural homology is likely a characteristic of hot stellar systems . |
due to its vector - space preserving property , random linear network coding @xcite can be viewed as transmitting subspaces over an operator channel @xcite . as such , error control for random linear network coding
can be modeled as a coding problem , where codewords are subspaces and the distance is measured by either the subspace distance @xcite or the injection metric @xcite .
codes in the projective space , referred to as subspace codes henceforth , and codes in the grassmannian , referred to as constant - dimension codes ( cdcs ) henceforth , have been both investigated for error control in random linear network coding .
using cdcs is sometimes advantageous since the fixed dimension of cdcs simplifies the network protocol somewhat @xcite .
the construction and properties of cdcs thus have attracted a lot of attention .
different constructions of cdcs have been proposed @xcite .
bounds on cdcs based on packing properties are investigated ( see , for example , @xcite ) , and the covering properties of cdcs are investigated in @xcite .
the construction and properties of subspace codes have received less consideration , and previous works on subspace codes ( see , for example , @xcite ) have focused on the packing properties . in @xcite , bounds on the maximum cardinality of a subspace code with the subspace metric , notably the counterpart of the gilbert bound , are derived .
another bound relating the maximum cardinality of cdcs to that of subspace codes is given in @xcite .
bounds and constructions of subspace codes are also investigated in @xcite . despite the previous works , two significant problems remain open .
first , despite the aforementioned advantage of cdcs , what is the rate loss of cdcs as opposed to subspace codes of the same minimum distance and hence error correction capability ?
since random linear network coding achieves multicast capacity with probability exponentially approaching 1 with the length of the code @xcite , the asymptotic rates of subspace codes and asymptotic rate loss of cdcs are both significant .
the second problem involves the two metrics that have been introduced for subspace codes : what is the difference between the two metrics proposed for subspace codes and cdcs beyond those discussed in @xcite ? note that the two questions are somewhat related , since the first question is applicable for both metrics .
the answers to these questions are significant to the code design for error control in random linear network coding . aiming to answer these two questions , our work in this paper focuses on the packing and covering properties of subspace codes .
packing and covering properties not only are interesting in their own right as fundamental geometric properties , also are significant for various practical purposes . first , our work is motivated by their significance to design and decoding of subspace codes .
since a code can be viewed as a packing of its ambient space , the significance of packing properties is clear .
in contrast , the importance of covering properties is more subtle and deserves more explanation .
for example , a class of nearly optimal cdcs , referred to as liftings of rank metric codes , have covering radii no less than their minimum distance and thus are not optimal cdcs @xcite .
this example shows how a covering property is relevant to the design of subspace codes .
the covering radius also characterizes the decoding performance of a code , since it is the maximum weight of a decodable error by minimum distance decoding @xcite and also has applications to decoding with erasures @xcite .
second , covering properties are also important for other reasons . for example , covering properties are important for the security of keystreams against cryptanalytic attacks @xcite .
our main contributions of this paper are that for both metrics , we first determine some fundamental geometric properties of the projective space , and then use these properties to derive bounds and to determine the asymptotic rates of subspace codes based on packing and covering .
our results provide some answers to both open problems above .
first , our results show that for both metrics optimal packing cdcs are optimal packing subspace codes up to a scalar if and only if their dimension is half of their length ( up to rounding ) , which implies that in this case cdcs suffer from a limited rate loss as opposed to subspace codes with the same minimum distance .
furthermore , when the asymptotic rate of subspace codes is fixed , the relative subspace distance of optimal subspace codes is twice as much as the relative injection distance .
second , our results illustrate the difference between the two metrics from a geometric perspective . above all
, the projective space has different geometric properties under the two metrics .
the different geometric properties further result in different asymptotic rates of covering codes with the two metrics .
with the injection metric , optimal covering cdcs can be used to construct asymptotically optimal covering subspace codes .
however , with the subspace metric , this does not hold . to the best of our knowledge ,
our results on the geometric properties of the projective space are novel , and our investigation of covering properties of subspace codes is the first one in the literature . note that our investigation of covering properties differs from the study in @xcite : while how cdcs cover the grassmannian was investigated in @xcite , we consider how subspace codes cover the whole projective space in this paper .
our investigation of packing properties leads to tighter bounds than the gilbert bound in @xcite , and our relation between the optimal cardinalities of subspace codes and cdcs is also more precise than that in @xcite .
our asymptotic rates based on packing properties also appear to be novel .
the rest of the paper is organized as follows .
section [ sec : preliminaries ] reviews necessary background on subspace codes , cdcs , and related concepts . in section [ sec : subspace ] , we investigate the packing and covering properties of subspace codes with the subspace metric . in section [ sec : injection ] , we study the packing and covering properties of subspace codes with the injection metric .
finally , section [ sec : conclusion ] summarizes our results and provides future work directions .
we refer to the set of all subspaces of @xmath0 with dimension @xmath1 as the grassmannian of dimension @xmath1 and denote it as @xmath2 ; we refer to @xmath3 as the projective space .
we have @xmath4 , where @xmath5 is the gaussian binomial @xcite .
a very instrumental result @xcite about the gaussian binomial is that for all @xmath6 : @xmath7 where @xmath8 represents the ratio of non - singular matrices in @xmath9 as @xmath10 tends to infinity . by definition , @xmath11 , where @xmath12 is the euler function .
furthermore , by the pentagonal number theorem , @xmath13 @xcite . finally , we also have @xmath14 , where @xmath15 is the partition number of @xmath16 @xcite . for @xmath17 , both the _ subspace metric _
* ( 3 ) ) @xmath18 and _ injection metric _
1 ) @xmath19 are metrics over @xmath20 . for all @xmath17 , @xmath21 and @xmath22 if and only if @xmath23 , and @xmath24 if and
only if @xmath25 or @xmath26 . a _ subspace code _ is a nonempty subset of @xmath20 .
the minimum subspace ( respectively , injection ) distance of a subspace code is the minimum subspace ( respectively , injection ) distance over all pairs of distinct codewords .
a subset of @xmath2 is called a constant - dimension code ( cdc ) .
a cdc is thus a subspace code whose codewords have the same dimension . since for cdcs @xmath22 , we focus on the injection metric when considering cdcs .
we denote the * maximum * cardinality of a cdc in @xmath2 with * minimum injection distance * @xmath27 as @xmath28 .
we have @xmath29 , @xmath30 and it is shown @xcite for @xmath31 and @xmath32 , @xmath33 the lower bound on @xmath28 in ( [ eq : bounds_ac ] ) is implicit from the code construction in @xcite , and the upper bounds on @xmath28 in ( [ eq : bounds_ac ] ) are from @xcite .
thus , cdcs in @xmath2 ( @xmath34 ) with minimum injection distance @xmath27 and cardinality @xmath35 proposed in @xcite are optimal up to a scalar ; we refer to these cdcs as kk codes henceforth .
the covering radius in @xmath2 of a cdc @xmath36 is defined as @xmath37 .
we also denote the * minimum * cardinality of a cdc with covering radius @xmath38 in @xmath2 as @xmath39 @xcite .
it was shown in @xcite that @xmath39 is on the order of @xmath40 , and an asymptotically optimal construction of covering cdcs is designed in ( * ? ? ?
* proposition 12 ) .
we first investigate the properties of balls with subspace radii in @xmath20 , which will be instrumental in our study of packing and covering properties of subspace codes with the subspace metric .
we first derive bounds on @xmath41 below . in order to simplify notations ,
we denote @xmath42 , which is related to the jacobi theta function @xmath43 by @xmath44 $ ] @xcite .
we remark that @xmath45 for all @xmath46 , and that @xmath47 is a decreasing function of @xmath48 and approaches @xmath49 as @xmath48 tends to infinity .
[ lemma : bounds_e ] for all @xmath10 , @xmath50 .
we have @xmath51 by ( [ eq : gaussian ] ) , which proves the lower bound . also , @xmath52 and hence @xmath53 by ( [ eq : gaussian ] ) . therefore ,
@xmath54 .
we observe that by ( [ eq : gaussian ] ) and lemma [ lemma : bounds_e ] , @xmath55 is the same as @xmath41 up to a scalar when @xmath56 or @xmath57 .
that is , the volume of @xmath58 , which is equal to that of @xmath59 when @xmath60 , dominates the volumes of other grassmannians .
this geometric property has significant implication to the packing properties of subspace codes .
we now determine the number of subspaces at a given subspace distance from a fixed subspace .
let us denote the number of subspaces with dimension @xmath61 at subspace distance @xmath27 from a subspace with dimension @xmath1 as @xmath62 .
[ lemma : ns ] @xmath62 is given by @xmath63 when @xmath64 is an integer , and @xmath65 otherwise .
for @xmath66 and @xmath67 , @xmath68
if and only if @xmath69 . thus there are @xmath70 choices for @xmath71 .
the subspace @xmath72 is then completed in @xmath73 ways .
we remark that this result in lemma [ lemma : ns ] is implicitly contained in ( * ? ? ?
* theorem 5 ) without an explicit proof .
it is formally stated here because it is important to the results in this paper .
we also denote the volume of a ball with subspace radius @xmath74 around a subspace with dimension @xmath1 as @xmath75 .
we now derive bounds on the volume of a ball with subspace radius .
since @xmath76 for all @xmath1 and @xmath74 , we only consider @xmath34 .
also , we assume @xmath77 , for only this case will be needed in this paper .
[ prop : bound_vs ] for all @xmath48 , @xmath10 , @xmath34 , and @xmath77 , @xmath78 , where @xmath79 the proof of proposition [ prop : bound_vs ] is given in appendix [ app : prop : bound_vs ] .
we remark that the lower and upper bounds on @xmath80 in proposition [ prop : bound_vs ] are tight up to a scalar , and that @xmath81 depends on both @xmath1 and @xmath74 .
we also observe that @xmath82 decreases with @xmath1 for @xmath34 .
that is , the volume of a ball around a subspace of dimension @xmath1 ( @xmath34 ) * decreases * with @xmath1 .
this observation is significant to the covering properties of subspace codes with the subspace metric .
figure [ fig : vs ] , where we show @xmath83 $ ] for @xmath84 , @xmath85 , @xmath86 , and @xmath87 , illustrates this observation . as a function of the dimension of its center and of its radius ] we are interested in packing subspace codes used with the subspace metric .
the maximum cardinality of a code in @xmath20 with minimum subspace distance @xmath27 is denoted as @xmath88 . since @xmath89 , we assume @xmath90 henceforth
. we can relate @xmath88 to @xmath28 .
first , we remark that @xmath91 for all @xmath48 , @xmath10 , and @xmath92 .
the claim is obvious for @xmath93 , and easily shown for @xmath94 by using ( [ eq : gaussian ] ) .
we also remark that @xmath95 . for all @xmath96
, we denote the maximum cardinality of a code with minimum subspace distance @xmath27 and codewords having dimensions in @xmath97 as @xmath98 . for @xmath99 , @xmath100 .
proposition [ prop : as_rd ] below compares @xmath88 to @xmath101 and shows that @xmath102 is a good approximate of @xmath88 .
[ prop : as_rd ] for @xmath103 , @xmath104 and for @xmath99 , @xmath105 . also , we have @xmath106 .
let @xmath36 be a code in @xmath20 with minimum subspace distance @xmath27 . for @xmath107
, we have @xmath108 ; therefore there is at most one codeword with dimension less than @xmath109 .
similarly , @xmath110 , therefore there is at most one codeword with dimension greater than @xmath111 .
thus @xmath105 for @xmath112 and @xmath113 .
since the code @xmath114 has minimum subspace distance @xmath115 , we obtain @xmath104 .
a cdc in @xmath2 with minimum injection distance @xmath116 has minimum subspace distance @xmath117 , and hence @xmath118 for all @xmath1 .
also , the codewords with dimension @xmath1 in a code with minimum subspace distance @xmath27 form a cdc in @xmath2 with minimum injection distance at least @xmath119 , and hence @xmath120 .
we compare our lower bound on @xmath88 in proposition [ prop : as_rd ] to the gilbert bound in ( * ? ? ?
* theorem 5 ) .
the latter shows that @xmath121 , where the average volume @xmath122 is taken over all subspaces in @xmath20 . using the bounds on @xmath123 in proposition [ prop : bound_vs ] , it can be shown that this lower bound is at most @xmath124 . on the other hand , proposition [ prop : as_rd ] and ( [ eq : bounds_ac ] ) yield @xmath125 .
the ratio between our lower bound and the gilbert bound is hence at least @xmath126 for all @xmath10 and @xmath27 .
therefore , our lower bound in proposition [ prop : as_rd ] is tighter than the gilbert bound in ( * ? ? ?
* theorem 5 ) .
the lower bound in proposition [ prop : as_rd ] is further tightened below by considering the union of cdcs in different grassmannians .
[ prop : lower_as ] for all @xmath48 , @xmath10 , and @xmath127 , we have @xmath128 , where @xmath129 . for @xmath130 ,
let @xmath131 be a cdc in @xmath132 with minimum subspace distance @xmath133 and cardinality @xmath134 and let @xmath135 .
we have @xmath136 , and we now prove that @xmath36 has minimum subspace distance at least @xmath27 by considering two distinct codewords @xmath137 and @xmath138 .
first , if @xmath139 , then @xmath140 ; second , if @xmath141 and @xmath142 , then @xmath143 by the minimum distance of @xmath144 . in order to characterize the rate loss by using cdcs instead of subspace codes , we now compare the cardinalities of optimal subspace codes and optimal cdcs with the same minimum subspace distance @xmath27 .
note that the bounds on the cardinalities of optimal cdcs in ( [ eq : bounds_ac ] ) assume the injection metric for cdc .
when @xmath27 is even , a cdc with a minimum subspace distance @xmath27 has a minimum injection distance @xmath109 .
when @xmath27 is odd , a cdc with a minimum subspace distance @xmath145 has a minimum injection distance @xmath146 .
thus , a cdc has a minimum subspace distance at least @xmath27 if and only if it has minimum injection distance at least @xmath116 .
hence , we compare @xmath88 and @xmath147 in proposition [ prop : as_v_ac ] below .
[ prop : as_v_ac ] _ ( comparison between optimal subspace codes and cdcs in the subspace metric ) .
_ for @xmath99 and @xmath148 , @xmath149 by ( [ eq : gaussian ] ) , proposition [ prop : as_rd ] , and ( [ eq : bounds_ac ] ) , we have @xmath150 .
also , proposition [ prop : as_rd ] and ( [ eq : gaussian ] ) also lead to @xmath151 where @xmath152 .
since @xmath153 and @xmath154 , we obtain @xmath155 where ( [ eq : as1 ] ) follows from ( [ eq : bounds_ac ] ) .
we now compare the relation between @xmath88 and @xmath28 in proposition [ prop : as_v_ac ] to the one determined in ( * ? ? ?
* theorem 5 ) .
the latter only provides the following lower bound on @xmath88 : @xmath156 .
the singleton bound on cdcs @xcite indicates that @xmath157 , which in turn satisfies @xmath158 by ( [ eq : gaussian ] ) .
hence the lower bound on @xmath88 in ( * ? ? ?
* theorem 5 ) is at most @xmath159 .
the ratio between our lower bound in proposition [ prop : as_v_ac ] and the lower bound in ( * ? ? ?
* theorem 5 ) is at least @xmath160 , and thus our lower bound in proposition [ prop : as_v_ac ] is tighter than the bound in ( * ? ? ?
* theorem 5 ) for all cases .
the bounds in proposition [ prop : as_v_ac ] help us determine the asymptotic behavior of @xmath88 .
we first define the rate of a subspace code @xmath161 as @xmath162 and @xmath163 where @xmath164 is the minimum subspace distance of a code , the asymptotic rate of a subspace code @xmath165 and of a cdc of given dimension @xmath166 can be easily determined .
[ prop : as ] _
( asymptotic rate of packing subspace codes in the subspace metric ) .
_ for @xmath167 , @xmath168 . for @xmath169 or @xmath170 , @xmath171 ; for @xmath172 , @xmath173
; for @xmath174 , @xmath175 . first , ( [ eq : bounds_ac ] ) and lemma [ lemma : bounds_e ] yield @xmath176 for @xmath169 .
since @xmath177 , we also obtain @xmath176 for @xmath172 .
second , ( [ eq : as_v_ac ] ) for @xmath178 and ( [ eq : bounds_ac ] ) yield @xmath179 .
propositions [ prop : as_v_ac ] and [ prop : as ] provide several important insights .
first , proposition [ prop : as_v_ac ] indicates that optimal cdcs with dimension being half of the block length up to rounding ( @xmath178 and @xmath180 ) are optimal subspace codes up to a scalar . in this case
, the optimal cdcs have a limited rate loss as opposed to optimal subspace codes with the same error correction capability .
when @xmath181 , the rate loss suffered by optimal cdcs increases with @xmath182 .
proposition [ prop : as ] indicates that using cdcs with dimension @xmath183 leads to a decrease in rate on the order of @xmath184 , where @xmath185 . since the rate loss increases with @xmath186 , using a cdc with a dimension further from @xmath187 leads to a larger rate loss .
the conclusion above can be explained from a combinatorial perspective as well .
when @xmath178 or @xmath180 , by lemma [ lemma : bounds_e ] , @xmath41 is the same as @xmath188 up to scalar .
thus it is not surprising that the optimal packings in @xmath20 are the same as those in @xmath2 up to scalar .
we also comment that the asymptotic rates in proposition [ prop : as ] for subspace codes come from singleton bounds .
the asymptotic rate @xmath189 is achieved by kk codes .
the asymptotic rate @xmath190 is similar to that for rank metric codes @xcite .
this can be explained by the fact that the asymptotic rate @xmath190 is also achieved by kk codes when @xmath178 , whose cardinalities are equal to those of optimal rank metric codes . in table
[ table : comparison ] we compare the bounds on @xmath88 derived in this paper with each other and with existing bounds in the literature , for @xmath84 , @xmath85 , and @xmath27 ranging from @xmath191 to @xmath192 .
we consider the lower bound in proposition [ prop : as_rd ] , its refinement in proposition [ prop : lower_as ] , and the lower bounds in @xcite and ( * ? ? ?
* theorem 5 ) described above , and the upper bound comes from proposition [ prop : as_rd ] .
note that proposition [ prop : as_v_ac ] is not included in the comparison since its purpose is to compare the cardinalities of optimal subspace codes and optimal cdcs with the same minimum subspace distance .
since bounds in propositions [ prop : as_rd ] and [ prop : lower_as ] and ( * ? ? ?
* theorem 5 ) depend on cardinalities of either related cdcs or optimal cdcs , we use the cardinalities of cdcs with dimension @xmath193 proposed in @xcite and @xcite as lower bounds on @xmath28 and the upper bound in @xcite on @xmath28 to derive the numbers in table [ table : comparison ] .
for example , the lower bound of proposition [ prop : as_rd ] is simply given by the construction in @xcite when @xmath194 , and given by the construction in @xcite for other values of @xmath27 .
table [ table : comparison ] illustrates our lower bounds in propositions [ prop : as_rd ] and [ prop : lower_as ] are tighter than those in @xcite and ( * ? ? ?
* theorem 5 ) .
the cardinalities of cdcs with dimension @xmath195 in @xcite and @xcite , displayed as the lower bound in proposition [ prop : as_rd ] , are quite close to the lower bound in proposition [ prop : lower_as ] , supporting our conclusion that the rate loss suffered by properly designed cdcs is smaller when the dimension is close to @xmath196 .
also , the lower and upper bounds in proposition [ prop : as_rd ] depend on @xmath197 , and hence the bounds for @xmath198 and @xmath199 are the same .
finally , the tightness of the bounds improves as the minimum distance of the code increases , leading to very tight bounds for @xmath200 . [ cols="^,^,^,^,^,^",options="header " , ] [ prop : km ] _ ( greedy bound for covering codes in the injection metric ) .
_ for all @xmath48 , @xmath10 , and @xmath38 , @xmath201 $ ] .
we finally determine the asymptotic behavior of @xmath202 by using the asymptotic rate @xmath203 . according to proposition [ prop : bound_vm ]
, the volume of a ball with injection radius is constant up to a scalar .
the consequence of this geometric result is that the greedy algorithm used to prove proposition [ prop : km ] above will produce asymptotically optimal covering codes in the injection metric .
however , since the volume of balls in the subspace metric does depend on the center ( see proposition [ prop : bound_vs ] ) , a direct application of the greedy algorithm for the subspace metric does not necessarily produce asymptotically optimal covering codes in the subspace metric .
[ prop : km ] _
( asymptotic rate of covering subspace code in the injection metric ) .
_ for @xmath204 , @xmath205 . for @xmath206 , @xmath207 . by proposition [ prop : km_2rho >
n ] , @xmath207 for @xmath206
. we have @xmath208 by lemma [ lemma : bounds_e ] and proposition [ prop : bound_vm ] .
this asymptotically becomes @xmath209 for @xmath210 .
similarly , proposition [ prop : km ] , lemma [ lemma : bounds_e ] , and proposition [ prop : bound_vm ] yield @xmath211\end{aligned}\ ] ] which asymptotically becomes @xmath212 for @xmath213 .
the proof of proposition [ prop : km ] indicates that the minimum cardinality @xmath202 of a covering subspace code with the injection metric is on the order of @xmath214 .
a covering subspace code is easily obtained by taking the union of optimal covering cdcs for all constant dimensions , leading to a code with cardinality @xmath215 . by @xcite ,
the cardinality of the union is on the order of @xmath214 .
thus , a union of optimal covering cdcs ( in their respective grassmannians ) results in asymptotically optimal covering subspace codes with the injection metric .
propositions [ prop : ks ] and [ prop : km ] as well as their implications illustrate the differences between the subspace and injection metrics .
first , the asymptotic rates of optimal covering subspace codes with the two metrics are different .
second , a union of optimal covering cdcs ( in their respective grassmannians ) results in asymptotically optimal covering subspace codes with the injection metric only , not with the subspace metric .
these differences can be attributed to the different behaviors of the volume of a ball with subspace and injection radius .
although @xmath216 , proposition [ prop : bound_vs ] indicates that @xmath80 decreases with @xmath1 ( @xmath34 ) , while according to proposition [ prop : bound_vm ] , @xmath217 remains asymptotically constant .
hence , for @xmath218 , the balls with subspace radius @xmath38 centered at a subspace with dimension @xmath1 have significantly smaller volumes than their counterparts with an injection radius .
therefore , covering the subspaces with dimension @xmath187 requires more balls with subspace radius @xmath38 than balls with injection radius @xmath38 , which explains the different rates for @xmath219 and @xmath220 . also , since the volume of a ball with subspace radius reaches its minimum for @xmath178 and @xmath58 has the largest cardinality among all grassmannians , using covering cdcs of dimension @xmath187 to cover @xmath58 is not advantageous .
thus , a union of covering cdcs does not lead to an asymptotically optimal covering subspace code in the subspace metric .
in this paper , we derive packing and covering properties of subspace codes for the subspace and the injection metrics .
we determine the asymptotic rates of packing and covering codes for both metrics , compare the performance of constant - dimension codes to that of general subspace codes , and provide constructions or semi - constructive bounds of nearly optimal codes in all four cases .
these results are briefly summarized in table [ table : results ] . despite these results ,
some open problems remain for subspace codes .
first of all , our bounds on the volumes of balls derived in lemma [ lemma : bounds_e ] and propositions [ prop : bound_vs ] and [ prop : bound_vm ] may be tightened .
although the ratio between the upper and lower bounds is a function of the field size @xmath48 which tends to @xmath49 as @xmath48 tends to infinity , it is unknown whether this ratio is the smallest that can be established .
this issue also applies to the bounds on packing subspace codes in propositions [ prop : as_v_ac ] and [ prop : am_v_ac ] , where the ratios between upper and lower bounds are similar functions of @xmath48 .
also , we only considered balls with radii up to @xmath221 , as only this case was useful for our derivations ; the case where the radius is above @xmath221 remains unexplored .
second , the bounds on covering codes in both the subspace and the injection metrics derived in this paper are only asymptotically optimal .
it remains unknown whether any of these bounds is tight up to a scalar .
third , the design of packing and covering subspace codes is an important topic for future work .
this is especially the case for covering codes in the subspace metric , as no asymptotically optimal construction is known so far .
finally , the aim of this paper was to derive simple bounds on subspace codes which are good for all parameter values , especially large values . on the other hand , a wealth of ad hoc bounds and heuristics can be used to tighten our results for small parameter values .
the authors are grateful to the anonymous reviewers and the associate editor dr .
mario blaum for their constructive comments , which have helped to improve this paper .
we now prove the bounds on @xmath80 for @xmath228 . by definition
, @xmath229 is a double summation of exponential terms .
the main idea of the proof is to determine the largest term in the summation : this not only gives a good lower bound , but the whole summation can also be upper bounded by that term times a constant .
first , by lemma [ lemma : ns ] , @xmath230 , where @xmath231 satisfies @xmath232 . thus @xmath233 by ( [ eq : gaussian ] ) , where @xmath234 .
hence , @xmath235 , where @xmath236 .
since @xmath237 is maximized for @xmath238 , we need to consider the following three cases .
* case i : @xmath239 .
we have @xmath240 and hence @xmath237 is maximized for @xmath241 : @xmath242 .
thus @xmath243 , and it is easy to show that @xmath244 since @xmath245 .
* case ii : @xmath246 .
we have @xmath247 and hence @xmath237 is maximized for @xmath248 : @xmath249 .
it is easily shown that @xmath250 for all @xmath251 and hence @xmath252 .
we also obtain @xmath253 .
* case iii : @xmath254 .
we have @xmath255 and hence @xmath237 is maximized for @xmath256 : @xmath257 .
thus @xmath258 , and it is easy to show that @xmath259 since @xmath260 .
first , @xmath278 .
we now prove the upper bound by determining the largest term in the double summation of @xmath217 . since @xmath279 , we assume @xmath34 without loss of generality .
the triangular inequality indicates that @xmath280 if @xmath281 or @xmath282 ; also , by definition of the injection distance , @xmath280 if @xmath283 .
we can hence restrict the range of parameters in the summation formula of @xmath217 as follows : @xmath284 by lemma [ lemma : nm ] and [ eq : gaussian ] , we have @xmath285 for @xmath286 and @xmath287 for @xmath288 , which with ( [ eq : range_vi ] ) yields @xmath289 where we make the following changes of variables : @xmath290 , @xmath291 , @xmath292 in ( [ eq : vi1 ] ) . since @xmath293 , we have @xmath294 . also , @xmath295 for @xmath296 , and hence @xmath297 ; similarly , we obtain @xmath298 .
hence , ( [ eq : vi1 ] ) leads to @xmath299 where we set @xmath300 and use @xmath301 in ( [ eq : vi3 ] ) .
10 t. ho , m. mdard , r. koetter , d. r. karger , m. effros , j. shi , and b. leong , `` a random linear network coding approach to multicast , '' _ ieee trans .
info . theory _
52 , no .
44134430 , october 2006 . | codes in the projective space and codes in the grassmannian over a finite field referred to as subspace codes and constant - dimension codes ( cdcs ) , respectively have been proposed for error control in random linear network coding . for subspace codes and cdcs ,
a subspace metric was introduced to correct both errors and erasures , and an injection metric was proposed to correct adversarial errors . in this paper , we investigate the packing and covering properties of subspace codes with both metrics .
we first determine some fundamental geometric properties of the projective space with both metrics . using these properties
, we then derive bounds on the cardinalities of packing and covering subspace codes , and determine the asymptotic rates of optimal packing and optimal covering subspace codes with both metrics .
our results not only provide guiding principles for the code design for error control in random linear network coding , but also illustrate the difference between the two metrics from a geometric perspective .
in particular , our results show that optimal packing cdcs are optimal packing subspace codes up to a scalar for both metrics if and only if their dimension is half of their length ( up to rounding ) . in this case ,
cdcs suffer from only limited rate loss as opposed to subspace codes with the same minimum distance .
we also show that optimal covering cdcs can be used to construct asymptotically optimal covering subspace codes with the injection metric only .
network coding , random linear network coding , error control codes , subspace codes , constant - dimension codes , packing , covering , subspace metric , injection metric . |
it is of great importance to identify strange stars ; a new window of distinguishing neutron stars and bare strange stars ( bsss ) has been proposed recently according to their sharp differences of surface conditions ( xu , zhang , qiao 2001 ; usov 2001a ) .
it is therefore essential to study the surface properties of bsss in details , e.g. , the degree of the thermal luminosity of a hot bss
. the surface electric field should be very strong ( @xmath0 v / cm ) near the bare quark surface of a strange star because of the mass difference of the strange quark and the up ( or down ) quark ( alcock et al . 1986 ) , which could play an important role in producing the thermal emission of bsss by the usov mechanism ( usov 1998 , usov 2001b ) , because the strange quark matter is a poor radiator of thermal photons at frequencies less than its plasma frequency ( @xmath1 mev ) ( alcock et al .
the basic idea of the usov mechanism is that @xmath2 pairs are created rapidly in a few empty quantum states with energy @xmath3 ( @xmath4 is the fermi energy , @xmath5 the electron mass ) due to the very strong electric field in an electron layer ( with a height of @xmath6 fm above quark surface ) ; the pairs subsequently annihilate into photons which are then thermalized in the electron layer k. ] around a bss .
this radiative mechanism has been applied tentatively to soft @xmath7 ray repeaters ( usov 2001c ; usov 2001d ) recently .
in addition , the strong electric field plays an essential role in forming a possible crust around a strange star , which has been investigated extensively by many authors ( e.g. , martemyanov 1992 ; kettner et al .
1995 ; huang & lu 1997 ; phukon 2000 ; see zdunik , haensel & gourgoulhon 2001 for the recent development with the inclusion of rotating and general - relativistic effects ) .
also it should be noted that this electric field has some important implications on pulsar radio emission mechanisms ( xu & qiao 1998 ; xu , qiao , zhang 1999 ; xu , zhang , qiao 2001 ) .
in fact the usov mechanism of pair production depends on many parameters ; it is therefore imperative to study the dependence of the process on these parameters . with some typical parameters chosen by usov in his calculations , the resultant thickness of the electron layer with electric field @xmath8 v / cm ( the critical field necessary for pair production ) , @xmath9 , is @xmath6 fm .
however the proper determination of the thickness @xmath9 should be done with the dynamical theory of the strange quark matter . because of the untractable nature of the quantum chromodynamics , some phenomenological models , i.e. , the mit bag model ( e.g. , jensen & madsen 1996 ) , the quark mass - density - dependent model ( e.g. , lugones & benvenuto 1995 ) , and the quark potential model ( e.g. , dey et al .
1998 ) , have been applied to the descriptions of the strange quark matter . in the bag model
, @xmath9 is a function of @xmath10 ( the coupling constant for strong interactions ) , @xmath11 ( the strange quark mass ) , and @xmath12 ( the bag constant ) .
also it should be noticed that the quark number densities , @xmath13 ( i = u , d , s for up , down and strange quarks , respectively ) , are assumed to be uniform below the quark surface , and therefore the quark charge density @xmath14 is constant near the surface .
the potential @xmath15 is usually chosen as 20 mev for typical cases .
however the quark number densities should not be uniform near the surface if boundary effects are included , since charge neutrality is broken there . in this paper
we improve the calculation of the electric field in the vicinity of a bss surface , using the popular bag model and the thomas - fermi model .
we investigate the electric characteristics of bsss for different values of @xmath16 .
initially we use typical parameters for @xmath17 mev and @xmath18 mev .
then the thickness , @xmath9 , of the electron layer where pairs can be created , is computed as functions of @xmath12 , @xmath11 , and @xmath16 .
it is found that usov s estimates are representative for bag - model parameters within a rather wide range .
boundary effect are also considered in this calculation .
the interesting electric properties of strange stars were first noted by alcock , farhi & olinto ( 1986 ) , who presented numerical calculations of the electric potentials of strange stars with or without crusts .
analytical solutions of the electron number density and the electric field was also formulated , which could be helpful in dealing with some physical processes near the quark surfaces of strange stars ( xu & qiao 1999 ) . in this section ,
the bag model is applied to calculate the quark charge density ; the thomas - fermi model is employed to find @xmath9 dependence on @xmath16 , @xmath11 , and @xmath12 , with the inclusion of boundary effects . in our calculation , the quarks and electrons near the surface of the strange star keep chemical equilibrium locally ; the relation between charge density and electric potential is described by the classical poisson equation .
the thermodynamic potentials @xmath19 as functions of chemical potentials @xmath20 ( i = u , d , s , e ) , @xmath11 and @xmath16 can be found in literature ( alcock et al . 1986 ) . in the chemical equilibrium of weak interaction , @xmath21 @xmath22 @xmath23 @xmath24 where @xmath25 is a measured height above the quark surface .
the quark number densities drop to zero on the surface , but they are not uniform below the surface if the boundary effects are considered .
the chemical potential @xmath26 can be determined by the condition that the pressure on the quark surface is zero , @xmath27 .
the pressure is @xmath28 ( alcock et al .
1986 ) . the boundary conditions for eq.([a6 ] )
are @xmath38 @xmath39 where @xmath40 is the electric potential in the deep core of the strange star , which is determined from eq.([a6 ] ) by letting the left hand side equal zero . by integrating eq.([a6 ] ) in
@xmath41 $ ] and @xmath42 respectively , with the boundary conditions , we get the expression of the electric field @xmath43 , where @xmath44 .
obviously @xmath45 is a function of @xmath16 , which can be obtained from the continuity of @xmath46 at @xmath47 . substituting @xmath45 into the equations of @xmath46 and then integrating them
, we get the expression of @xmath32 . for @xmath48 , we find @xmath49 @xmath50 @xmath51 several numerical results are obtained , which are shown in table 1 and figs.1 - 4 .
first , we study the electric characteristics of a bss for different values of @xmath16 , assuming and @xmath12 , @xmath52 . in this assumption @xmath53 mev . ]
@xmath54 mev , @xmath17 mev , and the renormalization point @xmath55 mev .
the quark and electron number densities \{@xmath13 } when @xmath56 , as well as potentials of @xmath40 and @xmath57 , are listed in table 1 for @xmath58 , respectively .
.the particle number densities far below the quark surface , and the potentials of @xmath59 and @xmath60 , for different values of @xmath16 . [ cols="^,^,^,^,^,^,^",options="header " , ] the electric potential @xmath32 and field @xmath46 are also plotted in figs.1 - 3 for this case .
we see from fig.1 that the potential curve becomes flatter when the coupling constant @xmath16 increases .
also it should be noted from figs.2 and 3 that the electric filed @xmath46 changes rather slowly for @xmath61 , but varies rapidly for @xmath62 .
this indicates the non - symmetric nature of the electric field with respect to the quark surface of @xmath47 .
secondly , we investigate numerically the thickness , @xmath63 , of the electron layer where pairs can be created , as functions of the bag parameters @xmath11 , @xmath12 , and @xmath10 . the results are shown in fig.4 .
it is found that the usov mechanism can work for bag parameters within a rather wide range .
the thickness @xmath9 could be large enough ( i.e. , @xmath64 fm ) as long as ( 1 ) @xmath12 is not too large and @xmath11 is not too small , or ( 2 ) @xmath12 is not too small and @xmath11 is not too large .
increasing of @xmath65 has an adverse influence on the usov mechanism .
for instance in fig.2 , the field @xmath46 can not exceed the critical field @xmath66 v / cm ( necessary for pair production ) if @xmath67 .
nevertheless previous dynamical calculations show that @xmath68 is less possible for a stable strange quark matter .
in conclusion we could expect that the pair emission process proposed by usov might appear in nature for acceptable bag parameters .
we have improved the calculation of the electric characteristics of bare strange stars with the inclusion of boundary effects ( i.e. , the effects of non - local neutrality near and below the quark surface ) . from our calculation , we find that the usov mechanism can work for bag parameters within a rather wide range . as shown in table 1 , both @xmath45 and @xmath40 , as well as their very small difference , decrease as @xmath16 increases . our results on the electric potential @xmath69 for @xmath62
is quit different from the previous calculation given by alcock et al.(1986 ) where the boundary effects were not included ( see fig .
it is shown that the strong electric field resides only about @xmath70 fm below the quark surface ( see fig .
3 ) , rather than @xmath71 fm obtained by alcock et al . ( 1986 ) . we can proof that @xmath46 is almost an exponential function of @xmath25 below the quark surface .
denote the right hand side of the eq.([a6 ] ) as @xmath72 . as @xmath73 , namely @xmath74
, we have @xmath75 . approximating @xmath72 as @xmath76
, we can obtain @xmath77z$ ] and @xmath78 $ ] for @xmath79 . _ acknowledgments .
_ this work is supported by national nature sciences foundation of china ( 10173002 ) and the special funds for major state basic research projects of china ( g2000077602 ) .
the authors sincerely thank dr .
shuangnan zhang for his comments and the improvement of the language . | the surface electric characteristics of bare strange stars are investigated with the inclusion of boundary effects .
the thickness of the electron layer where pairs can be spontaneously created is calculated as a function of the bag parameters .
we find that previous estimates are representative for bag parameters within a rather wide range , and therefore our results favor the thermal radiation mechanism of bare strange stars advanced by v. v. usov . |
the comb - like models have been introduced to investigate anomalous diffusion in low - dimensional percolation clusters @xcite .
it means that the mean square displacement ( msd ) has power - law dependence on time @xmath3 @xcite .
an elegant form of equation which describes the diffusion on a comb - like structure was introduced by @xcite @xmath4 where @xmath5 is the probability distribution function ( pdf ) , @xmath6 is the diffusion coefficient in the @xmath7 direction with physical dimension @xmath8=\mathrm{m}^{3}/\mathrm{s}$ ] , and @xmath9 is the diffusion coefficient in the @xmath10 direction with physical dimension @xmath11=\mathrm{m}^{2}/\mathrm{s}$ ] .
the @xmath12 function in the diffusion coefficient in the @xmath7 direction implies that the diffusion along the @xmath7 direction occurs only at @xmath13 .
thus , this equation can be used to describe diffusion in the backbone ( at @xmath13 ) where the teeth play the role of traps . nowadays ,
comb models have many applications .
they have been used for the understanding of continuous @xcite and discrete @xcite non - markovian random walks .
there are generalizations of this equation by introducing time fractional derivatives and integrals in ( [ diffusion like eq on a comb ] ) @xcite .
such generalized comb - like models have been used to describe anomalous diffusion in spiny dendrites , where the msd along the @xmath7 direction has a power - law dependence on time @xcite , or for describing subdiffusion on a fractal comb @xcite , the mechanism of superdiffusion of ultra - cold atoms in a one dimensional polarization optical lattice @xcite as a phenomenology of experimental study @xcite , and to describe diffusion processes on a backbone structure @xcite .
different generalizations of the comb model have been shown to represent more realistic models for describing transport properties in discrete systems , such as porous discrete media @xcite , electronic transport in semiconductors with a discrete distribution of traps , cancer development with definitely fractal structure of the spreading front @xcite , infiltration of diffusing particles from one material to another @xcite , description of diffusion of active species in porous media @xcite , etc .
furthermore , in @xcite it is shown that in a comb - like model a negative superdiffusion occurs due to the presence of an inhomogeneous convection flow . in this paper
we consider a generalization of eq.([diffusion like eq on a comb ] ) where we allow that diffusion along the @xmath7 direction may occur on many backbones , located at @xmath14 , @xmath15 , @xmath16 .
this means that we have a comb grid where @xmath0 can be arbitrarily large , even infinity .
the governing equation for such a structure is given by where @xmath17 are structural constants such that @xmath18 .
the initial condition is given by @xmath19 and the boundary conditions for @xmath5 and @xmath20 , @xmath21 are set to zero at infinity , @xmath22 , @xmath23 .
one can easily verify that for @xmath24 , @xmath25 , and @xmath26 eq.([diffusion like eq on a comb two delta ] ) becomes ( [ diffusion like eq on a comb ] ) .
the physical dimensions of @xmath27 and @xmath9 for a finite number of backbones are the same as those in eq.([diffusion like eq on a comb ] ) .
the case of a fractal structure of backbones will be described by an appropriate generalization of eq.([diffusion like eq on a comb two delta ] ) .
the motivation to introduce such a model is to describe the diffusion of solvents in thin porous films @xcite . such a product structure of backbones times comb is an idealization of more complex comb - like fractal networks , as they may appear , e.g. , in certain anisotropic porous media or anisotropic biological tissue .
the paper is organized as follows . in sec .
ii we analyze the pdf and the msd in both directions for the force free case .
anomalous diffusive behavior @xmath28 appears in the @xmath7 direction due to the comb structure of the system .
general results for the msd in the case of a finite number of backbones @xmath0 are presented .
we also investigate the effects of an external constant force , applied along the backbones , on the particle behavior . in sec .
iii we consider an infinite number of backbones .
it is shown that an infinite number of backbones , different from the case of a finite number of backbones , changes the transport exponent .
deviations from the standard msd @xmath29 were observed recently in combs with ramified teeth as well , due to teeth with a fractal structure @xcite .
the summary is given in sec .
we apply a laplace transform ( @xmath30=\hat{f}(s)$ ] ) to eq.([diffusion like eq on a comb two delta ] ) , and then a fourier transform with respect to the @xmath7 ( @xmath31=\tilde{f}(\kappa_{x})$ ] ) and @xmath10 ( @xmath32=\bar{f}(\kappa_{y})$ ] ) variables .
thus , we obtain @xmath33 where @xmath34 . from relation ( [ diffusion like eq on a comb fourier - laplace transform ] ) , the inverse fourier transform with respect to @xmath35 yields @xmath36 in the setting of a comb model , the nontrivial and interesting motion is along the backbones , i.e. , along the @xmath7 direction , while the @xmath10 direction is an auxiliary subspace .
therefore , integrating the motion in the @xmath10 direction , we analyze the pdf @xmath37 . by integration of eq.([diffusion like eq on a comb two delta ] ) with respect to @xmath10 and performing the laplace transform with respect to time @xmath38 , and the fourier transform with respect to @xmath7 , one obtains @xmath39.\end{aligned}\ ] ] from the pdf ( [ p1 general ] ) we calculate the msd along the @xmath7 direction by the following formula : @xmath40\right|_{\kappa_{x}=0}.\end{aligned}\ ] ] from relations ( [ diffusion like eq on a comb fourier - laplace transform inverse f ky])-([msd general ] ) for the msd we derive @xmath41\nonumber\\&&=\frac{\mathcal{d}_{x}}{\sqrt{\mathcal{d}_{y}}}\sum_{j=1}^{n}w_{j}\left[\frac{2}{\sqrt{\pi}}t^{1/2}e^{-\frac{|l_{j}|^{2}}{4\mathcal{d}_{y}t}}\right.\nonumber\\&&\left.-\frac{|l_{j}|}{\sqrt{\mathcal{d}_{y}}}\mathrm{erfc}\left(\frac{|l_{j}|}{\sqrt{4\mathcal{d}_{y}t}}\right)\right],\end{aligned}\ ] ] where @xmath42 is the complementary error function @xmath43 @xcite . for @xmath24
it follows that @xmath44.\end{aligned}\ ] ] for the long time scale when @xmath45 , @xmath46 , the msd reads @xmath47 which means that all backbones contribute in the msd .
in contrast to this , on a short time scale , when @xmath48 , @xmath46 , one finds that the main contribution in the msd is due to the first backbone , i.e. , @xmath49 this result is expected since for short times the particles move mainly in the first backbone because they had not enough time to reach the other ones by diffusion in the @xmath10 direction .
this can be easily verified by considering diffusion along the @xmath10 direction .
we analyze the pdf @xmath50 , for which we find that @xmath51 i.e. , @xmath52 .
for the msd along the @xmath10 direction one finds a linear dependence on time @xmath53 , i.e. , normal diffusion along the @xmath10 direction .
therefore , the probability to find the particle at the first backbone is @xmath54 ( @xmath24 ) , while at the second backbone it is @xmath55 , and so on . since for the short time scales , @xmath56 , we conclude that the main contribution in the msd for short times is due to the displacements in the first backbone . from relation ( [ diffusion like eq on a comb msd exact n l1 - 0 ] ) for @xmath25 , @xmath26 , and @xmath24 ( which means one backbone )
we obtain the msd for the comblike model ( [ diffusion like eq on a comb ] ) @xmath57 these results are supported by graphical representation in fig . [ comb ] of the msd in the case of two backbones and five backbones .
it is assumed that the first backbone is at @xmath13 and all the other backbones are at distances equal to @xmath58 , @xmath59 , @xmath60 , @xmath61 . from relations ( [ diffusion like eq on a comb msd exact l1=0,n t->large ] ) and ( [ diffusion like eq on a comb msd exact l1=0,l->infty ] )
we conclude that any finite number of backbones does not change the transport exponent in the short and long time limit . in the intermediate times
there is more complicated behavior of the msd given by relation ( [ diffusion like eq on a comb msd exact n l1 - 0 ] ) .
the crossover time scales separating the behavior at short , intermediate , and long times are given by @xmath62 and @xmath63 .
in the presence of a constant external force @xmath64 along the backbones we arrive at the following fokker - planck equation @xmath65\nonumber\\&&\times p(x , y , t)+\mathcal{d}_{y}\frac{\partial^{2}}{\partial y^{2}}p(x , y , t),\end{aligned}\ ] ] where @xmath66 is the mobility .
one can compute the first moment as a function of time , @xmath67,\nonumber\\\end{aligned}\ ] ] where by comparing it with relation ( [ diffusion like eq on a comb msd exact n ] ) we conclude that the generalized einstein relation is fulfilled @xcite , @xmath68 where @xmath69 .
to introduce a fractal structure of the backbones we go back to eq.([diffusion like eq on a comb two delta ] ) and replace the summation @xmath70 with summation over a fractal set @xmath71 , i.e. , @xmath72 , which means that the backbones are at positions @xmath10 which belong to the fractal set @xmath71 with fractal dimension @xmath73 .
a simple toy example , which corresponds to an infinite fractal set , can be treated as follows . in relation (
[ diffusion like eq on a comb msd exact n ] ) we calculate @xmath74 .
one should recognize that fractal sets ( like a cantor set ) are uncountable .
therefore , the last expression is purely formal and its mathematical realization corresponds to integration to fractal measure @xmath75 such that @xmath76 is the fractal density @xcite , and @xmath77 . here we note that @xmath27 is a generalized diffusion coefficient with physical dimension @xmath8=\mathrm{m}^{3-\nu}/\mathrm{s}$ ] that absorbs the dimension of fractal volume or measure @xmath78 . that finally yields the following integration : @xmath79 for the msd , we obtain from ( [ diffusion like eq on a comb msd exact n ] ) @xmath80 i.e. , anomalous diffusive behavior with the transport exponent equal to @xmath81 .
thus , the fractal set @xmath71 of the infinite number of backbones changes the transport exponent , from @xmath2 to @xmath82 . for @xmath83
the msd becomes @xmath84 , which is consistent with expectations , and for @xmath85 , we are back to the finite-@xmath0 case .
indeed , the fractal dimension of any finite number of discrete points is @xmath85 .
we further consider a random fractal set @xmath86 $ ] , with finite limits . from relation ( [ diffusion like eq on a comb msd exact n ] ) , in the same way as in ( [ infinite fractal mesh ] ) , for a finite integration in @xmath87 $ ] , one finds a result in the form of an incomplete @xmath88 function @xmath89 @xcite , @xmath90 thus , the msd becomes @xmath91 again , for @xmath83 the normal diffusive behavior along the @xmath7 direction appears , i.e. , @xmath92 . here
we note that the result for the msd ( [ msd infinite fractal mesh ] ) can be obtained in the framework of fractional integration as well . by integration of eq.([diffusion like eq on a comb two delta ] ) over @xmath10 and using the summation on the fractal set as above in this section , for the pdf @xmath93 one obtains @xmath94 the laplace transform to ( [ p1 fractal ] ) yields @xmath95 by representing the solution @xmath96 in the following way : @xmath97
, i.e. , @xmath98 , for the @xmath99 we find @xmath100 from the other side , by using the previous approach of summation , we have @xmath101 by substituting relation ( [ summation p lj ] ) in ( [ p1 fractal laplace ] ) , we obtain @xmath102 from this , the inverse laplace transform yields the following time fractional diffusion equation : @xmath103 where @xmath104 is the caputo time fractional derivative of order @xmath81 @xcite .
from here we easily obtain the msd @xmath105 that is of form ( [ msd infinite fractal mesh ] ) . the solution for the pdf @xmath93
can be represented in terms of the fox @xmath106 function @xmath107 @xcite , @xmath108,\ ] ] where @xmath109 is the generalized diffusion coefficient with physical dimension @xmath110=\mathrm{m}^{2}/\mathrm{s}^{(1+\nu)/2}$ ] .
therefore , as shown , the infinite number of backbones changes the transport exponent .
the asymptotic behavior of @xmath93 ( [ p1 fractal structure ] ) for @xmath111 is of the form @xcite : @xmath112,\end{aligned}\ ] ] i.e. , it has non - gaussian behavior . for @xmath83
it turns to gaussian behavior as it is expected and as it was shown by analysis of the msd . additionally to the msd we calculate the @xmath113th moment @xmath114 thus for the fourth moment it follows that @xmath115 the calculation of the fourth moment is useful to discriminate subdiffusive processes with identical msds , e.g. , subdiffusion due to different fractal structures or different mechanisms @xcite ( see also @xcite ) . for the even moments we obtain @xmath116 from which we can find the following interesting relation : @xmath117 where @xmath118 is the one parameter mittag - leffler function @xcite
finally , we show how the fractal structure @xmath71 relates to the fractional riesz derivative @xcite .
let us consider the fractal structure of backbones in eq.([diffusion like eq on a comb two delta ] ) separately . in the fourier - fourier @xmath119 space it reads @xmath120 where @xmath121 is the weierstrass function @xcite .
it can be obtained by the following procedure @xcite : let us use @xmath122 , where @xmath123 , @xmath124 .
thus @xmath125 now @xmath126 and @xmath127 are dimensionless scale parameters .
therefore @xmath128 where @xmath129 , and @xmath130 , and for convenience , we choose @xmath131 . from here
one obtains @xmath132 neglecting the last term since @xmath124 , therefore the scaling @xmath133 means that @xmath134 , where @xmath135 is the fractal dimension .
thus , for relation ( [ eq in ff space ] ) we have @xmath136 this integration is the riesz fractional derivative @xcite .
in this paper we introduce a diffusion equation for a comb structure where the displacements in the @xmath7 direction are possible along many backbones , even an infinite number of backbones , and we call this system by grid comb . we analyze the msd and we show that by adding a finite number of backbones , the transport exponent in the long time limit does not change . differently from that
, an infinite number of backbones changes the transport exponent . considering a fractal structure of backbones with fractal dimension @xmath137 we obtained the dependence of the transport exponent on @xmath137 .
we stress that the performed analysis is exact more precisely , that the evaluation of the contribution of the fractal structure @xmath71 to anomalous diffusion is exact .
note that the first attempt to take into account a fractal structure of traps was performed in @xcite in the framework of a coarse graining procedure of the fokker - planck equation that leads to the fractional differentiation in the real space .
in contrast to that , in the present analysis we are able to perform an exact analysis for the fractal structure @xmath71 .
this also relates to exact fractional differentiation in the reciprocal fourier space . in conclusion
, it should be admitted that a comb model is a toy model that can be solved exactly and establishes a relation between geometry and the transport exponent .
as is recently found it also corresponds to the real physical realization in experiments on calcium transport in spiny dendrites ( see @xcite and references therein ) .
the grid - comb model , suggested here as the generalization of the comb model , establishes an exact relation between a complicated fractal geometry and the transport exponent as well . another strong motivation of the model , also related to the result , is that in the framework of this model it is possible to infer an exactly fractional derivative related to fractal geometry .
all these points are important for the understanding of anomalous transport in heterogeneous material , in particular to describe diffusion of solvents in thin porous films @xcite , or in another two - dimensional material like graphene @xcite .
a.i . would like to thank the max - planck institute for the physics of complex systems in dresden , germany for financial support and hospitality , as well as the support by the israel science foundation ( isf-1028 ) . | a grid comb model is a generalization of the well known comb model , and it consists of @xmath0 backbones . for @xmath1
the system reduces to the comb model where subdiffusion takes place with the transport exponent @xmath2 .
we present an exact analytical evaluation of the transport exponent of anomalous diffusion for finite and infinite number of backbones .
we show that for an arbitrarily large but finite number of backbones the transport exponent does not change .
contrary to that , for an infinite number of backbones , the transport exponent depends on the fractal dimension of the backbone structure . |
to power the central black hole by accretion , bright quasars must consume @xmath5 solar masses during their lifetimes roughly 1% of the stellar mass in a bright elliptical galaxy or 10% of the gaseous mass of a bright spiral .
even with the assumption of a large host galaxy , an efficient mechanism is needed to channel gas to the central source .
this problem of fuelling quasars is normally divided into three parts : the movement of gas from galactic scales to the inner few hundred parsecs , the instabilities of a self gravitating disk that transports the gas to a compact accretion disk and the detailed dynamics and radiation mechanisms of the final accretion . here , we consider the first problem of moving gas from galactic scales into the inner few hundred parsecs . at redshifts of 2 where the number densities of quasars peak , this has been associated with the dynamical chaos of galaxy formation ( c.f .
haehelt and rees 1993 ) . at lower redshifts , interactions of galaxies leading to coalescence or merging have been proposed ( hernquist 1989 ) .
these mergers can drive more than 10% of the gas to the center .
the final host galaxy is the product of a merger , therefore it would be brighter than average .
there are two observations that suggest a third mechanism at intermediate redshifts @xmath6 .
the environment of quasars has been observed to change with redshift ( yates , miller and peacock 1989 ; hereafter ymp89 , yee and ellingson 1993 ; hereafter ye93 ) .
quasars at higher redshifts are in abell richness class 0 - 1 clusters of galaxies an environment that is considerably richer than that of lower redshift quasars ( the break point is at @xmath7 in ymp89 and @xmath8 in ye93 ) .
bahcall , kirhakos and schneider ( 1995 ; hereafter bks ) used hst to image eight luminous quasars at redshifts between 0.15 and 0.3 . only three quasars in their sample have candidate hosts that are as luminous as , the characteristic luminosity in the schechter ( 1976 ) luminosity function ( lf ) .
the other five hosts must be fainter than to have escaped detection .
after observations of additional quasars , bahcall et al .
( 1997 ; hereafter bkss97 ) conclude that the luminous quasars studied in this paper occur preferentially in luminous galaxies " .
they are rejecting the null hypothesis " that all galaxies are equally likely to have quasars ( e.g. a hypothesis that states that draco and the milky way are equally likely to host quasars ) .
their conclusion results because at least half of all galaxies are @xmath9 magnitudes fainter than whereas the dividing line for their sample of quasar hosts is @xmath10 within their errors .
this depends slightly on the logarithmic slope of the lf .
since most lfs are weakly diverging at the faint end @xmath11 , @xmath12 , a faint end cutoff is required to to define an average luminosity " and that average is normally 2 to 3 magnitudes brighter than the cutoff or @xmath9 magnitudes fainter than .
however , galaxies brighter than @xmath13 contain half of all the luminosity , where the range takes into accounts the uncertainty in the faint end slope and cutoff of the luminosity function .
this dividing line of luminosity is consistent with the bkss97 midpoint of quasar hosts within their errors .
so , quasars do nt obviously prefer brighter galaxies any more than stars do .
previously , mcleaod and rieke ( 1995 ) suggested a linear relation between the absolute magnitude of the quasar and its host .
bkss97 find no such relation other than what can be attributed to the obvious bias from detection limits .
the simplest summary of the observations to date is that quasars and galaxies may be related much as stars and galaxies ; the probability of finding either in a galaxy is proportional to the galaxies luminosity but their individual luminosities are not determined by the luminosity of their host . hence , _ we need a mechanism that will produce quasars with a frequency per unit luminosity rather than a mechanism that only operates in bright galaxies as would be the case if mergers were the dominant trigger_. we have found a mechanism that is extremely efficient at channeling gas into the center of sub - galaxies that live in clusters . galaxy harassment " drives dynamical instabilities that send most of the gas into the central few hundred parsecs of the harassed galaxies . in 2 , we present detailed hydrodynamical simulations that show these effects , while we look at the model s predictions in 3 .
we use treesph ( hernquist and katz 1989 , katz , weinberg and hernquist 1996 ) to examine the fate of a high resolution galaxy with @xmath14 particles in each of three components : gas , stars and dark matter .
the simulations were originally performed to follow the morphological evolution of galaxies in clusters ( moore 1996 , moore , katz and lake 1996b ) , so the parameters were chosen to be typical of the disturbed galaxies seen by hst at redshifts of @xmath15 .
we present three simulations , two with circular velocities of @xmath16 and one with a circular velocity of @xmath17 . using the fisher tully ( 1981 ) relation , the luminosity of a galaxy with a circular velocity , @xmath18 is @xmath19 .
the @xmath20 model galaxies have exponential disks with scalelengths of 2.5 kpc and scaleheights of 200 pc .
the smaller galaxy is scaled by using the faber - jackson relation and preserving the ratio of stars , gas and dark matter .
the disks are constructed with a toomre ( 1964 ) `` stability '' parameter @xmath21 and run in isolation for 2 gyr before being set into orbit in the cluster .
the gaseous disk is initially on cold circular orbits .
the dark halo of each galaxy is a spherical isothermal with a core radius of @xmath22 kpc , and is tidally truncated at the pericenter of the galaxy s orbit within the cluster . at 4 disk scalelengths ( 10 kpc ) ,
the initial ratio of dark matter to stars to gas is 11:4:1 . at 8 disk scalelengths ( 20 kpc ) ,
the ratios are 20:5:1 and the total mass is @xmath23 .
the cluster model is based on coma .
its one dimensional velocity dispersion is @xmath24 and the mass within the virial radius ( 1.5 mpc ) is @xmath25 . for an @xmath26 of 250 ,
the total cluster luminosity within this radius is @xmath27 .
the other galaxies in the cluster are drawn from a schechter ( 1976 ) luminosity function parameterized using @xmath28 and @xmath29 including all galaxies brighter than @xmath30 ( @xmath31 and @xmath32 throughout this paper ) .
this produces a model cluster that has 950 galaxies brighter than the magellanic clouds , but only 31 brighter than @xmath1 .
the masses and tidal radii of the other galaxies are determined by taking an isothermal model with a dispersion given by the faber - jackson relation and tidally limiting it at the galaxy s pericentric distance .
they are then modeled by spheres with a softening length equal to half of their tidal radius .
most simulations have no interpenetrating " collisions . however , any such collision will be more gentle than a collision with a realistic model of a galaxy that is more centrally concentrated .
white and rees ( 1978 ) speculated that the dark halos were stripped from galaxies within clusters ; we find that the dominant stripping mechanisms are tides and high speed encountersgalaxy harassment " ( moore , katz and lake 1996b ; hereafter mkl96b ) .
if galaxies are initially tidally limited by the cluster potential , bright galaxies retain more than half of their mass when followed for 5 gyr . for greater self - consistency , we reduced the mass of each perturber by 25% ,
the average loss over a hubble time ( mkl96b ) and we left the tidal / softening radius fixed . in the accompanying video , the perturbing galaxies are shown as green dots located at their centers . the mean ratio of a perturbing galaxy s apocenter to its pericenter in our cluster model is roughly 6-to-1 ( this ratio is even larger in infinite isothermal spheres with isotropic velocities ) ; a galaxy found at a radius of 450 kpc will have a mean orbital radius of 400 kpc and a typical pericenter ( @xmath33 ) that is slightly greater than 150 kpc .
we assign galaxies masses of @xmath34 corresponding to mass - to - light ratios , @xmath35 .
the luminous parts of elliptical galaxies have mass - to - light ratios of @xmath36 ( van der marel 1991 ) , so our perturbing galaxies have very modest extended dark halos .
the fraction of the cluster s density attached to galaxies varies with radius from zero at the center to nearly unity at the virial radius .
it is @xmath1020% at the mean orbital radius of our simulated galaxies .
the rest of the cluster mass is in a smoothly distributed background represented by a fixed analytic potential .
further details of the cluster model can be found in mkl96b .
we have intentionally made some conservative assumptions to ensure that our results are robust . for a model galaxy on a fixed orbit ,
the havoc wreaked by harassment depends on the square of the masses of the largest galaxies encountered .
the most massive galaxies are giant ellipticals that are far less prone to harassment owing to their high internal densities , yet we have reduced the masses of all galaxies by the same time averaged value of 25% .
at a fixed mean orbital radius , galaxies on elongated orbits experience greater harassment . we follow galaxies that have apo / peri ratios of 2 ( _ i.e. _ apocenter at 600 kpc , pericenter at 300 kpc ) , whereas the typical value is @xmath37 . as a result ,
our model galaxies avoid extremes of the cluster distribution and start with large dark halo masses determined by the tidal limit at their atypically large pericenters .
both effects serve to underestimate the effects of harassment .
the strongest encounters do not necessarily occur near the center of the cluster .
clearly , the frequency of encounters scales with the density of galaxies . however , there are two counterbalancing effects .
if the galaxies are all tidally limited , they are more massive at larger cluster radii .
secondly , the relative velocity of encounters decreases in the outer parts of the cluster and encounters are stronger at fixed impact parameter .
as long as the encounter timscale is short compared to the internal dynamical time of the galaxy , the gas dynamical effects are strongest outside the central region of a cluster .
however , galaxies near the center of the cluster are more strongly damaged by the cluster s tidal field .
care was taken to ensure that our initial models were sensible and stable by simulating galaxies on circular and elliptical orbits in smooth cluster potentials before examining the effects of harassment by other galaxies .
the disk of a galaxy on circular orbit at 450 kpc in a fixed cluster potential remains stable for 10 gyrs .
a galaxy on an eccentric orbit with an apocenter of 600 kpc and pericenter of 300 kpc becomes bar unstable after the first pericentric passage and a bar persists for 5 billion years .
each passage through pericenter results in the loss of a small fraction of dark halo material , but the gas and stars remain attached .
the most dramatic evolution owes to strong encounters when the other perturbing galaxies are included ( one simulation is shown on an accompanying video ) .
the first strong encounter usually causes a bar instability of greater strength than that induced by only the cluster s tidal field .
the continued heating of the disk by collisions transforms the galaxy into a spheroidal galaxy that matches the brightness profile and velocity dispersions of the dwarfs in nearby clusters ( moore , lake and katz 1997 ) .
perhaps the most stunning feature of the evolution is the rate that gas is driven into the center of the galaxy , as shown in figures 1 and 2 .
figure 1 shows snapshots of the evolution of gas disk ( also shown as the last sequence in the accompanying video ) .
the angular momentum of the gas is decreased by the torques of passing collisions and internal torques between the distorted disk and the halo . when combined with radiative cooling , this drives a large fraction of the gas into the center of the galaxy as shown in figure 2 . in our first three simulations , we found that @xmath1090% , 80% and 40% of the gas is driven to the center within 3 gyrs . in the first two cases ,
half of the mass is transferred in an interval of 100 - 200 myr .
we note one last problem in making detailed comparisons to observations .
we link the collisional deformation of a galaxy to the channeling of gas into the middle of the galaxy .
the response to the jolt of a high speed fly - by encounter occurs over the galaxy s internal orbital time , @xmath38 myr ( for a half light radius of 5.6 kpc and circular velocity of @xmath16 ) .
the three dimensional velocity dispersion in the cluster is @xmath39 , so each galaxy moves @xmath40 kpc through the cluster as it responds to their encounter .
therefore , one can determine neither the galaxy that stimulated the encounter nor the location where the encounter occurred .
we havent included star formation in our gas dynamical simulations , a weakness that we share with all past models of quasar fueling by gas in galaxies .
the lack of star formation enables the gas to radiate the random energy resulting from the impulsive torques that diminish its angular momentum .
if the material were all instantly turned into stars , it would not continue its inward flow and the central source would be extinguished .
this caveat applies to all other models that use gas dyanmical simulations such as feuling with mergers ( _ cf .
_ hernquist ) .
however , we offer some observational evidence that the material does remain gas for a fairly long time . in an accompanying paper ( moore , lake and katz 1997 )
, we examine the remnants and compare them to spheroidal galaxies .
we will examine the possibility of black hole hunting " in nucleated spheroidals in the next section .
galaxy harassment only occurs in clusters where the impact velocities are too large to permit merging . by invoking this process to explain the qso hosts below , we make four clear predictions that will each be explored in the next sections : * quasars with low luminosity hosts should be in clusters where harassment occurs
, * resolved hosts should appear disturbed , * regions where harassment is ongoing should show an enhanced quasar frequency * black holes should exist in some nucleated spheroidal galaxies early studies of the environments of quasars concluded that they were not in rich clusters ( roberts , odell and burbidge 1977 , stockton 1978 , french and gunn 1983 ) . that picture has changed and it is now clear that many low redshift quasars are in rich clusters .
the majority of these may lie at the edge of the cluster ( @xcite , green and yee 1984 , yee 1989 ) , but there are a few such as h 1821 + 643 ( schneider 1992 ) and 3c 206 ( yee 1989 ) that appear to lie at the center .
many other quasars show excesses of close optical companions ( hutchings , crampton and campbell 1984 , dahari 1984 , heckman 1984 and yee 1986 ) .
heckman ( 1984 ) obtained redshifts for 15 optical companions and found that 14 out of 15 were physical companions not projections .
several groups have found that the clustering properties of quasars change rapidly with redshift .
ymp93 find that quasars with redshifts @xmath0 occur in abell richness class 0 - 1 clusters of galaxies , an environment that is roughly three times richer than that of the lower redshift quasars .
ye93 reach a similar conclusion , but place the redshift break slightly higher ( @xmath41 ) .
fisher ( 1996 ; hereafter fbks ) looked at a sample of ( @xmath42 ) with a mean redshift of @xmath43 .
they conclude that the quasars reside in stuctures that have at least half as many members as richness class r=0 abell clusters .
so , there seems to be a smooth trend with redshift .
these results agree qualitatively with our model , where the chaos of galaxy formation remains the reason for the observed peak of quasar number densities at a redshift of 2 and merging could dominate at the present day . in the standard hierarchical clustering model , kaufman ( 1995 )
has shown that the infall rate of field galaxies into clusters peaks at redshifts of 0.30.5 . at these redshifts , infalling galaxies
are harassed to produce the central gas flow needed to create a quasar .
the real test will be a measurement of the evolution of the galaxy galaxy and quasar galaxy correlation functions to @xmath8 .
we can also examine what is known about the environments of the quasars in the bks95 sample .
bks95 found 7 companions within 25 kpc of their targets , whereas they estimate the chance occurrence should be 0.375 per target .
we found a wealth of other information on the environments of these quasars . _
pg 0953 + 414 : _ this is one of the richest environments in the green and yee ( 1984 ) sample .
they see 25 galaxies in their field of 140 " ( a radius of 1 mpc centered on the quasar at @xmath44 ) . bks95
detect one companion , there are none evident on the h - band image of mcleod and rieke ( 1994 , hereafter mcr ) . _ pg 1116 + 215 : _ this quasar lies at the edge of a relatively rich cluster discovered by green and yee ( 1984a ) and studied by ellingson , green and yee ( 1991 ) .
bks95 found one companion within 25 kpc . _
pg 1202 + 281 : _
bks95 find two companions .
there is a bright companion that can be clearly seen on mcr s h - band image .
we found no studies of it s broader environment . _
3c 273 : _ bks95 detect a host with a brightness that is greater than , but they see no companions .
mcr see the jet to nw in their h - band frame and a second feature that is just s of w. stockton ( 1980 ) found four galaxies within 250 kpc with velocity differences of -80 , 300 , 530 , and 510 @xmath45 .
this quasar is probably in a poor cluster . _
pks 1302 - 102 : _ bks95 found two companions within 25 kpc .
the limit set by bks95 was just 0.4 magnitudes fainter than .
this quasar was also observed by disney ( 1995 ) who detected a host galaxy using a different filter and detector ( the faint object camera rather than the wide field / planetary camera ) .
when they apply a standard color correction for the elliptical galaxy they detected , they find their detected galaxy is 0.2 magnitudes fainter than the bks95 limit .
this galaxy has been included in several studies of the environments of quasars .
green and yee ( 1984 ) find 10 galaxies on their 140 " field frames .
its cross - correlation with galaxy counts is typical of an abell richness class 0 cluster , but the significance of the cross - correlation is marginal ( yates , miller and peacock 1989 , yee and ellingson 1993 ) . _ pg 1307 + 085 :
_ bks95 did not detect any companions within 25 kpc , but yee ( 1987 ) found one that is just 41 kpc away .
the h - band image of mcr shows two patches of luminosity that are offset from the position of the quasar . _
pg 1444 + 07 : _
bks95 detected a host galaxy with a magnitude of .
it appears to have a bar and ring .
the h - band image in mcr shows luminosity that is offset from the qso .
this is the only quasar in the bks95 sample that was included in green and yee s ( 1984 ) sample , but no rich cluster was found .
_ 3c323.1 : _ oemler , gunn and oke ( 1972 ) declare that this is a qso in a rich cluster of galaxies " .
it lies 6.5 arcmin from the center of a compact zwicky cluster .
this is slightly less than 1 abell radius .
bks95 find a companion 6.9 kpc away that was previously found by stockton ( 1982 ) to have a velocity difference with respect to the qso of @xmath10150 @xmath45 .
three companions can be seen on mcr s h - band image . in summary ,
three of the quasars are known to live in rich clusters of galaxies : pg 0953 + 414 , pg 1116 + 215 , and 3c323.1 . a fourth ( pks 1302 - 102 ) appears to be in a cluster , but the evidence is marginal .
the only quasars that have deep wide field images and do not appear to be in clusters have hosts that are @xmath10 .
the galaxy host of 3c273 is brighter than and it is found in an environment that might be more conducive to merging than harassment ( the velocity dispersion appears to be less than 500 @xmath45 ) .
the only quasar in the green and yee ( 1984 ) survey that does not appear to be in a cluster , pg 1444 + 407 , has the second most luminous host ( @xmath10 ) in the bks95 sample .
more recently , bkss97 completed a study of 20 quasar hosts . in the original bks95 sample ,
half of the quasars had reshifts between 0.239 and 0.236 .
while the newer sample is 2.5 times larger , it added only 2 more galaxies to the redshift range where we expect harassment to become important .
examining the two new quasars in this range , one is observed to be interacting ( 0316 - 416 ) and the other is an elliptical ( pg1004 + 130 ) .
bks95 only found candidate hosts for the 3 qsos : pg 1116 + 215 , 3c 273 and pg 1444 + 407 .
only the host of 3c 273 is clearly as bright as and it appears to be an elliptical slightly offset from the position of the qso .
the hosts of the other two galaxies are about a half magnitude fainter than when bks95 use aperture magnitudes and slightly brighter than if they employ an exponential disk model
. the pg 1116 + 215 host appears to have partial rings while the pg 1444 + 407 host appears to have a bar and a ring - like structure .
having failed to confirm past detections of qso hosts , bks95 are circumspect in their descriptions of their images .
however , we are encouraged by the structures described for the two less luminous hosts . strong bars and rings
are often seen in harassed galaxies .
the phenomena of galaxy harassment was discovered in our study of the destruction of substructure in mergers ( mkl96b ) . in moore ( 1996 ) , we first applied it to explain the evolution of cluster populations discovered by butcher and oemler ( 1978 ) .
harassment is clearly occurring in these clusters .
if this is the cause of nuclear activity in quasars found in sub - galaxies , then we expect that the frequency of agns should be higher in clusters at @xmath46 that show the butcher - oemler effect .
we find the data to check this prediction to be extremely puzzling .
when they first embarked on their decade long spectroscopic study of these clusters , dressler and gunn ( 1982 , 1983 , see also dressler 1987 ) reported that the frequency of agns in these clusters was much higher than the 1% found in @xmath47 clusters ( dressler , thompson and shectman 1985 ) .
a high fraction of agns was confirmed by lavery and henry ( 1989 ) , but couch and sharpless ( 1985 ) found that only 1 in 112 of the blue galaxies they examined were agns .
we find no clear consensus on this issue and this remains a testable prediction of the model .
we do note that our first and third predictions have some redundancy .
quasars at intermediate redshifts could not lie in clusters rich enough to be classified by abell ( yates , miller and peacock 1989 , and yee and ellingson 1993 ) if nuclear activity is not enhanced in clusters . with sensible assumptions about their duty cycle ,
the dead remnants of quasars are inferred to have number densities comparable to galaxies brighter than ( c.f . haehelt and rees 1993 and references therein ) .
this has stimulated a search for massive objects in the centers of bright galaxies ( c.f .
kormendy and richstone 1996 ) .
our candidate hosts have a much greater spatial frequency and are associated with quasars at lower redshifts where they are relatively rare .
hence , a tiny fraction of the spheroidals are expected to have massive nuclei .
however , the low velocity dispersions of the spheroidal galaxies should make it easier to detect dark matter in their cores .
peterson and caldwell ( 1993 ) and nieto ( 1990 ) have conducted the largest studies of spheroidals and dwarf ellipticals .
all together , there are fewer than two dozen galaxies . of these , ngc 4486b
( nieto 1990 ) is one of the most unusual .
it has an absolute magnitude , @xmath48 , but its central velocity dispersion is measured to be @xmath49 , falling to @xmath50 when averaged over its half light radius .
it s mass - to - light ratio , is on the high end , @xmath51 , but hardly extreme .
the detection of low level nuclear activity in spheroidals would also be of great interest .
recent hst results have challenged the conventional picture that quasars are associated with high luminosity hosts . we have presented a model galaxy harassment that rapidly channels gas into the centers of low luminosity hosts .
this model has several predictions .
we find evidence that most if not all of the quasars with sub - hosts are in the high density environments that lead to harassment .
the hst images of host candidates by bks95 shows tantalizing evidence of the distortions associated with harassment . in attempting to test the prediction of a higher frequency of agns in butcher - oemler clusters
, we find some support amidst active controversy .
this research was supported by nasa s high performance computing and communications earth and space sciences program , nasa s astrophysical theory program and nasa s long term space astrophysics program .
our search for environmental information was aided by the nasa extragalactic database ( ned ) .
bahcall , j. n. kirhakos , s. , saxe , d. h. and schneider , d. p. 1995
, , 479 , 642 ( bkss97 ) bahcall , j. n. kirhakos , s. and schneider , d. p. 1995
, , 450 , 486 ( bks95 ) bender , r. and nieto , j .- l . 1990 , , 239 , 97 butcher , h. and oemler , a. 1978 , , 219 , 18 couch , w. and sharples , r.m .
1987 , , 229 , 422 dahari , o. 1984 , , 89 , 966 disney , m. , 1995 , nature , 376 , 150 dressler , a. and gunn , j. e. 1982 , , 263 , 533 dressler , a. and gunn , j. e. 1983 , , 270 , 7 dressler , a. 1987 , nearly normal galaxies : from the planck time to the present , s. m. faber , new york : springer - verlag , 265 dressler , a. , thompson , i.b . and shectman , s.a . 1985 , , 288 , 481 .
fisher , j. r. and tully , b. 1981 astron .
47 , 139 french , h.b and gunn , j.e .
1983 , , 270 , 7 hutchings , j. b. , crampton , d. and campbell , b. 1984 , , 280 , 41 kormendy , j. and richstone , d. , 1996 , anual rev .
, 33 , 581 ellingson , e. , green , r. f. and yee , h. k. c. 1991 , , 378 , 476 fisher , k. b. , bahcall , j. n. , kirhakos , s. and schneider , d. p. 1996
, , submitted .
green , r. f. and yee , h. k. c. 1994 , , 54 , 495 haehnelt , m.g . and rees , m.j .
1993 , , 263 168 .
heckman , t. m. , bothun , g. d. , balick , b. and smith , e.p .
1984 , , 89 , 958 hernquist , l. 1989 , nature , 340 , 687 hernquist , l. and katz , n. 1989 , , 70 , 419 katz , n. , weinberg , d. and hernquist , l. 1996 , , 105 , 19 kauffmann , g. 1995 , , 274 , 153 le fvre , o. , hudon , d. , lilly , s.j . ,
crampton , d. , hammer , f. and tresse , l. 1996 , , 461 , 534 lavery , r.j . and henry , p. 1988
, , 330 , 596 mcleod , k. k. and rieke , g. h. 1994 , , 431 , 137 moore , b. , katz , n. , lake , g. , dressler , a. and oemler , a. 1996 , nature , 379 , 613 .
moore , b. , katz , n. and lake , g. , 1996 , , 457 , 455 ( mkl96b ) moore , b. , lake , g. and katz , n. 1997 , , this volume nieto , j.l , davoust , e. , bender r. , and prugniel , p. 1990 , astron.astro .
, 230 , 17 oemler , a. , gunn , j. e. and oke , j. b. 1972 , , 176 , l47 .
peterson , r. c. and caldwell , n. 1993 , , 105 , 1411 .
schechter , p. l. 1976 , , 203 , 297 schneider , d. p. , bahcall , j. n. , gunn , j. e. and dressler , a. 1992 , 103 , 1047 stockton , a. 1980 , objects of high redshift , g.o .
abell and p. j. e. peebles , dordrecht : reidel , 133 stockton , a. 1982 , , 257 , 33 toomre , a. 1964 , 139 , 1217 van der marel , r.p .
, 1991 , , 253 , 710 white , s.d.m . , and rees , m. 1978 , , 183 , 341 yates , m. g. , miller , l. and peacock , j. a. , , 240 , 129 yee , h. k. c. 1987 , observational evidence of activity in galaxies , e. ye .
khachickian , k. j. fricke and j. melnick , dordrecht : reidel , 363 .
yee , h. k. c. and green , r. f. 1991 , , 280 , 79 yee , h. k. c. , ellingson , e. , green , r. f. and prichet , c. j. 1989 , the epoch of galaxy formation , c. s. frenk , dordrecht : kluwer , 185 .
yee , h. k. c. and ellingson , e. 1993 , 411 , 43 * figure 1 * snapshots of the evolution of the gas disk in one of the harassed galaxies that we simulated . the full time evolution can be seen in the last segment of the video .
the disk shown is the one where 90% of the gas is transported into the central kpc ( see figure 2 ) .
* figure 2 * the time evolution of the fraction of gas that lies within 1 kpc of the center of the galaxy . on average ,
more than half of the gas lies within this radius within a few gyr in one of the three cases , 90% of the gas is in the center . in the two cases where the most mass is pushed into the center ,
about half of it is transferred during an interval of just 100 - 200 myr . | we have simulated disk galaxies undergoing continual bombardment by other galaxies in a rich cluster .
galaxy harassment " leads to dramatic evolution of smaller disk galaxies and provides an extremely effective mechanism to fuel a central quasar . within a few billion years
after a small disk galaxy enters the cluster environment , up to 90% of its gas can be driven into the inner 500 pc . up to half of the mass can be transferred in a burst lasting just 100 - 200 myr .
this transport of gas to the center of galaxy is far more efficient than any mechanism proposed before .
galaxy harassment was first proposed to explain the disturbed blue galaxies in clusters seen in clusters at ( @xmath0 ) , the butcher oemler effect " .
quasars at the same reshifts lie in more clustered environments than those at lower redshift .
recent hst observations find that roughly half of all observed quasar host galaxiess are fainter than , with many of these less luminous hosts occuring at redshifts @xmath0 .
we examine 5 quasars that are claimed to have low luminosity hosts and find that 3 are in rich clusters of galaxies , the fourth may be in a cluster but the evidence for this is marginal .
the environment of the fifth has not been studied .
epsf @xmath1 0.5truecm * george lake@xmath2 , neal katz@xmath3 and ben moore@xmath4 * * 1 .
* astronomy department , university of washington , seattle , wa 98195 , usa
* 2 .
* department of physics & astronomy , university of massachusetts , amherst , usa * 3 .
* department of physics , university of durham , durham , uk |
in the current era of observational astronomy the size of galaxy datasets means that number statistics starts to lose its spot as number one source of uncertainty in galaxy evolution studies . with the availability of large wide field galaxy surveys as the sloan digital sky survey ( sdss * ? ? ? * ) or 2df galaxy redshift survey , and the deep , space based high resolution projects like the hubble ultra deep field , goods , gems and stages projects and the cosmic evolution survey ( cosmos ) with their @xmath9 to @xmath10 galaxies , other error sources become vital to understand .
if we want to understand the buildup of galaxies with their intricately linked evolution in stellar and black hole mass , luminosities , colors , morphological types , and the alternations between interaction and relaxation , we have to understand what the tools we apply really measure .
any given galaxy will look different when viewed with different instruments or when located at different redshifts , due to cosmological dimming and changes in rest - frame bandpass .
this makes the qualitative or quantitative classification of e.g. galaxy morphology non - trivial to compare for different redshifts or filters .
as one example faint disks visible at one redshift will fade from the observer s view at larger distances , not only affecting eyeball - classifications of morphology but also automatic classifier software .
moreover , low surface brightness signs of interaction and structures , such as bars which are not prominent in the rest - frame ultraviolet , will not be visible at higher redshfits . if the evolution of galaxies is to be studied via merger statistics , morphology - segregated evolution , star formation histories , and type - specific luminosity functions , all morphological classifiers have to be calibrated to deliver comparisons of similar quantities at all redshifts , taking into account bandpass - dependent properties and changes in signal - to - noise ratio . from the analysis of hst survey data
we know that the average surface brightness of the disc galaxy population fades with time ( e.g. * ? ? ?
* ; * ? ? ?
the change is approximately 1 - 1.5mag depending on rest - frame bandpass over the redshift interval @xmath11 . to some degree
this surface brightness evolution counters the cosmological dimming and helps detecting low surface brightness features .
thus , if this effect is not taken into account , predictions about the recoverability of structural parameters or classifications are overly pessimistic .
for a few projects in the past , codes were created that would include some of these effects for specific applications or datasets ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. however , to our knowledge no codes or datasets that include geometrical and cosmological bandpass shifting effects are currently publicly available . in the present article
we describe the creation of artificial image data sets in the range @xmath0 , computed from low-@xmath8 sdss galaxies by artificial `` redshifting '' .
since we refrain from adding any evolutionary models but purely apply cosmological changes in angular size , surface brightness and filter bandpass , this dataset shows exactly how such galaxies will appear when observed from cosmological distances , at which redshift certain features become undetectable , and any quantitative classifier procedure can be tested for dependency on redshift effects .
our code ferengi , `` full and efficient redshifting of ensembles of nearby galaxy images '' , and the present datasets were originally created for the morphological classification of active and inactive galaxies in the gems @xcite , stages @xcite , and cosmos @xcite projects .
the datasets for the hst acs image characteristics of the gems and cosmos project are freely available from our website . as many studies might benefit from such data we also make the ferengi code available on the same webpage for others to use with different input samples , redshifts , and instrument characteristics . in the following
we describe the input data in section [ sec : input ] , the redshifting procedure including the basic cosmological formulae that enter , and the bandpass shift in section [ sec : redshifting ] , and the creation of realistic images from this information in section [ sec : noise ] including example results ( section [ sec : results ] ) .
next , we present a series of tests characterising the robustness and accuracy of the code ( section [ sec : tests ] ) . in section [ sec : conclusions ] we discuss limitations of this procedure . all examples and cosmology - dependent numbers given here are computed assuming a flat universe with @xmath12 and @xmath13 .
the basic idea is to convert images of well resolved , low - redshift galaxies to images simulating the same galaxies at higher redshift .
input and output images can differ in assumed redshift , pixel size , point spread function ( psf ) and noise properties .
we explicitely compute two versions , with and without luminosity / surface brightness evolution terms .
when purely considering instrumental and cosmological effects , users of such datasets can apply his or her tools to the exactly same galaxies , only located at different redshifts for calibration purposes . when including evolution , more realistic galaxies are created . for the current study
we have implemented a linear scaling with redshift to make sources brighter at high redshift .
for this project we require input galaxy data with two main properties : i ) sufficient sampling : the combination of distance to the galaxy , pixel size and width of the psf must be sufficient for the target pixel size and psf width at the lowest target redshift ( e.g. @xmath14 ) .
ii ) information about the spectral energy distribution : the task needs homogeneous multiband imaging used for interpolation between filters to compute fluxes in the target filter band at the desired redshift without extrapolation .
the best source for such data are nearby galaxies from the sdss .
this survey provides imaging data in its own @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 filters with 0396/pixel sampling .
we compute the maximum distance for input galaxies from the two requirement to i ) simulate images with 005 and 003 pixel size at a minimum redshift of @xmath15 , and ii ) to have an output psf not already broader than the @xmath16 ( fwhm ) psf of the acs wfc drizzled images in the f606w or f814w filters .
figure [ fig : sampleselection ] shows the resulting limit in @xmath17 and the limit from the width of the psf for a sample of sdss galaxies and their psf conditions .
both criteria select a similar set of galaxies , corresponding to limiting recession velocities of @xmath18 and @xmath19 kms@xmath3 for @xmath20 and @xmath21 , respectively , and @xmath22 output pixel size . for pixel size 005 the limits change to @xmath23 and @xmath24 kms@xmath3 .
these technical aspects define the boundary conditions of our sample selection .
we emphasise that at this point we do not aim to select a sample that is complete or representative for the general population of galaxies in a statistical sense .
the sample selection only intends to span a large range of morphological types and common features , targeting large , rather luminous galaxies that could still be observed at larger @xmath8 . with this intention
we make a selection of galaxies from the sdss survey area , via the hyperleda @xcite catalogue facility , with the following parameters : recession velocity @xmath25 corrected for virgo infall ( 208 kms@xmath3 ) , apparent isophotal diameter @xmath26 arcmin , and apparent magnitude @xmath27 , and added a total @xmath28-band magnitude cut at @xmath29
. images were taken from the data release 4 ( dr4 ) of the sdss @xcite .
as the sdss imaging data is not targeted on individual galaxies , a fraction of about 40% of the selected galaxies extends beyond the borders of the 680@xmath30590 arcsec@xmath31 sdss field of view .
we excluded such galaxies as input after visual inspection . in total we selected a sample of 96 galaxies with @xmath32 km s@xmath3 and very heterogeneous morphologies .
figure [ fig : sample ] shows the hubble diagram of the input sample . in the same figure and in table [ tab : sample ] we give the distribution of morphological classification according to the _ third reference catalogue of bright galaxies _
( rc3 , * ? ? ? * ) to illustrate the sample composition .
note , that the class `` peculiar '' was manually selected for showing signs of ongoing mergers , or close or overlapping galaxies .
lc class&number of galaxies + e & 5 + s0 & 8 + early s & 16 + late s & 40 + irr & 5 + pec & 22 + as in the local universe few galaxies show strong signs of interactions or even mergers , we gradually extended our selection range from @xmath33 out to @xmath32 .
at larger distances we only included `` peculiar''-looking sources like strong interactions or mergers .
however , we also included some unrelated objects that have small projected radii or even overlap .
our intention here was to demonstrate the increasing difficulty at high redshift to morphologically discern interaction from projection effects .
the ferengi redshifting procedure has two components , cosmological angular size and surface brightness changes on one side and bandpass shifting on the other .
after computation of these cosmological effects we correct for psf effects and add noise . when converting from an input image with pixel size @xmath34 of a galaxy at redshift @xmath35 to an output redshift @xmath36 and desired pixel size @xmath37 both angular sizes and surface brightnesses have to be modified for distance and cosmological effects . using @xmath38 for small angles , the angular size @xmath39 of a given linear dimension changes as @xmath40 with the luminosity distance @xmath41 .
since the input image will be rebinned to the output pixel size , angular sizes in units of pixels @xmath42 , and thus image rebinning factors , are @xmath43 the flux in each pixel is subject to surface brightness dimming .
if we require the galaxy absolute magnitude to be conserved @xmath44 we find a relation for the ratio of observed fluxes @xmath45 of the in- and output images @xmath46 which gives the standard bolometric surface brightness dependence of @xmath47 .
the finite filter width is taken care of in the next component .
note , that in order to account for the bandpass shift it is imperative to match object positions in the input images .
this matching is a complex process , which potentially includes shifting , rotation and scaling of the input data .
moreover , the psfs of the data should match . in the case of strongly varying psfs colour terms
may be introduced in the bandpass shifting .
when combining e.g. galex and sdss data , to extend the wavelength baseline to the uv , the users would have to prepare the input image accordingly .
they have to shift all images to a common position , scale the galex image to the sdss pixel scale , possibly rotate the galex image to match the sdss frame and smooth all sdss images to the galex fwhm .
the so - processed images are then input to ferengi .
thus , preparing the input appropriately , ferengi is not limited in the combination of telescopes , filters , or instruments .
while the above rebinning and flux rescaling takes care of all geometrical effects , bandpass shifting and stretching still has to be added .
the cosmological redshift will ( by definition ) shift the observed restframe bandpass for a given observing filter as a function of redshift as well as change the size of a given filter .
in addition we want to have the option to produce images for a range of optical observed filters . for this task we input spectral information in the form of multiband imaging .
a combination of observers filter and redshift defines the desired rest - frame filter curve . for each pixel of the input frame
we calculate the expected flux for this rest - frame filter , by interpolating between multiband information on this pixel .
this task is facilitated by the routine k_correct(v4.14 , * ? ? ?
* available for idl and c ) that was written to fit spectral templates to a set of multiband images .
we use k_correct to determine a best template for individual pixels in an image , and from this the flux in each pixel for a given filter .
we so construct a rest - frame filter image . this bandpass shift and the above size - rebinning are interchangeable in order . to minimize the computational effort and to reduce noise we first recomputed the ( coarser ) redshifted images and then applied the bandpass shift .
also , only pixels with flux exceeding 2 times the rms of the background are input to k_correct ( optional feature ) .
the remaining pixels receive a flux - weighted @xmath48-correction computed from the bright pixels . while this could become more important at low s / n levels , it showed not to have a significant effect for the currently chosen bright sample of galaxies .
note , the template set incorporated in k_correct stems from @xcite models .
these model templates cover the rest - frame wavelength range from uv to ir ( 600 to 320@xmath49 m ) . as long as the input filters together with the targeted output redshift fall within this range any combination is possible .
two essential ingredients for realistic images are i ) to mimic the resolution of the real data by convolution with an appropriate psf and ii ) to generate realistic noise levels for galaxy and background .
the input sdss galaxies were observed under varying seeing conditions with an average psf width of 14 fwhm corresponding to @xmath13.5 sdss camera pixel .
after application of the redshifting procedure the intrinsic psf is of finite width and non - negligible . while for point- or composite sources as type i agn the detailed knowledge about the psf is essential , the requirements are not so stringent for extended galaxies
however , as the simulated dataset is meant to calibrate quantitative morphological analysis methods and programs , it is important to create a psf that is sufficiently similar to the real psf for the task in question . in light of this
we attempt to reach a final psf as close as possible in shape and width to the acs psf in the desired band .
this involves the creation of a convolution kernel that will produce the acs psf shape from the input psf shape .
as the width and geometric shrinking of the psf depends both on input and target redshift , and the input psf varies , this kernel needs to be recomputed for each input galaxy and output redshift .
this is done by a transformation into fourier space , suppression of noise by wiener filtering , and subsequent division of the spectra of the two psfs .
after transformation of this quotient back into the spatial domain we receive the function that is needed to convolve the input image to reach a psf close to the acs psf .
this is mathematically equivalent to a deconvolution of the output psf with the desired input psf . for the comparably high s / n psf in these data
this works fairly well .
note , that the filter of the input psf is chosen from the set of sdss filters minimising the wavelength difference to the desired rest - frame band , thus also minimising colour gradients in the reconstructed psf .
however the seeing is wavelength dependent and can also vary on the short timescales between the integration in the different sdss filter bands .
this can still result in small differences between the actual psf present for a reconstructed galaxy image and the reconstructed psf image . in order to remove this wavelength dependency
we choose a well sampled , not saturated star in each sdss filter as psf .
we convolve the sdss images with a kernel ( as described above ) that generates a round gaussian psf with a fwhm for all filters of 1.1@xmath30 the largest extent of the input psfs .
thus , psf resolution effects and colour gradients are kept at an absolute minimum .
a comparison of the reconstructed and target acs psf shows a difference of less than 0.1% in each individual pixel .
this procedure is limited to cases where the widths of in- and output psf are sufficiently different . if not
, noise will introduce artifacts as ringing patterns or mathematical ghost images near bright sources , which is clearly not desirable .
this occurs primarily when the input psf resolution becomes comparable to the output psf width
so the convolution function becomes very narrow
i.e. at the low redshift end .
in order to facilitate proper modelling of the redshifted galaxy images we discourage the use of the publicly available gems or cosmos psf , but suggest to prefer the reconstructed psf instead .
although this might not be crucial for determining two - dimensional galaxy light profiles , it might make a significant difference in the case of agns and their host galaxies .
likewise , we recommend using scaled versions of the reconstructed psf to be put on top of redshifted galaxy images in order to create artificial agns .
noise in the output images has two main sources , sky background and the galaxy flux itself . for one - orbit exposures as for cosmos and gems the sky background noise dominates over the readout noise ( @xmath15.2 @xmath50 for the acs wide - field camera ) after a few 100 seconds of integration , in the case of broad band filters .
thus the latter is negligible .
also , at 0.0022 @xmath50 @xmath51 the acs / wfc dark current does not play a significant role .
the background noise in our simulations is not created from random numbers , but the noise - added galaxy images are added onto blank sky taken from the observed data itself ( for acs data reduction for gems and cosmos see * ? ? ?
* ; * ? ? ?
in this way reduction signatures like correlated noise and intrinsic small variations in the otherwise empty regions of the sky are also present in the simulated data .
we extract several 60@xmath52 square fields from the data . while we choose comparably empty regions of the sky , the acs camera is so efficient over one orbit for both the cosmos and gems filters that there exists no `` empty '' sky over more than 1020@xmath52 regions .
to remove prominent remaining objects we replace the corresponding pixel regions with unique other small patches of blank sky .
this process does not involve any filtering , only replacing . in this way the resulting empty sky fields are very good random representations of empty sky regions .
if for an application the sensitivity of a measurement to small residual back-/foreground objects is to be tested , the user can add his / her favorite contamination or even use random patches of uncleaned sky as background .
noise has to be added also to the redshifted galaxy images themselves .
the s / n per pixel of the input galaxies is high by definition of the sample .
after redshifting and psf adaptation their photon noise is negligible compared to the noise expected from a single orbit hst acs exposure .
the resulting images are scaled in flux corresponding to e.g. 2028 seconds ( cosmos ) integration time , in the respective output filter . the galaxy image in units of electrons per pixel then has random poisson noise added , with a @xmath53 corresponding to the galaxy flux per pixel .
subsequently , the galaxy is added on top of the empty background images . in a statistical sense
this procedure is not strictly correct , as the photon noise addition should be applied to the sum of sky and galaxy , but can not be avoided if real sky images are to be used .
however , the resulting higher noise only appears in regions of the final image where sky background and galaxy contribute equally in noise , so in the faint isophote regime of the galaxy .
c p3.4 cm p3.8 cm p4.2 cm c c 0 & fit real images using galfit & galfit parameters & & @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 & - + 1 & from output of 0 create local & artificial sdss galaxies & double - check with 0 & @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 & - + & smooth model images & fit & & & & + 2 & create high - z smooth & smooth model images at high - z & galfit response to decreasing s / n & @xmath4 , @xmath5 , @xmath6 & [ fig : ell01 ] + & model images & fit & & & & + 3 & redshifting of output from 1 & redshifted , non-@xmath48-corrected & impact of downscaling , psf effects & @xmath4 , @xmath5 , @xmath6 & [ fig : ell02 ] + & ( @xmath48-corr .
disabled ) & fit & smooth model images & & & + 4 & redshifting of output from 1 & fully redshifted , @xmath48-corrected & impact of @xmath48-correction & acs filter & [ fig : ell03 ] + & ( @xmath48-corr . enabled ) & fit & smooth model images & & & + 5 & redshifting of 0 & fit & redshifted real images & in comparison with 4 , & acs filter & [ fig : ell06 ] + & & & impact of morphology & & +
artificially redshifted galaxies are created for a redshift range out to @xmath54 .
the starting redshift was taken for each galaxy either @xmath15 or the minimum redshift at which both input psf width and pixel size begin to match the output psf and pixel size .
we give images for steps of @xmath55 out to @xmath56 and @xmath57 beyond that , thus a maximum of 15 output redshifts per input galaxy .
simply shifting local galaxies out to high redshift makes them look rather faint in comparison to real average galaxies at such distances . in order to reflect the brightness increase of high redshift sources we put in a crude mechanism to introduce evolution
this optional feature allows to make galaxies brighter as a linear function of redshift : @xmath58 setting @xmath59 would make a galaxy 1 mag brighter at redshift @xmath60 as it would normally be . yet
, this option is not meant to be a substitute for real morphological or photometrical evolution and does not replace stellar evolution codes .
the reason for putting in such a simple functional form is the application of galaxy classification by eye .
if one is to re - identify galaxies shifted to high redshift , the task is made increasingly unfair compared to real galaxies , which on average do become brighter at higher redshift , if one does not apply artificial brightening .
a few resulting example images for cosmos are shown in figure [ fig : examples ] .
all images , for both gems and cosmos are made available electronically .
in order to demonstrate the accuracy of ferengi and characterise its limits , we perform a number of tests .
we use the programme galfit @xcite to determine structural parameters , which is often used for parameter estimation and morphological classification of galaxy samples , particularly in survey applications .
the programme is ( potentially ) susceptible to s / n changes and morphological @xmath48-correction , like any other fitting code .
yet , @xcite have shown that galfit is very robust and does not exhibit systematic biases when fitting radial surface brightness models to artificial 2d srsic - type .
for @xmath61 the profile is a simple exponential curve , typical of disc galaxies , for @xmath62 the profile becomes a de vaucouleurs exponential @xmath63 profile . ]
light profiles . in the following section we make use of this robustness and apply galfit to various sets of simulated 2d srsic profiles to determine the dependence of structural parameters on the application of ferengi .
in addition to galfit we use a code described in @xcite to create 2d srsic profiles mimicking our observed galaxies .
this procedure features a fine subsampling of the inner pixels allowing highly accurate flux calibration .
convolution is done with the original psf as reconstructed from the sdss , after smoothing the individual five bands to a round gausssian configuration .
poisson noise and a real sdss sky background image are added .
such simulated images are created for all five sdss bands @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 .
we conduct the simulations in six different steps to check for possible systematics introduced at each step .
the summary of each step is shown in table [ tab : tests ] .
we show by way of illustration an example of an elliptical galaxy ( fig . [ fig : example_ell ] ) .
it has an almost de vaucouleurs - type light profile ( @xmath42 increases from 3.2 to 3.6 going from @xmath4 to @xmath6 band ) .
the size of the galaxy is constant across the three bands ; the magnitude covers a range of over 2.5 magnitudes
@xmath4 being faintest and @xmath6 brightest , as expected for a red elliptical .
@xmath64 we start by using galfit to fit a srsic profile to the set of sdss images belonging to the objects .
this provides the basis for simulating the smooth model images .
@xmath64 with the galfit parameters we create artificial model images using the procedure by @xcite described above , resembling smooth versions of the sdss galaxy images .
a fit to these smooth images in all bands confirms the stability of the process by retrieving almost exactly the input values .
@xmath64 next we create artificial model images of the same galaxies , as they would appear at high redshift ( our set of output redshifts given in section [ sec : results ] ) using the parameters from before , in the @xmath4 , @xmath5 , and @xmath6 bands .
we do not apply ferengi here , but only convert the linear scales according to redshift .
this will quantify the response of galfit to the decreasing s / n as a result of cosmological surface brightness dimming .
figure [ fig : ell01 ] demonstrates how galfit performs as a function of redshift for the example galaxy ( no . 2 from table [ tab : tests ] ) .
the higher the redshift , the lower is the s / n of the object and its surface brightness .
quite expectedly , the quality of the recovered fit parameters degrades with redshift . as the @xmath4-band has the lowest s / n of all sdss filters , galfit fares worst there ; @xmath6-band behaves best , having the highest s / n .
the chosen galaxy being rather red , this effect is even amplified .
as the galaxy is rather bright , overall , the deviations in the parameters from their input values are rather small . in order to allow a comparison with gems , we compute the average apparent surface brightness of our sample galaxies with @xmath65 in @xmath4 , @xmath5 and @xmath6 . from @xcite
we obtain the corresponding galfit errors in gems .
the results are listed in table [ tab : gemserrors ] .
ccccc z & band & @xmath66 & @xmath67 & @xmath42 + & & [ mag ] & [ % ] & [ % ] + & @xmath4 & 0.05 @xmath68 0.42 & -0.06 @xmath68 0.41 & -0.03 @xmath68 0.27 + 0.2 & @xmath5 & 0.01 @xmath68 0.15 & 0.00 @xmath68 0.22 & -0.01 @xmath68 0.21 + & @xmath6 & 0.01 @xmath68 0.13 & 0.00 @xmath68 0.20 & -0.01 @xmath68 0.20 + & @xmath4 & 0.10 @xmath68 0.50 & -0.08 @xmath68 0.48 & -0.05 @xmath68 0.28 + 0.4 & @xmath5 & 0.02 @xmath68 0.20 & -0.01 @xmath68 0.27 & -0.01 @xmath68 0.22 + & @xmath6 & 0.01 @xmath68 0.17 & 0.00 @xmath68 0.24 & -0.01 @xmath68 0.21 + & @xmath4 & 0.18 @xmath68 0.61 & -0.15 @xmath68 0.50 & -0.06 @xmath68 0.31 + 0.8 & @xmath5 & 0.03 @xmath68 0.32 & -0.04 @xmath68 0.35 & -0.02 @xmath68 0.26 + & @xmath6 & 0.02 @xmath68 0.27 & -0.03 @xmath68 0.32 & -0.02 @xmath68 0.25 + @xmath64 now we use ferengi on the simulated smooth @xmath4 , @xmath5 , and @xmath6 band model images to create versions of these images appropriate for higher redshifts .
the srsic fits to this set of images we use to determine the influence of conversion to a new pixel grid and related psf on different galaxy parameters , as a function of redshift .
note that the @xmath48-correction code is _ disabled _ for this purpose .
the image set contains the flux of the sdss filters transformed to a new pixel size , psf , background noise properties and the cosmological effect of surface brightness dimming
. for the example galaxy the left hand side of figure [ fig : ell02 ] shows the same as figure [ fig : ell01 ] ; the right hand side has the values from figure [ fig : ell01 ] subtracted in order to indicate the extra influence of the re - gridding process .
we find no additional systematics and the scatter hardly increases .
@xmath64 after that we _ enable _ @xmath48-correction , thus fitting an sed to all five sdss bands and extracting the flux at the position of the observed acs filter , in addition to the down - scaling and surface brightness dimming .
this provides a set of images representing the simulated sdss galaxies as they would look at high redshift including all cosmological effects ( no . 4 from table [ tab : tests ] ) .
the result for the example galaxy is shown in figure [ fig : ell03 ] . in the top panels we plot the original input values . also shown
are the fit values from using the @xmath48-correction code . in order to provide a rough estimate of the deviation from the expected values we fit the five values from the simulated sdss images with a polynomial and subtract this fit .
going into all the details of modelling galaxy photometry and structure in multiple wavebands ( or even continuously ) is well beyond the scope of this paper .
therefore , we regard the lower panel not as a proper estimate of the error budget , but rather an indication of systematic trends . points missing from the lower panel were left out in order not to overstretch the plotting axis and to focus on the main region of interest .
@xmath64 finally , we run ferengi on the real sdss images including @xmath48-corrections ( no . 5 from table [ tab : tests ] ) .
the galfit results for this series reveals in comparison to the redshifted simulations ( no . 4 from table
[ tab : tests ] ) the impact of morphology on the whole process ( see figure [ fig : ell06 ] ) .
as the performed calculations are virtually identical any differences in the output must result from irregularities in the real data , such as tidal features , spiral arms , dust lanes , etc .
the example galaxy was chosen to be rather featureless . therefore , a perfect match is obtained . yet , in the majority of cases we find systematic differences : either redshift dependent deviations ( originating from irregularities ) or global shifts towards higher or lower values . at the highest redshift ( @xmath69 )
the impact of morphology might decrease again as virtually all features are smeared out .
yet , the reason for global offsets is of technical nature : the drastic change in resolution and depth when transforming a low redshift image ( @xmath17 of a few thousand km / h ) to high redshift ( @xmath70 ) results in the rebinned psf not being nyquist sampled any more .
this causes shifts when performing fourier transformations .
the effect is more pronounced at low ( @xmath70 ) than at high ( @xmath69 ) redshift , because eventually the psf becomes pointlike .
we applied ferengi to all 96 sample galaxies .
this allows us to quantify systematic biases of the redshifting code in some detail . in figure
[ fig : mean01 ] we show the average deviation from the input values for simulations of high redshift galaxies and downscaled versions of simulated galaxies ( no . 2 & 3 from table [ tab : tests ] ) .
the errors and systematic offsets are not significantly different . for the given observational setup ( redshifting of sdss images to cosmos and gems ) roughly at @xmath71 larger deviations in particular in @xmath4-band occur .
note , that these depatures from the expected mean are seen in the simulated images as well .
therefore , they must originate from the specific galfit setup , but not ferengi itself . in order to compare the results of the high redshift simulations and the downscaling - only simulations ( no . 2 & 3 from table [ tab : tests ] ) with the @xmath48-corrections - enabled setup ( no .
4 ) , we plot as a function of redshift the measurements in the sdss filters closest to the rest - frame acs filter ( figure [ fig : mean02 ] ) .
interestingly , the error bars in the right panel of figure [ fig : mean02 ] ( no . 4 ) at the highest two redshift bins are _ smaller _ than without @xmath48-corrections ( no .
3 ) or even in the pure simulations ( no . 2 ) .
the reason for this is , that the sed fitting uses all five sdss bands and therefore introduces some information extrapolated from @xmath5 , @xmath6 , @xmath7 and @xmath8 ( the closer bands being weighted more strongly than the redder bands ) to improve the low s / n @xmath4-band data .
however , this implies also that template mismatches might change the morphological appearance of the object to some extent . as the sed - fitting code
is deeply embedded in k_correct we do not attempt to characterise the impact any further .
moreover , we find , that the @xmath48-correction code on average introduces stronger fluctuations and increases the error bars slightly at lower redshifts . systematic deviations to lower brightnesses , radii or srsic indices are not statistically significant .
we take from figures [ fig : mean01 ] and [ fig : mean02 ] that ferengi has no unwanted systematic effects on measured morphological and photometrical parameters of simulated galaxies . the deviations from theoretical values
are well within the uncertainties expected from , in this case , galfit , also in absolute terms .
both linear scales as well as magnitudes are well reproduced .
this includes the bandpass shifts induced by redshifting and different filters , as is demonstrated in the right panels in figures [ fig : mean02 ] : the k_correct module correctly interpolates between input filter images .
we can not exclude issues when _ extrapolating _
k_correct templates , but we explicitly restrict the output range of redshifts to only use template _ interpolation _ , between bands .
we make ferengi publically available from our webpage , as are redshifted sets of images for redshifts @xmath72 and different hst acs filters .
primarily these are created for the gems ( f606w and f850lp at 003/pixel ) and cosmos ( f814w at 005/pixel ) projects , but the code can be modified and used for other filters .
the user can extend the input sample of images ( also provided on the webpage ) , to other objects , filter curves , and output redshifts , and in principle also to other data sources .
we leave it up to the user to update ferengi with respect to the @xmath48-correction module when including other input filter bands .
as stated , templates are used for interpolation between input filters .
they can theoretically also be extrapolated beyond the input wavelength interval , if one has faith in the templates .
however , we strongly advise against this for a different reason : due to the limited s / n available for the pixel - by - pixel @xmath48-correction that is computed , a mild extrapolation might be acceptable , but the noise of the output pixels and systematic deviations will obviously increase with distance to the last supported wavelength . as various templates might be degenerate when interpolating , there might be striking differences on extrapolation , resulting in large errors in the output flux . as a second caveat we note that stars in the input images are not treated separately in ferengi or in the provided datasets .
_ any _ flux in the input images , be it stars or back-/foreground galaxies , will be treated as if it were at the distance of the target and thus be redshifted .
while distant background galaxies usually disappear , stars will be represented with the output psf , resembling a faint dense star field ( the density of stars is enhanced quadratically with decreasing linear scales as the simulation redshift increases ) .
of even greater concern are foreground ( or closer background ) galaxies . in the output , their physical properties
are calculated falsely as their assumed distance is not correct . whether the resulting images will in the end be reliable representations of reality , strongly depends on the quality of input images , the type of galaxies , and their correspondence with the templates used .
the authors gratefully acknowledge numerous helpful discussions with chien y. peng and hans - walter rix .
this publication makes use of the sloan digital sky survey , http://www.sdss.org .
we selected the sdss data using the hyperleda database ( http://leda.univ-lyon1.fr/intro.html ) .
this work is also based on observations taken with the nasa / esa _ hubble space telescope _
, which is operated by aura .
support for the gems project ( http://www.mpia.de/gems/gems.htm ) was provided by nasa through grant number hst - go-9500 .
the hst cosmos treasury program was supported through nasa grant hst - go-09822 ( http://cosmos.astro.caltech.edu ) .
m.b . was supported in part by the austrian science foundation fwf under grant p18416 .
k.j . was supported by the german dlr under project number 50 or 0404 and by the german dfg with grant schi 536/3 - 1 within the priority program 1177 . , m. , rix , h .- w . , somerville , r. s. , bell , e. f. , huler , b. , peng , c. y. , borch , a. , beckwith , s. v. w. , caldwell , j. a. r. , heymans , c. , jahnke , k. , jogee , s. , mcintosh , d. h. , meisenheimer , k. , snchez , s. f. , wisotzki , l. , & wolf , c. 2005 , , 635 , 959 , g. , de vaucouleurs , a. , corwin , 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 ) rix , h .- w . ,
barden , m. , beckwith , s. v. w. , bell , e. f. , borch , a. , caldwell , j. a. r. , huler , b. , jahnke , k. , jogee , s. , mcintosh , d. h. , meisenheimer , k. , peng , c. y. , snchez , s. f. , somerville , r. s. , wisotzki , l. , & wolf , c. 2004 , apjs , 152 , 163 | we describe the creation of a set of artificially `` redshifted '' galaxies in the range @xmath0 using a set of @xmath1100 sdss low redshift ( @xmath2 kms@xmath3 ) images as input .
the intention is to generate a training set of realistic images of galaxies of diverse morphologies and a large range of redshifts for the gems and cosmos galaxy evolution projects .
this training set allows other studies to investigate and quantify the effects of cosmological redshift on the determination of galaxy morphologies , distortions and other galaxy properties that are potentially sensitive to resolution , surface brightness and bandpass issues .
we use galaxy images from the sdss in the @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 filter bands as input , and computed new galaxy images from these data , resembling the same galaxies as located at redshifts @xmath0 and viewed with the hubble space telescope advanced camera for surveys ( hst acs ) . in this process
we take into account angular size change , cosmological surface brightness dimming , and spectral change .
the latter is achieved by interpolating a spectral energy distribution that is fit to the input images on a pixel - to - pixel basis .
the output images are created for the specific hst acs point spread function and the filters used for gems ( f606w and f850lp ) and cosmos ( f814w ) .
all images are binned onto the desired pixel grids ( 003 for gems and 005 for cosmos ) and corrected to an appropriate point spread function .
noise is added corresponding to the data quality of the two projects and the images are added onto empty sky pieces of real data images .
we make these datasets available from our website , as well as the code ferengi : `` full and efficient redshifting of ensembles of nearby galaxy images '' to produce datasets for other redshifts and/or instruments . |
the statistical properties of linear polymers near an impenetrable short range attractive boundary have been extensively studied for a long time .
the most frequently applied model for a polymer chain has been the self avoiding walk ( saw ) model ( that is , the walk without self intersections ) , so that steps of the walk have been identified with monomers that comprise the polymer , while the solvent surrounding has been represented by a lattice .
here we assume that polymer is immersed in a good solvent , and interacts only with an adsorbing surface bounding the polymer container , so that for strong enough monomer - surface interaction the polymer undergoes phase transition from desorbed to adsorbed phase .
since the polymer adsorption is a surface critical phenomenon , it has been possible to describe various polymer quantities in terms of power laws described by concomitant critical exponents .
early investigations of polymer behavior near attractive surfaces dealt with polymer chains immersed in homogeneous spaces with planar adsorbing boundaries ( see @xcite for a review ) .
these studies have been subsequently extended to polymers immersed in porous ( inhomogeneous ) media , modeled by fractal lattices embedded in two - dimensional @xcite and three - dimensional @xcite space . in these studies , almost exclusively , only two critical exponents have been studied , that is , the end to
end distance critical exponent @xmath5 and the crossover exponent @xmath6 ( that governs the number of contacts between the polymer and the surface ) .
however , a complete picture about the adsorption problem requires knowledge of surface critical exponents that describe numbers of polymer configurations grouped according to the different ways of anchoring to the adsorbing boundary . in terms of the self avoiding random walk ( saw ) model of linear polymers , these exponents are defined by the following formulas for numbers of possible different configurations averaged over the number of sites on the impenetrable surface @xmath7 valid for large number @xmath8 of saw steps .
here @xmath9 , and @xmath10 , are numbers of all possible saws with both , one , and no ends grafted on the boundary respectively , @xmath11 is temperature dependent connectivity constant and @xmath12 , and @xmath3 , are concomitant surface critical exponents that take different values in various polymer phases .
so far , surface critical exponents have been studied mostly for saws near the boundary surfaces of two and tree - dimensional euclidean spaces .
these studies were performed using various techniques including series enumeration @xcite , conformal invariance theory @xcite , coulomb gas method @xcite , field theoretical approach @xcite , and monte carlo simulations @xcite .
on fractals , the surface critical exponents where studied only for saws immersed in a good solvent on two - dimensional fractal lattices @xcite . in this paper
we study the surface critical exponents for the polymer chain situated on fractals that belong to the three - dimensional ( 3d ) sierpinski gasket ( sg ) family .
each member of the sg family is labeled by an integer @xmath0 ( @xmath1 ) , and it is assumed that one side of each sg fractal is impenetrable adsorbing wall . by applying an exact renormalization group ( rg ) method for the saw model that includes monomer - surface interactions we have calculated critical exponents @xmath13 , @xmath14 , and @xmath3 , for @xmath15 , 3 and 4 fractals . this paper is organized as follows . in section [ druga ]
we describe the 3d sg fractals for general scaling parameter @xmath0 , and introduce the self - avoiding walk model in the case when a boundary of 3d sg fractal is an adsorbing surface . then , we present the framework of the general rg method for studying the polymer adsorption problem on these fractals . in section [ rezultati ]
we display the exact results for the studied critical exponents @xmath2 , and @xmath3 for @xmath15 , 3 and 4 fractals , in different polymer regimes .
all obtained results are summarized , discussed and compared with related previous results in section [ diskusija ] .
finally , some technical details are given in the appendix .
in this section we are going to expound on the renormalization group ( rg ) approach of calculating the critical exponents @xmath2 , and @xmath3 for the adsorption problem of saws immersed in a solvent modeled by fractals belonging to the 3d sg family of fractals . here
we give a brief summary of their basic properties .
we start with recalling the fact that each member of 3d sg fractal family is labeled by an integer @xmath16 and can be constructed in stages . at the first stage ( @xmath17 ) of the construction
there is a tetrahedron of base @xmath0 containing @xmath18 upward oriented unit tetrahedrons .
the subsequent fractal stages are constructed recursively , so that the complete self - similar fractal lattice can be obtained as the result of an infinite iterative process of successive @xmath19 enlarging the fractal structure @xmath0 times , and replacing the smallest parts of enlarged structure with the initial ( @xmath17 ) structure . in the case under study
, we take that one of the four boundaries of the 3d sg fractal is impenetrable adsorbing surface ( wall ) , which is itself a 2d sg fractal with the fractal dimension @xmath20/\ln b$ ] , whereas the fractal dimension of the complete 3d sg fractal is @xmath21/{\ln b}}$ ] . in order to describe the effect of attractive ( adsorbing ) surface
, one should introduce two boltzmann factors : @xmath22 , and @xmath23 , where @xmath24 is the energy of a monomer lying on the adsorbing surface , and @xmath25 is the energy of a monomer in the layer adjacent to the surface . if we assign the weight @xmath26 to a single step of the saw walker , then the weight of a walk having @xmath8 steps , with @xmath27 steps on the surface , and @xmath28 steps in the layer adjacent to the surface , is @xmath29 ( see figure [ fig1 ] ) .
3d sg fractal at the first stage of construction , with an example of the saw path starting from the adsorbing surface depicted by the yellow area .
the steps on the adsorbing surface and in the adjacent layer are weighted by the factors @xmath30 and @xmath31 , respectively . here
@xmath32 is the energy of a monomer lying on the adsorbing wall ( @xmath33 ) , and @xmath34 is the energy of a monomer that appears in the layer adjacent to the wall .
the depicted saw path corresponds to a polymer configuration with one end ( red point ) grafted on the adsorbing surface and has the weight @xmath35 . ]
the weighting factors defined in the foregoing paragraph allow us to introduce the following global generating functions @xmath36 where @xmath37 ( @xmath38 ) represents the average number ( over all sites of adsorbing wall ) of @xmath8-step saws with @xmath27 steps on the surface and @xmath28 steps in the layer adjacent to the wall provided one ( both ) end(s ) of the walk is ( are ) anchored to the wall , while @xmath39 is the number of saws with no ends anchored to the wall . if we assume that , for large @xmath8 , the numbers @xmath40 , @xmath41 and @xmath42 behave in accordance with the laws defined in ( [ defgamma ] ) , then the leading singular behavior of the generating functions are of the form @xmath43 in the vicinity of the critical value @xmath44 , _
i.e. _ when @xmath26 approaches @xmath45 from below . , @xmath46 and @xmath47 ) , within the @xmath48th stage of 3d sg fractal structure .
thus , for example , the @xmath49 represents the situation when saw starts from the impenetrable surface and leaves the @xmath48th stage fractal lattice at its vertex which does not lie in the surface .
the interior details of the @xmath48th stage fractal structure , as well as details of the chains , are not shown .
small red circles depict the saw starting points from the attaching surface , while the blue ones denote the saw starting points that are in the bulk part of tetrahedron lying on the surface . ] to calculate the surface critical exponents @xmath2 , and @xmath3 , we have found that it is helpful to define three kinds of restricted generating functions that provide a complete description of the generating functions @xmath50 , @xmath46 and @xmath47 .
these function are : the traversing saw generating functions @xmath51 , @xmath52 , @xmath53 , @xmath54 and @xmath55 ; one - leg saw functions @xmath56 , @xmath57 , @xmath58 and @xmath59 ( which are depicted in figure [ fig2 ] ) ; and , the two - leg saw functions @xmath60 ( which are not presented in figure [ fig2 ] , because they are not relevant for the critical exponent calculation @xcite ) . using these restricted generating functions , one can express each global generating function in the following manner .
we start with the function @xmath50 , that can be written in the form @xmath61}^{r+1 } } & \biggl(&\sum_{i=1}^{2}\sum_{j=1}^{2}\>f_{ij}c_i^{(r)}c_j^{(r)}+ \sum_{i=1}^{2}\sum_{j=1}^{3}\>g_{ij}c_i^{(r)}d_j^{(r ) } \nonumber\\ % & + & \sum_{i=1}^{3}\sum_{j=1}^{3}\>h_{ij}d_i^{(r)}d_j^{(r)}+ \sum_{m}e_m e_m^{(r)}\biggr)\ > , \label{g11 } \end{aligned}\ ] ] where the coefficients @xmath62 , @xmath63 , @xmath64 and @xmath65 are polynomials in @xmath66 and @xmath55 .
similarly , the generating functions @xmath67 and @xmath68 can be written as @xmath69}^{r+1 } } \biggl[\sum_{i=1}^{2}\sum_{j=3}^{4}\>p_{ij}c_i^{(r)}c_j^{(r)}+ \sum_{i=1}^{2}\sum_{j=4}^{6}\>q_{ij}c_i^{(r)}d_j^{(r ) } \nonumber\\ % \fl & + & \sum_{i=3}^{4}\sum_{j=1}^{3}\>r_{ij}c_i^{(r)}d_j^{(r ) } + \sum_{i=1}^{3}\sum_{j=4}^{6}\>s_{ij}d_i^{(r)}d_j^{(r)}+ % \sum_{i=1}^{2}c_i^{(r)}(p_{i}c^{(r)}+q_{i}d^{(r)})\nonumber\\ \fl&+ & \sum_{i=1}^{3}d_i^{(r)}(r_{i}c^{(r)}+s_{i}d^{(r ) } ) % + \sum_{m}t_m e_m^{(r)}\biggr]\ > , \label{g1 } \end{aligned}\ ] ] and @xmath70}^{r+1 } } \biggl[\sum_{i=3}^{4}\sum_{j=3}^{4}\>p'_{ij}c_i^{(r)}c_j^{(r)}+ \sum_{i=3}^{4}\sum_{j=4}^{6}\>q'_{ij}c_i^{(r)}d_j^{(r ) } \nonumber\\ % \fl & + & \sum_{i=4}^{6}\sum_{j=4}^{6}\>s'_{ij}d_i^{(r)}d_j^{(r)}+ % \sum_{i=3}^{4}c_i^{(r)}(p'_{i}c^{(r)}+q'_{i}d^{(r ) } ) + \sum_{i=4}^{6}d_i^{(r)}(r'_{i}c^{(r)}+s'_{i}d^{(r ) } ) \nonumber\\ \fl&+&p'(c^{(r)})^2+q'c^{(r)}d^{(r)}+s'(d^{(r)})^2 + % \sum_{m}t'_m e_m^{(r)}\biggr]\ > , \label{gs } \end{aligned}\ ] ] where , again , the coefficients standing with one and two leg generating functions on the right - hand side , are polynomials in traversing functions @xmath66 and @xmath55 . due to the self - similarity of fractals , restricted generating functions obey recursive relations , which can be interpreted as rg equations @xcite . for arbitrary @xmath48
, these equations for the bulk restricted partition functions ( for any @xmath71 ) have the form @xmath72 where @xmath73 and @xmath74 are non - negative integers , whereas coefficients @xmath75 , @xmath76 , @xmath77 and @xmath78 are polynomials in @xmath79 and @xmath80 , neither of them depending on @xmath48 @xcite .
for surface traversing functions , @xmath81 and @xmath55 , rg equations have the form @xmath82 where on the left - hand side we have used the prime symbol as a superscript for the @xmath83-th order generating functions , and no indices on the right - hand side for the @xmath48-th order functions . the numbers @xmath84 , @xmath85 , and @xmath86
do not depend on @xmath48 @xcite .
analogously , we can construct additional recursion relations for surface one - leg generating functions @xmath87= { \mathcal m_s}\left [ \matrix{c_1\cr c_2\cr d_{1}\cr d_{2}\cr d_{3}}\right]\ , , \quad \qquad
\left [ \matrix{c_3'\cr c_4'\cr d_{4}'\cr d_{5}'\cr d_{6}'}\right]= { \mathcal m_s}\left [ \matrix{c_3\cr c_4\cr d_{4}\cr d_{5}\cr d_{6}}\right]+{\mathcal m_b}\left [ \matrix{c\cr d\cr 0\cr 0 \cr 0}\right]\>,\ ] ] where elements of matrices @xmath88 and @xmath89 ( @xmath90 denotes the surface functions , while @xmath80 the bulk ones ) are polynomials in @xmath79 , @xmath80 , @xmath91 , @xmath92 and @xmath93 functions . starting with the initial conditions @xmath94 which correspond to the elementary tetrahedron @xmath95 one can iterate rg relations ( [ eq : rga ] ) , ( [ eq : rgb ] ) , and ( [ eq : rga1])-([eq : rgb1 ] ) for traversing saw function in order to establish the phase diagram of the polymer system .
in addition , we define the initial condition for one - leg generating functions @xmath96 that are needed for studying large @xmath48 behavior of these functions ( through relations ( [ eq : rgc ] ) , ( [ eq : rgd ] ) and ( [ c1d6 ] ) ) , and consequently to determine the surface critical exponents @xmath13 , @xmath14 and @xmath3 , from the singular parts of functions @xmath50 , @xmath46 and @xmath47 .
to perform described procedure we need to know all rg equations for a specific fractal .
we have been able to complete the exact form of required rg transformations and carry out a comprehensive analysis for the first three members ( @xmath15 , 3 and 4 ) of the 3d sg family of fractals .
the rg equations ( [ eq : rga])-([eq : rgb1 ] ) were found in previous studies @xcite , and in this work we have determined the additional rg transformation ( [ c1d6 ] ) by applying an exact enumeration method . for @xmath15 fractal these are given in the appendix , while for @xmath97 and 4 fractals they can be obtained upon request to the authors ( to be more precise , for the critical exponents calculation only transformations defined with @xmath98 were needed , and consequently only that transformations are quoted ) .
we note that computer enumeration and classification of all saw configurations required to build rg transformations for one - leg generating functions , was done in a few seconds for @xmath15 and @xmath97 fractals , while in @xmath99 case it took 7 hours on a pc with i5 intel microprocessor . details of the performed rg analysis together with the specific results for @xmath15 , 3 and 4
fractals are presented in the next section .
numerical analysis of rg equations for traversing saw functions showed that for each value of @xmath102 ( between 0 and 1 ) there exists a critical value of @xmath103 , such that for values of @xmath104 smaller than @xmath105 polymer is in desorbed state , whereas for @xmath106 it is adsorbed at the surface . precisely at the critical value @xmath107
the transition from adsorbed to desorbed phase occurs . in the following subsections
all established polymer regimes will be reviewed separately and for each of them the surface critical exponents will be evaluated . for weak monomer - surface interactions @xmath109 , and critical value of the fugacity @xmath110 ( which does not depend on the values of @xmath104 and @xmath102 ) the parameters ( @xmath111 ) tend to @xmath112 , when @xmath113 , which indicates that polymer , stays away from the attractive surface @xcite .
this state is referred to as desorbed phase , determined by the rg fixed point @xmath114 the mean squared end - to - end distance @xmath115 of the polymer chain scales with its length @xmath8 as @xmath116 , where the critical exponent @xmath5 is equal to @xmath117 and @xmath118 is the largest eigenvalue of the rg transformation ( [ eq : rga ] ) and ( [ eq : rgb ] ) linearized in the vicinity of the corresponding fixed point @xmath119 @xcite . in order to calculate surface critical exponents
@xmath13 , @xmath14 and @xmath3 , one should investigate singular behavior of the generating functions ( [ eq : g11])([eq : gs ] ) , for which it is first necessary to analyze rg transformations ( [ eq : rga1])([c1d6 ] ) in the vicinity of the bulk fixed point ( [ eq : desorbedsawfp ] ) .
after large number of rg iterations , the traversing functions behave as @xmath120 where @xmath121 is irrelevant eigenvalue of ( [ eq : rga1])([eq : rgb1 ] ) calculated at the fixed point ( [ eq : desorbedsawfp ] ) .
also , one - leg generating functions have the following large @xmath48 behavior @xmath122 where @xmath123 is the relevant eigenvalue of the matrix @xmath124 \>,\ ] ] made out of the coefficients @xmath125 and @xmath126 appearing in rg equations ( [ eq : rgc ] ) and ( [ eq : rgd ] ) for the pure bulk functions @xmath56 and @xmath57 , and @xmath127_{11}^*\>,\ ] ] asterisk denoting that the value of the polynomial @xmath128_{11}$ ] ( appearing in rg equations ( [ c1d6 ] ) ) is evaluated in the fixed point ( [ eq : desorbedsawfp ] ) .
we remind here that the eigenvalue @xmath123 determines the bulk critical exponent @xmath129 which governs the singular behavior @xmath130 of the generating function @xmath131 for all possible saws in the bulk ( away from adsorbing boundary ) , where @xmath132 is the average number ( over all starting points ) of such @xmath8-step saws @xcite . from ( [ d21 ] ) and ( [ d2 ] ) , we perceive that , for large @xmath48 , behavior of @xmath133 , @xmath134 and @xmath135 , depends on mutual relation between the specific values of @xmath136 , @xmath123 and @xmath137 .
calculated values of @xmath136 and @xmath123 , for @xmath15 , 3 and 4 fractals , are given in table [ tabela1 ] , while the particular values for @xmath137 are : @xmath138 , @xmath139 and @xmath140 . since for each studied fractal @xmath141 and @xmath142 one finds that @xmath143const , @xmath144 , @xmath145 , and @xmath146 @xmath147 .
then , from ( [ gs ] ) it follows that on the large scale @xmath148 behavior of the generating function @xmath149 is determined by the term containing @xmath150 ^ 2 $ ] , and consequently @xmath151 since @xmath152 , and @xmath153 , the critical behavior @xmath154 follows , with @xmath155 by examining the large @xmath48 behavior of the terms in the sum ( [ g11 ] ) , representing the generating function @xmath50 , one concludes that the term containing @xmath156 ^ 2 $ ] dominates , so that the largest term in the sum behaves as @xmath157^r$ ] .
however , @xmath158 is smaller than @xmath159 in each studied @xmath0 case , implying that @xmath50 remains finite at the critical point .
therefore , one should inspect the large scale behavior of the derivative @xmath160 . by finding this derivative from ( [ g11 ] )
, one finds that it can be expressed as an infinite sum , similar to the sum on right - hand side of ( [ g11 ] ) , but with terms that , apart from rg parameters , also contain their derivatives with respect to @xmath26 . on the other hand , by finding derivative of the rg equations ( [ eq : rga])([c1d6 ] ) , one can find recurrence relations for the rg parameters derivatives , and analyze their large @xmath48 behavior .
from such an analysis , it follows that derivatives of the traversing rg parameters behave as @xmath161 , as well as @xmath162 , whereas @xmath163 , @xmath164 , and @xmath165 .
these findings , together with the established behavior of the rg parameters , imply that @xmath166 with @xmath167 in all cases , meaning that the function @xmath168 diverges on the large scale @xmath148 . on the other hand , in the vicinity of the critical point , the following relation @xcite is satisfied @xmath169 whereupon follows @xmath170 finally , using the established behavior of the rg parameters and @xmath50 , from ( [ g1 ] ) one can , in a similar way , find that on large scale the generating function @xmath171 behaves as @xmath172 implying that @xmath173 with @xmath174 for all studied fractals , the obtained specific results of these critical exponents , in desorbed polymer phase , are listed in table [ tabela1 ] .
llllllllll + @xmath0 & @xmath175&@xmath176&@xmath177 & @xmath178 & @xmath136 & @xmath179 & @xmath13&@xmath180&@xmath181 + 2 & 0.4294&0.0500&0.6740&4.2069 & 1.2883 & 1.4461&-0.5756&0.5751&1.7259 + 3 & 0.3420&0.0239&0.6542&11.477&1.7206 & 1.5352&-0.4206&0.7094&1.8394 + 4 & 0.2899&0.0122&0.6411&25.679 & 2.3381 & 1.6165&-0.2793&0.8289&1.9370 + + @xmath0 & @xmath175&@xmath176&@xmath177 & @xmath178 & @xmath136 & & @xmath13&@xmath180&@xmath181 + 2 & 0.4294&0.0500&0.6740&4.2069 & 3.1441 & & 1.1596&1.4427&1.7259 + 3 & 0.3420&0.0239&0.6542&11.477&6.7759 & & 1.2118&1.5256&1.8394 + 4 & 0.2899&0.0122&0.6411&25.679 & 12.4404 & & 1.2668&1.6019&1.9370 + + @xmath0 & @xmath182&&@xmath183 & @xmath184 & & + 2 & 0.6180 & & 0.7986&3.1456 & & + 3 & 0.5511 & & 0.7936&6.6395 & & + 4 & 0.5063 & & 0.7884&11.645 & & + when @xmath104 is increased beyond @xmath105 , and for @xmath186 @xcite rg parameters flow towards the new fixed point that describes the adsorbed polymer chain @xmath187 where @xmath182 is the fixed point for the corresponding two dimensional sg fractal @xcite . in this fixed point polymer chain
displays the features of a polymer system situated on two - dimensional sg fractal , and consequently all surface critical exponents are equal to the corresponding value of @xmath179 for 2d sg fractals ( _ i.e. _ @xmath188 ) , whose values were found in @xcite , and are quoted in table [ tabela1 ] . precisely at @xmath107 behavior of the polymer abruptly changes . for this value of @xmath104 , and @xmath110 ( the same as for the desorbed phase ) , the symmetric special fixed point
is reached @xmath190 which corresponds to the surface attached chain , when a balance between the attractive polymer surface interaction and an effective entropic " repulsion sets in . in this case
, we have found the following large @xmath48 behavior for traversing functions @xmath191 whereas the one - leg functions behave as @xmath192 where @xmath123 has the same value as in desorbed saw phase ( _ i.e. _ it is the eigenvalue of ( [ lambdab ] ) ) , and @xmath136 is the largest eigenvalue of the matrix @xmath98 , evaluated at the symmetric fixed point ( [ eq : symetric ] ) . using already described approach , one can establish the same singular behavior of the generating functions with the relations ( [ ekspbulkgs ] ) , ( [ ekspbulkgama11 ] ) , and ( [ ekspbulkgama1 ] ) , for the surface critical exponents @xmath193 , and @xmath14 , respectively , as in the case of the desorbed chain . the obtained specific results for surface critical exponents , together with related eigenvalues needed for their evaluation , are given in table [ tabela1 ] .
in this paper we have studied the configurational properties of linear polymers in the vicinity of adsorbing wall of fractal containers modeled by self - avoiding walks on 3d sg family of fractals .
each member of the 3d sg fractal family has a fractal impenetrable 2d adsorbing boundary ( which is , in fact , 2d sg fractal surface ) and can be labeled by an integer @xmath0 ( @xmath1 ) . in this
model interactions between monomers and adsorbing wall are described by parameters @xmath22 , and @xmath23 , where @xmath24 is the energy of a monomer lying on the adsorbing surface , and @xmath25 is the energy of a monomer in the layer adjacent to the surface .
depending of the values of @xmath104 and @xmath102 the polymer chain can exists in one of three possible states : adsorbed , desorbed and surface attached @xcite . for the first three members ( @xmath15 , 3 and 4 ) of the 3d sg fractal family we have performed the exact rg analysis to evaluate the surface critical exponents @xmath13
, @xmath14,and @xmath3 that govern the numbers of all possible polymer configurations with both , one , and no ends grafted on the adsorbing boundary , respectively
. to evaluate specific values of @xmath2 , and @xmath3 , one needs to know the values of the end - to - end distance critical exponent @xmath5 and eigenvalues @xmath123 and @xmath136 of matrices @xmath88 and @xmath194 defined by ( [ c1d6 ] ) and ( [ lambdab ] ) respectively .
the exact values for @xmath5 and @xmath123 have been already found @xcite , for fractals with @xmath15 , 3 and 4 , and it has remained to calculate the values of @xmath136 , which requires knowledge of the elements of the matrix @xmath98 .
the required elements are polynomials , that can be formed by enumeration and classification of saws , described by restricted partition functions @xmath195 , @xmath196 , @xmath197 , @xmath198 and @xmath199 ( see figure [ fig2 ] ) , and we have found that this is feasible for @xmath200 . the obtained specific results for studied surface critical exponents @xmath13 , @xmath14 , and @xmath3 are given in table [ tabela1 ] .
values of the critical exponents @xmath13 , @xmath14 and @xmath3 , for desorbed ( @xmath201 ) and surface attached polymer ( @xmath202 ) as functions of @xmath203 ( @xmath15 , 3 and 4 ) .
full lines only serve as guides for the eyes .
the @xmath204 euclidean values @xmath205 , and @xmath206 , for desorbed polymer phase ( @xmath201 ) , that spring from two findings @xcite , are depicted by dashed horizontal lines , black for @xmath13 and red for @xmath180 .
the dashed horizontal blue line corresponds to the @xmath204 euclidean value of @xmath207 . ] for the sake of a better assessment of the global behavior of the surface critical exponents as functions of the scaling parameter @xmath0 , we present our results in figure [ fig : gamma ] .
it can be seen that the following chain of inequalities is satisfied @xmath208 this is a quite plausible result since the generating function @xmath50 describes saws with a stronger constraint than those described by the generating function @xmath46 , which implies the inequality @xmath209 , for both desorbed and surface attached chain .
also , we see that for each @xmath0 the inequalities @xmath210 , and @xmath211 are valid , implying that the number of saw configurations with one or two ends anchored to the wall are larger in the surface attached regime then in the desorbed one . here
we note that the inequality @xmath209 , for both desorbed and surface attached chain , were established for saws grafted on one edge of fractals belonging to 2d sg fractal family , whereas for @xmath212 it was obtained that @xmath213 @xcite .
furthermore , for @xmath201 , we see that @xmath13 and @xmath180 are monotonically increasing functions of @xmath0 , and can be compared with their euclidean counterparts . in desorbed
phase the exponent @xmath13 for @xmath99 is larger than the corresponding three
dimensional euclidean value @xmath214 that springs from the field - theory approach result @xmath215 @xcite and monte carlo simulations finding @xmath216 @xcite ( black dashed line in figure [ fig : gamma ] ) , whereas @xmath180 surpasses the euclidean value @xmath217 ( evaluated from @xmath218 @xcite and @xmath219 @xcite ) , for @xmath220 . in the case of the surface attached extended chain ( @xmath202 ) , both critical exponents , @xmath14 and @xmath13 are also monotonically increasing functions of @xmath0 , being always larger then the corresponding euclidean values for @xmath221 ( obtained as an average of @xmath222 @xcite , and @xmath223 @xcite ) and @xmath224 ( average of @xmath225 @xcite and @xmath226
the behavior of the critical exponent @xmath3 ( whose values are the same in both the desorbed chain phase and the surface attached chain region ) , is similar to the behavior of @xmath14 and @xmath13 , that is , @xmath3 is monotonically increasing function of @xmath0 , which for @xmath220 surpasses its euclidean counterpart @xmath227 ( calculated from scaling relation @xmath228 @xcite , where we have put the last estimates for three dimensional euclidean values of @xmath229 @xcite and @xmath230 @xcite ) .
finally , we have tested the scaling relation @xmath231 which was proposed in @xcite as a modification of the scaling relation @xmath232 , obtained for euclidean containers @xcite . here
@xmath233 is the fractal dimension of the solvent and @xmath234 is the fractal dimension of the attracting surface .
putting the data from table [ tabela1 ] into ( [ barber ] ) , one can see that this scaling relation is exactly satisfied for both desorbed and surface attached polymers . at the end
, one may pose the question about the possible behavior of the surface critical exponents @xmath2 , and @xmath3 for larger @xmath0 ( when the sg fractal dimension approaches the euclidean value 3 ) . to answer this question one have to apply some other method , such as monte carlo renormalization group method which appeared to be very precise and quite efficient method for critical exponents calculation on finitely ramified fractals , which will be the matter of an independent study .
in this appendix we give the exact rg equations ( [ c1d6 ] ) for the surface one - leg parameters @xmath195 , @xmath196 , @xmath197 , @xmath198 and @xmath199 in the case of the @xmath15 3d sg fractal .
we have found that these equations have the following form : @xmath235 | we study the adsorption problem of linear polymers , when the container of the polymer solvent system is taken to be a member of the three dimensional sierpinski gasket ( sg ) family of fractals .
members of the sg family are enumerated by an integer @xmath0 ( @xmath1 ) , and it is assumed that one side of each sg fractal is impenetrable adsorbing boundary .
we calculate the critical exponents @xmath2 , and @xmath3 which , within the self avoiding walk model ( saw ) of polymer chain , are associated with the numbers of all possible saws with one , both , and no ends grafted on the adsorbing impenetrable boundary , respectively . by applying the exact renormalization group ( rg ) method , for @xmath4 , we have obtained specific values for these exponents , for various type of polymer conformations .
we discuss their mutual relations and their relations with other critical exponents pertinent to saws on the sg fractals .
_ keywords _ : renormalisation group ; solvable lattice models ; critical exponents and amplitudes ( theory ) ; polymers , polyelectrolytes and biomolecular solutions |
the number of known extrasolar planets orbiting nearby stars is now more than 100 ( see e.g. , the website : http : + exoplanets.org ) . because of strong observational selection effects most of the systems discovered first consisted of a jovian planet in a very close orbit around the parent star .
multiple planet systems have been discovered now , as have systems with jovian planets or low mass brown dwarfs with orbital radii more like those of our own solar system .
the question of how these diverse solar systems formed has generated much interest . as in our own solar system , there are also many interesting questions about the fate of these systems when their parent stars reach the end of their evolution .
however , it is difficult to discover planets around luminous , bloated giant stars . thus , if this study is to advance in the near future , unique and strong observational signatures of planets around evolved stars must be found .
@xcite and @xcite ( also see @xcite and @xcite ) have suggested that the elliptical , bipolar and even non - axisymmetric shapes of planetary nebulae could be the result of their binary nature , with the binary companion being a substellar object in many cases .
@xcite has also proposed that jets in planetary nebulae could be formed as a result of the engulfment of giant planets or brown dwarfs in the envelope of agb stars .
@xcite recently proposed that such engulfments could modify the mass loss of the agb star .
they estimated that at least 3.5% of solar type stars would be affected .
these phenomena provide strong observational signatures , but the disadvantage is that the substellar companion must disappear in most cases ( and in a short time ) to produce them .
we have suggested that other observable signatures may result from the accretion of agb wind or extended atmosphere material onto closely orbiting giant planet or brown dwarf companions .
these include enhancing sio maser emission in mira variables @xcite , and producing short time - scale optical variations @xcite , such as reported in hipparcos observations of miras @xcite .
previously ( @xcite , henceforth paper i ) we presented some preliminary numerical hydrodynamical models of the accretion process .
these signatures do not require the companion to be swallowed , but the optical variations are more subtle and difficult to detect .
the masers , in particular , may have multiple causes , so the role of companions is difficult to decipher .
moreover , the companion masses and radii are restricted by the need to have a strong enough accretion interaction to produce the observable phenomena .
when solar - type stars evolve to the agb stage , their outer atmospheres expand out to nearly an astronomical unit . during parts of this evolutionary stage they experience radial pulsations as mira or semi - regular variable stars , suffer thermal pulses , and generate winds that increase with time .
the models of @xcite show how pulsational shocks steepen as they run down the pressure gradient outside the photosphere , utimately lifting material into an oscillating extended atmosphere .
further out , where gravitational periods are longer , individual gas elements are not able to complete a cycle between shocks , and a wind develops @xcite .
radiation pressure on dust grains also contributes to driving the wind ( e.g. , @xcite , @xcite , @xcite ) .
the dominant pulsational modes in miras are radial , so it is likely that the wind is spherically symmetric , unless the atmosphere has been spun up or otherwise modified by interaction with a close companion .
the discovery of concentric shells around some young planetary nebulae provides strong evidence for circular symmetry in the winds ( e.g. , @xcite and references therein ) .
any companions within the large optical photosphere of the mira star will be swallowed - that is , heated and dispersed as they are dragged inward . companions orbiting at distances of greater than about 10 au from the central star will accrete , but accretion from the quite steady and relatively low - density wind at these radii is unlikely to generate any very strong observational signatures .
( that is , unless the companion is of stellar mass , see @xcite and @xcite . )
the best place to produce time - varying observable effects with accretion onto low - mass companions is closer to the star , where they can interact with pulsational shocks in the extended atmosphere . according to the @xcite models , pulsational shocks , and quasi - ballistic motions of gas elements with radial velocity sign changes ,
persist out to radii of about @xmath1 in typical miras . in @xcite and paper i we found , from energy estimates and preliminary models ,
that even this close to the central star , it is unlikely that steady accretion onto giant planets and low - mass brown dwarfs would produce observable effects . in this region
the pulsational shocks modulate the amount of material that sweeps into the companion s sphere of influence , though not by enough to give rise to large emission bursts .
captured material has little specific angular momentum relative to the companion , so we expect it to settle into a disc of radius @xmath2 much less than the capture radius @xmath3 .
( for clarity , we will refer to radii or distances from the companion with a capital letter r , and distances from the mira with the lower case r. ) because the surrounding medium is oscillating in radius , the sign of the accreted angular momentum is expected to change periodically .
( as discussed below , the situation is in fact more complicated than this . )
thus , we expect a disc to build up near the companion with the gas orbiting in one sense , but to be disrupted when gas orbiting in the opposite sense begins to rain down .
the resulting angular momentum cancellation would drive infall onto the planet , with a burst - like energy release .
the transient disc essentially serves as an energy reservoir , allowing substantial burst - like releases .
the models presented below do not spatially resolve the accretion disc , but they can provide some information on its development .
we note in passing that , since we are only considering companions that are much less massive than the star , the accretion process is not roche overflow , but rather direct hydrodynamic capture from the surrounding medium .
similarly , estimates of the tidal circularization or in - spiral time ( see section 2 of @xcite and references therein ) , show that tidal effects will only be important over the brief mira lifetime ( about @xmath4 ) if the companion orbits within about @xmath5 stellar radii of the star .
the oscillating extended atmosphere extends much farther out than this .
the orbital radius adopted in the models below is 3.2 times the radius of a solar mass mira ( e.g. , @xcite ) .
( there is controversy in the literature about mira radii , but larger mira radii would also imply larger oscillatory zones , and not change the conclusion that tidal effects are only important in the lower part of this zone . )
the bow shock , wake and accretion phenomena in this problem are examples of the bondi - hoyle - lyttleton class of wake accretion flows ( @xcite , @xcite , @xcite ) . in the last 20 years there has been a great deal of study of the problem of accretion in x - ray binaries , where a compact object orbits within the wind of a massive star ( e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , & @xcite ) .
these flows are very similar to those of the present work - unstable , oscillating wakes and bursty accretion of low angular momentum gas are also common in them
. we will make specific comparisons to this work below .
however , we note that because the time - dependent hydrodynamics is driven by `` external '' pulsational forcing in the present problem , it is fundamentally different than that of the flip - flop internal instability found in models of x - ray binary stars . in the earlier work we also found that companions must be considerably more massive than jupiter to produce observable bursts , i.e. , at least 10 times as massive . the brown dwarf / planet boundary
is usually defined as the critical mass for deuterium burning , about 13 @xmath0 ( jupiter masses ) . in the models below we consider companion masses in the range 10 - 50 @xmath0 . as the companion mass
is increased through this range we find an interesting transition in the wake and accretion hydrodynamics .
if we further increase the companion mass much above this , e.g. , @xmath6 the companion would no longer be substellar , and we would be entering the domain of the symbiotic star phenomenon , which is beyond the scope of this paper .
once we have restricted consideration to systems with brown dwarf companions orbiting within 10 au of a mira variable , we must return to the question of how common are such systems ?
the question is especially important in light of the two points : 1 ) most of the recently discovered brown dwarfs are young , freely floating in star - forming regions ( e.g. , @xcite , @xcite , @xcite ) , and 2 ) for several years after the discovery of extrasolar planets , it appeared that there was a `` brown dwarf desert . ''
the latter point refers to the fact that as extra - solar planets were being discovered , brown dwarf companions were not found ( see review of @xcite and references therein ) .
the difficulty posed by this point has been mitigated recently .
specifically , the first 50 extrasolar planets discovered included only 3 objects with @xmath7 , but all had orbital radii much less than @xmath8 .
however , the second 50 planets discovered include 5 objects of about this mass or somewhat greater , and all have orbital semi - major axes of greater than @xmath8 .
thus , near the planet / brown dwarf boundary , the desert is less deserted .
some specific systems are also noteworthy .
these include the hr 7672 ( hd 190406 ) system , with a solar analog star .
the mass of its companion is estimated to be in the range of about @xmath9 , and its orbital semi - major axis is about 14 au @xcite .
it provides an example of a high mass brown dwarf companion of a solar type star .
the more massive of the two companions in the hd 168443 system has @xmath10 , and a semi - major axis of 2.9 au ( eccentricity of e = 0.20 ) @xcite , which very much fits the range of this study .
@xmath11 draconis ( hd 137759 ) is a roughly solar mass k giant star with a companion minimum mass of @xmath12 , and whose semi - major axis is about 1.3 au ( eccentricity 0.7 , see @xcite ) .
this planet , the first discovered orbiting a giant , may be doomed when the parent reaches the agb stage .
these examples show that while systems with closely orbiting brown dwarf companions may be quite rare , they do exist . in the following sections we investigate the accretion hydrodynamics they generate at the end of the parent stars life , and the consequences for observation .
the models were produced with a smoothed particle hydrodynamics code , which is described in paper i , and @xcite .
since the code is documented in these papers , here we will only describe the initial conditions , boundary conditions , and scaling issues relevant to the present models .
first of all , the gravities of the parent star and companion were approximated as softened point masses .
the sph smoothing length in these models was fixed in space and time to a value of 0.1 length units .
the unit of length is taken to be 1.0 au , and the time unit is 1.0 yr . the gravitational softening length of the companion was set to 0.08 length units .
most of the runs described below used 157,500 gas particles distributed in a complete ( @xmath13 ) annulus in the orbital plane of the companion .
initially , the particles were placed on a circular grid with constant radial separations between particle circles .
a fixed number of particles were placed on each circle , since this gives a @xmath14 density profile in two dimensions , which is like the observed @xmath15 profile in three dimensions .
the particles were also given a slight random offset from this grid , and a small random velocity ( 5of the sound speed in each of the 2 dimensions ) , to produce a smooth distribution .
particle radial and azimuthal velocities were initialized to zero , except for the random part .
the parent simulation code is three dimensional , but the present simulations are two dimensional . in the model presented in paper
i , an isothermal equation of state was used with an initial temperature of @xmath16 .
the simulations described below also used a polytropic or gamma law equation of state , @xmath17 . runs with different values of @xmath18
were carried out to test the effects of this factor .
the mira atmosphere and wind models of @xcite show that the presence of a small amount of dust can so enhance the cooling behind the pulsational shocks that the extended atmosphere is nearly isothermal .
however , bowen models with no dust show large temperature variations .
thus , the isothermal case may be the most relevant , but is not the only case of interest .
the first boundary condition , for the inner boundary , is a wall that is moved inward and outward on a cycloidal trajectory , like that of a ballistic particle at the same distance from the star .
the radial range of this motion is from r = 2.0 to about 2.5 au , and its period is about one time unit . as in the model of paper
i , we neglect the effect of radiation pressure on grains .
then , as noted in that paper , pulsational waves generated from below with the piston - like boundary , and a zero pressure outer bound , reproduce the main features of the flow in bowen s ( dust - free ) models , including , approximately , the velocity profile .
in the model of paper i the outer boundary was an impervious , reflecting wall , that was moved steadily outward .
this motion avoided reflection shocks from the outer boundary , and was responsible for the low pressure that allowed the development of a wind at larger radii .
however , since a fixed number of particles were used , the gas density steadily decreased , so that the model could not be run for a long time . in the models presented below we used a fixed outer boundary .
when gas particles reached this outer bound they were removed from the grid and reinserted just above the lower boundary , at the same azimuth , and then moved by a small random displacement to smooth any clumps .
their velocities were reset to values close to that of the inner piston .
shock reflections were generally small with this boundary procedure .
the procedure does not , however , completely smooth large density changes ( e.g. , behind shocks ) , so there will be some feedback effects .
no growing instabilities resulted from this feedback , but the gas behind the pulsational shocks did acquire a more complex , multi - layered appearance . since the companion was only followed through one orbit , the bow shock was not affected .
the final boundary condition is applied on a circle of radius r = 0.05 au ( 106 jupiter radii ) around the companion .
large numbers of particles are captured by the companion .
as noted in the introduction , the particles have quite small values of specifc angular momentum relative to the companion , so we expect them to fall into a very small region around the companion .
this region can not be resolved by the present code , so their fate is determined by the boundary condition . in the model of paper
i they were transferred to the centre of the region , assigned the same velocity as the companion , and thenceforth traveled as a single clump with the planet .
we will call this case the clump condition . in the models presented below
, particles found within the boundary radius are removed , and reinjected into the simulation at the lower boundary , just like those passing through the outer boundary .
we call this the vacuum condition
. when the clump condition is used the particles in the clump still exert a hydrodynamic pressure on particles outside the boundary , computed according to the usual sph algorithm . in the paper i model this pressure was initially small , but grew in magnitude as the clump grew . eventually , this pressure resisted accretion , and particles accumulated on the circular boundary until their pressure and gravitational force overwhelmed the resistance .
this effect was responsible for the burst - like accretion described in that paper .
there is no pressure resistence when the vacuum condition is used , and any burst - like accretion must be due to other causes , see below . in reality ,
pressure on the outer accretion flow from a small accretion disc is not realistic . on the other hand ,
such resistance might be generated by other means .
for example , if the companion has a strong magnetic field , the magnetosphere would provide magnetic pressure against the partially ionized accretion flow .
energy release from accetion onto the companion surface may generate a hot corona , and coronal pressure around the companion .
the clump condition crudely represents such a effects ( for a limited duration ) .
however , the vacuum condition is more realistic on the relatively large scales modeled here , and ultimately resisting pressures should be modeled on the smaller scales where they are generated .
we begin each of our model runs with the companion starting on a counterclockwise orbit from a position near the x - axis , and thus , moving through the first quadrant . in a very short time
compared to the orbital period the companion draws gas particles around it into mildly supersonic motion with it . a classic bondi - hoyle - lyttleton bow shock and wake structure
soon develop ( @xcite , @xcite ) .
figure [ f1 ] shows six snapshots through the course of one mira pulsation cycle ( also see fig .
[ f1 ] of paper 1 ) , at a time when the companion has moved into the second quadrant .
particles in the wake and the ( incomplete ) bow shocks are highlighted in the insets , which show magnified views of the region near the companion , and in which only particles with supersonic ( i.e. , greater than 1.0 velocity units ) azimuthal velocities are plotted .
initially the gas particles have zero azimuthal velocities and only acquire them through interaction with the companion .
the wakes are truncated at the outer boundary , but can clearly be seen for a length of more than an au . also at the beginning of a model run , the piston at the lower boundary of the annular grid is started outward . as this and subsequent pulsational shocks move radially out through the grid they set up a pattern of radial motions much like those seen in the @xcite atmospheric models .
we model a minimal radial range needed to capture this basic kinematics , in contrast to bowen s models .
the snapshots in figure [ f1 ] show the progress of one shock from the lower piston to the outer boundary .
recycling material at outer boundary back to the inner boundary does not seem affect the flow much in regions away from the companion .
flow profiles are quite similar to those of paper i , where a moving / reflecting boundary was used .
this is probably because the pulsational shocks are weak by the time they reach this bound in both cases .
nonetheless , the shocks driven by the lower piston do acquire some additional structure from the remnant waves reinserted , but with some smoothing , from the outer bound , as can be seen in figure [ f1 ] .
note that a layer of particles `` stick '' to the lower boundary ( a well known effect in sph ) .
we inject the recycled particles just above this layer .
it is clear from the figure [ f1 ] snapshots and their associated insets that the radial shocks are able to move the companion wake .
this is reasonable since the wake column density along a radial line is of the same order as that of the gas immediately behind a pulsational shock .
the wake is stretched outward when pulsational shocks sweep over it .
the infall in the postshock rarefaction wave leads to reformation of the wake , and leaves it pointing in a slightly inward direction .
it is also apparent from figure [ f1 ] that the wake interaction disturbs the pulsation shock and postshock flow .
this is especially apparent in the fourth and fifth snapshots of figure [ f1 ] , where the shock is disturbed over a distance of nearly 2.0 au behind the companion . in this section
we consider the mean properties of the accretion flow onto the companion , in the next section we consider its more detailed structure .
this accretion is characterized by two radial scales : an accretion radius ( and bhl accretion cylinder ) from which it is gravitationally captured , and an inner scale equal to the radius of the accretion disc around the planet , where the material settles .
as noted above and in paper i our models do not resolve the inner scale .
this means that we can not account for the possibility that some captured material may be subsequently pushed back out by dynamical processes at the inner scale , before being accreted onto the companion .
however , we expect this to be a small fraction of the captured gas . in the simulation code , any particle that comes within a radius of @xmath19 of the companion
is assumed to be accreted , and is removed and recycled .
figure [ f2 ] shows this estimate of the number of particles captured versus time in five model runs with companion masses of @xmath20 .
the classical bhl accretion radius is , @xmath21 where @xmath22 is the companion mass , c is the sound speed in the ambient gas , and @xmath23 is the relative velocity of the companion and the ambient gas , i.e. , the sum of the companion s orbital velocity and the current ambient radial velocity ( see e.g. , @xcite ) .
( note that the latter velocity is constantly changing in direction and magnitude , so that the concept of an upstream accretion cylinder out of which gas is accreted becomes somewhat confused . )
the adopted value of @xmath3 is about equal to @xmath24 in the @xmath25 case , and smaller than @xmath24 for larger companion masses .
( in the limit that the relative velocity equals the companion orbital velocity , @xmath26 where a is the companion orbital radius , and @xmath27 the mira mass . )
thus , the bow shock and accretion cylinder is resolved only for the larger mass companions .
the cumulative accretion in each of the models plotted in figure [ f2 ] is roughly linear with time . in that figure each curve derived from the model data
is paired with a line derived from a least squares fit .
the deviations between the model data and the fitted line are small in all cases , though interesting as we will see in the next subsection .
this implies that the accretion is on average constant , despite the radial pulsations in the extended atmosphere .
figure [ f3 ] compares the accretion rates of the different models .
specifically , it shows the average accretion rate over the duration of each run , plotted against the mass of the companion .
in addition to the data points a line fit is shown .
this line simply connects the best - resolved @xmath28 case to the origin , since we demand zero accretion for a zero mass companion .
most of the other models lie close to this line except for the poorly resolved @xmath29 model . as in the classic bhl problem
most of the accretion in the more massive companions comes through the wake . in the bhl problem
the accretion radius increases with the companion mass , and the accretion rate should increase with the square of this radius .
that is , with the square of the companion mass .
the goodness of the linear fit in figure [ f3 ] appears to represent the two - dimensional bhl result .
however , in this problem , the near linear fit may be more generally applicable .
the oscillatory motions of the ambient gas deliver material to the companion from a greater range of radial distances than @xmath30 .
the companion effectively has a greater reach in the radial direction , though this `` reach '' still seems to scale with the companion mass .
the increase of the size of the accretion cylinder in the direction perpendicular to the orbital plane with increasing mass ( i.e. , the increase in @xmath24 ) is smaller and contributes less to the accretion increase than the increase in the radial direction .
if this is correct , then in three dimensions , the accretion rate will continue to scale more linearly than quadratically with mass through the brown dwarf mass range .
high resolution , three - dimensional simulations are needed to resolve this question . it is also of interest to compare the numerical mass accretion rates to bhl rates .
e.g. , the mean mass accretion rate for a @xmath31 companion is about @xmath32 , though we will see below that burst rates can be much higher .
if we adopt a value for the particle mass of @xmath33 , as in paper i ( adjusted for the different particle numbers used here , and assuming a vertical capture distance of @xmath34 ) , then we get the following conversions : @xmath35 the stellar mass loss rate is assumed to be of order @xmath36 .
( this is the mass loss rate of the standard dust - free model of @xcite , and that used in @xcite , though it is about an order of magnitude less than the typical mira mass loss rate . )
if we compute the bondi rate , @xmath37 using the average circumstellar gas density @xmath38 , an efficiency of @xmath39 , the sound speed at the companion s radius ( @xmath40 ) , and set the relative velocity of the companion equal to its orbital velocity we get @xmath41 .
thus , the numerical rate nearly equals the bondi rate .
this seems somewhat high , yet it is comparable to those @xcite and @xcite . however , if the bhl scaling with mass does not hold in three - dimensions , then accretion rates in this problem will be very different than bhl rates at the extremes of the brown dwarf mass range . finally , we note that the @xmath42 run was repeated with several different values of @xmath18 , the exponent in the equation of state .
we found hardly any change in the accretion rate as @xmath18 was varied .
the analytic interpolation formula of @xcite also shows little deviation from the bhl accretion rates at mach numbers greater than a few , when @xmath18 is varied from 4/3 to 5/3 .
numerical studies of the flip - flop instability indicate that there is no instability at @xmath43 , so the accretion is steady ( see @xcite , sec .
this contrasts greatly with the present externally driven case . in this section
we consider time - dependent aspects of the accretion more carefully .
we begin with figure [ f4 ] , which shows the instantaneous accretion rate ( @xmath44 ) in three of the models .
we note the presence of sharp , high amplitude peaks in the accretion rate .
there are also sharp decreases relative to the small , but nonzero , mean accretion rate . in all cases
the accretion bursts are clustered .
the cumulative effect of these clusters is presumably responsible for the deviation from the mean lines in the curves of figure [ f2 ] .
a second feature of figure [ f4 ] is the fact that the burst clusters are nearly periodic at low companion mass , but appear to become progressively more random , and frequent , as the companion mass increases . at low masses ,
the burst period is about equal to the pulsation period , and in fact , the bursts are coincident with pulsational shocks overrunning the companion .
it is natural that the companion should accrete more from dense post - shock layers ( which is also when the radial velocity changes sign ) .
the bursts are more random with high mass companions because , as we will describe below , of the processing of this material through the longer wake . before proceeding we
note the similarity of some of these features to those of simulations of flip - flop accretion in x - ray binaries .
instability and quasi - periodic bursts are certainly a characteristic of accretion in this problem as well ( e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
since the flip - flop instability is self - excited , much of the discussion of this literature has focussed on the question of what circumstances are required for its development .
in particular , issues such as the effect of pressure and velocity gradients across the mach cone have been considered in some detail .
numerical studies have been plagued by spatial size and temporal limitations , as well as numerical resolution problems ( see @xcite , @xcite , and references therein ) .
the present problem has the advantage of being driven by strong , external perturbations , so we do nt have the difficulty of modeling the slow growth of small amplitude perturbations under the internal instability . as a result of this fundamental difference in the two processes , the details of the accretion burst phenomenon
are also different . in the published flip - flop simulations
the amplitude of accretion bursts relative to the mean never exceeds a factor of about 5 .
figure [ f4 ] shows much larger burst amplitudes .
there also seems to be no equivalent to the systematic change in the character of the accretion bursts as companion mass is increased , as shown in figure [ f4 ] .
however , it appears that nonlinear trends with companion ( e.g. , stellar black hole ) mass have not yet been investigated in the flip - flop problem .
we noted above that models with larger mass companions draw gas in from a larger range of radii .
figures [ f5 ] and [ f6 ] provide many more details about this process for the @xmath45 and @xmath46 models , respectively .
these figures show the partial trajectories of particles that come within a radius of @xmath47 of the companion , and their subsequent fate .
the value of this cutoff radius is arbitrary , and was chosen to provide an adequate particle sample .
specifically , once the particles within the cutoff radius were identified ( at the times indicated ) , their positions relative to the companion at 4 previous output times were used ( with the present position ) to produce the curves .
since the companion position is fixed in these plots , the primary component of these trajectories is the relative azimuthal motion toward the companion .
( note : the companion is located in the second quadrant at these times . ) in figure [ f5 ] and the first panel of figure [ f6 ] the time shown is when a pulsational shock is just about to overrun the companion .
this is evident in the abrupt trajectory changes visible on the right side of the companion .
the second panel of figure [ f6 ] shows a time of about half a pulsational cycle after that of the first panel , when no shock is near the planet . at that time the trajectories are smoother , and more symmetrically distributed .
figure [ f6 ] not only shows that many more particles are drawn in by the @xmath46 companion , but also resolves the capture process quite well . a large fraction of the accreted particles circle around the companion , impact the wake , and are drawn down from the inner wake .
the trajectories coming from above ( behind in the orbital sense ) are especially interesting .
some particles still escape through the outer wake , but they are a much smaller fraction with the @xmath46 companion . with this information , we can now understand the patterns of instantaneous accretion in the different models . in the low mass models accretion bursts
result from enhanced feeding associated with the passage of pulsational shocks . because gas accretes through a larger wake in the higher mass cases
, we would expect longer duration burst clusters .
there is also a possibility of echo bursts following passage of a pulsational shock , if relatively large amounts of gas are deposited in the wake , and its accretion is significantly delayed .
the middle panel of figure [ f4 ] , for the @xmath42 model , does indeed show wider burst clusters , and bursts appearing at new phases . when the companion mass is increased to @xmath46
the wake becomes so massive that the passage of pulsational shocks does not greately enhance the accretion .
there is still some hint of periodicity in the bursts in the third panel of figure [ f4 ] , but the duration of the burst clusters is comparable to the time between them . to further probe the dependence of the accretion process on companion mass ,
it is worth looking at the wakes in more detail . like
the insets of figure [ f1 ] , figure [ f7 ] shows only those gas particles that have acquired significant azimuthal velocities , though in figure [ f7 ] we show the 5 models at about the same time , rather than snapshots of one model .
the wakes are much longer and more massive when the companion mass is large .
in fact , in the case of the most massive companions the graphs underestimate the extent of the wake , because wake particles reach the outer bound and are recycled .
we note that the particle clumps seen in the lower centre part of the graphs for @xmath48 are apparently bow shock particles that have moved to the outer bound and have been recycled .
figure [ f7 ] also shows us that the bow shock is barely detectable in the low mass cases , and asymmetric in the high mass cases .
the phase dependence of this asymmetry can be seen in figure [ f1 ] .
gas behind the bow shocks is pushed outward , like wake gas , by the passage of pulsation shocks . following these passages ,
material sweeping past the inner side of the companion encounters little resistance .
material moving forward toward the outer side of the companion impacts both bow shock and wake .
in paper i we suggested that the angular momentum vector , relative to the companion , of material accreted out of the radially oscillating circumstellar gas would change signs with pulsation phase .
figure [ f8 ] shows the mean specific angular momentum of the accreted gas as a function of time in the @xmath49 models .
this figure confirms the fact that the accreted angular momentum does have both positive and negative signs at different times , though the positive ( counterclockwise ) sign seems predominant .
moreover , the sign changes are often quite rapid .
what is even more striking about figure [ f8 ] is the contrast with figure [ f4 ] .
it appears that the values of accreted angular momentum are quite random at low companion mass , where the mass accretion is periodic , and the angular momentum accretion is periodic at high companion mass , where the mass accretion is much less regular .
a somewhat more careful reading is that at low mass the amount of accreted angular momentum is just low .
evidently , even with the passage of pulsational shocks , gas is accreted relatively symmetrically in the low mass case , with little net angular momentum .
some of this effect may be due to low resolution of asymmetries in the low mass case .
however , the size of the accretion radius is smaller than the thickness of the dense post - shock layer in the low mass case , and more comparable in the high mass case . thus , density asymmetries are not noticeable on the accretion scale in the low mass case , but are in the high mass case .
the representative particle velocity vectors shown at two times in figure [ f9 ] provide more detail on the wake hydrodynamics responsible for the periodic angular momentum accretion in the @xmath50 model .
the right panel shows a time just after a pulsational shock has passed through the gas , when the velocity vectors have an outward radial velocity component ( toward the upper left ) , except in the vicinity of the companion .
the left panel shows the velocity field at a time about midway between shock passages , when most of the velocities have a radial infall component ( toward the lower right ) . in the left panel
we see gas flowing around the inner side of the companion , and encountering a shock where the velocity vectors get smaller and turn . in the post - shock wake
this material streams around and behind the companion , where it encounters a shock separating it from material flowing in from the outer side of the companion .
the latter stream is turned aside .
the vortex likely formed in this sense because gas coming around on this side encounters the wake and is shocked .
the outer stream hits the wake at a glancing angle an is deflected .
the inner stream finally spirals onto the companion in a counterclockwise sense . in the right panel
the situation is reversed .
it is the outer stream that impacts the wake , is shocked , and streams around and onto the companion in a clockwise sense . in this case , when the pulsational shock has just passed , the wake and its vortex structure is more compressed .
shock passage moves and compresses the wake quite abruptly , accounting for the rapid drop in accreted angular momentum in the last plot of figure [ f8 ] .
after that the radial flow slows and reverses , and it takes most of the cycle to reform the vortex of the left panel of figure [ f9 ] .
this accounts for the relatively slow rise of angular momentum in the @xmath46 model . in sum , the change in the nature of angular momentum accretion from quite random to nearly periodic is associated with the increasing role of accretion out of the more distant parts of the wake , and the formation of wake vortices with alternating senses of rotation .
this is in contrast to the mass accretion , which is periodic in low mass models because of its immediate sensitivity to the passage of pulsational shocks . as in the case of mass accretion , there are both similarities and differences in the angular momentum accretion of the x - ray binary flip - flop case and the present case . like the present case
, the value of the accreted specific angular momentum is highly variable in the flip - flop models ( e.g. , @xcite , @xcite ) .
the variations can be of comparable magnitude to those in the present case , and also include sign changes in the accreted angular momentum .
the variations in the three - dimensional models of @xcite are not as regular as the high mass cases shown in figure [ f8 ] , but the two - dimensional simulations of @xcite are very similar .
however , these flip - flop models are mostly adiabatic , and we recall that there is no flip - flop instability in the isothermal case .
in this section we consider the ramifications of the results of the previous section for observation . before beginning we recall that the models above were focussed on a particular region of parameter space .
specifically , we considered stellar companions in the mass range of @xmath51 , orbiting at a radius of @xmath52 from an approximately @xmath53 mira , a region in the extended stellar atmosphere where pulsational motions are still important . in the discussion below we
will also compare the results above to those of paper i , and note possible observational effects beyond this parameter range . in @xcite , in paper i , and in section 1 above we considered
the possibility that accretion processes onto a companion might help explain the short time - scale optical variability in miras studied by @xcite in hipparcos data . in this subsection
we reconsider this issue in light of the new model results .
the models simulate the flow on the outer or accretion scale , and provide no direct information about the inner scale , near the companion . however , the properties of the captured material , like its time dependent angular momentum , allow us to infer a good deal about the behavior on the small scale .
consider first the low mass companions where the accreted angular momentum varies relatively stochastically , with a low mean value .
for example , in the case of a @xmath45 companion , the specific angular momentum is about @xmath54 ( see figure [ f8 ] ) .
if gas elements with this value of h settle into a circular accretion disc around the companion , without loss or gain of angular momentum , then the radius of that disc would be of about , @xmath55 where @xmath56 is the radius of jupiter . in the example with a @xmath45 companion , and @xmath57
, this equation yields @xmath58 .
thus , at the low end of the mass range considered above , the significant fraction of the accreted gas with specific angular momentum somewhat smaller than this value ( or companion mass or radius somewhat larger ) can fall directly onto the companion surface .
the rest may be found in a small , and continually disturbed , accretion disc .
( we note that such discs are much smaller than those of mastrodemos & morris ( 1998 , 1999 ) , whose stellar mass companions , located in steady outflows , accrete gas with higher values of the specific angular momentum . )
we discussed mass accretion rates in section 3.2 . with those rates , and if we assume that accreted particles hit the surface of a @xmath42 companion of radius about @xmath59 with about free - fall velocity , and that essentially all of their kinetic energy is converted into optical luminosity , then the magnitude of this luminosity is , @xmath60 this value is for accretion bursts and is correspondingly smaller for the low accretion rates between bursts .
it also assumes an optimistic conversion of accretion energy to optical photons . on the other hand
, it would be substantially increased for a somewhat more massive companion in the denser wind of a more evolved mira .
it would be reduced in the case of greater star / companion separations .
these results are in agreement with those of our earlier papers , where we found that there would probably not be sufficient energy in the mean accretion onto giant planets to produce optical flashes .
we conjectured that gas with angular momentum of one sign might be accreted for about half a pulsation cycle , and stored in a disc .
then in the second half of the cycle a comparable mass of gas with angular momentum of opposite sign might rain down and flush the disc onto the companion .
the large mass of stored gas could produce luminosities large enough to be observable at stellar minimum light . in the case of x - ray binaries ,
this disc storage and flushing has long been considered as a possible explanation for the rapid changes in the luminosity and period of x - ray pulsars @xcite .
more recently , @xcite estimated that in favorable cases , angular momentum accretion reversals could be responsible for abrupt spin changes in some x - ray pulsars .
figure [ f8 ] suggests that for the lower companion masses there is no periodicity in the accreted angular momentum , so the simple storage mechanism will not work .
most of the time the accreted angular momentum has a positive sign , so most of the time some fraction of this material adds to the small disturbed disc mentioned above .
occasionally there are bursts of negative angular momentum accretion .
if these bursts contain enough mass , or their angular momentum is sufficiently negative , then they might dump enough material onto the companion to produce an observable burst . in numerical models
such random bursts are partially the product of noise resulting from small particle numbers .
turbulence or similar effects would likely generate them in the circumstellar environment .
another possibility is that pressure from hot gas or magnetic fields could inhibit the flow to small scales until the accreted mass builds up to a critical level .
the result could be a burst of quite large mass ( and small angular momentum ) , like those in the model of paper i. interaction with a rotating magnetosphere could also add or subtract angular momentum to accretion disc gas .
the study of such processes is beyond the scope of this paper . in the case of large companion masses ,
figure [ f4 ] shows larger mass accretion rates , and figure [ f8 ] indicates quasi - periodic angular momentum accretion .
thus , the prospects for storage , followed by large bursts are much better . according to the estimate of equation ( 1 ) , with h values from figure [ f8 ] ,
the typical accretion disc radii are still small , of order a few times @xmath56 .
however , with the deeper gravitational potential , and higher mass accretion rates in bursts ( accumulated in storage ) , the luminosity estimates from equation ( 2 ) can exceed a solar luminosity .
this is more than enough to give a detectable signal near minimum light in a large amplitude mira , as noted in @xcite .
in fact , storage may not be required for observable optical effects at companion masses of about @xmath46 or higher .
they might also be possible at fairly large radii where the ambient gas density is smaller and where pulsational shocks are weak and there is essentially no radial motion .
in this case the optical enhancement would be relatively constant , with no bursts , which would make it harder to distinguish from the mira minimum luminosity .
however , we would expect variations over the companion orbital period , as the companion is more or less hidden by the extended mira atmosphere ( unless its orbital plane coincided with the plane of the sky ) . to clarify the issues raised in the previous paragraphs it is worth extending the numerical results with a simple semi - analytical model of the inner accretion disc .
we adopt the simplifying assumption that all particles accreted with negative angular momentum and a number of disc particles carrying an equal amount of positive angular momentum fall immediately onto the companion .
after such events , the accretion disc is assumed to consist of all the remaining particles , with each carrying an equal part of the remaining cumulative angular momentum . with these approximations
the model yields the amount of infall onto the planet and the angular momentum in the disc as functions of time .
figure [ f10 ] shows some results of applying this simple model to data output at every timestep from the numerical hydrodynamical model , in the cases of the @xmath61 models . in each case , the rapidly varying line shows the mass added to the disc versus time , which is negative if material falls onto the companion .
the smoother solid curve shows the net specific angular momentum in the disc , and the dashed curved shows the net particle number in the disc .
comparison of these particle numbers with those of figure [ f2 ] for the outer accretion shows that a significant fraction of the particles do fall onto the companion in this model .
the values of the specific angular momenta generally seem to settle to a limiting value in the three cases , though this is not entirely clear in the @xmath45 model . in the other two cases ,
this quantity first peaks , and then settles to a lower final value . comparison shows that , in all cases , the final value for the model in figure [ f10 ] is slightly higher than the mean in figure [ f8 ] . presumably this is because the mean value of the accreted positive angular momentum is much greater than the mean negative value . finally , we can use figure [ f10 ] and the first equality of equation ( 5 ) to get another estimate of the magnitude of energetic outbursts from the planet .
the typical timestep of the numerical model is about 1.5 days , and from figure [ f10 ] , the mass of infall onto the companion is typically a burst of about 30 particles in the @xmath62 models , respectively .
these values translate into a mass accretion value of @xmath63 , and a luminosity of about @xmath64 for the @xmath42 model .
the luminosity would be a few times greater for the @xmath46 model and mass infalls can also sometimes be a few times greater . moreover ,
since the free - fall time at the mean disc radius is only a few hours , the numerical models may overestimate the infall time of a mass cluster , and thus , underestimate the luminosity .
luminosities of order @xmath65 seem quite possible .
note , in particular , that the estimates of equation ( 2 ) are based on a relatively low mira mass loss rate ( @xmath66 ) and the corresponding mean ambient density . as discussed above
infall events come in clusters , so we would expect an outburst to be sustained for times of several days to several weeks , as seen in the hipparcos data . in previous papers ( struck - marcell 1988 , willson and struck 2001 , paper i ) we noted that the region of sio maser emission in miras with such emission coincides with the pulsating region of the extended atmosphere .
those works also discussed how the presence of a giant planet or brown dwarf companion at similar orbital radii could effect the sio emission , e.g , , with clumping due to turbulence in the wake to produce density enhancements needed for the masing , and with relatively strong local magnetic fields around the companion and in the wake to help explain the high polarizations of some maser stars .
the new models provide more detail on wake structure , especially in the case of high mass companions . on the other hand , information on the structure of clumps
is limited , because that will depend on detailed cooling and heating processes not included in our isothermal equation of state . despite this limitation ,
figure [ f7 ] still shows the presence of substantial inhomogeneities in the wakes .
the vortices in figure [ f9 ] provide an even more striking example of inhomogeneous flow .
the wake bounding shocks visible in figure [ f9 ] , combined with the au length of the wakes in figure [ f7 ] , show that there are substantial shock surfaces .
wake and pulsational shocks are of comparable amplitude in the models above .
thus , the wake environment should be just as favorable as that of the pulsational shocks for exciting sio molecules . moreover ,
the last three panels of figure [ f1 ] show that the companion and the wake can noticably ripple the outer edge of pulsational shocks , and perturb the gas behind that edge , over a large azimuthal angle . as a result , wakes may also enhance clumping behind pulsational shocks . @xcite
have suggested that pulsational shocks are mediated by strong magnetic fields ( of order 50 g ) , and that clumps result from the action of the parker instability in those shocks .
one obvious difference between the two models is that companion induced clumping occurs around an orbital ring , which could be highly flattened in some views .
the hartquist and dyson clumping , like clumping in the pulsational shocks , would be spherically symmetric , in the absence of rotational flattening , and probably be seen as always circular , with limb brightening .
the maser spots of _ o ceti _
( mira ) appear to have a fairly flattened distribution , which could be a ring viewed edge - on @xcite .
the range of sizes in the wake and wake inhomogeneities in the substellar companion model ( i.e. , of order tenths of an au up to a few au ) also seem in accord with vlbi observations of sio maser spots . for example , masers spots subtending an angle of 3 milliarcseconds , near a mira at a distance of about 300 pc , have a physical extent of about 1 au .
( these numbers derive from @xcite study of tx cam , but are also appropriate for _ o ceti _ , see @xcite . )
the shortest time - scales reported for variations in sio maser line profiles are about 10 - 20 days @xcite .
this corresponds to a distance of a few hundredths of an au for a roughly sonic velocity of about 5.0 km / s , and still less than 0.1 au for shocks propagating at 10 km / s .
thus , such rapid changes must be associated with quite small scale dynamical phenomena , such as a shock passage , collisions between small clumps , or interactions on the small scale near a companion . on the other hand
, lifetimes of specific spectral features can range up to several hundred days , which would correspond to a sonic traversal distance of close to an au .
such motions could include either traversal of a companion wake or the development of a clump in a hydromagnetic instability .
in this subsection we consider some more speculative possibilities for detecting planetary or brown dwarf companions to miras . these include effects related to the presence of : 1 ) a hot corona around the companion , and 2 ) intermittent jets launched from the companion .
we have described above how mira wind material may fall onto the companion and an accretion disc with a typical radius of several times that of the companion radius , at about free fall velocities . with the companion mass range considered in this paper ,
these velocities are of order 300 km / s .
collisions between gas elements with such relative velocities could generate gas temperatures of millions of kelvins in at least a fraction of the gas .
at the high densities of the companion surface , and even the accretion disc surface , cooling is probably very efficient .
however , less dense gas in the atmospheres of the companion and the disc can be heated either in the collision or by absorbing the cooling radiation .
in paper i we estimated that , if converted to ionizing photons , the accretion luminosity could ionize a sphere of radius of order @xmath67 around the companion ( depending on many poorly known parameters ) .
the pressure scale height of gas with a temperature of a few million degrees is also of order @xmath68 .
thus , this material is deep within the capture radius of the companion , so any passing circumstellar gas will be captured , and will not sweep the corona out .
we would expect the coronal gas to be of quite low density relative to its surroundings .
for example , we can require that its free cooling time be greater than about 1 yr . , a typical time between accretion bursts or pulsational wave passages .
then with a high temperature cooling rate of about @xmath69 ( e.g. , @xcite ) , we find that the coronal density must be less than about @xmath70 . given this density and a volume corresponding to the radius above
, we find that the emission associated with coronal cooling is negligible ( @xmath71 ) .
generally , we expect that to obtain a detectable coronal emission measure would require a material density that is high enough to give a cooling time of much less than a year .
thus , while a corona is an interesting component of the companion region , is not directly observable .
most of the accretion energy will be emitted from transient hot spots on the disc and companion .
the companion may have a magnetic field , and magnetosphere , like that of jupiter .
if so , it would have an important effect on the corona and the accretion disc .
first of all , we note that the magnetic pressure , given a surface field of about 3.0 g , would be of the order , @xmath72 at a distance r from the companion centre .
this is generally comparable to or greater than the thermal pressure of a maximum density corona given by , @xmath73 when hot gas is created in accretion impacts it will begin to expand out into the corona , but it will also be constrained by the magnetic field .
if the companion and its magnetosphere are rotating like jupiter , then the centrifugal force will be important , e.g. , in driving a centrifugal wind . @xcite and
@xcite have also discussed the formation of companion winds around agb stars , however , the case they considered was that of a white dwarf or main sequence companion , which is able to spin up the mira atmosphere , and induce strong roche lobe accretion onto the companion .
this induces a strong fast wind .
the cases considered in this paper generally fail to meet soker & rappaport s condition for strong tidal interaction , though the most massive brown dwarfs orbiting within a few au would be affected .
therefore we would not expect as strong a wind as described in their work .
nonetheless , once the accretion disc builds up to a substantial mass , the enviroment around it my be quite similar to that around a magnetic protostar .
the x - wind theory of shu and collaborators ( e.g. , @xcite and references therein ) may be applicable
. however , the mass and energy in such a variable wind would be much less than in the protostellar case , so it seems unlikely that these effects would be directly observable .
in the models described above we have studied the interaction between the extended , pulsating atmosphere and wind of a mira variable and a brown dwarf or giant planet companion .
our numerical models resolved the time - dependent bow shock and wake structure created by the companion .
the outermost parts of the accretion onto the companion were also resolved .
we obtained the following results concerning the gas flow .
\1 ) the accretion rate scales roughly linearly ( in two dimensions ) with the companion mass .
( see fig .
[ f3 ] and section 3.2 . )
\2 ) the equation of state , or more precisely the value of the exponent @xmath18 of the adopted barotropic equation , has little effect on the accretion rate . naively , this is because accretion is the result of gravitational capture under conditions where the thermal pressure is small .
\3 ) one of the most interesting results above is the transition from periodic to chaotic mass accretion , and from chaotic to periodic angular momentum accretion as the companion mass is increased through the range considered in this paper ( @xmath51 ) .
( see figs .
[ f4 ] and [ f8 ] , and sections 3.3 and 3.4 . )
\4 ) the wakes grow substantially as the companion mass is increased through this range .
wake structure , and in particular , clumping could have an important effect on sio maser emission , which derives from the same region .
( see fig .
[ f6 ] and section 4.3 ) .
while we do not resolve the flow on the inner scale ( @xmath74 ) , we did monitor the mass and angular momentum accreting into this region . with these data and a few simplifying assumptions
, we examined the properties of a simple semi - analytic model for this flow ( section 4.2 ) . using this model we obtained the following additional results for flow on the inner scale .
\5 ) the effective boundary conditions between the inner and outer scales are important . in paper
i we adopted a boundary condition equivalent to the assumption that it was difficult for accreting gas to penetrate to the inner scale .
this lead to strong bursts of accretion .
this case requires coronal or magnetospheric pressure to be strong enough to resist the infalling gas .
estimates of the corona derived from our semi - analytic model , and of magnetospheres comparable to that of jupiter , suggest that that would not be the case with companions like those studied in this paper .
\6 ) the semi - analytic model suggests that an accretion disc would form around the companion on a scale of order a few companion radii ( see section 4.4 ) .
\7 ) the pattern of angular momentum accretion ( positive and negative ) suggests that accretion bursts are likely .
they may be strong enough to generate observable effects in the optical ( see section 4.1 ) .
\8 ) energy dissipation from the impact of free - falling gas elements on the companion and the accretion disc may generate a thin hot corona around the companion , and possibly a coronal wind ( see section 4.4 ) .
it is unlikely that these would be directly observable . in sum ,
the situation studied in this paper is a previously unstudied type of close binary interaction .
this interaction is different from that of the more well known symbiotic and cataclysmic type binaries in several respects .
first of all , the close companion in those systems is generally a more massive white dwarf or m dwarf , and the interaction may involve mass transfer via roche lobe overflow , or interacting winds , rather than direct accretion out of a mira wind .
a second aspect of the present study is the inclusion of a radially oscillating environment , which we found has profound effects on the accretion process .
the oscillating environment has not generally been considered in models of other types of close agb binaries
. however , as @xcite and @xcite have pointed out , a close massive companion would exert significant tidal forces on the mira atmosphere .
moreover , the mira would have a significant motion around the centre of mass . these and other relatively energetic phenomena make it unlikely that such systems would be classified as ordinary miras .
such cases deserve further study .
several other issues raised above also merit further study .
the first is the thermohydrodynamics of the inner scale .
fully self - consistent numerical modeling with heating and cooling and magnetic fields included would be very interesting , if not entirely feasible at present .
including heating , cooling and magnetic effects in the wake flow would also be very interesting , especially to see the impact of these processes on sio maser emission .
however , the most urgent need is for more observational constraints on the models .
the models above provide a number of observationally testable predictions .
the observations published to date provide intriguing hints , but are not sufficient to provide strong evidence for the existence of such systems or strong tests of the models .
we are grateful to g. h. bowen for many helpful conversations on miras , and to an anonymous referee for numerous helpful suggestions .
this research has made extensive use of the nasa ipac extragalactic database ( ned ) . | we present numerical hydrodynamical modeling of the effects of a giant planet or brown dwarf companion orbiting within the extended atmosphere and wind formation zone of an approximately solar - mass mira variable star .
the large - scale , time - dependent accretion flows within the radially oscillating and outflowing circumstellar gas around miras are related to bondi - hoyle - lyttleton flows , but have not , to our knowledge , been previously modelled .
the new models presented in this paper illustrate the changes in accretion and wake dynamics as the companion mass is varied over a range from 10 to 50 jupiter masses ( @xmath0 ) , and generalize the results of the single model we presented in an earlier paper .
the character of the accretion onto the companion changes greatly as the companion mass is increased . at the lowest companion masses considered here ,
a low continuous rate of mass accretion is punctuated by large , nearly periodic bursts of accretion .
when the companion mass is large , the mass accretion has both a continuous part , and a rapidly varying , nearly stochastic part .
surprisingly , the angular momentum of the accreted gas shows an opposite trend with mass , varying nearly periodically at large companion masses , and stochastically at low masses .
these trends can be understood as the result of the interplay between the shocks and radial oscillations in the circumstellar gas , and the wake flow behind the companion .
boundary conditions also affect the character of the accretion .
the equation of state , however , is found to have little effect , at least for gamma - law gases , with gamma in the range from 1 to 5/3 .
models with accretion bursts may produce observable optical brightenings , and may affect sio maser emission , as we suggested in previous papers
. interruptions of continuous accretion , or shadowing effects , could give rise to bursts of dimming in the optical .
such dimming effects are likely to be correlated with bursts and optical flashes , helping to explain some rather mysterious hipparcos observations .
[ firstpage ] stars : variables : stars agb and post - agb accretion , accretion discs masers circumstellar matter planetary systems : stars winds . |
phenomenological supersymmetry allows to incorporate the perturbative standard model in a theory which solves the hierarchy problem all the way up to the planck scale , with a potentially successful description of gauge coupling unification . the consistency with the electroweak precision tests ( ewpt ) of the standard model ( sm ) with a relatively light higgs boson adds support to this view , making the test of the minimal supersymmetric standard model ( mssm ) a crucial task of the lhc
this is a meaningful straight path to be followed for particle physics in the next few years . while remaining in the context of phenomenological supersymmetry , however , the lack of signals so far both in the higgs and in the flavour sectors have raised and continue to raise questions . that the higgs problem of the mssm be a naturalness problem is too well known to be recalled here in detail : the sensitivity of the fermi scale , as determined by the higgs potential of the mssm , to the average stop masses makes it unnatural to raise the mass of the lightest scalar , @xmath2 , much above the tree level bound , @xmath3 , in potential conflict with the lep bounds .
is the flavour problem a naturalness problem as well ? given the little we understand about flavour , this is not the easiest question to answer .
let us take the view , however , as put forward by many authors in the nineties @xcite - @xcite , that the supersymmetric flavour problem may have something to do with a hierarchical structure of s - fermion masses : the first two generations significantly heavier than the third one .
how much heavier can now become a naturalness problem , depending on the bounds that the sfermion masses have to satisfy @xcite@xcite . here
we argue about the possibility that the two issues , the higgs problem " and the flavour problem " , be related naturalness problems , that may be addressed at the same time by properly extending the mssm .
let us insist on the supersymmetric flavour problem in connection with a hierarchical s - fermion spectrum .
as well known , without degeneracy nor alignment between the first two generations of squarks , @xmath4 , the consistency with the @xmath5 transitions , both real and especially imaginary , would require values of @xmath4 far too big to be natural .
relatively mild assumptions on all the s - fermion masses of the first two generations , on the other hand , as recalled later , allow to satisfy the various flavour constraints by smaller values of @xmath6 that may be considered if they are natural or not , hence the potential connection with the higgs mass problem .
in formulae , the two naturalness constraints ( @xmath7 is the amount of fine tuning as defined in the usual way @xcite , @xmath8 is the average stop mass ) : @xmath9 @xmath10 must be considered together and the corresponding bounds might be reduced to an acceptable level by pushing up the theoretical value of @xmath11 , on which the level of fine tuning depends at least quadratically with the @xmath12 mass or with any of the soft mass parameters for the higgs doublets does not change the naturalness constraints on @xmath8 or on @xmath6 , at least as long as the other physical higgs bosons are not too close in mass to the lightest one , @xmath2 , as we consider in the following for good phenomenological reasons . on this , see e.g. @xcite . ] . ways to push up @xmath11 even by a significant amount , between 200 and 300 gev , have already been put forward @xcite - @xcite . whether and how the flavour problem can also be attacked in this manner is a model dependent question that we are going to analyze in various cases proposed in the literature . in summary , and as an anticipation , we seek for models where a typical spectrum like the one shown in fig .
[ spettro2 ] can be naturally implemented .
gev and @xmath13 tev.,width=377 ]
a way to summarize the potential connection between the supersymmetric flavour problem and hierarchical s - fermion masses is the following or viceversa . ] . * without degeneracy nor alignment the bounds that the first two generations of squark masses would have to satisfy to be compatible with the flavour constraints , mostly from @xmath14 transitions , are in the hundreds of tev , with weak dependence on the much lighter gaugino masses . on the other hand ,
if we assume degeneracy and alignment of order of the cabibbo angle , i.e. in terms of the standard notation : @xmath15 and @xmath16 , then the bounds are significantly reduced to : @xmath17 @xmath18 furthermore if @xmath19 ( or @xmath20 ) , these bounds are replaced in the strongest cases by : @xmath21 @xmath22 from cp conserving or cp violating effects respectively .
* the exchange of the third generation of s - fermions may also produce too big flavour effects unless the off - diagonal @xmath23 are small enough .
if for example we assume a correlation between the off - diagonal elements and the ratio of the diagonal masses of the type : @xmath24 a dominant constraint comes from @xmath25 mixing : @xmath26 similar or weaker constraints are obtained from the electric dipole moments . [ cols="^,^ " , ] the way in which the lsp in the mssm can acquire the observed relic abundance to allow its interpretation as a dm candidate is well known . as observed in @xcite ,
after lep constraints are taken into account , the correct prediction for the dm density requires special relations among parameters , justifying the terminology of well - tempered " neutralino .
this is neatly illustrated in fig .
[ dm1 - 2 ] on the left hand side , which is appropriate to the well - tempered " bino / higgsino case , i.e. for large ( and irrelevant ) @xmath27 : to obtain the observed relic abundance , @xmath28 and @xmath29 should be pretty close to each other . in the same plot , which is for @xmath30 gev and @xmath31 , the regions are also shown to which the direct detection searches are either currently sensitive @xcite or should become sensitive in a near future @xcite . to draw these contours
we assume everywhere a standard dm density in the halo of our galaxy .
these sensitivity regions are therefore directly relevant only where they overlap with regions of correct relic abundance .
the effect of the larger @xmath11 is clearly visible in the same fig .
[ dm1 - 2 ] on the right hand side , which is appropriate to @xmath32susy for @xmath33 gev and @xmath34 , while @xmath27 is still kept large . in both plots of fig .
[ dm1 - 2 ] it is @xmath35 gev . the effect of the @xmath36 threshold , only visible in the figure on the right , is due to the @xmath37 behaviour of the @xmath38 coupling , negligible in the case of the mssm for @xmath39 . fig .
[ dm3 - 4 ] shows two other cases for @xmath32susy . on the left hand side
everything is as in fig .
[ dm1 - 2 ] right , except for @xmath40 gev . on the right hand side , for @xmath41 gev and @xmath34 , @xmath27 is lowered to 200 gev .
the raise of @xmath11 has also a clear and well known effect on the direct detection cross sections , dominated by @xmath2-exchange and therefore proportional to @xmath42 @xcite@xcite .
this effect is relatively compensated in the low @xmath27 case by a significant change in the lsp composition .
can it be that the higgs mass problem and the flavour problem contain a unique message ?
this is the question we have addressed in this work . notwithstanding
the validity of the standard mssm approach , which makes its test a crucial task of the lhc , we believe that this is a meaningful question .
truly enough it rests on the notion of naturalness , which can hardly be viewed as the basis of any theorem .
the lack of any serious understanding of the flavour pattern is another difficulty we face .
yet the possibility that the higgs mass problem and the flavour problem point to an extension of the mssm needs to be given serious consideration .
the basic simple idea that we pursue is that a lightest higgs boson naturally heavier than in the mssm renders at the same time more plausible that the supersymmetric flavour problem has something to do with a hierarchical structure of the s - fermion masses , a connection often invoked in the past . at first the constraints set by
the lack of flavour signals would seem to require values of the masses of the first two generations totally incompatible with naturalness .
however the combination of mild flavour assumptions with a relaxation of the naturalness constraints by an order of magnitude thorough a heavier higgs boson than normal can change the situation . the concrete proposal that we make , which should at least be taken as an example ,
consists of the following .
with degeneracy and alignment between the first two generations of s - fermions controlled by a parameter of the order of the cabibbo angle and a ratio of @xmath43 between @xmath44 and @xmath45 , in one direction or another , even the hardest flavour constraints can be satisfied by @xmath46 tev . in turn
these masses are natural if they are born degenerate , at least within @xmath47 multiplets , at a scale @xmath48 below @xmath49 tev and a modification of the higgs sector , which remains perturbative up to the same scale , raises the lighest higgs boson mass in the @xmath50 gev range , e.g. like in @xmath32susy . no matter what produces it , a non standard supersymmetric spectrum like the one shown in fig .
[ spettro2 ] is brought into focus . even before any detailed investigation , which we believe is well worth doing ,
the following phenomenological consequences clearly emerge : * the abundance of top , even generally more than bottom quarks , in the gluino decays , giving rise to a distinctive signature in gluino pair production , which could be detected already in the early stages of the lhc . * the appearance of the very much non mssm - like golden mode decay of the lighest higgs boson , @xmath51 , although with a reduced branching ratio @xcite@xcite relative to the sm one with the same higgs boson mass . * a distinctive distortion of the relic abundance of the lightest neutralino , again relative to the mssm , due to the @xmath52-channel exchange of the heavier higgs boson in the lsp annihilation cross section , with an lsp which needs no longer be well - tempered " .
several other phenomenological features , more or less tied to @xmath32susy , are present , which may be useful to study carefully .
we thank andrea romanino and marco nardecchia for help on sect .
[ hier ] , as well as gino isidori and gilad perez for their comments on related issues .
this work is supported in part by the european programme unification in the lhc era " , contract pitn - ga-2009 - 237920 ( unilhc ) , by the miur under contract 2006022501 and by the swiss national science foundation under contract no .
200021 - 116372 .
m. dine , r. g. leigh and a. kagan , phys .
d * 48 * ( 1993 ) 4269 , http://arxiv.org/pdf/hep-ph/9304299[_arxiv:hep-ph/9304299_ ] .
p. pouliot and n. seiberg , phys .
b * 318 * ( 1993 ) 169 , http://arxiv.org/pdf/hep-ph/9308363[_arxiv:hep-ph/9308363_ ] .
a. pomarol and d. tommasini , nucl .
b * 466 * , 3 ( 1996 ) , http://arxiv.org/pdf/hep-ph/9507462[_arxiv:hep-ph/9507462_ ] .
r. barbieri , g. r. dvali and l. j. hall , phys .
b * 377 * , 76 ( 1996 ) , http://arxiv.org/pdf/hep-ph/9512388[_arxiv:hep-ph/9512388_ ] .
a. g. cohen , d. b. kaplan and a. e. nelson , phys .
b * 388 * , 588 ( 1996 ) , http://arxiv.org/pdf/hep-ph/9607394[_arxiv:hep-ph/9607394_ ] .
r. barbieri , l. j. hall and a. romanino , phys .
b * 401 * ( 1997 ) 47 , http://arxiv.org/pdf/hep-ph/9702315[_arxiv:hep-ph/9702315_ ] .
j. r. espinosa and m. quiros , phys . rev .
* 81 * ( 1998 ) 516 , http://arxiv.org/pdf/hep-ph/9804235[_arxiv:hep-ph/9804235_ ] .
n. polonsky and s. su , phys .
b * 508 * , 103 ( 2001 ) , http://arxiv.org/pdf/hep-ph/0010113[_arxiv:hep-ph/0010113_ ] .
j. a. casas , j. r. espinosa and i. hidalgo , jhep * 0401 * ( 2004 ) 008 , http://arxiv.org/pdf/hep-ph/0310137[_arxiv:hep-ph/0310137_ ] .
a. brignole , j. a. casas , j. r. espinosa and i. navarro , nucl .
b * 666 * ( 2003 ) 105 , http://arxiv.org/pdf/hep-ph/0301121[_arxiv:hep-ph/0301121_ ] .
r. harnik , g. d. kribs , d. t. larson and h. murayama , phys .
d * 70 * ( 2004 ) 015002 , http://arxiv.org/pdf/hep-ph/0311349[_arxiv:hep-ph/0311349_ ] .
s. chang , c. kilic and r. mahbubani , phys .
d * 71 * ( 2005 ) 015003 , http://arxiv.org/pdf/hep-ph/0405267[_arxiv:hep-ph/0405267_ ] .
p. batra , a. delgado , d. e. kaplan and t. m. p. tait , jhep * 0402 * , 043 ( 2004 ) , http://arxiv.org/pdf/hep-ph/0309149[_arxiv:hep-ph/0309149_ ] .
a. birkedal , z. chacko and y. nomura , phys .
d * 71 * ( 2005 ) 015006 , http://arxiv.org/pdf/hep-ph/0408329[_arxiv:hep-ph/0408329_ ] .
a. maloney , a. pierce and j. g. wacker , jhep * 0606 * ( 2006 ) 034 , http://arxiv.org/pdf/hep-ph/0409127[_arxiv:hep-ph/0409127_ ] .
k. s. babu , i. gogoladze and c. kolda , http://arxiv.org/pdf/hep-ph/0410085[_arxiv:hep-ph/0410085_ ] .
a. delgado and t. m. p. tait , jhep * 0507 * ( 2005 ) 023 , http://arxiv.org/pdf/hep-ph/0504224[_arxiv:hep-ph/0504224_ ] .
r. barbieri , l. j. hall , y. nomura and v. s. rychkov , phys .
d * 75 * , 035007 ( 2007 ) , http://arxiv.org/pdf/hep-ph/0607332[_arxiv:hep-ph/0607332_ ] . g. f. giudice , m. nardecchia and a. romanino , nucl .
b * 813 * ( 2009 ) 156 , http://arxiv.org/abs/arxiv:0812.3610[_arxiv:0812.3610_ ] .
lodone , http://arxiv.org/abs/arxiv:1004.1271[_arxiv:1004.1271_ ] .
n. arkani - hamed and h. murayama , phys . rev .
d * 56 * ( 1997 ) 6733 http://arxiv.org/abs/arxiv:hep-ph/9703259[_arxiv:hep-ph/9703259_ ] .
k. agashe and m. graesser , phys .
d * 59 * ( 1999 ) 015007 http://arxiv.org/abs/arxiv:hep-ph/9801446[_arxiv:hep-ph/9801446_ ] .
r. barbieri and d. pappadopulo , jhep * 0910 * ( 2009 ) 06 , http://arxiv.org/abs/arxiv:0906.4546[_arxiv:0906.4546_ ] .
r. m. barnett , j. f. gunion and h. e. haber , phys .
b * 315 * ( 1993 ) 349 , http://arxiv.org/pdf/hep-ph/9306204[_arxiv:hep-ph/9306204_ ] .
m. guchait and d. p. roy , phys .
d * 52 * , 133 ( 1995 ) , http://arxiv.org/pdf/hep-ph/9412329[_arxiv:hep-ph/9412329_ ] .
m. toharia and j. d. wells , jhep * 0602 * ( 2006 ) 015 , http://arxiv.org/pdf/hep-ph/0503175[_arxiv:hep-ph/0503175_ ]
. b. s. acharya , p. grajek , g. l. kane , e. kuflik , k. suruliz and l. t. wang , http://arxiv.org/abs/arxiv:0901.3367[_arxiv:0901.3367_ ] .
l. cavicchia , r. franceschini and v. s. rychkov , phys .
d * 77 * ( 2008 ) 055006 , http://arxiv.org/abs/arxiv:0710.5750[_arxiv:0710.5750_ ] .
n. arkani - hamed , a. delgado and g. f. giudice , nucl .
b * 741 * ( 2006 ) 108 , http://arxiv.org/pdf/hep-ph/0601041[_arxiv:hep-ph/0601041_ ] .
z. ahmed _ et al .
_ [ the cdms - ii collaboration ] , http://arxiv.org/abs/arxiv:0804.0622[_arxiv:0912.3592_ ] .
e. aprile , l. baudis and f. t. x. collaboration , pos * idm2008 * , 018 ( 2008 ) , http://arxiv.org/abs/arxiv:0902.4253[_arxiv:0902.4253_ ] .
r. barbieri , m. frigeni and g. f. giudice , nucl .
b * 313 * ( 1989 ) 725 .
m. drees and m. nojiri , phys .
d * 48 * , 3483 ( 1993 ) , http://arxiv.org/pdf/hep-ph/9307208[_arxiv:hep-ph/9307208_ ] . | taking a bottom - up point of view and focussing on the lack of signals so far in the higgs and in the flavour sectors , we argue in favour of giving consideration to supersymmetric extensions of the standard model where the lightest higgs boson has a mass between 200 and 300 gev and the first two generations of s - fermions are above 20 tev . after examining the simplest extensions of the minimal supersymmetric standard model that allow this in a natural way , we summarize the main consequences of this pattern of masses at the lhc and we analyze the consequences of a heavier than normal higgs boson for dark matter . 1.0 cm * a non standard supersymmetric spectrum * 1.0 cm riccardo barbieri@xmath0 , enrico bertuzzo@xmath0 , marco farina@xmath0 , paolo lodone@xmath0 and duccio pappadopulo@xmath1 + _ @xmath0 scuola normale superiore and infn , piazza dei cavalieri 7 , 56126 pisa , italy _
+ _ @xmath1 institut de thorie des phnomnes physiques , epfl , ch1015 lausanne , switzerland_1.0 cm |
core - collapse supernovae are energetic explosions that mark the death of stars more massive than @xmath20 m@xmath8 .
they are extremely important in several areas of astrophysics , including the nucleosynthesis of chemical elements , energy feedback that affects the evolution of galaxies , the formation of compact object remnants , the production of high - energy particles , as tracers of recent star - formation in galaxies , and as test cases of massive stellar evolution .
supernovae discovered in nearby galaxies have been particularly important for testing our physical understanding of the explosions and to establish direct links between the events and their progenitors , which puts tight constraints on stellar evolution theory .
the two best - studied nearby core - collapse supernovae , the peculiar type ii sn 1987a in the lmc and the type iib sn 1993j in m81 , had progenitors that were detected in pre - explosion images . both were very interesting cases that challenged theoretical expectations based on stellar evolution models that predicted red supergiants as their progenitor stars : sn 1987a had a @xmath21 m@xmath8 blue supergiant progenitor star ( e.g. , arnett et al .
1989 ) and sn 1993j had a @xmath22 m@xmath8 yellow supergiant progenitor ( e.g. , aldering et al .
1994 ) , which turned out to be a massive binary system ( e.g. , maund et al .
2004 ) . after more than a decade of work , smartt et al .
( 2009 ) presented a thorough study of nearby ( @xmath23 mpc ) type ii - plateau supernovae with deep pre - explosion imaging ( see also li et al . 2005 ) , which allowed them to constrain the progenitors of type iip supernovae to be red supergiants with initial main - sequence masses in the range @xmath24 m@xmath8 .
this result is in conflict with well - studied red supergiant samples in the galaxy and the magellanic clouds ( e.g. , levesque et al .
2006 ) , which also contain massive supergiants with @xmath25 m@xmath8 , and a salpeter initial mass function ( imf ) for high - mass stars .
this red supergiant problem " can be alleviated if high - mass red or yellow supergiants produce other spectroscopic or photometric type ii supernova sub - types ( e.g. , smith et al .
2009 ; elias - rosa et al . 2009 , 2010 ) , or if high - mass supergiants experience strong winds due to pulsations ( e.g. , yoon & cantiello 2010 ) .
very recently , walmswell & eldridge ( 2011 ) have proposed circumstellar dust as a possible solution .
it can also be explained by failed supernovae , an hypothesis that would have several deep implications ( e.g. , kochanek et al . 2008 ) .
in addition , there are known selection biases that have not been discussed in depth or accounted for .
of the 18 nearby type ii supernovae with fairly secure progenitor ( either single , binaries , or compact clusters ) detections , two of the 18 events were initially discovered by the galaxy - targeted lick observatory supernova search ( loss ; filippenko et al .
2001 ) , and 14 of the 18 were initially discovered by dedicated amateur astronomers that mostly concentrate their search efforts in big galaxies .
only 2 of these supernovae were found in dwarf galaxies ( sn 1987a and 2008ax ) and all the type iips ( 10 of the 18 ) were found in the disks of spiral galaxies , mostly in grand - design spirals . the existing selection bias against finding nearby core - collapse supernovae in dwarf galaxies could be important because @xmath26 of the local star formation rate density is in galaxies with absolute magnitude @xmath27 mag ( e.g. , james et al .
2008 ; young et al . 2008 ; williams et al . 2011 ) .
since metallicity is a key parameter in massive stellar evolution and stellar death ( e.g. , prieto et al . 2008a ; modjaz 2011 , and references therein ) , changes in the fractions of different type ii spectroscopic subtypes with metallicity ( e.g. , arcavi et al . 2010 ) could affect our census of progenitors in local samples . also , possible variations of the stellar imf with environment ( e.g. , meurer et al .
2009 ; but see , e.g. , myers et al .
2011 and references therein ) can introduce complications to the progenitor analyses .
recently , horiuchi et al .
( 2011 ) discussed in detail related effects of incompleteness in the existing supernova rate measurements and how they affect their use as star formation rate indicators .
in this paper we present the discovery , follow - up observations , and analysis of the nearby type ii sn 2008jb ( @xmath28 , @xmath29 ) in a southern dwarf - irregular galaxy at @xmath30 mpc ( prieto et al .
this bright @xmath31 mag supernova was missed by targeted southern supernova surveys , like chase ( pignata et al . 2009 ) and amateur searches , mainly because the host galaxy eso 302@xmath014 is a fairly low - luminosity dwarf and was not included in the catalogs of galaxies that are surveyed for supernovae .
we are able to recover the optical light curves of this supernova from two surveys that do not target individual galaxies , but rather scan large areas of the sky : the catalina real - time transient survey ( crts ; drake et al .
2009 ; djorgovski et al . 2011 ) and the all - sky automated survey ( asas ; pojmanski 1998 ) . in section [ sec1 ] we discuss the observations , including the optical discovery and follow - up , archival mid - infrared imaging , optical spectroscopy , and other archival data . in section [ sec2 ] we present detailed analysis and discussion of the optical and mid - ir observations . in section
[ sec3 ] we discuss the results and conclusions .
we adopt a distance of 9.6 mpc to eso 302@xmath014 ( lee et al .
2009 ) , estimated from the virgo - inflow corrected recession velocity and @xmath32 km s@xmath33 mpc@xmath33 .
sn 2008jb was discovered in crts archival data from the siding spring survey ( sss ) 0.5 m schmidt telescope through the snhunt project .
snhunt is an open survey for transients in nearby galaxies that uses image subtraction in images obtained by the crts .
a new transient was recovered in 2010 oct .
26 ( i d snhunt12 ) after running the difference imaging pipeline in archival data from the sss .
sn 2008jb is first detected in unfiltered sss images obtained on 2008 nov .
23 , at @xmath34 mag .
figure 1 shows the first detection of sn 2008jb in the sss images .
it was also detected in 10 images obtained between 2008 dec . 3 and 2010 jan .
we used the image subtraction software described in freedman et al .
( 2009 ) in order to obtain clean images of the supernova ( see fig . 1 ) .
the image used as a template for the subtractions was a combination of images obtained in 2007 .
we performed psf photometry using daophot ( stetson 1992 ) on the difference images .
the final calibration of the supernova photometry was done relative to asas @xmath5-band data of stars in the field ( see discussion below ) .
table 1 gives the @xmath5-band calibrated photometry from the sss data .
we also include 3@xmath35 upper limits on the magnitudes of the supernova obtained from the images just before the first detection and after the last detection . the supernova was also detected in archival images collected using the 7 cm asas south telescopes in las campanas observatory , chile . the first detection in asas south data is from 2008 nov
19 ( 4 days earlier than crts ) at @xmath36 mag in the @xmath37-band .
the asas images were processed with the reduction pipeline described in @xcite .
the @xmath5 and @xmath37 magnitudes are tied to the johnson @xmath5 and cousins @xmath37 scale using tycho @xcite and landolt @xcite standard stars .
we used aperture photometry ( 2 pixel radius ) from the normal reduction pipeline to obtain the magnitudes of the supernova in asas images and estimated @xmath38 upper limits before and after the first and last detection epochs , respectively . unlike the case of the sss images where there is contamination from the nearby star - forming region at faint magnitudes , the much shallower asas data ( @xmath39 mag )
does not require image subtraction .
the asas @xmath5 and @xmath37 magnitudes of sn 2008jb are presented in table 2 .
figure 2 shows the light curve of the supernova from asas and crts data .
we obtained late - time observations of sn 2008jb on 2010 nov .
7 using the inamori - magellan areal camera & spectrograph ( imacs ; dressler et al .
2006 ) on the 6.5 m magellan i ( baade ) telescope at las campanas observatory .
we obtained @xmath40 sec images of the field of sn 2008jb with the f/2 camera ( @xmath41 per pixel ) using the @xmath15-band filter under good weather and seeing conditions ( @xmath42 ) .
since we lack the deep template image necessary to obtain the supernova flux at late times , these data are only used to study the host galaxy environment .
we plan to obtain more late - time observations and the analysis of these data will be presented in a future publication .
we gathered spectroscopic observations of sn 2008jb on 2011 jan 6 using the imacs on the 6.5 m baade telescope at las campanas observatory .
the observations consisted of @xmath43 sec spectra obtained using the 300 l / mm grism ( range @xmath44 ) and a @xmath45 slit aligned at the parallactic angle , obtained under clear weather and good seeing ( @xmath46 ) .
this setup gives a fwhm resolution of @xmath47 at 6500 .
the data were reduced using standard techniques in iraf , which included basic data reduction ( overscan and bias subtraction , flat - fielding ) , 1d spectrum extraction , wavelength calibration using a henear arclamp obtained after the science observations , and flux calibration using a flux standard observed on the same night .
the spectrum goes through two ccd detectors separated by a gap of @xmath48 .
we used a simple linear interpolation at the edges of the ccds .
the final spectrum of sn 2008jb is shown in figure 3 .
we also obtained the spectrum of an region in the host galaxy at @xmath49 pc ( @xmath50 ) south of the supernova position on 2011 jan .
2 with the wide - field reimaging ccd camera ( wfccd ) on the 2.5-m du pont telescope at las campanas observatory .
the spectrum consisted of @xmath43 sec exposures with the 400 l / mm grism , which gives a fwhm resolution of 8 and continuous coverage in the @xmath51 range .
the data were reduced using standard tasks in iraf similar to the reduction procedure of the imacs supernova spectrum .
the final spectrum of the region is shown in figure 4 .
we measured the fluxes of the most prominent emission lines in the spectrum using gaussian profiles and a custom perl data language ( pdl ) fitting routine , including the balmer recombination lines ( h@xmath52 @xmath53 4862 and h@xmath9 @xmath53 6563 ) and forbidden emission lines ( [ ] @xmath54 4959 , 5007 , [ ] @xmath54 6713 , 6731 , and ) .
the fluxes of the optical emission lines are presented in table 3 . the host galaxy of sn 2008jb , eso 302@xmath014 ,
was observed with spitzer / irac ( fazio et al . 2004 ) using the 3.6 and 4.5 @xmath55 m channels in warm spitzer observations obtained as part of the spitzer survey of stellar structure in galaxies ( s4 g ; sheth et al .
they obtained two epochs of imaging during cycle 6 ( pid 61060 ) separated by @xmath56 month , in 2009 sep . 2 and oct . 10 .
we retrieved the post - bcd images of eso 302@xmath014 from the spitzer heritage archive , which are fully reduced and flux - calibrated .
the supernova is clearly detected as a bright , variable mid - ir source in these images , at a position consistent with the optical coordinates .
the mid - ir detections are shown in figure 5 .
we measured the 3.6 and 4.5 @xmath55 m fluxes of the source using aperture photometry with a @xmath57 aperture radius ( @xmath58 annulus for sky determination ) and applied the aperture corrections and flux conversion factors listed in the irac instrument handbook .
the irac fluxes are presented in table 4 . the wide - field infrared survey explorer ( wise ; wright et al .
2010 ) preliminary data release includes observations of the field of sn 2008jb obtained at 3.4 , 4.6 , 12 , and 22 @xmath55 m between 2010 jan .
23 and jan . 30 .
we searched for the wise data in the nasa / ipac infrared science archive ( irsa ) . the supernova is clearly detected at 3.4 and 4.6 @xmath55 m ( see figure 5 ) , marginally detected at 12 @xmath55 m , and undetected at 22 @xmath55 m .
we retrieved the photometry of source j035144.63 - 382700.6 from the wise catalog and list the fluxes in table 4 .
we did not attempt to measure photometry in individual wise images , but rely on the catalog fluxes measured from the combined data .
the effective date of the observations is 2010 jan .
we searched for other existing archival data of eso 302@xmath014 obtained before the discovery of sn 2008jb to help characterize the progenitor and galaxy environment .
the survey for ionization in neutral gas galaxies ( singg ; meurer et al .
2006 ) obtained @xmath15-band and h@xmath9 narrow - band images of the host galaxy using the ctio 1.5-m telescope on 2000 oct .
28 ( seeing was @xmath59 in @xmath15 ) .
we retrieved calibrated images from singg ( including h@xmath9-subtracted images ) through the nasa extragalactic database ( ned ) .
the galaxy evolution explorer ( galex ; martin et al .
2005 ) ultraviolet space telescope observed the field of eso 302@xmath014 on 2004 nov . 18 .
we retrieved calibrated nuv and fuv images ( gr6 data release ) from the galex online archive .
we also searched for pre - explosion imaging data in the _ hst _ , gemini , and eso archives .
unfortunately , there are no deep pre - explosion images obtained with these facilities .
in the following sections we present analysis and discussion of the results from the light curves , spectra , mid - ir emission , and host galaxy environment .
we present a summary of derived physical properties obtained from the optical light curves and host galaxy in table 5 and table 6 , respectively . in order to make light curve comparisons and derive intrinsic physical properties , we first need to estimate the total reddening along the line of sight to the supernova . for the galactic reddening
, we use a ccm reddening law with @xmath60 ( cardelli , clayton & mathis 1989 ) and @xmath61 mag from the reddening maps of schlegel et al .
we estimate the intrinsic reddening in the host galaxy using the balmer decrement measured from the spectrum of the region in eso 302@xmath014 ( see table 3 ) .
we assume an intrinsic case b recombination balmer flux ratio @xmath62 , which is appropriate for an region at a typical electron temperature and density ( storey & hummer 1995 ) .
then we assume an smc reddening law from gordon et al .
( 2003 ) for the host galaxy , which is appropriate given its star - formation rate , absolute magnitude , and metallicity ( see section 3.4 ) .
the resulting intrinsic color excess is @xmath63 mag and extinction @xmath64 mag , which we will use in all subsequent analysis .
the values obtained using an lmc reddening law , @xmath65 mag and @xmath66 mag , and galactic ccm reddening law , @xmath67 mag and @xmath68 mag , are consistent with our adopted reddening .
we note that for regions its usually assumed @xmath69 ( calzetti 2000 ) , but we adopt conservatively @xmath70 because of the young age of the region ( see section 3.4 ) .
the @xmath5- and @xmath37-band light curves of sn 2008jb resemble those of type ii - plateau supernovae , the most common kind of core - collapse events in nearby galaxies ( e.g. , li et al . 2011 ) . in figure 6
we compare the absolute @xmath5-band light curves of different core - collapse supernovae with sn 2008jb , including : the low - luminosity type iip 2005cs ( pastorello et al . 2009a ) , the luminous type iip 2004et ( maguire et al . 2010 ) , and the low - luminosity type iil 1999ga ( pastorello et al .
the absolute magnitudes of sn 2008jb lie in between sn 2005cs and sn 2004et , and appear consistent with sn 1999ga and normal type iip supernovae ( e.g. , li et al . 2011 ) . in the comparison
we have adopted the host distances , total extinctions , and explosion times presented in the published studies .
we estimate an approximate explosion time for sn 2008jb as the midpoint between the last pre - discovery @xmath5-band non - detection from crts and the first @xmath37-band detection from asas , which gives @xmath71 ( 2008 nov .
the asas data samples well the initial plateau , which lasts @xmath72 days in the @xmath37-band .
this plateau duration is in the observed range of plateaus in type iip supernovae , typically between @xmath73 days ( e.g. , bersten & hamuy 2009 ) .
the @xmath37-band light curve shows a slow linear decay of @xmath74 mag / day and the @xmath5-band light curve declines faster at @xmath75 mag / day in the initial phase , which is slightly faster than well - studied type iip supernovae where the initial decay slope in the plateau is slower at redder bands ( e.g. , poznanski et al .
2009 ; dandrea et al .
it is interesting to note the similarities of the initial light curve decline of sn 2008jb with the light curve of the low - luminosity type iil sn 1999ga ( pastorello et al .
2009b ) , as shown in figure 6 .
the @xmath5-band light curves of type iil supernovae typically have faster initial decline slopes than sn 2008jb and sn 1999ga ( e.g. , barbon et al .
1979 ; fig . 3 of pastorello et al .
2009b ) .
the absolute magnitudes close to the explosion date are @xmath76 mag and @xmath77 mag . at mid plateau ( @xmath78 days ) , the absolute magnitudes are @xmath79 mag and @xmath80 mag .
the evolution of the @xmath81 color , from @xmath82 mag close to explosion to @xmath83 mag at mid plateau , is also fairly consistent with the evolution of well - studied type iip supernovae , which traces small changes in the color temperature of the photosphere ( e.g. , hamuy et al . 2001 ) .
it is worth noting that the unreddened color at mid - plateau of sn 2008jb is very close to the @xmath84 mag ridgeline that nugent et al .
( 2006 ) derived from a large sample of type iip supernovae , supporting our reddening estimation . after the @xmath3 day plateau ,
the light curve drops @xmath4 mag in 32 days ( at @xmath85 mag / day ) to a late - time linear decay of @xmath75 mag / day .
we do not have @xmath37-band observations that show the transition phase , but the asas non - detections at these epochs are consistent with the @xmath5-band light curve shape .
the @xmath5-band drop to a late - time linear decay is seen in all type iip supernovae , but it is typically stronger ( @xmath86 mag in the @xmath5-band ; e.g. , maguire et al .
2010 ) . in figure 6
we show that the late - time linear decay of sn 2008jb is similar to other type ii supernovae and reasonably consistent with the @xmath6co to @xmath6fe radioactive decay slope .
hamuy ( 2003 ) studied the physical properties of a sample of well - observed type ii supernovae and derived the @xmath6ni masses produced in the explosions from their late - time light curves , assuming full trapping of @xmath87-ray photons produced by the radioactive decay of @xmath6co .
using eq . 1 in his study evaluated at day 115 after explosion ( the first light curve point at the radioactive tail )
we obtain a @xmath6ni mass of @xmath88 m@xmath8 .
this result assumes a bolometric correction of @xmath89 mag in the @xmath5-band , which is calculated from the well - studied sn 1987a and sn 1999em at nebular phases ( hamuy 2003 ) .
the estimated uncertainty does not include a systematic error in the distance to the host .
the late - time spectrum of sn 2008jb , obtained @xmath90 years after the explosion , shows a prominent h@xmath9 emission feature ( see figure 3 ) , supporting the classification of this transient as a type ii .
the h@xmath9 feature is broad ( @xmath91 ) , flat - topped , boxy , and blueshifted ( centered at @xmath92 km s@xmath33 ) , with the blue edge of the line at @xmath93 km s@xmath33 and red edge at @xmath94 km s@xmath33 after correcting for the host galaxy s recession velocity .
the spectrum does not show any other strong emission features characteristic of normal type ii supernovae at late times ( e.g. , [ ] and [ ] ) , except for a tentative low s / n feature centered at @xmath95 with @xmath96 km s@xmath33 . if the spectrum is heavily smoothed , there is a `` bump '' that shows up more clearly at @xmath97 , which could correspond to low - level emission associated with the [ ] doublet at @xmath54 7291 , 7323 .
there is an increase in flux to the blue of @xmath98 , which could in part be explained by contamination from nearby sources , although the s / n in the blue part of the spectrum decreases also due to flatfielding errors . the broad and boxy h@xmath9 emission feature detected in sn 2008jb
is not typically observed in late - time spectra of type ii supernovae .
we used spectra of normal type iips in the suspect database to measure the fwhm at @xmath99 days after explosion , in the nebular phase , and measure h@xmath9 line widths of @xmath100 km s@xmath33 , significantly lower velocities than sn 2008jb .
the peculiar sn 2007od ( andrews et al .
2010 ; inserra et al .
2011 ) is the only example we could find in the literature of a type iip with a comparably broad and blueshifted h@xmath9 component at late - times . in this case , however , the line also showed multiple narrower peaks on top of the broad profile that indicated clear signs of interaction between the ejecta and the circumstellar medium ( csm ) from the projenitor wind , which are not seen in sn 2008jb .
other less common type ii supernovae have shown very broad emission lines at late times .
the type iil sn 1979c in m100 ( e.g. , branch et al .
1981 ) showed a broad @xmath101 km s@xmath33 h@xmath9 profile one year after discovery ( cappellaro et al .
1995 ) . also the type iil sn 1980k showed an h@xmath9 profile with @xmath102 km s@xmath33 eight years after the explosion ( fesen & becker 1990 ) , although at earlier epochs comparable to our observations of sn 2008jb the line widths were significantly narrower ( e.g. , capellaro et al . 1995 ) .
the well - studied type iib sn 1993j showed a strong h@xmath9 emission feature at @xmath103 year after explosion , with velocity and profile shape fairly similar to sn 2008jb ( filippenko et al .
1994 ; matheson et al .
the boxy feature observed in sn 1993j was interpreted as indication of circumstellar interaction in a spherical shell ( matheson et al .
2000 ) , which supported the evidence from the radio and x - ray observations ( e.g. , fransson et al . 1996 and references therein ) .
patat et al . (
1995 ) had also shown that the luminosity of the h@xmath9 line @xmath104 year after the explosion was in excess of the expectations from radioactive @xmath6co decay , and the most likely source of extra energy was the ejecta - csm interaction .
chugai ( 1991 ) presented models of the evolution of the h@xmath9 line luminosity as a function of time for type ii supernovas assuming radioactive decay as the energy source .
we used these models to test if the measured luminosity of the h@xmath9 emission line in sn 2008jb at @xmath105 days after explosion is consistent with heating from radioactive decay or extra energy is needed to explain it .
we use the parametrization of the models presented in pastorello et al .
( 2009b ; their fig
. 9 ) , scaled to a nickel mass of @xmath106 m@xmath8 , and extrapolated the curves linearly at late - times ( in log l versus time ) .
figure 7 shows the models with different assumptions for the total ejected mass from the progenitor ( @xmath107 m@xmath8 ) and the measured h@xmath9 luminosity of sn 2008jb after correcting for galactic and internal extinction . there is a clear excess in h@xmath9 emission luminosity with respect to the models at @xmath105 days by a factor of @xmath108 , suggesting the existence of an extra energy source ( e.g. , from ejecta interaction with the progenitor wind ) or a more massive progenitor star ( @xmath109 m@xmath8 ) .
unfortunately , we do not have early spectra to trace the evolution in h@xmath9 emission luminosity with time to differentiate between these scenarios . also , the gap between ccd chips is right at the wavelength of the h@xmath9 line , so we are unable to detect a narrow component that would be a clear indication of circumstellar interaction .
we clearly detect mid - ir emission from sn 2008jb in three spitzer and wise epochs obtained between 286 and 434 days after the first optical detection ( see figure 5 ) .
the red @xmath110 - [ 4.5]$ ] spitzer color of @xmath111 mag indicates a rising spectral energy distribution at mid - ir wavelengths , a clear sign of warm dust emission . in order to characterize the evolution of the optical and mid - ir emission
, we fit the sed of the supernova using two black - bodies , a hot component to fit the optical @xmath5-band data and a warm component to fit the mid - ir data .
since we only have single - band optical data at late - times , we assume an effective temperature of @xmath112 k for the hot component , which is a typical temperature measured in normal type ii supernovae in the nebular phase .
the results of the two component black - body fits are presented in table 7 , and figure 8 shows the model fits to the sed as a function of time .
the total integrated luminosity decreases by a factor of 6.8 in @xmath113 days , which is equivalent to 0.014 mag / day .
the luminosity contributed by the warm black - body component , which is better constrained from the observed sed , decreases by a factor of 5.5 in 148 days , or @xmath75 mag / day .
the consistency with the late - time @xmath5-band decline and the @xmath6co decay slope is interesting and argues for radioactive decay as the dominant energy source of dust heating .
the temperature of the warm black - body component is @xmath11 k in the three epochs where we have mid - ir imaging .
the wise data from day 434 clearly shows that the sed peaks between 4.6 and 12 @xmath55 m , supporting our conclusion .
we also find that the radius of the warm black - body component decreases by a factor of @xmath114 between the first and last epochs with mid - ir data , between 765 au ( day 286 ) and 237 au ( day 434 ) .
the total mid - ir luminosity can be used to estimate the amount of dust needed to explain it ( e.g. , dwek et al .
1983 ; prieto et al . 2009 ) ; we find a dust mass of @xmath115 m@xmath8 assuming a black - body spectrum and carbon dust composition .
several nearby type ii supernovae have shown excess mid - ir emission at late times , which is interpreted as the presence of warm dust either newly formed or pre - existing in the progenitor csm ( or a combination ) .
some examples of nearby type iip events with clear signs of warm dust emission include sn 2003gd ( sugerman et al .
2006 ; meikle et al .
2007 ) , sn 2004et ( kotak et al . 2009 ) , sn 2004dj ( szalai et al .
2011 ; meikle et al .
2011 ) , sn 2007it ( andrews et al .
2011 ) , and sn 2007od ( andrews et al .
the classic , luminous type iil sn 1979c and sn 1980k showed infrared excesses at late times ( dwek 1983 ; dwek et al .
1983 ) . also , a large fraction of type iin supernovae show mid - ir emission at late times , which has been associated with pre - existing csm dust ( e.g. , fox et al .
2011 , and references therein ) . in the low - luminosity end
, sn 2008s - like events have dusty massive star progenitors and also show signs of dust during the transients ( e.g. , prieto et al . 2008b , 2009 ; thompson et al .
2009 ; kochanek 2011 ) .
the typical dust masses needed to explain the mid - ir emission in normal type ii supernovae are @xmath116 m@xmath8 ( but see matsuura et al .
2011 for the discovery of a large reservoir of cold dust in sn 1987a ) and fairly consistent with the range derived here for sn 2008jb .
the observations for several of the well - studied type iis have been usually explained by newly formed dust in the ejecta .
it seems unlikely that this can explain sn 2008jb .
the main reason for this is that the shock velocities inferred from the black - body fits to the mid - ir detections are in the range of @xmath117 km s@xmath33 ( depending on the epoch ) , which is quite low compared to the observed velocity of the h@xmath9 emission line . on the other hand ,
a decreasing black - body radius as a function of time has been seen in sn 2008s and ngc 300-ot ( also the type iip sn 2007it and sn 2007od ) , and has been explained by kochanek ( 2011 ) using a model in which dust reforms in the progenitor wind .
this model , however , requires very high densities and low velocities , which are not observed in sn 2008jb .
the host of sn 2008jb , eso 302@xmath014 , is a star - forming irregular galaxy similar to the magellanic clouds .
in particular , it has a total @xmath118-band absolute magnitude , @xmath119 mag fainter than the smc , with a low uv derived ( total ) current star - formation rate ( sfr ) of 0.03 m@xmath8 yr@xmath33 ( lee et al .
the total stellar mass of the galaxy is @xmath120 m@xmath8 , estimated from @xmath121 , @xmath122 , and the mass - to - light ratios as a function of @xmath123 presented in bell & de jong ( 2001 ) , assuming a salpeter imf . the total neutral hydrogen ( ) mass is @xmath124 m@xmath8 ( meurer et al .
2006 , scaled by our adopted distance ) .
these masses imply high sfrs per unit stellar mass , @xmath125 and @xmath126 , which are typical for local star - forming dwarf galaxies ( e.g. , lee et al .
2011 ) .
we use the spectrum we obtained of the region at @xmath127 pc from the position of sn 2008jb to measure the local oxygen abundance using the strong forbidden oxygen lines and hydrogen recombination lines . using the emission line fluxes reported in table 3 and correcting for host galaxy extinction , we estimated the [ ] @xmath535007/h@xmath52 and [ ] /h@xmath9 line ratios .
then we used the @xmath128 and @xmath129 calibrations from pettini & pagel ( 2004 ; pp04 ) to estimate oxygen abundances of @xmath130 ( @xmath128 ) and @xmath131 ( @xmath129 ) .
these uncertainties are statistical and do not include the @xmath132 dex error in the oxygen abundance calibration ( pp04 ) . for comparison ,
the average oxygen abundance of regions in the smc from russell & dopita ( 1990 ) are @xmath133 ( @xmath128 ) and @xmath134 ( @xmath129 ) .
the environment of sn 2008jb in eso 302@xmath014 has a low oxygen abundance compared to the environments of nearby type iip supernovae ( @xmath23 mpc ) used to constrain progenitor properties from deep the pre - explosion observations ( smartt et al .
2009 ) , although there are also normal core - collapse supernovae discovered in relatively nearby galaxies that have fairly low - metallicity environments ( e.g. , prieto et al . 2008b ; anderson et al .
the metallicity here is similar to the environments of long - duration grb hosts ( e.g. , stanek et al .
2006 ; levesque et al . 2010 ) and the hosts of the most luminous core - collapse supernovae that are being discovered in galaxy - blind surveys ( e.g. , kozlowski et al . 2010 ; stoll et al .
2011 ) , which are bound to contain a population of very massive stars . by closely examing the local host galaxy properties we can constrain the progenitor properties ( see , e.g. , anderson & james 2008 , 2009 ) .
we have relatively deep optical images obtained at late times with magellan / imacs , archival uv data from galex and h@xmath9 from singg .
figure 9 shows a mosaic with the magellan @xmath15-band and galex fuv image , scaled to fit the whole galaxy ( top panels ) and the region where the supernova exploded ( lower panels ) .
we see that sn 2008jb exploded in a large star - formation complex , the brightest and highest surface brightness star - forming region within the galaxy in the optical and fuv .
it is composed of at least two resolved clusters or stellar associations that are well separated in the magellan @xmath15-band image .
interestingly , the h@xmath9 emission ( red contours in the lower panel of fig .
9 ) is offset and outside the brightest optical and fuv emission , forming an apparent ring with a projected diameter of @xmath135 pc .
we can estimate an approximate age for the star - forming complex in eso 302@xmath014 using the ratio of h@xmath9 to fuv luminosities , which is a sensitive age indicator in an instantaneous burst of star formation ( e.g. , stewart et al .
2000 ; snchez - gil et al .
we use the starburst99 ( leitherer et al .
1999 ) models to generate a grid of single - age clusters with total masses between @xmath136 m@xmath8 and @xmath137 m@xmath8 , in steps of @xmath138 m@xmath8 .
we choose a salpeter imf with stellar masses between 0.1 and 100 m@xmath8 and geneva stellar evolution models with @xmath139 , consistent with the measured oxygen abundance of the region .
figure 10 shows the expected ratio of h@xmath9-to - fuv luminosities as a function of the fuv luminosities for these models .
we label the ages of the clusters between 1 myr and 50 myr .
the total measured h@xmath9-to - fuv ratio @xmath140 and fuv luminosity @xmath141 erg s@xmath33 hz@xmath33 of the star - forming complex , corrected by galactic and host galaxy extinction , are shown in the figure .
we obtain an age of @xmath142 myr and total mass of @xmath143 m@xmath8 .
we also used the integrated @xmath15-band flux ( extinction corrected ) of the star - forming region to independently check this mass estimate .
we derive a total mass of @xmath144 m@xmath8 from the @xmath15-band flux and the mass - to - light ratio obtained from the starburst99 models , which is fully consistent with the fuv estimate .
if the progenitor star of sn 2008jb was formed in this star - formation episode 9 myr ago , then we infer an initial main - sequence mass of @xmath145 m@xmath8 for the progenitor from the geneva stellar evolution models with extended mass - loss used in starburst99 ( lejeune & schaerer 2001 ) .
this estimate is obtained as the maximum main - sequence mass of a star from an isochrone of 9 myr of age .
we note that this estimate is based only in the total h@xmath9 and fuv fluxes of the region , and a more detailed analysis of the seds of the individual stellar clusters is in preparation . the equivalent width ( ew ) of the h@xmath9 ( @xmath146 ) and h@xmath52 ( @xmath147 ) emission lines in the spectrum of the region
can also be used to estimate an age for the star - forming region using the starburst99 models ( e.g. , stasiska & leitherer 1996 ; schaerer & vacca 1998 ; leloudas et al . 2011 ) .
essentially , the ew of the recombination lines gives an estimate of the young over the old stellar population ( see leitherer 2005 , and references therein ) .
we obtain an age of @xmath148 myr from h@xmath9 ew and @xmath149 myr from h@xmath52 ew .
these estimates are @xmath150 myr ( @xmath151 using the statistical uncertainty ) younger than the age derived from the h@xmath9-to - fuv ratio of the whole star - forming region .
the differences might imply multiple star - formation episodes within the star - forming region , however , this would need to be tested with a detailed study of the stellar populations within the region .
we can estimate the sfr of the star - forming region using the fuv luminosity with the calibrations of lee et al .
( 2009 ) , finding @xmath152 m@xmath8 yr@xmath33 , or about 30% of the total sfr of the dwarf galaxy .
we note that the h@xmath9-estimated sfr is a factor of @xmath153 lower than the fuv value , a systematic difference seen in dwarf galaxies at low sfrs which lacks a satisfactory explanation ( e.g. , lee et al .
2009 ; see also fumagalli et al .
we have presented the discovery , follow - up observations , and analysis of sn 2008jb , a bright type ii supernova in the metal - poor , southern dwarf irregular galaxy eso 302@xmath014 at @xmath30 mpc .
this @xmath31 mag supernova was found in archival data obtained by the crts and asas all - sky surveys .
this transient was missed by galaxy - targeted supernova surveys like chase and by amateur astronomers mainly because the host galaxy is a low - luminosity dwarf , @xmath119 mag fainter than the smc , and targeted surveys use catalogs that are incomplete for small galaxies ( e.g. , leaman et al .
2011 ) .
sn 2008jb has @xmath5- and @xmath37-band light curves similar to normal type iip supernovae , with peak magnitude @xmath154 mag , but it can also be classified as an intermediate case between type iip and type iil due to its faster initial decay , perhaps similar to sn 1999ga ( pastorello et al .
it shows a @xmath155 day plateau and a fast @xmath4 mag decline to a late - time decline slope of @xmath75 mag / day in the @xmath5-band .
this decline is consistent with the radioactive decay of @xmath6co to @xmath6fe , and argues for @xmath88 m@xmath8 of @xmath6ni synthesized in the explosion that is powering the lightcurve .
we detect mid - ir emission from sn 2008jb @xmath10 months after explosion in three epochs of archival spitzer and wise data .
the mid - ir emission has an sed with black - body temperature of @xmath11 k , characteristic of the warm dust emission seen in some nearby core - collapse supernovae and luminous transients ( e.g. , kotak et al .
2009 ; prieto et al .
the evolution of the mid - ir emission with time is consistent with the products of radioactive decay heating the dust .
the characteristic mid - ir dust radius shrinks with time , an evolution that is not typically seen in normal type iis ( some exceptions are sn 2007it and sn 2007od ; andrews et al .
2010 , 2011 ) .
we obtained a spectrum of sn 2008jb about 2 years after the explosion .
it displays a very broad ( @xmath156 km s@xmath33 ) , boxy , and flat - topped h@xmath9 emission line , leading to its type ii supernova spectroscopic classification .
the broad and boxy line profile seen in h@xmath9 is quite unusual for normal type iip supernovae at late times , but has been seen in some objects like sn 1993j , sn 2007od , and also a few well - studied type iil supernovae .
we find that the h@xmath9 line luminosity is in excess of the expected luminosity from radioactive @xmath6co decay predicted by the models of chugai ( 1991 ) for total ejected masses @xmath157 m@xmath8 .
this indicates that there is an external source of energy , like ejecta - csm interaction , and/or the mass of the progenitor star is
@xmath158 m@xmath8 .
we do not see clear signs of ejecta - csm interaction like narrow lines or irregularities in the h@xmath9 profile .
it would be interesting to obtain late time x - ray and radio observations in order to have independent constraints on the importance of ejecta - csm interaction in sn 2008jb .
we studied the host galaxy environment of sn 2008jb in eso 302@xmath014 with optical spectra and imaging . using the spectrum of an region at @xmath49 pc from the supernova site we measure an oxygen abundance of @xmath159 ( pp04 @xmath129 method ) and @xmath160 ( pp04 @xmath128 method ) from the strong nebular emission lines , which is similar to the smc and one of the lowest measured metallicities of local core - collapse supernovae environments ( e.g. , in the lower 3% of measured oxygen abundances of the core - collapse sample of anderson et al . 2010 ) .
the supernova exploded in a large star - forming complex with strong optical and galex fuv emission which is surrounded by a large h@xmath9 ring with @xmath161 pc .
the h@xmath9-to - fuv ratio of this region is consistent with a stellar population with an age of @xmath17 myr derived from starburst99 modeling and single - age stellar population models .
this age implies a supernova progenitor mass of @xmath145 m@xmath8 , assuming a single star ( but see , e.g. , smith et al .
2011a for the importance of binary progenitors ) , if the progenitor formed in this star - forming episode .
the equivalent width of the h@xmath9 and h@xmath52 emission lines in the spectrum of the region give another constraint on the age of the region of @xmath162 myr , @xmath150 myr younger than the estimate from h@xmath9-to - fuv ratio .
we note that the star - formation history could be more complicated than a single burst , and we plan to study the region in detail using multiwavelength data in a future study .
in particular , it would be interesting to include high - resolution data from hst .
large expanding h@xmath9 shells ( supershells with radii @xmath163 pc ) have been observed and studied in many nearby star - forming dwarf galaxies and have typical dynamical ages of @xmath30 myr ( e.g. , martin 1998 ) , which is fairly consistent with the age we derive here from the h@xmath9 and fuv emission .
these structures are produced by the combined effect of many supernova explosions and winds from massive stars ( e.g. , chakraborti & ray 2011 ) , and are the likely precursors of galactic winds . in a sense , we may perhaps be witnessing supernova feedback in real - time in this star - forming region .
the bias to large star - forming galaxies is clearly present in the samples of nearby core - collapse supernovae used for progenitor studies ( e.g. , smartt et al .
2009 ) , and has been discussed in detail as a possible explanation for the discrepancy between measured local supernova rates and predicted supernova rates from galaxy star - formation rates ( e.g. , horiuchi et al .
2011 ) . since the environments of nearby core - collapse supernovae used for progenitor studies generally miss dwarf galaxies because they were not included in the original searches , the progenitor properties and conclusions drawn from these samples regarding stellar evolution are not complete . in particular , the dearth of high - mass ( @xmath164
m@xmath8 ) progenitor stars of type iip supernovae ( red supergiant problem " ) could be alleviated if these progenitor stars prefer lower - metallicity environments .
for example , this could be caused by evironmental variations in the stellar imf ( e.g. , meurer et al .
2009 ) or by changes in the fraction of type ii spectroscopic subtypes as a function of metallicity due to stellar evolution ( e.g. , arcavi et al .
indeed , we find that the properties of the spectrum of sn 2008jb are more consistent with a massive progenitor ( but see discussion about supernova modeling in , e.g. , smartt et al . 2009 ; bersten & hamuy 2009 ; bersten et al .
2011 ) , and the star - forming region where it was found has a young age compared with the ages of detected type iip progenitors ( @xmath165 myr ) .
interestingly , a strong preference for low - metallicity hosts is observed in long grbs ( e.g. , stanek et al .
2006 ) and luminous core - collapse supernovae ( e.g. , neill et al . 2011 ; stoll et al . 2011 ) , which have been linked with massive star progenitors ( @xmath166 m@xmath8 ) .
sn 2008jb offers the unique chance of studying in detail a nearby type ii supernova with host properties similar to long grbs and the most luminous core - collapse supernovae .
the mapping between different classes of massive stars and their supernovae is not yet fully understood .
special insights are expected to be obtained when unusual explosions can be connected to unusual progenitor stars and galaxy hosts .
nearby objects are especially useful in terms of larger fluxes for an extended time after explosion , better spatial resolution for progenitor studies , and improved prospects for detection by new messengers like gamma rays ( e.g. , timmes & woosley 1997 ; horiuchi & beacom 2010 ) , neutrinos ( e.g. , ando & beacom 2005 ; kistler et al . 2011 ) , and gravitational waves ( e.g. , ott 2009 ) . in addition , data from these objects are needed for a comprehensive understanding of the nearby universe .
it is difficult to find the nearest supernovae in small host galaxies with searches that target individual ( generally large ) galaxies , like loss , chase , and also amateur efforts . and
it is difficult to find them with volume - based searches such as ptf ( rau et al .
2009 ) and pan - starrs ( kaiser et al . 2002 ) that have a deep , but relatively small survey area with good cadence .
a shallower all - sky survey with excellent cadence , like asas , will help us find nearby supernovae in all kinds of environments , including low - metallicity dwarf galaxies like the host of sn 2008jb ( see also khan et al . 2011 , stoll et al .
2011 , for other supernovae studied with asas ) . upgrades that will significantly increase the sensitivity and response speed of asas are underway .
we thank john mulchaey for obtaining one of the magellan images presented in this work , chris kochanek for detailed comments , rupali chandar , crystal martin , and linda watson for discussions , and chris burns for providing his image subtraction code .
we also thank the anonymous referee for a careful reading of the manuscript .
we are indebted to the staff of las campanas observatory for their assistance .
jlp acknowledges support from nasa through hubble fellowship grant hf-51261.01-a awarded by stsci , which is operated by aura , inc . for nasa , under contract nas 5 - 2655 .
jfb is supported by the national science foundation career grant phy-0547102 .
gp is supported by the polish mnisw grant n203 007 31/1328 .
ksz and dms are supported in part by nsf grant ast-0908816 . support for mc
is provided by the ministry for the economy , development , and tourism s programa iniciativa cientfica milenio through grant p07 - 021-f , awarded to the milky way millenium nucleus ; by proyecto basal pfb-06/2007 ; by fondap centro de astrofsica 15010003 ; and by proyecto fondecyt regular # 1110326 .
the crts is supported in part by the nsf grant ast-0909182 .
this research has made extensive use of the nasa / ipac extragalactic database ( ned ) which is operated by the jpl , caltech , under contract with nasa .
lcc 4774.21 & @xmath167 & + 4794.12 & 13.58 & 0.08 + 4804.14 & 13.86 & 0.08 + 4835.09 & 14.26 & 0.08 + 4864.96 & 14.61 & 0.08 + 4896.96 & 16.03 & 0.08 + 5074.28 & 18.36 & 0.09 + 5114.17 & 18.81 & 0.09 + 5147.21 & 19.53 & 0.11 + 5163.15 & 19.06 & 0.11 + 5180.08 & 19.81 & 0.12 + 5208.05 & 20.23 & 0.16 + 5238.93 & @xmath168 & + lcc + 4785.66 & @xmath169 & + 4794.68 & 13.80 & 0.25 + 4797.69 & 13.80 & 0.28 + 4802.64 & 13.65 & 0.21 + 4804.73 & 14.13 & 0.17 + 4807.59 & 13.49 & 0.16 + 4809.65 & 13.89 & 0.30 + 4817.64 & 13.88 & 0.24 + 4822.77 & 14.02 & 0.35 + 4825.65 & 13.85 & 0.34 + 4846.61 & 14.35 & 0.24 + 4850.73 & 14.16 & 0.37 + 4853.64 & 14.29 & 0.33 + 4868.60 & 14.83 & 0.28 + + 4787.65 & @xmath170 & + 4789.70 & 13.89 & 0.20 + 4793.70 & 13.61 & 0.15 + 4795.70 & 13.17 & 0.15 + 4797.71 & 13.20 & 0.16 + 4799.72 & 13.45 & 0.15 + 4801.68 & 13.32 & 0.15 + 4803.65 & 13.38 & 0.16 + 4805.65 & 13.50 & 0.16 + 4807.64 & 13.72 & 0.15 + 4809.65 & 13.60 & 0.15 + 4811.68 & 13.88 & 0.16 + 4815.67 & 13.39 & 0.16 + 4817.67 & 13.51 & 0.15 + 4819.67 & 13.65 & 0.15 + 4821.66 & 13.47 & 0.15 + 4823.66 & 13.54 & 0.15 + 4825.66 & 13.65 & 0.16 + 4827.67 & 13.46 & 0.15 + 4829.68 & 13.42 & 0.16 + 4833.64 & 13.64 & 0.15 + 4835.62 & 13.63 & 0.16 + 4837.62 & 13.66 & 0.15 + 4841.64 & 13.46 & 0.15 + 4845.64 & 13.61 & 0.15 + 4849.63 & 13.78 & 0.15 + 4855.63 & 13.69 & 0.16 + 4859.62 & 13.52 & 0.15 + 4861.62 & 13.76 & 0.16 + 4865.61 & 14.00 & 0.15 + 4867.60 & 13.94 & 0.17 + 4871.61 & 13.81 & 0.15 + 4875.61 & 14.10 & 0.16 + 4879.60 & 13.83 & 0.16 + 4883.59 & 14.12 & 0.15 + 4885.60 & @xmath171 & + lcl h@xmath52 & @xmath172 + @xmath173@xmath174 $ ] 4959 & @xmath175 + @xmath173@xmath174 $ ] 5007 & @xmath176 + h@xmath9 & @xmath177 + @xmath173@xmath174 $ ] 6583 & @xmath178 + @xmath173@xmath174 $ ] 6713 & @xmath179 + @xmath173@xmath174 $ ] 6731 & @xmath179 + lcccc
2455076.12 & 3.6 @xmath55 m & @xmath180 & @xmath181 & spitzer / irac + 2455076.12 & 4.5 @xmath55 m & @xmath182 & @xmath183 & spitzer / irac + 2455113.71 & 3.6 @xmath55 m & @xmath184 & @xmath185 & spitzer / irac + 2455113.71 & 4.5 @xmath55 m & @xmath186 & @xmath187 & spitzer / irac + 2455223.63 & 3.4 @xmath55 m & @xmath188 & @xmath189 & wise + 2455223.63 & 4.6 @xmath55 m & @xmath190 & @xmath191 & wise + 2455223.63 & 12 @xmath55 m & @xmath192 & @xmath193 & wise + 2455223.63 & 22 @xmath55 m & @xmath194 & @xmath195 & wise + lcl sn name & sn 2008jb & cbet 2771 + ra ( j2000 ) & 03@xmath19651@xmath19744@xmath19866 & + dec ( j2000 ) & @xmath199@xmath20027@xmath20100@xmath2021 & + spectroscopic type & ii & broad h@xmath9 in spectrum + @xmath203 & 0.009 mag & schlegel et al .
( 1998 ) + @xmath204 & @xmath205 mag & from balmer decrement + @xmath206 & 2454782.0 & explosion time + @xmath207 & @xmath208 mag & @xmath5 and @xmath37 at maximum + @xmath209 & @xmath210 mag & absolute @xmath5 mag at maximum + @xmath211 & @xmath212 mag & unreddened color at maximum + @xmath213 & @xmath214 mag & @xmath5 and @xmath37 at mid plateau + @xmath215 & @xmath216 mag & absolute @xmath5 mag at mid plateau + @xmath217 & @xmath218 mag & unreddened color at mid plateau + duration of `` plateau '' & @xmath155 days & from @xmath37 + linear decline slope ( @xmath219 days ) & 0.013/0.007 mag day@xmath33 & @xmath5 and @xmath37 + linear decline slope ( @xmath220 days ) & 0.013 mag day@xmath33 & @xmath5 + lcl name & eso 302@xmath014 & + ra ( j2000 ) & 03@xmath19651@xmath19740@xmath1988 & paturel et al .
( 2003 ) + dec ( j2000 ) & @xmath199@xmath20027@xmath20112@xmath2024 & paturel et al .
( 2003 ) + morphological type & ib(s)m & rc3 + heliocentric velocity & @xmath221 km s@xmath33 & koribalski et al .
( 2004 ) + distance modulus & 29.91 mag & using @xmath222 mpc + @xmath223 & @xmath224 mag & lauberts & valentijn ( 1989 ) + @xmath225 & @xmath226 mag & lauberts & valentijn ( 1989 ) + @xmath227 & @xmath228 mag & absolute @xmath118 mag + @xmath229 & @xmath230 mag & unreddened color + oxygen abundance 1 & @xmath231 & pp04 o3n2 method + oxygen abundance 2 & @xmath232 & pp04 n2 method + star formation rate & 0.03 @xmath233 yr@xmath33 & lee et al .
( 2009 ) + stellar mass & @xmath234 m@xmath8 & m / l from bell & de jong ( 2001 ) + mass & @xmath235 m@xmath8 & meurer et al . | we present the discovery and follow - up observations of sn 2008jb , a core - collapse supernova in the southern dwarf irregular galaxy eso 302@xmath014 at .
this nearby transient was missed by galaxy - targeted surveys and was only found in archival optical images obtained by the catalina real - time transient survey and the all - sky automated survey .
the well sampled archival photometry shows that sn 2008jb was detected shortly after explosion and reached a bright optical maximum , @xmath1 mag ( @xmath2 ) .
the shape of the light curve shows a plateau of @xmath3 days , followed by a drop of @xmath4 mag in @xmath5-band to a slow decline with the approximate @xmath6co decay slope .
the late - time light curve is consistent with @xmath7 m@xmath8 of @xmath6ni synthesized in the explosion .
a spectrum of the supernova obtained 2 years after explosion shows a broad , boxy h@xmath9 emission line , which is unusual for normal type iip supernovae at late times .
we detect the supernova in archival spitzer and wise images obtained @xmath10 months after explosion , which show clear signs of warm ( @xmath11 k ) dust emission .
the dwarf irregular host galaxy , eso 302@xmath014 , has a low gas - phase oxygen abundance , @xmath12 ( @xmath13 z@xmath8 ) , similar to those of the smc and the hosts of long gamma - ray bursts and luminous core - collapse supernovae .
this metallicity is one of the lowest among local ( @xmath14 mpc ) supernova hosts .
we study the host environment using galex far - uv , @xmath15-band , and h@xmath9 images and find that the supernova occurred in a large star - formation complex .
the morphology of the h@xmath9 emission appears as a large shell ( @xmath16 pc ) surrounding the fuv and optical emission .
using the h@xmath9-to - fuv ratio , and fuv and @xmath15-band luminosities , we estimate an age of @xmath17 myr and a total mass of @xmath18 m@xmath8 for the star - formation complex , assuming a single - age starburst .
these properties are consistent with the expanding h@xmath9 supershells observed in many well - studied nearby dwarf galaxies , which are tell - tale signs of feedback from the cumulative effect of massive star winds and supernovae .
the age estimated for the star - forming region where sn 2008jb exploded suggests a relatively high - mass progenitor star with initial mass @xmath19 m@xmath8 , and warrants further study .
we discuss the implications of these findings in the study of core - collapse supernova progenitors . |
the objective of this research is a quantitative , first - principles description of the energy deposition by a bare ion in diamond .
the energy loss per unit path length of a massive , punctiform , charged particle with charge number @xmath2 to a target is@xcite @xmath3 with @xmath4 the projectile velocity in the target rest frame , @xmath5 the elementary charge unit , and @xmath6 the vacuum permittivity . for reciprocal lattice vectors
@xmath7 , @xmath8 is the @xmath9 ( `` head '' ) component of the inverse dielectric matrix defined by @xmath10 with respect to @xmath11 , the wave - vector and frequency dependent microscopic dielectric matrix . in terms of the irreducible polarizability @xmath12 ,
the dielectric matrix is @xmath13 and an integral over the first brillouin zone ( bz ) @xmath14 @xmath15 are the band energies , and @xmath16 are fermi occupation numbers ( @xmath17 or 2 for @xmath15 above or below the fermi energy @xmath18 ) .
the matrix elements in eq .
( [ piint ] ) are @xmath19 we determine the ground - state electronic structure of solids within density functional theory ( dft ) as established in the kohn - sham ( ks ) variational procedure and implemented in the computational package gtoff.@xcite the results of the all - electron , full - potential calculations are eigenfunctions @xmath20 , expressed as linear combination of gaussian type orbitals ( gto s ) , and eigenvalues @xmath15 .
the bloch functions are finally expanded in a ( truncated ) plane wave ( pw ) series , @xmath21 to represent @xmath22 as a simple sum over products of expansion coefficients @xmath23,@xcite where @xmath24 is the volume of a unit cell ( uc ) .
the equivalence of the gto and pw representations is maintained by monitoring the accumulated norm for each @xmath25 relative to the exact values , @xmath26 we subdivide the integration region of the integral ( [ piint ] ) into @xmath27-space tetrahedra .
recursive further subdivision of a given tetrahedron into smaller tetrahedra is done if @xmath28 is not constant over all four vertices .
next comes linearization of the product of the matrix elements in the numerator and of the energy denominator inside each tetrahedron for each @xmath29 .
the resulting approximated integral is evaluated analytically.@xcite the product @xmath30 is calculated and , in compensation , the term @xmath31 in the denominator of eq .
( [ dedx ] ) is dropped .
the dielectric function is tabulated for @xmath32 commensurate with the uniform mesh of wave vectors used in the underlying gtoff calculation but covering higher bz s as well as the first .
[ values outside the @xmath33-meshes are replaced by the vacuum response , @xmath34 . ]
@xmath35 is linearized inside each @xmath32-space tetrahedron .
multiplied by the linear factor @xmath36 , the integrals ( [ dedx ] ) over tetrahedra are done analytically , then summed .
diamond is studied at the experimental lattice parameter , @xmath37 for the cubic unit cell , where @xmath38 denotes one bohr .
the moruzzi - janak - williams parameterization@xcite of the hedin - lundqvist local - density approximation ( lda ) to the exchange - correlation potential is used .
small , highly contracted basis sets as used in ref .
are generally insufficient to calculate the real parts of dielectric functions and subsequently the energy loss functions .
we started from partridge s @xmath39 set,@xcite contracted the seven tightest @xmath40-functions and the four tightest @xmath41-functions , removed the most diffuse @xmath40 and the two most diffuse @xmath41-functions to avoid approximate linear dependencies , and added a full set of three @xmath42-functions with exponents equal to the remaining most diffuse @xmath41-functions .
( without @xmath42-orbitals the total energy would rise by 0.24 ev / atom . )
site centered @xmath40- and @xmath43-type fitting functions were used with exponents for the @xmath40-types as in ref . , and for the @xmath43-types @xmath44 and @xmath45 as in ref . .
space group and lattice type are those of silicon ; hence we may refer to a prior gtoff study@xcite for the symmetry properties of the fitting functions . the density of states ( dos ) computed with this @xmath46 basis ( 84 basis functions in the primitive uc ) as shown in fig .
[ fig1.ps ] is stable up to @xmath47 ev above the fermi energy with respect to further de - contraction .
band gaps are too small compared with experiments , as usual for the lda and known from other dft calculations on diamond.@xcite the lowest 24 bands were expanded into 531 plane waves , a cut - off at @xmath48 , with a norm in eq .
( [ sumpw ] ) above @xmath49 for the @xmath50-shell electrons ( excitations from which were excluded in the subsequent calculations of @xmath51 and @xmath52 anyway ) , and a norm above @xmath53 for the remaining 22 bands .
a first result is the all - electron momentum density ( emd ) @xmath54 are the states with long wavelengths and reveal the macroscopic symmetry of the crystal system , the 4-fold axis of the cubic system here . if @xmath55 is of the order of half a reciprocal lattice vector , the interference with the next nearest neighbors in the lattice becomes visible ; the eight foothills in the figure may be interpreted as a projection of the four corners of a carbon tetrahedron onto the @xmath56 plane complemented by the inversion@xcite @xmath57 if @xmath55 is large , the spherical symmetry of the core states prevails .
the element @xmath58 of the dielectric matrix is shown in fig .
[ fig3.ps ] .
the main absorption peak at 11 ev for @xmath59 corresponds to direct transitions from the top of the valence to the bottom of the conduction band at x and l.@xcite local field effects are illustrated in fig .
[ fig4.ps ] .
@xmath60 repeats some values of fig .
[ fig3.ps ] ; @xmath61 $ ] includes an estimate of the local field by calculating @xmath62 as the inverse of a @xmath63 dielectric matrix which contains @xmath64 and the eight vectors of the closest shell in the bcc reciprocal lattice .
the reduction of the values without lfe ( @xmath65 , open symbols ) compared to those with lfe ( @xmath66 , filled symbols ) is of the order reported by van vechten and martin@xcite ( without their `` dynamical correlations '' ) .
the different sign of the effect for frequencies above and below the peak has been noticed before.@xcite the differences are even smaller for the energy loss function .
hence the energy loss in the next paragraph was calculated from @xmath58 alone . to integrate the energy loss function as described by eq .
( [ dedx ] ) , the bz mesh was reduced to @xmath67 points , i.e. , 35 points in the irreducible brillouin zone ( ibz ) .
the result is shown in fig .
[ fig5.ps ] .
the dielectric matrix was tabulated on a @xmath68 mesh in @xmath32-space ( parallelepiped with three edges of length @xmath69 ) , and the stopping power was integrated over the superset of all @xmath32-values obtained from these via point group operations with seitz symbol @xmath70 , @xmath71 it is about 015% lower than experimental results@xcite for @xmath72 which are shown in fig .
[ fig6.ps ] .
( @xmath73 m / s is one atomic unit of velocity .
values within the lpda represent intgrals of the all - electron density of gtoff fitting functions weighted with the stopping number of the feg in the rpa . ) reduction of the integrated @xmath32-region to the parallelepiped decreases the stopping cross section by approximately 10% at @xmath74.@xcite within the framework of linear dielectric response an underestimation is appropriate though , because terms of @xmath75 will add about 15% to @xmath76 at @xmath77.@xcite a new result is the `` ionic '' band gap , the non - linear suppression of @xmath76 for @xmath78 , which is approximately the value extracted from@xcite @xmath79 for a band gap of @xmath80 ev .
( @xmath81 is one rydberg . ) by way of contrast , calculations within the local - plasma - density approximation usually integrate over volume elements in real space that are parameterized by the homogeneous electron gas , and inevitably yield @xmath82 at low velocities .
however , an attempt at experimental verification of this reduction of the energy loss in a wide - gap material is handicapped by the additional nuclear energy loss , which has an estimated maximum of @xmath83 ev@xmath84atoms at @xmath85.@xcite note that eq .
( [ vgap ] ) predicts contributions from @xmath50-shell excitations to start at @xmath86 , which are not included here .
they have been estimated to shift the maximum to higher velocities by about @xmath87,@xcite and to grow until @xmath88.@xcite for a thorough discussion of the underlying approximations ( linear response , neglect of channeling and local field factors etc . ) and the implementation see r.j .
mathar , s.b .
trickey , and j.r .
sabin , nucl .
instrum .
methods b * 155 * , 249 ( 1999 ) ; 17th werner brandt workshop on charged particle penetration phenomena , may 89 , 1997 , charlottesville , va ( unpublished ) .
adler , phys .
rev . * 126 * , 413 ( 1962 ) .
( 2.11)(2.12 ) are inconsistent with eqs .
( 2.7)(2.10 ) .
either fields @xmath89 must be replaced by charge densities @xmath90 in ( 2.11)(2.12 ) , or all factors of squared wavenumbers be dropped . because no bond - centered basis functions are used
, the charge fitting functions do not reproduce the quasi - forbidden ( 222 ) x - ray reflex : z.w .
lu , a. zunger , and m. deutsch , , 9385 ( 1993 ) ; m.a .
spackman , acta crystallogr .
a * 47 * , 420 ( 1991 ) ; t. takama , k. tsuchiya , k. kobayashi , and s. sato , _ ibid . _ * 46 * , 514 ( 1990 ) .
s. fahy , x.w .
wang , and s.g .
louie , , 1631 ( 1988 ) ; m.r .
salehpour and s. satpathy , , 3048 ( 1990 ) ; a. onodera , m. hasegawa , k. furuno , m. kobayashi , y. nisida , h. sumiya , and s. yazu , _ ibid . _ * 44 * , 12176 ( 1991 ) ; m.p .
surh , s.g .
louie , and m.l .
cohen , _ ibid . _
* 45 * , 8239 ( 1992 ) ; s. logothetidis , j. petalas , h.m .
polatoglou , and d. fuchs , _ ibid . _ * 46 * , 4483 ( 1992 ) .
w. kferbck , w. rssler , v. necas , p. bauer , m. pealba , e. zarate , and a. arnau , , 13275 ( 1997 ) .
note that their theoretical values are already multiplied with charge state fractions ; the h@xmath91 curve in their fig . 1 is not the analogue of our results .
the experimental value is @xmath92 ev : j. nithianandam and j.c .
rife , , 3517 ( 1993 ) ; y. ma , n. wassdahl , p. skytt , j. guo , j. nordgren , p. d. johnson , j .- e .
rubensson , t. boske , w. eberhardt , and s.d .
kevan , , 2598 ( 1992 ) .
using @xmath93 @xmath27-points in the bz ( 213 in the ibz ) .
the direct gap at @xmath94 is 5.59 ev , the indirect gap 4.13 ev , and the width of the four valence bands with eight electrons 21.29 ev .
the two bands with four core electrons at @xmath95 ev are not shown.@xcite ] for @xmath99 and @xmath100 .
the underlying dft calculation is based on a @xmath101 mesh of @xmath27-points in the bz ( 98 points in the ibz ) , which creates for this particular @xmath32-direction commensurate values at @xmath102 as shown here for @xmath103 . ] ( open symbols ) and of @xmath104 ( filled symbols ) .
@xmath32 is parallel to @xmath105 with @xmath106 ( @xmath107 , squares ) and @xmath108 ( @xmath109 , circles ) as in fig .
[ fig3.ps ] .
note that refs . and refer to the optical limit @xmath110 . ]
-axis ( 0 103 ev ) .
the lowest 28 bands were included in the sum over band - pairs in eq .
( [ piint ] ) .
the line refers to the free electron gas @xcite with a density equivalent to 4 electrons per diamond atom . ] | we calculate the `` head '' element or the @xmath0-element of the wave - vector and frequency - dependent dielectric matrix of bulk crystals via first - principles , all - electron kohn - sham states in the integral of the irreducible polarizability in the random phase approximation .
we approximate the macroscopic `` head '' element of the inverse matrix by its reciprocal value , and integrate over frequencies and momenta to obtain the electronic energy loss of protons at low velocities . numerical evaluation for diamond targets
predicts that the band gap causes a strong non - linear reduction of the electronic stopping power at ion velocities below @xmath1 atomic units . |
the scale - dependence of mass - to - light @xmath0 ratio is always an important topic of cosmology .
the @xmath0 ratio on scales of galaxies , double galaxies , groups of galaxies , and galaxy clusters provided the first approach of evaluating the mean mass density of the universe ( e.g. faber & gallagher 1979 ) . a most striking result in this study
is that @xmath0 ratio increases with the scale of the system considered ( e.g. rubin 1993 ) .
this has inspired the searching for dark matters in objects on scales larger than clusters . yet , the @xmath0 ratio appears to be flat on the scales of rich clusters and larger . this result has been employed to propose that the @xmath0 ratio on scales larger than clusters asymptotically reaches a constant , which is approximately equal to the median @xmath0 ratio of rich clusters , i.e. @xmath3 @xmath4 @xmath5 ( bahcall , lubin & dorman 1995 ) .
this asymptotic @xmath0 ratio yields @xmath6 , suggesting a low density universe ( bahcall & fan 1998 ) .
that is , the behavior of @xmath0 ratio on large scales is directly related to the cosmological parameter @xmath7 .
the evidence for a flat @xmath0 on large scale is still weak .
the observed virgo cluster infall motion seems to be the only evidence of the flatness on the scale of the local supercluster ( @xmath8 10 @xmath2 mpc ) ( e.g. yahil , sandage & tammann , 1980 ) .
however , a direct measure of the mass - to - light ratio of the corona borealis supercluster gives @xmath9 .
namely , the @xmath0 ratio on scale @xmath8 20 @xmath2 mpc is higher than that of rich clusters by a factor of 2 ( small et al . 1998 ) . on the other hand ,
galaxy clusters with abell radius generally are dominated by elliptical galaxies at their centers , while larger structures have a higher fraction of spiral galaxies .
therefore , it is uncertain if the morphological segregation leads to a variation of the @xmath0 ratio of structures larger than clusters . in this _
, we try to approach this problem by analyzing the observed scaling of clustering on scales beyond clusters .
we have found recently that the richness function ( rf ) and the mass function ( mf ) of objects identified by multi - resolution analysis are scale invariant from the abell radius to about 20 @xmath2 mpc ( xu , fang and wu 1998 , hereafter xfw ) .
we will show that the scale dependence of @xmath0 ratio is completely determined by the scaling factors of rf and mf of these multi - scale identified objects . if the mass field traced by the observed galaxies can match with popular models of cdm cosmogony
, one can calculate the @xmath0 s scale dependence by comparing the observed rf scaling with the model - predicted mf scaling . using the data of the iras 1.2 jy galaxy survey
, we find that the @xmath0 ratio is approximately flat on scales up to , at least , 24 @xmath2 mpc .
in 2 we calculate the mass functions of multi - scale identified clusters from simulation and calculate their scaling indices . in 3 we derive the richness functions and their scaling indices of multi - scale identified clusters from iras 1.2 jy galaxy survey . in 4 we study the scale dependence of mass - to - light ratio by matching the scaling behaviors of theoretical mass functions with those of observational richness functions .
5 is our conclusion .
the model samples of mass distribution used in this paper are similar to xfw .
they are generated by n - body simulations with the p@xmath10 m code developed by y.p .
jing ( jing & fang 1994 ) .
three popular models of the cdm cosmogony , i.e. the standard cdm ( scdm ) , the flat low - density cdm ( lcdm ) models and the open cdm model ( ocdm ) , are employed .
the hubble constant @xmath4 ( @xmath11 km / sec / mpc ) ) , mass density @xmath12 , cosmological constant @xmath13 and the normalization of the power spectrum of the density perturbation @xmath14 are chosen to be ( 0.5 , 1.0 , 0.0 , 0.62 ) for scdm , ( 0.75 , 0.3 , 0.7 , 1.0 ) for lcdm , and ( 0.75 , 0.3 , 0.0 , 1.0 ) for ocdm . the simulation parameters are : 1 . )
box size @xmath15 mpc ; 2 . ) number of simulation particles @xmath16 ; and 3 . ) the effective force resolution @xmath17 mpc . in this case , each particle has a mass of @xmath18 , which is small enough to resolve reliably objects with mass @xmath19.we produce 8 realizations for each model . for the simulation results ,
the clusters on scale @xmath1 and with mass larger than @xmath20 are identified by the multi - resolution ( or multi - scale ) analysis based on discrete wavelet transformation ( dwt ) ( fang & thews 1998 ) .
it has been shown that the dwt - identified clusters on the abell scale , i.e. @xmath21-clusters , are the same as those given by _
friends - of - friends _ ( fof ) identification ( xfw ; jing & fang 1994 ) . that is , in terms of the @xmath21-cluster identification , the dwt method is as good as the fof .
thus , the dwt method can be reliably used to systematically identify objects on scales @xmath22 ( xu , fang & deng 1999 ) . with the multi - scale identification of simulation samples ,
we have model - predicted the mass function of @xmath1-clusters , @xmath23 , which is the number density of @xmath1-clusters with masses larger than @xmath20 .
the mfs of @xmath1-clusters with @xmath24 , 3 , 6 , 12 and 24 @xmath2 mpc in models of the scdm , lcdm and ocdm are shown in fig .
the middle part of mass function , roughly in the range @xmath25 @xmath26 mpc@xmath27 , is more reliable , as all the poisson errors are relatively small .
it can be seen from fig .
1 that all the mfs have a similar shape .
this is because of the mf scale invariance ( xfw ) as @xmath28 where the scaling factor @xmath29 is only a function of the ratio @xmath30 , but independent of @xmath20 .
obviously @xmath31 when @xmath32 .
the scaling factor @xmath33 is an important statistic description of clustering on large scales .
we apply the dwt multi - resolution identification to the redshift survey of the infrared astronomical satellite ( iras ) galaxies with flux limit of @xmath34 ( fisher et al .
the iras sample is of 2 ( physical space ) + 1 ( redshift space ) dimension .
a dwt algorithm for identifying @xmath1-clusters from the 2 + 1 dimension samples has been developed ( xfw ; xu , fang & deng 1999 ) . to minimize the radial effects
, the data are divided into redshift intervals of @xmath35 $ ] , @xmath36 $ ] and @xmath37 $ ] km s@xmath38 .
we then identify @xmath1-clusters in each interval on scales @xmath39 and
@xmath40 @xmath2 mpc .
the sampling rate of the iras sample is low .
the iras clusters identified on abell scale @xmath21 contain much fewer galaxies than optical clusters , and iras galaxies do not populate the clusters in the same manner as optical galaxies ( fisher et al .
1995 ; webster , lahav & fisher 1997 ) .
this gives rise to higher poisson errors .
however , if we are interested in clusters larger than the abell scale , this problem would be softened as structures contain more member galaxies . meanwhile , iras galaxies are mostly of late - type .
it traces clusters on various scales @xmath1 uniformly .
therefore , the samples of iras @xmath1-clusters may be less affected by morphological segregation , and good for studying the scale - dependence of structures . with the identified @xmath1-clusters of iras galaxies
, one can find the richness function of iras @xmath1-clusters , @xmath41 , which is the number density of @xmath1-clusters containing @xmath42 members of iras galaxies in the redshift shell @xmath43 .
the rfs of @xmath1-clusters with @xmath24 , 3 , 6 , 12 and 24 @xmath2 mpc are shown in fig .
2 , in which the errors are poissonian ( xu , fang & deng 1999 ) . despite the errors are large , it is clear that the shapes of all the rfs are basically the same . this feature can be described as a rf scaling invariance in each @xmath43 shell , @xmath44 the scaling factor @xmath45 is a function of the scale ratio @xmath30 , but independent of @xmath46 .
obviously @xmath47 when @xmath48 .
the scaling factor @xmath49 of iras @xmath1-clusters as a function of @xmath30 is presented in fig .
it shows that @xmath49 is also independent of @xmath43 within the 1-@xmath50 error bars .
by definition , the mass - to - luminosity ratio for iras @xmath1-clusters is the ratio between mass and luminosity of @xmath1-clusters .
one can introduce the @xmath0 ratio of the iras @xmath1-clusters by a mass - number relation as @xmath51 where @xmath20 is the total mass of a @xmath1-cluster , and @xmath52 the total number of member galaxies of the @xmath1-cluster , defined by @xmath53 where @xmath54 is the selection function ( fisher et al .
the definition of @xmath55 by eq.(3 ) has implicitly assumed that @xmath55 , or the @xmath0 ratio , is independent of the mass of the objects , since the @xmath0 ratio of @xmath21-clusters is found to be mass - independent on average ( bahcall & cen 1993 ) .
such independence may be caused by the fact that the scale of luminosity - related hydrodynamical process are smaller than that of clusters .
thus , luminosity functions and @xmath0 ratios are statistically independent of environment on cluster scale
. therefore , it would be reasonable to assume that the mass - independence of the @xmath0 ratio still hold on scales larger than clusters .
the number - mass conversion coefficient @xmath55 is proportional to the mass - to - light ratio @xmath56 for iras galaxies on scale @xmath1 .
we have then @xmath57 using eqs . ( 3 ) and ( 4 ) , one can find the mass function of iras @xmath1-clusters @xmath58 as @xmath59 therefore , the rf scaling ( 2 ) will further yield a mf scaling of iras @xmath1-clusters as @xmath60 where the mf scaling factor @xmath61 is given by @xmath62 or @xmath63 namely , the scale - dependence of the @xmath0 ratio is completely determined by the ratio between the mf and rf scaling factors of the iras @xmath1-clusters .
however , despite that we have the observed scaling factor @xmath45 , we are in lack of the observed @xmath64 .
this problem can be overcome by matching the iras galaxy traced mass field with popular models of cdm cosmogony . with properly selected parameters ,
cdm cosmogony models have been found to provide good approximation to many observational properties of large scale structures , especially the mass function of optical and x - ray clusters ( e.g. jing & fang 1994 ; bahcall , fan & cen 1997 , and references therein ) . on the other hand ,
some individual clusters and superclusters from optical surveys have been identified from the iras data ( fisher et al .
1995 ; webster et al .
therefore , it is reasonable to assume that the mass field of cdm cosmogony model provides a description for the mass field traced by the iras galaxies .
thus , the statistical properties of the simulated @xmath1-clusters should be the same as that of the iras @xmath21-clusters .
we have then @xmath65 i.e. the scaling factors of @xmath23 and @xmath66 are the same .
using eqs .
( 9 ) and ( 10 ) we have @xmath67 thus , the @xmath56 ratio can be determined by the ratio between the model - predicted @xmath29 and the observed @xmath45 .
the ratio @xmath68 as a function of @xmath30 is plotted in fig .
4 . for lcdm and ocdm ,
the value of this ratio does not depart from 1 with a significance larger than 1-@xmath50 in the entire range of scales from 3 to 24 @xmath2 mpc . in the case of scdm , only at @xmath69 24 @xmath2 mpc , the ratio shows a difference from 1 with 2 @xmath50 level .
particularly , in the redshift shell [ 2500,5000 ] km s@xmath38 , the ratio is very flat for all the three models .
thus , for iras sample , the @xmath0 ratio is basically flat on large scales from 3 to 24 @xmath2 mpc , i.e. @xmath70 this result is _ weakly _ dependent on specific parameters of models of the cdm cosmogony .
the @xmath71 goodness is larger than 98% for all the three cdm models . nevertheless , in fig .
4 , we can see a slight , but systematic increase of @xmath0 with scale .
a linear fitting by @xmath72 yields @xmath71 goodness larger than 99.9% for all the three models .
the least - square fitted slope @xmath73 is found to be @xmath74 for ocdm , @xmath75 for lcdm , and @xmath76 for scdm .
that is , the @xmath0 ratio on scale @xmath77 mpc is probably larger than that of @xmath21 by a factor about 1.7 for lcdm and ocdm .
( or , a factor of 2.9 for scdm . ) these results for ocdm and lcdm are consistent with the @xmath0 ratio measurement of the corona borealis supercluster .
namely , @xmath0 scale - dependence given by iras galaxies is consistent with optical data .
the rf scaling factor of multi - scale identified @xmath1-clusters is fundamentally important as a statistical measure for clustering on scales larger than clusters . since these rfs depend only on the number of member galaxies of @xmath1-clusters , they can be directly detected from a galaxy redshift survey regardless the mass determination of individual objects .
the @xmath0 ratio of gravitationally clustering systems which consist of luminous objects and dark matter depends only on the scale of the system but not on total mass ( bahcall and cen 1993 ) .
this is actually a scaling of cosmic gravitationally clustering .
therefore , the scale dependence of @xmath0 ratio is closely related to other scaling indices of these objects .
we have shown that the scale behavior of @xmath0 is determined by the rf and mf scaling factors of @xmath1-clusters . by comparing the mass function scaling of simulated samples with the richness function scaling of iras 1.2 jy galaxies ,
we conclude that the assumption of @xmath0-flatness is basically correct in the range from 1.5 to 24 @xmath2 mpc .
this conclusion is weakly dependent on the cosmological parameters we have considered . | it has been suggested that the mass - to - light ( @xmath0 ) ratio of gravitationally clustering objects is scale - independent on scales beyond galaxy clusters , and may also be independent of the mass of the objects . in this paper , we show that the scale behavior of @xmath0 ratio is closely related to the scaling of cosmic structures larger than clusters .
the scale dependence of the @xmath0 ratio can be determined by comparing the observed scaling of richness function ( rf ) of multi - scale identified objects with the model - predicted scaling of mass function ( mf ) of large scale structures . using the multi - scale identified clusters from iras 1.2 jy galaxy survey ,
we have made comparisons of the observed rf scaling of iras @xmath1-clusters with the mf scalings given by simulations of three popular models scdm , lcdm and ocdm .
we find that , the m / l ratio basically is scale - independent from the abell radius up to about 24 @xmath2mpc , while it seems to show a slight , but systematical , increase over this scale range .
this result is weakly dependent on the cosmological parameters . |
cosmological models with cold dark matter plus a @xmath3 term ( @xmath3cdm ) may explain most of the current astronomical observations ( see , e.g. , @xcite for some recent reviews ) .
however , from the theoretical viewpoint it is really difficult to reconcile the small value required by observations ( @xmath4 ) with estimates from quantum field theories ranging from 50 - 120 orders of magnitude larger , as well as to explain why this is exactly the right value that is just beginning to dominate the energy density of the universe today .
these issues make a complete cancellation of @xmath3 ( from some unknown symmetry of nature ) seem a plausible possibility and have also motivated a number of alternative explanations for the cosmic acceleration ( for some of these alternative scenarios , see @xcite ) .
one of these possibilities , possibly the next simplest approach toward an accelerating model for the universe , is to work with the idea that the dark energy component is due to a minimally coupled scalar field @xmath5 which has not yet reached its ground state and whose current dynamics is basically determined by its potential energy @xmath6
clearly , however , such a procedure can not provide a model - independent parameter space to be compared with the observational data .
another way , widely explored in the literature , is to build a phenomenological functional form for the dark energy equation of state ( eos ) , i.e. , the ratio of its pressure to its energy density , @xmath7 , in terms of its current value @xmath8 and of its time - dependence @xmath9 , and study its cosmological consequences as well as possible constraints on its behavior from observations .
usually , these parameterizations have not only the standard @xmath3cdm scenario ( @xmath10 ) but also the so - called @xmath11cdm model ( @xmath12 as particular cases , so that constraints on its parameters may provide more accurate consistency checks to the original models . examples of some eos parameterizations recently explored are ( see also @xcite ) : @xmath13 an interesting aspect worth mentioning is that it is difficult to obtain the above parameterizations from usual scalar field dynamics since they are not limited functions , i.e. , the eos parameter does not lie in the interval @xmath14 $ ] . in other words , this amounts to saying that when extended to the entire history of the universe , @xmath15 , the three parameterizations above are divergent functions of the redshift ( see also @xcite for a discussion ) parameterizations given by eq .
( 1 ) can be expressed as a single and generalized eos function , i.e. , @xmath16 , where the parameter @xmath17 takes , respectively , the values @xmath18 and @xmath19 @xcite . ] . in ref .
@xcite we investigated some cosmological consequences of a new phenomenological parameterization : @xmath20 or , equivalently,@xmath21 which has the same linear behavior at low redshifts presented by the parameterizations discussed above but with the advantage of being a limited function of @xmath1 throughout the entire history of the universe ( see also @xcite for a recent analysis of a coupled quintessence model driven by a dark energy component parameterized by ( [ myparameterization ] ) ) .
our goal in this paper is twofold .
first , to derive a scalar field description for the dark component whose eos parameter is given by eq .
( [ myparameterization ] ) . to that end
, we use the theoretical method of constructing the dark energy potential @xmath6 directly from the effective eos , as developed in ref . @xcite .
we also generalize this algorithm to include the so - called phantom case .
second , to place constraints on the ( @xmath8 , @xmath22 ) plane of parameterization ( [ myparameterization ] ) from current observational data to check which class of dark energy is observationally prefered .
we use some of the most recent type ia supernovae ( sne ia ) observations , namely , the so - called union2 sample of 557 events @xcite , the nearby + sdss + essence + snls + hubble space telescope ( hst ) set of 288 sne ia discussed in ref .
@xcite ( which we refer to as sdss compilation ) .
we consider two sub - samples of this latter compilation that use salt2 @xcite and mlcs2k2 @xcite sn ia light - curve fitting method . along with the sne
ia data , and to help break the degeneracy between the dark energy parameters we also use measurements of the baryonic acoustic oscillation ( bao ) peak at @xmath23 and 0.6 @xcite and the current estimate of the cmb shift parameter of ref .
@xcite . 0.1 in
let us first consider a homogeneous , isotropic , spatially flat cosmologies described by the friedmann - robertson - walker ( frw ) flat line element , @xmath24 , where we have set the speed of light @xmath25 .
in such a background , the friedmann equation for a dark matter - dark energy dominated universe is given by @xmath26 where @xmath27 stands for the non - relativistic ( dark plus baryonic ) matter .
the energy density and pressure of the dark energy field are given by @xmath28 @xmath29 where @xmath30 stands for quintessence ( @xmath31 ) @xcite and phantom fields ( @xmath32 ) @xcite , respectively . by combining eqs .
( [ rhophi ] ) and ( [ pphi ] ) , we obtain @xmath33 and @xmath34 or still , in terms of @xmath1 , @xmath35 so that , @xmath36 where the negative ( positive ) signs stands to @xmath37 ( @xmath38 ) . here
, we adopt the negative sign . by defining @xmath39 and @xmath40 and taking into account that @xmath41 , we have @xmath42 and @xmath43\omega_{\phi,0}f(z),\ ] ] where @xmath44 and @xmath45 stands for the time - dependent part of the energy density of the dark component , which , for parameterization ( [ myparameterization ] ) , is given by @xmath46 note also that eqs .
( [ phitilde ] ) and ( [ vphitilde ] ) are valid for both quintessence and phantom fields . by combining numerically these equations and taking into account the above constraints , we show in figures 1a and 1c the resulting potential @xmath47 for parameterization ( [ myparameterization ] ) for quintessence and phantom regimes , respectively .
figures 1b and 1d show the evolution of the dark energy field as function of the redshift .
the selected points used to plot the curves belong to quintessence and phantom families with @xmath48 and @xmath49 and follow the constraints given in sec .
we note that , in contrast with a canonic scalar field in which the potential increases with the redshift ( panel 1b ) , for the phantom case shown in panel 1d @xmath47 decrease with @xmath1 .
this result can be more easily understood by considering eq .
( [ vphitilde ] ) , i.e. , for @xmath50 , @xmath51 and @xmath52 , so that for phantom fields @xmath53 when @xmath54 .
before proceeding to the observational analyses on the @xmath55 parametric plane , it is worth mentioning that the parameters @xmath8 and @xmath22 of ( [ myparameterization ] ) are subject to the following constraints : @xmath56 and @xmath57 for a quintessence - like behavior and @xmath58 and @xmath59 for phantom fields .
these bounds allow us to divide the parametric plane ( @xmath55 ) in defined regions associated to distinct classes of dark energy models that can be confronted with current observational data .
these regions are shown in the figures 2a-2c , where the area of early dark energy dominance corresponds to the constraint @xmath60 , required to ensure @xmath61 at @xmath62 .
the blank regions indicate models that at some point of the cosmic evolution , @xmath63 , have switched or will switch from quintessence to phantom behaviors or vice - versa . in order to discuss the current observational constraints on @xmath8 and @xmath22 we use two of the most recent sne ia compilations available , namely , the union2 sample of ref .
@xcite and the two compilations of the sdss collaborations discussed in ref .
the union2 sample is an update of the original union compilation .
it comprises 557 data points including recent large samples from other surveys and uses salt2 for sn ia light - curve fitting @xcite .
the sdss compilation of 288 sne ia uses both salt2 and mlcs2k2 light - curve fitters and is distributed in redshift interval @xmath64 @xcite .
along with the sne ia data , and to diminish the degeneracy between the dark energy parameters @xmath65 , @xmath8 and @xmath22 , we also use the results of current bao and cmb experiments . for the bao measurements , we use the three estimates of the bao parameter @xmath66 at @xmath67 , @xmath68 and @xmath69 , as given in table 2 of ref
. @xcite . in this latter expression , @xmath70^{1/3}$ ] is the so - called dilation scale , defined in terms of the dimensionless comoving distance @xmath71 . for the cmb ,
we use only the measurement of the cmb shift parameter @xcite @xmath72 where @xmath73 . in our analyses , we minimize the function @xmath74 , which takes into account all the data sets mentioned above and marginalize over the present value of the hubble parameter @xmath75 . for simplicity ,
the combinations union2 + bao + cmb , sdss ( salt2 ) + bao + cmb and sdss ( mlcs2k2 ) + bao + cmb are denoted by c1 , c2 and c3 respectively .
panels 2a-2c show the results of our statistical analyses . in order to compare the theoretical frame with the observational constraints discussed above the three - dimensional parameter space
@xmath76 has been projected into the plane @xmath55 .
panels 2a , 2b and 2c show contours of @xmath77 and 6.17 arising from the joint analyses c1 , c2 and c3 , respectively . with exception of c3
, we note that no dark energy behavior is preferred or ruled out by observations , although the largest portion of the confidence contours lies into the blank region , which indicates a possible crossing of the so - called phanton divide line at some point of the cosmic evolution ( see @xcite for a discussion ) .
ccccc + & @xmath78&@xmath8&@xmath22&@xmath65 + + c1 .............
&545.02&@xmath79&@xmath80&@xmath81 + c2 .............
&249.14&@xmath82&@xmath83&@xmath81 + c3 ............. &242.09&@xmath84&@xmath85&@xmath86 + in table i we summarize the results of our statistical analyses . for the sake of comparison
, we also performed the c1 , c2 and c3 analyses for the so - called cpl parameterization , @xmath87 @xcite .
we have found very similar values to those shown in table i. this means that , although different from the theoretical viewpoint cpl parameterization blows up exponentially in the future as @xmath88 for @xmath89 whereas parameterization ( 1 ) is a limited function of @xmath1 @xmath90 both parameterizations provide very similar description for the current observational data .
we have examined theoretical and observational aspects of the eos parameterization given by eq .
( [ myparameterization ] ) @xcite . by following the method of constructing the quintessence potential directly from the effective equation of state function developed in ref .
@xcite , we have derived the scalar field description for this @xmath0 parameterization and extended our results for the case of phanton fields in which @xmath91 . we also have performed a joint statistical analysis involving some of the most recent cosmological measurements of sne ia , bao peak and cmb shift parameter . from a pure observational perspective , we have shown that both quintessence and phantom behaviours are acceptable regimes . in agreement with recent analyses
, it has been shown that the largest portion of the confidence contours arising from these observations lies in the region of models that have crossed or will eventually cross the so - called phanton divide line at some point of the cosmic evolution .
v. sahni and a. a. starobinsky , int .
j. mod .
d * 9 * , 373 ( 2000 ) ; p. j. e. peebles and b. ratra rev .
* 75 * , 559 ( 2003 ) ; t. padmanabhan , phys . rept . * 380 * , 235 ( 2003 ) ; e. j. copeland , m. sami and s. tsujikawa , int . j. mod
d15 * , 1753 ( 2006 ) ; j. s. alcaniz , braz . j. phys . * 36 * , 1109 ( 2006 ) .
astro - ph/0608631 ; b. ratra & m. s. vogeley , pasp * 120 * , 235 ( 2008 ) ; a. silvestri & m. trodden , rept .
* 72 * , 09690 ( 2009 ) .
j. a. frieman _ et al .
_ , phys .
lett . * 75 * , 2077 ( 1995 ) ; r. r. caldwell , r. dave and p. j. steinhardt , phys .
lett . * 80 * , 1582 ( 1998 ) ; r. r. caldwell and e. v. linder , phys .
lett . * 95 * , 141301 ( 2005 ) ; f. c. carvalho _ et al .
_ , phys .
* 97 * , 081301 ( 2006 ) .
astro - ph/0608439 .
y. wang and p. m. garnavich , astrophys .
j. * 552 * , 445 ( 2001 ) ; c. r. watson and r. j. scherrer , phys .
d * 68 * , 123524 ( 2003 ) ; p.s .
corasaniti et al . , phys .
d * 70 * , 083006 ( 2004 ) ; v. b. johri , astro - ph/0409161 ; h. k. jassal , j. s. bagla , and t. padmanabhan , mon . not .
astron . soc . *
356 * , l11 ( 2005 ) ; y. wang , phys . rev . *
d 77 * , 123525 ( 2008 ) .
a. r. cooray and d. huterer , astrophys .
j. * 513 * , l95 ( 1999 ) ; p. astier , phys .
b , * 500 * , 8 ( 2001 ) ; m. goliath
_ et al . _ ,
. astrophys .
* 380 * , 6 ( 2001 ) ; j. weller and a. albrecht , phys .
rev d * 65 * , 103512 ( 2002 ) . m. m. phillips , astrophys .
j. * 413 * , l105 ( 1993 ) ; a. g. riess , w. h. press , and r. p. kirshner , astrophys .
j. * 438 * , l17 ( 1995 ) ; s. jha , a. g. riess , and r. p. kirshner , astrophys . j. * 659 * , 122 ( 2007 ) . | we investigate theoretical and observational aspects of a time - dependent parameterization for the dark energy equation of state ( eos ) @xmath0 , which is a well behaved function of the redshift @xmath1 over the entire cosmological evolution , i.e. , @xmath2 . by using a theoretical algorithm of constructing the quintessence potential directly from the effective eos parameter
, we derive and discuss the general features of the resulting potential for this @xmath0 function .
since the parameterization here discussed allows us to divide the parametric plane in defined regions associated to distinct classes of dark energy models , we use the most recent observations from type ia supernovae , baryon acoustic oscillation peak and cosmic microwave background shift parameter to check which class is observationally prefered .
we show that the largest portion of the confidence contours lies into the region corresponding to a possible crossing of the so - called phanton divide line at some point of the cosmic evolution . |
cosmological relaxation of the higgs mass has been proposed recently as an alternative solution to the weak scale hierarchy problem @xcite . in this scenario , a pseudo nambu - goldston boson @xmath0
is coupled to the standard model higgs doublet @xmath1 , scanning the higgs mass from a large initial value to the small present value .
this scalar field , often referred as relaxion , has a potential including the piece enforcing the relaxion to move to scan the higgs mass , and also a periodic barrier potential to stop the relaxion at the position giving @xmath2 .
more specifically , the relevant scalar potential is given by [ potential ] ( ^2 - ) |h|^2 -c_0 + v_b , where @xmath3 denotes the higgs mass cutoff scale , @xmath4 corresponds to the relaxion excursion required to scan the higgs mass - square from @xmath5 to @xmath6 , @xmath7 is a positive coefficient of order unity as suggested by the naturalness argument , and the barrier potential @xmath8 generically takes the form : @xmath9 with a higgs - dependent amplitude @xmath10 where @xmath11 is determined by the scale where @xmath8 is generated , as well as by the involved couplings . imposing the stationary condition to the potential ( [ potential ] ) , one finds [ stationary ] ~ , where @xmath12 gev and @xmath13 @xcite for the relaxion vacuum value @xmath14 in the present universe
. there can be two ways to generate the barrier potential as discussed in the literatures @xcite .
the minimal scenario is to generate @xmath8 through the relaxion coupling to the qcd anomaly , i.e. @xmath15 , which would result in @xmath16 , where @xmath17 mev is the qcd scale and @xmath18 is the up - quark yukawa coupling . in this case , @xmath19 is identified as the qcd vacuum angle @xmath20 and therefore constrained as @xmath21 alternatively , the barrier potential can be generated by a new physics around the weak scale , yielding for instance @xmath22 with @xmath11 around the weak scale . to implement the relaxation mechanism , the amplitude of the barrier potential is required to be bounded as @xmath23 @xcite , where @xmath12 gev is the higgs vacuum value in the present universe .
then the stationary condition ( [ stationary ] ) shows that the relaxion mechanism transmutes the weak scale hierarchy @xmath24 to another hierarchy @xmath25 .
although the latter hierarchy can be technically natural , it may require an explanation for its origin .
this issue has been addressed in @xcite , proposing a scheme to generate an exponential hierarchy @xmath26 based on models with @xmath27 axions @xcite . a key ingredient of the relaxation scheme is a mechanism to dissipate away the relaxion kinetic energy which is originating from the initial potential energy of @xmath28 .
it is usually assumed that the relaxion loses its kinetic energy by the hubble friction during the inflationary period .
then the scheme requires a rather large number of inflationary e - foldings @xcite , which is estimated as @xcite [ efolding_1 ] n_e ~()^2 for the case that the barrier potential is induced by new physics , and [ efolding_2 ] n_e ~ 10 ^ 24()^4 for the other case that the barrier potential is induced by low energy qcd . the above result and
the relaxion scale hierarchy ( [ stationary ] ) show that the scenario with qcd - induced barrier potential requires a huge e - folding number and also a big hierarchy among the relaxion scales .
as a too large e - folding number might cause a severe fine - tuning problem , in the following we will focus on the scenario that the barrier potential is generated by new physics around the weak scale , yielding [ barrier ] v_b=-_b^4(h)(/f)=-_b^2 |h|^2(/f ) with @xmath29 . by the same reason
, we will be more interested in the case that @xmath11 is somewhat close to the weak scale . in the relaxation scenario , to avoid a fine - tuning of initial condition
, the relaxion is assumed to be stabilized before the inflation is over . if the temperature during the reheating phase is well below the weak scale which corresponds to the scale where the barrier potential is generated , the relaxion dynamics after the reheating is trivial .
it stays there without changing the higgs mass selected during inflation .
however , if the universe experiences a high temperature @xmath30 after inflation , the electroweak gauge symmetry is restored and the barrier potential disappears . then the relaxion starts to roll again until the temperature cools down to a critical temperature @xmath31 where the barrier potential is developed again , and such subsequent evolution may ruin the successful implementation of the relaxation mechanism . on the other hand , high reheating temperature @xmath32
is often favored for viable cosmology , in particular for baryogenesis .
it is therefore an interesting question if the cosmological relaxation mechanism can be compatible with such high reheating temperature . in this paper , we wish to examine such possibility within the relaxion scenario in which the barrier potential ( [ barrier ] ) is generated by new physics near the weak scale . to proceed ,
let us first consider the case that there is no additional dissipation of the relaxion energy other than those by the hubble friction . during the period
when @xmath8 is negligible , e.g for radiation - dominated period with @xmath33 , solving the equation of motion determined by ( [ potential ] ) , one finds that the relaxion speed behaves as ( t ) t ( ) ^1/2 , where @xmath34 , @xmath35 is the number of relativistic degrees of freedom at @xmath36 , and we use the relation ( [ stationary ] ) for the last expression . as the relaxion speed is increasing in time , to stop the relaxion by the barrier potential developed around the time @xmath37
, one needs [ condition_1 ] ( t_c ) _ b^2 .
one can also make sure that if this condition is satisfied , the relaxion is successfully stabilized within a few hubble time from @xmath37 with a total excursion @xmath38 , and therefore the corresponding change of the higgs mass is negligible . on the other hand
, the condition ( [ condition_1 ] ) puts a lower bound on the relaxion decay constant @xmath39 , given by [ bound_f ] ( ) ^1/2 , where we used @xmath31 .
although it is possible that the dynamics to generate @xmath8 involves a small coupling , so @xmath40 , such a small @xmath41 is disfavored as it requires a bigger e - folding number ( [ efolding_1 ] ) for a given value of the higgs mass cutoff @xmath3 . for @xmath42 which is more
favored in view of ( [ efolding_1 ] ) , the bound ( [ bound_f ] ) suggests that @xmath39 should be at least comparable to the planck scale .
the above observation implies that one needs an additional mechanism to dissipate the relaxion energy to make the scheme compatible with @xmath43 for the more interesting parameter range with @xmath42 and @xmath44 .
it is well known that a rolling scalar field @xmath0 can lose its kinetic energy through gauge field production induced by the coupling , x _ ^ , [ anomalous_coupling ] where @xmath45}$ ] is an abelian gauge field strength and @xmath46 is its dual . in the presence of this coupling
, a rolling @xmath0 develops tachyonic instability of @xmath47 , causing an exponential growth of @xmath47 for certain range of wave number .
this provides an effective frictional force to the motion of @xmath0 , which has been applied recently to the relaxion dynamics in the early universe @xcite . in this paper
, we explore the possibility of high reheating temperature in the relaxation scenario in which the coupling ( [ anomalous_coupling ] ) is mainly responsible for the relaxion energy dissipation after reheating . as we will see , in case that @xmath47 is identified as the @xmath48 gauge boson of the standard model ( sm ) , due to its large thermal mass , the gauge field production is not efficient enough to slow down the relaxion motion in most of the parameter space allowed by other constraints . on the other hand ,
if @xmath47 is identified as a dark gauge boson with negligible thermal mass , the gauge field production can efficiently slow down the relaxion motion , allowing the relaxion to be successfully stabilized after the electroweak phase transition for a wide range of model parameters .
this paper is organized as follows . in section [ sec : production_gauge_field ] , we review the gauge field production by a rolling scalar field with the coupling ( [ anomalous_coupling ] ) , and apply the results for the relaxion dynamics at @xmath30 .
our primary concern is to identify the parameter region which allows the reheating temperature @xmath43 without modifying the standard cosmology after reheating . for this
, we estimate the relaxion excursion and terminal speed at the time when the relaxion is stopped by a barrier potential developed at @xmath31 .
we provide also numerical results to cross check our analytic estimation . in section [ sec : discussion ] , we discuss additional constraints on the scenario discussed in the previous section , and section [ sec : conclusion ] is the conclusion .
in the presence of the coupling ( [ anomalous_coupling ] ) , a background evolution of relaxion develops tachyonic instability of the abelian gauge boson @xmath47 @xcite .
let us begin with the case without any light @xmath49 charged particle , in which the gauge field production turns out most efficient . in this case
, there is no thermal mass of @xmath47 even at high temperature limit , and then the equation of motion for @xmath47 in the expanding universe is given by x_+ ( k^2 a k ) x_= 0 , where @xmath50 denotes the helicity state , @xmath51 is the scale factor of the expanding universe with the metric @xmath52 and @xmath53 and @xmath54 for the conformal time @xmath55 and the physical time @xmath56 . assuming @xmath57 , the positive helicity state experiences tachyonic instability for the wave number @xmath58 . using the wkb approximation
, we find that the corresponding gauge field modes grow as x_+(k ) ~ , with the growth rate determined as ^2= ak - k^2 [ tachyon_mass1 ] for @xmath59 . here
the lower bound on @xmath60 is required for the validity of wkb approximation , @xmath61 .
note that the gauge field production is dominated for the modes with @xmath62 .
so a rolling relaxion produces gauge fields with comoving wave number @xmath63 , and those gauge fields will eventually modify the evolution of relaxion . to see the interplay between the gauge field production and the relaxion evolution , we consider the relaxion equation of motion = - 3 h - - x _
^ , [ eom_scalar ] where the ensemble average of gauge fields is given by x _ ^ & = & dk k^3 ( | x_+|^2 - |x_-|^2 ) + & & , [ gauge_friction ] where @xmath64 is the wave length at which the gauge field production is maximized .
the above equation of motion shows that the produced gauge fields provide additional frictional force to the relaxion motion .
then the relaxion speed reaches at its terminal value around the time when the accelerating force @xmath65 in is balanced by the last frictional force term .
we already noticed that the gauge field production is most efficient for @xmath62 .
then , equating with @xmath65 , the relaxion terminal speed is estimated as _ term= h f , [ term_v ] where the dimensionless coefficient @xmath66 mildly depends on various factors such as @xmath67 , @xmath68 , and the initial condition for @xmath47 .
we performed a numerical analysis to examine the relaxion evolution for the model parameters in table [ tab : parameter_set_1 ] , and depict the result in figure [ fig : vel_no_plasma ] .
our result shows that the relaxion speed indeed approaches to the form ( [ term_v ] ) with @xmath69 as indicated in fig .
[ fig : terminal_velocity ] . .sample model parameters for numerical analysis .
[ cols="^,^,^,^,^",options="header " , ] in figure [ fig : param ] , we show the viable parameter region for our scenario in the absence ( left ) and presence ( right ) of dark plasma of the sm singlet but @xmath49-charged particles . for the purpose of illustration
, we adopt the dark sector temperature @xmath70 with @xmath71 .
there are two primary requirements : conditions for successful relaxation , i.e. or discussed in the previous section , and additional constraints for @xmath72 discussed in this section .
other constraints from astrophysics and terrestrial experiments @xcite are overlaid in the same plot .
first of all , in the presence of dark plasma providing a thermal mass @xmath73 , our scenario works only on a very limited region of parameter space as described in figure .
[ fig : param ] .
the viable parameter regions shrink even more if we increase @xmath74 .
this essentially forbids the @xmath49 gauge boson to be identified as the sm @xmath48 gauge boson for the most of parameter space , as stated in the end of previous section . on the other hand
, our mechanism works for a reasonably wide range of parameter space in the absence of dark plasma .
in particular , @xmath41 can be as large as @xmath75 which is preferred in view of the inflationary model building .
such a large @xmath41 is constrained further by the fifth force experiments and stellar evolution in globular clusters .
our scenario may be tested if the sensitivity of these experiments is significantly improved in the future .
we finally comment on the possible perturbations of the relaxion field around homogeneous background . as in the conventional relaxation scenario @xcite
, we assume that the electroweak scale is selected during the long period of inflation .
then the background value of relaxion at the beginning of reheating is highly homogeneous . on the other hand ,
since the production of @xmath49 bosons depends on the gauge field wave number , one may suspect that an inhomogeneity of @xmath0 might be developed consequently .
however , it turns out that due to the negative feedback working between the relaxion speed and the gauge field production , the relaxion excursion in homogenous and isotropic background is stable against perturbations .
eventually the relaxion field enters into the terminal regime , and thereafter the growth of perturbations is by no means possible as @xmath76 and the resulting gauge field production continuously decay . we thus conclude that sizable perturbations in the relaxion and @xmath49 gauge fields can hardly be produced in our scenario .
in this paper , we examined if the cosmological relaxation of the higgs boson mass , which was proposed recently as an alternative solution to the weak scale hierarchy problem , can be compatible with high reheating temperature well above the weak scale .
as the barrier potential disappears at high temperature , the relaxion rolls down further after the reheating , which may ruin the successful selection of the right higgs boson mass .
as it can provide a working scheme over a wide range of model parameters , we focus on the scenario that the relaxion is coupled to a dark @xmath49 gauge boson as @xmath77 . in the presence of this coupling , the background relaxion evolution causes tachyonic instability of the @xmath49 gauge boson , leading to an explosive gauge field production .
then the relaxion is slowed down soon after the gauge field production , and can be re - stabilized by the barrier potential developed at lower temperature around the electroweak scale . to identify the working parameter region
, we estimate the relaxion excursion after the reheating and impose the condition not to produce too much @xmath49 gauge bosons , as well as other observational constraints on the model parameters .
we have examined this for two different cases .
the first is the case that there is no thermal plasma of @xmath49-charged particles , so no thermal mass of the @xmath49 gauge boson .
the second is the case with dark plasma providing a thermal gauge boson mass @xmath78 . in the first case ,
the gauge field production is most efficient , and therefore a successful re - stabilization of the relaxion can be achieved over a wide range of model parameters , including the one with @xmath79 gev which is favored in view of the required inflationary e - folding number in ( [ efolding_1 ] ) . in the second case ,
the thermal mass @xmath80 suppresses the gauge boson production , and then the relaxion can be successfully re - stabilized only for a limited parameter range with smaller @xmath41 . through this study
, we have also shown that @xmath49 can hardly be identified as the @xmath48 of the sm under the assumption that the universe has been radiation - dominated until when the relaxion is re - stabilized .
if we adopt the possibility that the universe is dominated by the relaxion energy density over certain period , @xmath49 might be identified as the @xmath48 for a limited range of model parameters .
although an interesting possibility , the analysis for this case is more involved , and we leave it to future work .
this work was supported by ibs under the project code , ibs - r018-d1 .
we would like to thank hye - sung lee , thomas flacke , sang hui i m , ryusuke jinno , and seokhook yun for useful discussions .
99 p. w. graham , d. e. kaplan and s. rajendran , phys .
lett . * 115 * , no .
22 , 221801 ( 2015 ) [ arxiv:1504.07551 [ hep - ph ] ] .
o. antipin and m. redi , jhep * 1512 * , 031 ( 2015 ) [ arxiv:1508.01112 [ hep - ph ] ] .
r. s. gupta , z. komargodski , g. perez and l. ubaldi , jhep * 1602 * , 166 ( 2016 ) [ arxiv:1509.00047 [ hep - ph ] ] .
j. r. espinosa , c. grojean , g. panico , a. pomarol , o. pujols and g. servant , phys .
lett . * 115 * , no .
25 , 251803 ( 2015 ) [ arxiv:1506.09217 [ hep - ph ] ] .
s. p. patil and p. schwaller , jhep * 1602
* , 077 ( 2016 ) [ arxiv:1507.08649 [ hep - ph ] ] .
j. jaeckel , v. m. mehta and l. t. witkowski , phys .
d * 93 * , no .
6 , 063522 ( 2016 ) [ arxiv:1508.03321 [ hep - ph ] ] .
b. batell , g. f. giudice and m. mccullough , jhep * 1512 * , 162 ( 2015 ) [ arxiv:1509.00834 [ hep - ph ] ] . o. matsedonskyi , jhep * 1601 * , 063 ( 2016 ) [ arxiv:1509.03583 [ hep - ph ] ] . l. marzola and m. raidal , arxiv:1510.00710 [ hep - ph ] . s. di chiara , k. kannike , l. marzola , a. racioppi , m. raidal and c. spethmann , phys .
d * 93 * , no .
10 , 103527 ( 2016 ) [ arxiv:1511.02858 [ hep - ph ] ] .
l. e. ibanez , m. montero , a. uranga and i. valenzuela , jhep * 1604 * , 020 ( 2016 ) [ arxiv:1512.00025 [ hep - th ] ] .
n. fonseca , l. de lima , c. s. machado and r. d. matheus , phys .
d * 94 * , no . 1 , 015010 ( 2016 ) [ arxiv:1601.07183 [ hep - ph ] ] .
a. fowlie , c. balazs , g. white , l. marzola and m. raidal , jhep * 1608 * , 100 ( 2016 ) [ arxiv:1602.03889 [ hep - ph ] ] . j. l. evans , t. gherghetta , n. nagata and z. thomas , jhep * 1609 * , 150 ( 2016 ) [ arxiv:1602.04812 [ hep - ph ] ] .
huang , y. cai , h. li and x. zhang , arxiv:1605.03120 [ hep - ph ] .
e. hardy , jhep * 1511 * , 077 ( 2015 ) [ arxiv:1507.07525 [ hep - ph ] ] .
t. higaki , n. takeda and y. yamada , arxiv:1607.06551 [ hep - ph ] .
t. kobayashi , o. seto , t. shimomura and y. urakawa , arxiv:1605.06908 [ astro-ph.co ] .
k. choi and s. h. i m , arxiv:1610.00680 [ hep - ph ] .
t. flacke , c. frugiuele , e. fuchs , r. s. gupta and g. perez , arxiv:1610.02025 [ hep - ph ] .
k. choi and s. h. i m , jhep * 1601 * , 149 ( 2016 ) [ arxiv:1511.00132 [ hep - ph ] ] .
d. e. kaplan and r. rattazzi , phys .
d * 93 * , no . 8 , 085007 ( 2016 ) [ arxiv:1511.01827 [ hep - ph ] ] .
j. e. kim , h. p. nilles and m. peloso , jcap * 0501 * ( 2005 ) 005 [ hep - ph/0409138 ] .
k. choi , h. kim and s. yun , phys .
d * 90 * , 023545 ( 2014 ) [ arxiv:1404.6209 [ hep - th ] ] .
a. hook and g. marques - tavares , arxiv:1607.01786 [ hep - ph ] . m. m. anber and l. sorbo , phys .
d * 81 * , 043534 ( 2010 ) [ arxiv:0908.4089 [ hep - th ] ] .
j. i. kapusta and c. gale , finite - temperature field theory : principles and applications , " cambridge university press , 2011 .
m. l. bellac , `` thermal field theory , '' cambridge university press , 2011 . | we examine if the cosmological relaxation mechanism , which was proposed recently as a new solution to the hierarchy problem , can be compatible with high reheating temperature well above the weak scale . as the barrier potential disappears at high temperature ,
the relaxion rolls down further after the reheating , which may ruin the successful implementation of the relaxation mechanism .
it is noted that if the relaxion is coupled to a dark gauge boson , the new frictional force arising from dark gauge boson production can efficiently slow down the relaxion motion , which allows the relaxion to be stabilized after the electroweak phase transition for a wide range of model parameters , while satisfying the known observational constraints . |
based on the general argument that the de broglie length of an ultrarelativistic particle is very small , here , we would like to solve the dirac equation @xmath166\psi=0\ ] ] by applying the wentzel - kramers - brillouin ( wkb ) method ( the same notation is employed as in the main text ) @xcite .
more precise conditions of validity of the present approach are provided below . according to the wkb method
, we look for a solution of the form @xmath120\varphi(x)$ ] @xcite and the dirac equation becomes @xmath167\varphi = i\hbar\gamma^{\mu}\partial_{\mu}\varphi.\ ] ] we first neglect the term proportional to @xmath168 in this equation .
the resulting equation @xmath169\varphi=0\ ] ] for the zero - order bi - spinor @xmath170 admits a non - trivial solution only if @xmath171=0 $ ] .
as it is known @xcite , this condition implies that @xmath121 has to satisfy the hamilton - jacobi equation @xmath172 and that it coincides with the classical action @xcite . by applying the method of characteristics ( see , e.g. ,
@xcite ) , the solution of the partial differential hamilton - jacobi equation is reduced to the solution of the ordinary differential lorentz equation .
then , the action can be constructed from the equation @xcite @xmath173 where all quantities are calculated along the electron s trajectory / characteristics , parametrized via the proper time @xmath36 . since we have already determined the electron s trajectory within the required approximation ( see eqs .
( 6)-(8 ) in the main text ) , it is straightforward to integrate the equation for the action @xmath174 by parametrizing the electron s trajectory via the variable @xmath61 .
after the integration has been carried out , one has to eliminate the initial quantities @xmath116 in terms of the quantities @xmath175 via eq .
( 5 ) in the main text .
the final expression of the action @xmath142 calculated up to terms proportional to @xmath85 is @xmath176 .
\end{split}\ ] ] now , the remaining task is to determine the bi - spinor @xmath177 , which solves the equation @xmath178\varphi_{p_0,\sigma_0}=0,\ ] ] where the four - momentum @xmath179 has to be intended as a function of the coordinates like the action , and where the quantum number @xmath180 indicates the electron s spin degree of freedom . by recalling the solution of the dirac equation for a free electron @xcite ,
we can already write @xmath177 in general as @xmath181 where @xmath182 are the pauli matrices , @xmath183 is an arbitrary spinor , and where the normalization factor @xmath139 has been introduced for convenience ( a unity quantization volume is assumed ) . in order to determine the spinor @xmath183
, we consider now eq . ( [ dirac ] ) at first order in @xmath168 and notice that it implies that @xmath184\gamma^{\nu}\partial_{\nu}\varphi_{p_0,\sigma_0}=0.\ ] ] by substituting the bi - spinor ( [ phi_0 ] ) in this equation , the latter reduces to the following equation for the spinor @xmath183 ( see also @xcite ) : @xmath185 this equation can also be solved via the method of characteristics by parametrizing the trajectory via the variable @xmath61 . by setting @xmath186\rho_{p_0,\sigma_0}(\tau),\ ] ] the spinors @xmath187 fulfill time - independent normalization conditions , which can be chosen as : @xmath188 .
also , the spin four - vector @xmath189 in units of @xmath168 , defined as @xmath190 , with @xmath191 for a state @xmath119 @xcite , is given in our case by @xcite @xmath192}\bm{p}(x)\right),\ ] ] with @xmath193 . by evaluating the four - vector @xmath194 along the electron s trajectory
, it can be shown that it satisfies the bargmann - michel - telegdi equation @xmath195 for a dirac electron with spin gyromagnetic factor equal to two @xcite . by proceeding analogously as in the case of the action , one obtains the following first - order expression of @xmath196 : @xmath197\}w_{0,p_0,\sigma_0 } , \end{split}\ ] ] where @xmath198 is the initial spin vector , normalized as @xmath199 ( see discussion below eq .
( [ eq_w ] ) ) and assumed to satisfy the equation @xmath200 .
finally , by expanding the resulting state @xmath201 up to terms proportional to @xmath85 in the expression @xmath202\varphi_{p_0,\sigma_0}(\tau,\bm{r})$ ] of the electron state up to first order in @xmath85 , we obtain : @xmath203\}\bigg ] \frac{u_{p_0,\sigma_0}}{\sqrt{2\varepsilon_0 } } , \end{split}\ ] ] where @xmath126 and @xmath127 is the usual constant free bi - spinor @xcite .
note that up to first order in @xmath85 it is @xmath204 thus , since the integral over the whole space of the quantity @xmath205 vanishes , it is @xmath206 .
note that the same normalization condition holds also in the ordinary coordinate space @xmath207 , as the operator @xmath208 can be expressed as a linear combination of the derivatives with respect to the ordinary coordinates . in order to determine the validity conditions of the approach presented above , we first request that the first - order corrections in @xmath85 in the action and in the pre - factor matrix in the state @xmath123 in eq .
( [ psi_0_sm ] ) are much smaller than the leading - order terms . equations ( [ s^1_sm ] ) and ( [ psi_0_sm ] ) indicate that , as expected , such conditions are in order of magnitude equivalent to those obtained for the validity of the classical approximated solution ( see discussion below eq .
( 8) in the main text ) .
more precise conditions also depend on the process to be investigated and on its formation region .
additional conditions for the validity of the above wave functions are obtained by estimating the size of the terms , which have been neglected in the original dirac equation during the derivation .
this can be carried out more easily by means of an alternative derivation of the wave function @xmath123 in eq .
( [ psi_0_sm ] ) , starting from the `` squared '' dirac equation @xcite @xmath209-m^2\psi-\frac{ie\hbar}{2}\sigma^{\mu\nu}f_{\mu\nu}\psi=0.\ ] ] by writing again the wave function @xmath119 in the form @xmath120\varphi(x)$ ] and by collecting the terms with the same power of @xmath168 , one obtains @xmath210\varphi\\ & \quad - i\hbar\left[2(\partial_{\mu}s+ea_{\mu})\partial^{\mu}+\square s+\frac{e}{2}\sigma^{\mu\nu}f_{\mu\nu}\right]\varphi\\ & \quad\quad-\hbar^2\square\varphi=0 , \end{split}\ ] ] where @xmath211 . at zero order in @xmath168 ,
we again obtain the hamilton - jacobi equation for the quantity @xmath121 , which we identify , as above , with the classical action .
thus , the four - vector @xmath179 is the corresponding classical electron four - momentum and , with this choice , eq .
( [ dirac_s_2 ] ) becomes @xmath212 now , our solution above corresponds to neglecting the term on the right - hand side proportional to @xmath168 .
this is a good approximation if the formal operator conditions @xmath213 are fulfilled .
we can `` estimate '' the differential and integral operators starting from the spatio - temporal extension of the background field and by applying the operators to the bi - spinor @xmath214\psi^{(1)}_{p_0,\sigma_0}(\tau,\bm{r})$ ] ( see eq .
( [ psi_0_sm ] ) ) . also , we specialize to the case of a focused gaussian laser field as that considered in the main text ( central angular frequency @xmath20 , central wavelength @xmath215 , waist size @xmath216 , rayleigh length @xmath217 , and pulse duration @xmath218 ) . for the sake of simplicity and recalling that @xmath17 in our units , we also assume that @xmath219 ( for experimental laser parameters of a tightly focused ti : sapphire laser : @xmath220 , @xmath221 , and @xmath222 , it is @xmath223 and @xmath224 ) . since @xmath225 , an inspection to eq .
( [ psi_0_sm ] ) indicates that @xmath226 .
analogously , it can be shown that @xmath227 , such that , being understood that the classical strong inequality @xmath132 is satisfied , the above conditions are fulfilled in order of magnitude if @xmath228 .
this condition is safely satisfied in the case of a tightly focused ( @xmath229 ) , optical ( @xmath131 ) laser field , as @xmath230 , with @xmath231 being the compton wavelength .
t. heinzl , a. ilderton , and m. marklund , phys .
b * 692 * , 250 ( 2010 ) .
s. meuren and a. di piazza , phys .
* 107 * , 260401 ( 2011 ) .
a. ilderton , phys .
rev . lett . * 106 * , 020404 ( 2011 ) .
mller and c. mller , phys .
b * 696 * , 201 ( 2011 ) .
f. mackenroth and a. di piazza , phys .
a * 83 * , 032106 ( 2011 ) .
d. seipt and b. kmpfer , phys .
a * 83 * , 022101 ( 2011 ) .
a. i. titov , h. takabe , b. kmpfer , and a. hosaka , phys .
108 * , 240406 ( 2012 ) .
f. ehlotzky , k. krajewska , and j. z. kamiski , rep .
. phys . * 72 * , 046401 ( 2009 ) .
r. ruffini , g. vereshchagin , and s. s. xue , phys . rep . * 487 * , 1 ( 2010 ) .
l. d. landau and e. m. lifshitz , _ quantum mechanics ( non - relativistic theory ) _ ( elsevier , oxford , 1977 ) .
w. pauli , helv .
acta * 5 * , 179 ( 1932 ) .
s. i. rubinow and j. b. keller , phys .
rev . * 131 * , 2789 ( 1963 ) .
v. p. maslov and m. v. fedoriuk , _ semi - classical approximation in quantum mechanics _ , ( d. reidel publishing company , dordrecht , 1981 ) .
t. heinzl and a. ilderton , opt
. commun . * 282 * , 1879 ( 2009 ) .
g. chriaux , _ et al .
_ , aip conf . proc .
* 1462 * , 78 ( 2012 ) .
x. wang , _ et al .
_ , nature commun . * 4 * , 1988 ( 2013 ) .
j. b. kogut and d. e. soper , phys .
d * 1 * , 2901 ( 1970 ) .
a. m. dirac , rev .
* 21 * , 392 ( 1949 ) .
y. i. salamin , g. r. mocken , and c. h. keitel , phys .
beams * 5 * , 101301 ( 2002 ) .
n. b. narozhny , s. s. bulanov , v. d. mur , and v. s. popov , phys .
a * 330 * , 1 ( 2004 ) .
a. l. troha et al . , phys .
e * 60 * , 926 ( 1999 ) .
_ r. weingartner , _ et al . _ ,
phys . rev .
beams * 15 * , 111302 ( 2012 ) .
_ a. i. akhiezer , v. f. boldyshev , and n. f. shulga , sov .
* 10 * , 19 ( 1979 ) ; a. i. akhiezer and n. f. shulga , phys . rep . * 234 * , 297 ( 1993 ) .
r. blankenbecler and s. d. drell , phys .
d * 36 * , 277 ( 1987 ) .
r. blankenbecler and r. l. sugar , phys .
d * 2 * , 3024 ( 1970 ) .
h. cheng and t. t. wu , phys .
* 22 * , 666 ( 1969 ) .
h. d. i. abarbanel and c. itzykson , phys .
* 23 * , 53 ( 1969 ) .
e. s. fradkin , d. m. gitman , and s. m. shvartsman , _ quantum electrodynamics with unstable vacuum _ , ( springer , berlin , 1991 ) . | the feasibility of obtaining exact analytical results in the realm of qed in the presence of a background electromagnetic field is almost exclusively limited to a few tractable cases , where the dirac equation in the corresponding background field can be solved analytically .
this circumstance has restricted , in particular , the theoretical analysis of qed processes in intense laser fields to within the plane - wave approximation even at those high intensities , achievable experimentally only by tightly focusing the laser energy in space . here , within the wentzel - kramers - brillouin ( wkb ) or eikonal approximation
, we construct analytically single - particle electron states in the presence of a background electromagnetic field of general space - time structure in the realistic assumption that the initial energy of the electron is the largest dynamical energy scale in the problem .
the relatively compact expression of these states opens , in particular , the possibility of investigating analytically strong - field qed processes in the presence of spatially focused laser beams , which is of particular relevance in view of the upcoming experimental campaigns in this field .
the predictions of qed have been confirmed with outstanding precision in numerous experiments .
the impressive agreement between the theoretical and the experimental value of the electron @xmath0-factor is customarily quoted as a prominent example @xcite .
however , the experimental scrutiny of qed becomes much less thorough when processes are involved , occurring in the presence of a strong background electromagnetic field , i.e. of the order of @xmath1 ( here @xmath2 and @xmath3 are the electron mass and charge , respectively ) @xcite .
the main reason is that these values largely exceed the field strengths available in laboratories . an important exception is represented by the electric field of highly - charged ions ( charge number @xmath4 , with @xmath5 ) at the typical qed length @xmath6 @xcite
indeed , numerous experiments on processes occurring in the presence of highly - charged ions @xcite have already successfully confirmed the predictions of qed .
correspondingly , advanced analytical methods @xcite , have been developed to interpret accurate experimental data beyond the exactly - solvable coulomb model of the ionic field .
modern high - power lasers represent an alternative source of intense electromagnetic fields structurally thoroughly different from atomic fields .
although the amplitude @xmath7 of the strongest laser pulse ever produced is about @xmath8 @xcite , it can be boosted to an effective strength @xmath9 in the rest - frame of ultrarelativistic particles colliding with the laser beam @xcite .
this principle has been exploited at slac to perform the so - far unique experimental campaign on strong - laser field qed @xcite , employing a laser with photon energy @xmath10 and amplitude @xmath11 , and an almost counter - propagating electron beam with energy of @xmath12 ( @xmath13 ) .
the relatively large pulse spot area ( @xmath14 ) allowed for the experimental results being well reproduced theoretically within the plane - wave field approximation .
approximating the laser field as a plane wave allows one to solve exactly the dirac equation in the resulting background electromagnetic field @xcite .
the corresponding electron single - particle states ( volkov states ) have been extensively employed to investigate different strong - field qed processes @xcite ( see also the recent reviews @xcite ) .
correspondingly , particle in cell ( pic ) codes including strong - field qed effects @xcite in the dynamics of laser - irradiated plasmas employ expressions of the qed rates calculated in the plane - wave ( local - constant - crossed - field ) approximation .
however , no analytical calculations in strong - field qed have been performed so far , which also include self - consistently the spatial focusing of the laser beam .
this is especially desirable as ultra - high intensities are attained nowadays by spatially focusing the laser energy almost down to the diffraction limit . in the present letter
we determine analytically the electron single - particle states in the presence of a strong background electromagnetic field of general space - time structure in the experimentally relevant case of an ultrarelativistic electron .
in the realistic assumption that the initial energy of the electron is the largest dynamical energy scale in the problem , we first determine the classical worldline of the electron and then we construct the corresponding quantum states in the wentzel - kramers - brillouin ( wkb ) or eikonal approximation @xcite .
the availability of such single - particle states and , in particular , their relatively compact expression open the possibility of investigating analytically and in a systematic way strong - field qed processes in the presence of intense background fields with complex space - time structure as , e.g. , those of tightly - focused laser beams . to date
, two methods have been developed to investigate qed processes in the presence of a virtually arbitrary background electromagnetic field .
however , the first one , based on the quasiclassical operator technique @xcite , allows one to obtain results only at the leading order in the quasiclassical , ultrarelativistic limit and does not contain a general prescription on how to calculate neither the amplitude of a generic qed process nor high - order corrections .
the second one , instead , employs the so - called `` trajectory - coherent - states '' ( tcs ) @xcite , which are relativistic electron wave - functions localized near the classical electron s trajectory .
however , the expression of the tcs is extremely cumbersome and of limited use for practical calculations .
we first consider the classical problem of an ultrarelativistic electron moving in a background electromagnetic field , described by the four - vector potential @xmath15 in the lorentz gauge @xmath16 ( here and below , units with @xmath17 are employed ) .
we work in the laboratory frame where the electron initial four - momentum is @xmath18 and we have in mind the case where the background electromagnetic field represents an intense , short , and tightly focused laser beam .
thus , we also assume that the field tensor @xmath19 is localized in space and time , that it has a maximum amplitude @xmath7 , and that it is characterized by a typical angular frequency @xmath20 , such that the classical nonlinearity parameter @xmath21 ( see @xcite for a manifestly covariant and gauge - invariant definition of the parameter @xmath22 ( see also @xcite ) ) satisfies the strong inequalities : @xmath23 .
the above assumptions well fit present and near - future experimental conditions envisaged to test strong - field qed with intense lasers .
in fact , even next - generation of @xmath24 ti : sa lasers @xcite ( central wavelength @xmath25 ) are realistically expected not to exceed a peak intensity of @xmath26 , corresponding to @xmath27 , @xmath28 , and @xmath29 .
such a field amplitude is effectively boosted to the critical value @xmath30 in the rest frame of an electron with an energy of @xmath31 , which is about six times @xmath32 .
in addition , electron beams with energies of about @xmath33 have been already demonstrated experimentally also with laser - plasma accelerators @xcite .
our starting point is the lorentz equation : @xmath34 , where @xmath35 is the electron four - momentum and @xmath36 is its proper time . according to the analytical solution of the lorentz equation in a plane - wave @xcite , the condition @xmath37 in the laboratory frame ensures that the electron will be only slightly deflected from its initial direction by the background field in the physically relevant situation where it is initially counterpropagating with respect to the laser field .
thus , rather than working with manifestly covariant equations it is convenient to introduce the light - cone coordinates @xmath38 , @xmath39 , and @xmath40 , with @xmath41 , and the quantities @xmath42 , @xmath43 , and @xmath44 , with @xmath45 ( in this respect , see also @xcite , where vacuum qed has been formulated by employing light - cone coordinates in the so - called `` infinite - momentum frame '' ) .
the quantities @xmath46 and @xmath47 introduced above are two unit - vectors perpendicular to @xmath48 and to each other , and such that @xmath49 .
an arbitrary four - vector @xmath50 can be expressed as : @xmath51 , where @xmath52 , @xmath53 , and @xmath54 ( note that @xmath55 ) . in the original light - cone notation @xcite , the direction @xmath48 was chosen as the `` third '' one , i.e. , @xmath56 .
however , for the sake of convenience in the use of the final results , we prefer to keep @xmath48 as an arbitrary unit vector .
the on - shell condition @xmath57 implies that @xmath58 and , in the physical situation of interest here , we require that the condition @xmath59 is satisfied in the laboratory frame , i.e. , that the quantity @xmath60 is the largest dynamical energy scale in the problem . by parametrizing the electron trajectory via the `` time '' @xmath61 , the three independent components of the lorentz equation can be written in the convenient form @xmath62 where the light - cone components of the field tensor have been expressed in terms of the electromagnetic field as @xmath63 , @xmath64 , @xmath65 and @xmath66 , with @xmath67 .
the idea now is to solve eqs .
( [ l_+])-([l_j ] ) iteratively by exploiting the appearance of different powers of the small quantity @xmath68 .
we assume that the light - cone components of @xmath69 have all the same order of magnitude and that the relative size of each term is determined by the power of the quantity @xmath70 .
if this is not the case , in fact , a careful analysis is required , as the mentioned hierarchy could be altered .
this is expected to occur more likely in the idealized case of highly symmetric fields .
for example , for a constant and uniform magnetic field perpendicular to @xmath48 , it is @xmath71 and the term proportional to @xmath70 in eq .
( [ l_j ] ) vanishes , unlike the one proportional to @xmath72 .
we also note that for a tightly focused gaussian beam counterpropagating with respect to the electron , it is @xmath73 @xcite .
now , the field components in eqs .
( [ l_+])-([l_j ] ) are calculated along the electron s trajectory . since @xmath74 , the following exact equations for the electron s `` spatial '' coordinates @xmath75 as functions of @xmath61 can be derived : @xmath76 we set the initial conditions at a given time @xmath77 as @xmath78 , with @xmath79 , and as @xmath80 ( recall that , by definition , @xmath81 and that the on - shell condition implies that @xmath82 ) .
we also assume that @xmath83 , with @xmath84 . by denoting the quantities calculated up to terms proportional to @xmath85 via the upper index @xmath86
, we have that @xmath87 where @xmath88 . by substituting the expression of @xmath89 in eqs .
( [ l_+])-([l_j ] ) and by integrating them , we obtain that @xmath90 , \end{split}\\ \label{u_j^1 } \begin{split } \bm{p}^{(1)}_{\perp}(\tau)=&\int_{\tau_0}^{\tau}d\tau'\bigg[e\bm{f}_p(\tau',\bm{r}_0)\\ & + \frac{e^2}{p_{0,+}}\bm{g}_p(\tau',\bm{r}_0)\cdot\bm{\nabla}_{\perp } \int_{\tau'}^{\tau } d\tau''\bm{f}_p(\tau'',\bm{r}_0)\\ & -\frac{e^2}{p_{0,+}}b_n(\tau',\bm{r}_0)(\bm{n}\times\bm{g}_p(\tau',\bm{r}_0))\bigg ] , \end{split}\\ \label{u_-^1 } p^{(1)}_-(\tau)=&\frac{m^2+e^2\bm{g}^2_p(\tau,\bm{r}_0)}{2p_{0,+}}.\end{aligned}\ ] ] the condition @xmath91 in the laboratory frame ensures that our approximated solution is accurate ( see eqs .
( [ l_+])-([l_j ] ) ) and it is fulfilled if @xmath92 . since tightly focused laser pulses are usually localized in a space ( time ) region of the order of a few laser central wavelengths ( periods ) , the above condition is equivalent in order of magnitude to the requirement @xmath37 in the relevant case of an electron initially counterpropagating with respect to the laser beam .
in order to highlight the qualitative novelties in the theoretical predictions brought about by the inclusion of the laser spatial focusing , in fig .
1 we plot the momentum @xmath93 in units of @xmath2 as a function of the quantity @xmath94 for an electron ( initial conditions @xmath95 and @xmath96 at @xmath97 ) initially counterpropagating with respect to a ti : sa , gaussian , @xmath98-pulse beam @xcite , linearly polarized along the @xmath99-direction with duration @xmath100 , spot radius @xmath101 , and peak intensity @xmath102 ( @xmath103 ) . in units of the electron mass as a function of @xmath104 for numerical values and details given in the text . ]
the continuous ( dashed ) line indicates the results of the numerical integration of the lorentz equation neglecting ( including ) the beam spatial focusing .
the dotted line , on top of the dashed one , corresponds to the analytical result from eq .
( [ u_j^1 ] ) ( discrepancies between the numerical and the analytical results arise at the third significant digit ) .
the final value of @xmath93 in the case of the gaussian beam is @xmath105 , whereas it vanishes within the plane - wave approximation ( see the inset in fig .
1 ) , according to the lawson - woodward theorem ( see , e.g. , @xcite ) . note that the corresponding divergence of @xmath106 is larger than already demonstrated electron beam divergences ( e.g. , of @xmath107 at a beam energy of @xmath108 @xcite ) .
in general , by setting @xmath109 in eqs .
( [ u_+^1])-([u_-^1 ] ) , the relation between the final four - momentum @xmath110 and the initial one @xmath111 can be obtained .
another qualitatively new feature brought about by the laser focusing in the above - mentioned physical setup is that the quantity @xmath112 is no more a constant of motion as in the plane - wave case and , indeed , the correction to the plane - wave result depends on the longitudinal electric field of the laser ( see eq .
( [ u_+^1 ] ) ) .
it is interesting to observe that by keeping only the leading - order term of each component of the four - momentum obtained above ( i.e. , by approximating @xmath113 , @xmath114 , and @xmath115/2p_{0,+}$ ] ) , the corresponding expression coincides with the exact four - momentum of an electron in a background plane - wave like field depending on @xmath61 and calculated at the initial coordinates @xmath116 .
this is in agreement with the classical result that an ultrarelativistic particle `` sees '' an arbitrary background field in its rest frame as a plane - wave - like field at leading order @xcite .
note that , although it might be more convenient to express the four - momentum obtained above in terms of the four - potential @xmath15 as in the plane - wave case , we prefer to employ the physical observable electromagnetic field .
we pass now to the quantum case and we consider the dirac equation @xmath117\psi=0 $ ] , where @xmath118 are the dirac matrices and @xmath119 is the bi - spinor electron wave function @xcite . based on the general argument that the de broglie length of an ultrarelativistic particle is very small , we apply the wkb method @xcite and look for a solution of the form @xmath120\varphi(x)$ ] , where @xmath121 turns out to be the classical electron action @xcite . for an electron with initial four - momentum @xmath111 and spin quantum number @xmath122 , the positive - energy wave function @xmath123 up to
the first order in @xmath85 reads ( see the supplemental material ( sm ) for a detailed derivation ) : @xmath124\}\bigg ] \frac{u_{p_0,\sigma_0}}{\sqrt{2\varepsilon_0 } } , \end{split}\ ] ] where @xmath125 , \end{split}\ ] ] where @xmath126 , where @xmath127 is the usual constant free bi - spinor @xcite , and where a unity quantization volume is assumed . in the sm it is shown that in the relevant case of a strong ( @xmath128 ) , tightly - focused ( @xmath129 ) , and short ( @xmath130 ) optical ( @xmath131 ) laser field
the only restrictive condition for the validity of the wave function @xmath123 is the classical one @xmath132 .
the wave function @xmath123 reduces to the one obtained in @xcite in the particular case of a background time - independent scalar potential ( see also @xcite ) . also , ultrarelativistic wave functions for scalar particles @xcite and two - particles scattering amplitudes @xcite have been derived in the context of high - energy scattering in qed in the leading - order eikonal approximation , which corresponds in our notation to neglect terms proportional to @xmath85 . however , keeping these terms is essential here , e.g. , to recover the plane - wave results .
in fact , if the background field is a plane - wave field depending on @xmath61 , the state in eq .
( [ psi_0 ] ) coincides within our approximations with the corresponding volkov state @xcite . in this respect ,
we note that the average spin @xmath133 ( see the discussion below eq .
( 11 ) in the sm ) in a volkov state describing an electron initially counterpropagating with respect to a linearly polarized plane wave never acquires a component along the magnetic field of the plane wave if it is initially along the electron momentum @xcite . whereas , by employing the wave function @xmath123 , this does occur in the case of a focused laser field . in the particular setup
mentioned below eq .
( [ u_-^1 ] ) , the average spin acquires an @xmath134-component @xmath135 $ ] , which vanishes identically in the corresponding plane - wave case . the negative - energy electron states @xmath136 can be obtained via the substitutions @xmath137 and @xmath138 in eq .
( [ psi_0 ] ) except that in @xmath139 , with the resulting quantity @xmath140 being the free negative - energy constant bi - spinor @xcite .
in-/out - states are obtained by performing the limit @xmath141 in eq .
( [ psi_0 ] ) and in the action @xmath142 , with the quantum numbers @xmath143 and @xmath122 corresponding to the asymptotic four - momentum and spin outside the field at @xmath141 .
once the single - particle positive- and negative - energy , in- and -out - states @xmath144 in ordinary coordinates have been determined ( the upper index @xmath86 from the single - particle states has been removed for the sake of notational simplicity ) , the matrix element @xmath145 of a typical process as nonlinear compton scattering can be calculated as ( see , e.g. , eq .
( 4.1.32 ) in @xcite ) @xmath146 here , the quantities @xmath147 and @xmath148 characterize the initial / final electron , whereas the emitted photon has four - momentum @xmath149 and polarization four - vector @xmath150 ( @xmath151 ) .
a semi - quantitative analysis of the matrix element @xmath145 already reveals new features in the focused - field case with respect to the plane - wave one ( and also to the locally constant - crossed - field one , which is relevant for pic codes ) .
first , unlike that in the plane - wave case , we can introduce here the concept of transverse formation region(s ) of radiation with respect to the laser propagation direction , analogous to the concept of `` impact parameter '' in , e.g. , electron - nucleus collision @xcite . in the quasiclassical limit , this can be physically understood as , unlike that in a plane wave , electron trajectories in a focused field differing only by the initial transverse position contribute in general with different phases to the radiation process .
now , eqs .
( [ psi_0])-([s^1 ] ) show that @xmath152u_{p_{i / f},\sigma_{i / f}}$ ] , with @xmath153 . in the quasiclassical ,
ultrarelativistic regime at @xmath128 , the matrix element @xmath145 can be evaluated approximately via the saddle - point method ( see , e.g. , @xcite ) . for any saddle point @xmath154 characterized by the conditions @xmath155 , with
@xmath156 $ ] , one can estimate the transverse formation regions @xmath157 , with @xmath158 , from the resulting quadratic term in @xmath159 in the exponent as @xmath160 , where all quantities are calculated at @xmath154 . moreover ,
contrary to a plane wave , a focused field can transfer momentum to the electron in principle along any direction .
the four - momentum transfer @xmath161 at each emission point @xmath154 from the field can be estimated from the relations @xmath162 and by employing the classical solution in eqs .
( [ u_+^1])-([u_-^1 ] ) .
finally , the focusing of the laser is expected to alter also the electron emission spectrum .
this can be already anticipated by estimating the `` instantaneous '' classical cut - off emission frequency @xmath163 , where @xmath164 is the curvature radius of the electron trajectory at the instant of emission @xcite . by calculating @xmath164 from eqs .
( [ u_+^1])-([u_-^1 ] ) , one estimates @xmath165 , with the second term inside the square root vanishing identically in the plane - wave case .
i am grateful to k. z. hatsagortsyan , d. v. karlovets , c. h. keitel , f. mackenroth , s. meuren , n. neitz , m. tamburini , and e. yakaboylu for useful discussions . |
* 1 . details on the experimental setup * the light source is a continuous - wave laser diode centered at 1560 nm which delivers 5 mw of power .
the two steps in power are generated by means of an electro - optic modulator ( eom , nir - mx - ln series , 20ghz of bandwidth , photline ) driven by an arbitrary waveform generator ( awg7000 , 50 ghz of bandwidth , tektronix ) .
the signal is pre - amplified in a semiconductor optical amplifier , the spontaneous emission in excess is removed by a spectral filter of 1 nm bandwidth ( full width half maximum ) , and further amplified with an erbium - doped fiber amplifier ( edfa ) .
the power can be tuned by means of a variable attenuator before being launched inside the optical fiber .
temporal traces are recorded with an optical sampling oscilloscope with a 1.6 ps resolution ( eyechecker , alnair ) .
overall , we obtain a clean step - like optical waveform with controllable power levels and durations , as well as a very short rise time ( @xmath104 rise time is @xmath66 ps ) , which is suitable to observe the dam - break phenomenon .
the repetition rate was fixed to 78 mhz that corresponds to a good trade off between a large enough value required by the sampling oscilloscope to get stable behavior and a sufficiently low value to obtain high peak power by using the edfa .
the distortions induced by the eom and the nonlinear response of the amplifiers are pre - compensated by finely adjusting the shape of the electrical signal delivered by the awg . as can be seen in fig .
s1 , in order to obtain the clean step - like signal at the fiber input ( as in fig .
2(a ) of the letter , and fig .
[ precompensation](d ) ) , it was necessary to drive the modulator with the signal shown in fig .
[ precompensation](c ) to avoid the unexpected overshoots at @xmath105 ps shown in fig .
[ precompensation](b ) , which result from driving the awg with the nearly ideal signal in fig .
[ precompensation](a ) .
w and different background fractions @xmath93 :
( a1,a2 ) r=0.15 ; ( b1,b2 ) 0.13 ; ( c1,c2 ) 0.11 ; ( d1,d2 ) 0.09 ; ( e1,e2 ) 0.07 ; ( f1,f2 ) 0.05 ; ( g1,g2 ) 0.03 ; ( h1,h2 ) 0.01 ; left column : experimental results .
right column : numerical simulations based on the nlse with linear losses .
the vertical dashed lines give the predicted delays ( gray and orange lines , edge of rw from eqs .
( 4 ) of the letter ; magenta and green lines , edges of dsw from eqs . ( 5 ) of the letter ) , respectively .
dashed - dot red lines is the numerical step - like input.,title="fig:",width=306 ] w and different background fractions @xmath93 : ( a1,a2 ) r=0.15 ; ( b1,b2 ) 0.13 ; ( c1,c2 ) 0.11 ; ( d1,d2 ) 0.09 ; ( e1,e2 ) 0.07 ; ( f1,f2 ) 0.05 ; ( g1,g2 ) 0.03 ; ( h1,h2 ) 0.01 ; left column : experimental results .
right column : numerical simulations based on the nlse with linear losses .
the vertical dashed lines give the predicted delays ( gray and orange lines , edge of rw from eqs .
( 4 ) of the letter ; magenta and green lines , edges of dsw from eqs . ( 5 ) of the letter ) , respectively .
dashed - dot red lines is the numerical step - like input.,title="fig:",width=309 ] , except that the data are collected with activated raman amplification , which allows to effectively compensate for the linear loss in the fiber.,title="fig:",width=306 ] , except that the data are collected with activated raman amplification , which allows to effectively compensate for the linear loss in the fiber.,title="fig:",width=306 ] an other crucial feature of the set - up is the loss compensation scheme . indeed , while the losses in the optical fiber are rather low ( 0.5 db / km ) , the long fiber length ( 15 km ) leads to 7.5 db ( @xmath106 ) of total attenuation .
this strongly affects the dynamics of the process as evident by contrasting fig .
[ no_compensation](a - h ) with fig .
[ with_compensation](a - h ) that depict the regimes with and without losses for different values of @xmath93 and the same input power @xmath83 w. the losses , besides substantially damping the output power ( compare vertical scales in figs .
[ no_compensation ] and [ with_compensation ] ) , determine a strong deformation of the characteristic plateau that separates the rw and the dsw , which is an essential trait of the fluid - like dam - break scenario . in order to compensate this deviation from the ideal fluid behavior ,
we compensate the fiber losses by taking advantage of raman amplification as employed in telecommunication systems @xcite . the aim is to get an effectively transparent fiber .
we used a raman pump of 2 w maximum power centered at 1482 nm ( @xmath107 thz detuning from the signal , which guarantees a large raman gain @xcite ) , which is launched and extracted by means of multiplexers at the fiber input and output . the pump counter - propagates with respect to the signal in order to minimise the pump to signal relative intensity noise transfer @xcite .
we carefully adjusted the raman pump power by looking at the output traces until we achieve both a flat plateau between the dsw and the rw and a peak power plateau that equals the input level .
the validity of the scheme is clear by comparing the left columns in fig .
[ no_compensation ] ( raman pump off ) with those in fig .
[ with_compensation ] ( raman pump on , loss compensated ) , and from the perfect agreement that we obtain with numerical simulations ( right columns ) based on the nlse with included linear loss in fig .
[ no_compensation ] ( a2-h2 ) and the pure ( conservative ) nlse in fig .
[ with_compensation ] ( a2-h2 ) . to summarise
, our fiber system accurately mimics the dispersive hydrodynamics described by the conservative nlse . : ( a ) 0.15 ; ( b ) 0.13 ; ( c ) 0.11 ( cavitation threshold ) ; ( d ) 0.09 ; ( e ) 0.07 ; ( f ) 0.05 ; ( g ) 0.03 ; ( h ) 0.01 . left column : experimental results ; right column : numerical simulations ( solid - black line is the output , dashed red line is the input).,title="fig:",width=302 ] : ( a ) 0.15 ; ( b ) 0.13 ; ( c ) 0.11 ( cavitation threshold ) ; ( d ) 0.09 ; ( e ) 0.07 ; ( f ) 0.05 ; ( g ) 0.03 ; ( h ) 0.01 .
left column : experimental results ; right column : numerical simulations ( solid - black line is the output , dashed red line is the input).,title="fig:",width=306 ] we report in fig .
[ vacuum_points ] the full sequence of close - up measurements ( with corresponding simulations ) of the dsw bottom for different values of @xmath93 ( fig .
4 in the letter contains a subset of such measurements ) . above threshold
@xmath108 the dsw envelope is monotone and no vacuum appears ( fig .
[ vacuum_points](a - b ) ) .
conversely , below threshold , a self - cavitating state ( vacuum ) @xmath109 appears , as highlighted by red circles in fig .
[ vacuum_points](c - e ) .
the large extension of the dsw due to the high power regime ( @xmath59 w ) allows us to observe with extremely good resolution the shift of the vacuum towards the leading ( linear ) edge of the dsw as @xmath78 is reduced .
a very good agreement is also achieved with numerical simulations .
all the simulations based on the nlse are made by means of the split - step method .
the elevating step from @xmath8 to @xmath67 is modelled by the input field @xmath110^{1/2}$ ] , where @xmath111 ps arises from the fit with the optical input trace at any run , and is due to the rise - time of the modulator .
similarly , for the decreasing step on the trailing ( dry - bed " ) edge we use @xmath112^{1/2}$ ] with @xmath113 ps due to slightly longer falling time of the modulator .
obvious combinations of the steps allow for modelling the full waveform as in fig .
[ precompensation](d ) .
we display in fig .
[ phase_intensity](a ) and ( b ) an example of the full evolution of the intensity and phase along the fiber , for @xmath114 , which corresponds to fig .
[ vacuum_points](f1,f2 ) .
the dot - dashed oblique lines in fig .
[ phase_intensity](a1 ) stand for the edges of the dsw and rw from eqs .
( 4 - 5 ) of the letter . from fig .
[ phase_intensity](a1 ) and fig .
[ phase_intensity](a2 ) reporting the output power profile , we clearly see the expanding plateau between the dsw and the rw .
we also notice that the top of the pulse ( peak power of 1w ) is eroded by the two rws ( the left and right rw are due to the wet - bed " and dry - bed " dynamics on the input leading and trailing edges , respectively ) , without any appreciable interactions between them ( this witnesses that the two scenarios do not mix up over the finite fiber length ) .
a close - up of the bottom of the shock fan is shown in the inset of fig .
[ phase_intensity](a2 ) . around the vacuum point ( @xmath115 ps )
, the numerics reveals a @xmath92 phase jump as shown in the inset of fig .
[ phase_intensity](b2 ) , consistently with the local nature of black solitons . across the vacuum
the chirp ( phase derivative ) @xmath25 exhibits strong peaks of different sign ( see fig .
[ phase_intensity](c ) ) .
unfortunately such fast phase features are not measurable with the present set - up and will be investigated in the future by means of phase reconstruction techniques .
level color plot of power ( a1 ) and phase ( b1 ) evolution along the fiber .
power ( a2 ) and phase ( b2 ) profiles at the fiber output . dashed - dot lines in ( a1 ) correspond to the analytical predictions for the dsw and rw edges , respectively [ eqs .
( 4 - 5 ) in the letter ] .
the insets of ( a2 ) and ( b2 ) zoom on the dsw bottom to emphasise the self - cavitating state ( null power ) and the corresponding phase jump of @xmath92 .
( c ) output frequency chirp @xmath25 , with close - up showing the change of sign of the chirp . ,
title="fig:",width=332 ] ( f2 ) .
level color plot of power ( a1 ) and phase ( b1 ) evolution along the fiber .
power ( a2 ) and phase ( b2 ) profiles at the fiber output . dashed - dot lines in ( a1 ) correspond to the analytical predictions for the dsw and rw edges , respectively [ eqs .
( 4 - 5 ) in the letter ] .
the insets of ( a2 ) and ( b2 ) zoom on the dsw bottom to emphasise the self - cavitating state ( null power ) and the corresponding phase jump of @xmath92 .
( c ) output frequency chirp @xmath25 , with close - up showing the change of sign of the chirp .
, title="fig:",width=332 ] since the shallow water equations ( swes ) are a two - dimensional vector conservation law , the dynamics of the general riemann initial value problem entails the superposition of a wave pair , where each of the wave can be of rarefaction wave ( rw ) type or classical shock wave ( sw ) type
. a classification into ( i ) rw - rw ; ( ii ) sw - sw ; ( iii ) rw - sw , can be made according to the values of the step in @xmath116 and @xmath25 .
such classification can be suitably extended for the dispersive problem ruled by the nlse @xcite .
the dam - break riemann problem corresponds to a step in the variable @xmath117 with @xmath118 , and leads to a rw - sw pair separated by a plateau with constant @xmath37 . for an increasing step ( @xmath119 ) , as in the experiment , the sw is left - going , heading towards @xmath31 ( whereas , it would be right - going for a decreasing step @xmath120 ) .
the first solution to the wet - bed " problem ( @xmath121 ) in the swes was reported by stoker @xcite .
essentially one can start from the rankine - hugoniot conditions for the sw , and using also the constancy of @xmath122 across the rw , one can eliminate @xmath91 to achieve an equation to be solved for @xmath123 .
then , returning to the starting equations , one can specify all the quantities of interest ( @xmath91 , sw velocity , edges of rw ) as a function of the quiescent states @xmath124 and @xmath38 . +
a slightly different formulation , which is directly suited for the extension to the dispersive case ruled by the full nlse , amounts to solve in terms of so called simple waves .
the starting point are the swes cast in diagonal form in terms of riemann invariants @xmath125 , @xmath126 where the eigenvelocities are @xmath127 . the dam - break problem with @xmath128
implies an increasing [ decreasing ] step for @xmath129 [ @xmath130 , as displayed in fig .
[ s6 ] ( steps from @xmath131 to @xmath132 , and from @xmath133 to @xmath134 , respectively ) . for this case ,
characterised by @xmath135 , we sketch in fig .
[ s6 ] the simple wave solutions of eqs .
( [ diag ] ) , such that one of the riemann invariant is constant and other one varies in self - similar way , i.e. @xmath136 . the solution involves , for @xmath137 , a constant @xmath138 and a rw for @xmath139 described by @xmath140 , whereas for @xmath141 , @xmath142 is constant , and @xmath143 describes the characteristic overturning ( see fig . [ s6 ] ) , which allows to introduce the sw through the rankine - hugoniot condition ( cyan step in fig .
[ s6 ] , see also fig . 1 of the letter ) .
the upper state of the shock is the plateau in @xmath144 , which is found by exploiting continuity in @xmath145 to be @xmath146 and @xmath147 ^ 2=(\sqrt{\rho_l}+\sqrt{\rho_r})^2/4 $ ] .
the approriate limits of the solution give the relevant values of the self - similar variable @xmath148 , @xmath149 shown in fig .
[ s6 ] , as follows : @xmath150 where we have exploited the fact that @xmath151 . it is important to emphasise that @xmath152 and @xmath153 in eqs .
( [ rw1]-[rw2 ] ) represent the edge velocities of the rw , which hold valid , due to the rw smoothness , also when dispersive corrections are accounted for .
conversely , when dispersive correction are applied to the sw , one needs to calculate the edge velocities by averaging over the dsw oscillation as explained below .
we anticipate that this gives a result markedly different from the expression of @xmath88 and @xmath154 in eqs .
( [ sw1]-[sw2 ] ) . in the presence of an undular shock or dsw
, the modulation theory based on whitham averaging can be exploited to describe the dsw as a modulated cn - oidal wave with parameters that varies slowly with respect to its natural periodicity .
the averaged whitham equations describe the slow variation of the parameters and allow to compute the velocities of the linear and soliton edge of the dsw . for the details on the method
, we refer the reader to the original literature @xcite and to the excellent reviews on the topic , in particular to refs .
below , we only briefly outline the steps that lead us to obtain eqs .
( 4 - 6 ) in our letter .
the whitham equations for the nlse can be expressed in the following diagonal form , first derived by pavlov @xcite , by introducing four riemann invariants @xmath155 , @xmath149 , ordered in such a way that @xmath156 , which are a suitable combination of the parameters of the cn - oidal wave @xcite @xmath157 here the velocities @xmath158 depends on combinations of @xmath159 and elliptic integrals of first ( @xmath160 ) and second ( @xmath161 ) kind . for instance the velocity @xmath162 , which will be relevant in the following , reads explicitly as @xmath163^{-1},\ ] ] where @xmath164 and @xmath165 . +
as shown above a left - going classical shock of the swes is characterised by a constant @xmath139 and a jump in @xmath166 from @xmath167 to @xmath168 ( with given left and right boundaries @xmath169 such that @xmath170 ) .
the corresponding dispersive regularisation of the shock , i.e. the dsw , can be described by a solution of eqs .
( [ weqs ] ) where only one of the riemann invariants varies . for the left - going dsw
, this solution is generated by the initial value @xmath171 with @xmath172 , @xmath173 , @xmath174 , and a step @xmath175 in @xmath176 from @xmath168 to @xmath167 ( obtained from the so - called data regularisation procedure , see @xcite ) .
the solution is such that @xmath177 remain constant at any @xmath178 , whereas @xmath179 gives rise to a smooth rw ( owing to the fact that @xmath175 is non - decreasing ) that depends on the self - similar variable @xmath30 .
indeed all whitham equations are formally satisfied when @xmath180 and @xmath181 , provided the equation @xmath182 is fulfilled . for @xmath183 ,
this implies @xmath184 .
the latter relation , once @xmath185 is expressed as @xmath186 according to eq .
( [ v2 ] ) , becomes a nonlinear equation in the only unknown @xmath187 , which can be solved to find the rw .
the velocity @xmath188 and @xmath189 of the leading and trailing edges of the dsw correspond to the edges of this rarefaction wave and can be calculated from the following limits ( see fig .
[ s7 ] ) @xmath190 to obtain final formulas that depend only on the quiescent states @xmath191 , we observe that , across the shock , one has a constant @xmath142 , as clear from fig . [ s6 ] . ) , where @xmath192 stay constant and @xmath193 varies smoothly in the range @xmath194 between the values @xmath195 and @xmath196 .
@xmath154 and @xmath88 correspond to the linear and soliton edges of the dsw , respectively .
the dashed red lines are the riemann invariants in the dispersionless limit.,width=302 ] moreover , since the minimum power of the cn - oidal wave turns out to be @xmath197 @xcite , the existence in the dsw of a cavitating or vacuum state @xmath198 requires @xmath199 .
therefore , in analogy with eqs .
( [ dsw1]-[dsw2 ] ) , we obtain the self - similar location of the vacuum by performing the following limit @xmath200^{-1 } \nonumber\end{aligned}\ ] ] where @xmath201 . finally , we obtain eqs .
( 5 ) and ( 6 ) in the letter by substituting in eqs .
( [ dsw1]-[dsw2 ] ) and ( [ vacuum ] ) , respectively , the boundary value of riemann invariants expressed in terms of @xmath116 , i.e. @xmath202 . g. a. el and m. a. hoefer , physica d * 333 * , 11 ( 2016 ) .
t. grava , in _ rogue and shock waves in nonlinear dispersive media _ , m. onorato , s. residori , f. baronio , eds . ,
lecture notes in physics ( springer , berlin , 2016 ) . a.m. kamchatnov , _ nonlinear periodic waves and their modulations - an introductory course _ , ( world scientific , singapore , 2000 )
. m. a. hoefer , m. j. ablowitz , i. coddington , e. a. cornell , p. engels , and v. schweikhard , phys .
a * 74 * , 023623 ( 2006 ) .
p. a. p. janantha , m. a. hoefer , and m. wu , arxiv:1610.08846 ( 2016 ) . | we investigate the temporal photonic analogue of the dam - break phenomenon for shallow water by exploiting a fiber optics setup .
we clearly observe the decay of the step - like input ( photonic dam ) into a pair of oppositely propagating rarefaction wave and dispersive shock wave .
our results show evidence for a critical transition of the dispersive shock into a self - cavitating state .
the detailed observation of the cavitating state dynamics allows for a fully quantitative test of the whitham modulation theory applied to the universal defocusing nonlinear schrdinger equation . _
introduction. _ the laser light propagating in nonlinear media often behaves as a photon fluid @xcite , sharing phenomena that characterise fluid flows such as rogue waves @xcite , instabilities @xcite , transition to turbulence @xcite , coherent @xcite and incoherent @xcite shock waves , and superfluid flow around obstacles @xcite . in regimes described by the defocusing nonlinear schrdinger equation ( nlse ) ,
a distinctive trait of the photon fluid evolution is the formation of dispersive shock waves ( dsws , or undular bores ) @xcite , fast oscillating wavetrains that spontaneously emerge from the tendency to develop a gradient catastrophe @xcite .
dsws are ubiquitous , being observed in other systems ruled by the nlse such as cold atom condensates @xcite and spin waves @xcite , as well as in other dispersive hydrodynamic settings involving , e.g. electrons @xcite , water waves @xcite and viscous fluid conduits @xcite .
a major breakthrough in the analytical description of the dsw is the whitham modulation theory @xcite , which , however , assumes the dsw to develop from step - like initial conditions ( the riemann problem ) , whereas experiments to date have been mostly concerned with smooth or periodic initial conditions
. therefore experiments devoted to investigate the dispersive riemann problem are of paramount importance for advancing the understanding of dispersive hydrodynamic flows , a fortiori for systems such as the nlse where the modulation theory predict critical transitions in the behavior of the shock @xcite .
+ in this letter , we exploit a fiber optics set - up to investigate experimentally the riemann problem associated with an initial step , in the temporal domain , in the optical power . in the absence of any frequency chirp across the jump such
a problem is isomorphic to the classic 1d dam - break problem of hydrodynamics @xcite .
indeed , we demonstrate that the light evolves as a fluid mimicking the basic features of the dam - break in shallow water , namely the decay into a shock and a rarefaction - wave ( rw ) pair , connected by an expanding plateau .
the dispersive character of the shock , however , leads to a critical transition , which is predicted in the framework of whitham theory @xcite . above a critical height of the jump , we report evidence for the onset of self - cavitation , i.e. the appearance of a null point in the optical power .
the full experimental characterisation of the cavitating state allows , for the first time , for a quantitative comparison with modulation theory .
_ theory of dispersive dam - break. _ the dam - break riemann problem that we investigate is described by the nlse that rules the evolution of the temporal envelope field @xmath0 along the fiber distance @xmath1 @xmath2 subject to a step initial condition in @xmath3 , i.e. @xmath4 for @xmath5 , and @xmath6 for @xmath7 , where the constant left and right power levels @xmath8 and @xmath9 define the bottom and the top of the optical dam in @xmath3 ( see fig . [ fig1](a ) ) . here
@xmath10 is the retarded time in a frame moving with group - velocity @xmath11 , whereas @xmath12 ps@xmath13/km and @xmath14 ( w km)@xmath15 are the dispersion and the nonlinear coefficient of our fiber , @xmath16 being the carrier frequency .
@xmath17 corresponds to the defocusing regime of the nlse .
the madelung transform @xmath18 allows to formulate the nlse in hydrodynamical form : @xmath19_t , \label{swe2}\end{aligned}\ ] ] where we set @xmath20 , @xmath21 with @xmath22 and @xmath23 . by neglecting the rhs containing higher - order derivatives ( quantum pressure term @xcite ) , eqs .
( [ swe1]-[swe2 ] ) constitute the dispersionless vector conservation law known as the shallow water equations ( swes ) @xcite .
the role of local water depth and longitudinal velocity are played , here , by the normalized power @xmath24 and chirp @xmath25 , whereas space and time have interchanged role .
the evolution of an initial ( @xmath26 ) step elevation from @xmath27 to @xmath28 @xcite , with @xmath29 identically vanishing , is the classic dam - break riemann problem .
the solution of the swes , first given by stoker @xcite , can be formulated in terms of the self - similar variable @xmath30 @xcite and involves a classical shock wave and a rw . as shown by the black solid line in fig .
[ fig1](a ) , the shock and the rw propagate in opposite directions ( towards @xmath31 and @xmath32 , respectively , or down- and up - stream directions in the hydrodynamic problem ) , being connected by an expanding plateau characterised by intermediate constant values @xmath33 , @xmath34 .
the step values @xmath27 and @xmath35 also fix the edge velocities of the smooth rw to the values : @xmath36 conversely , the jump from @xmath37 to @xmath38 constitutes a classical shock which moves with velocity @xmath39 , deriving from the well - known rankine - hugoniot condition @xcite .
however , such shock is regularized into an oscillating dsw by the effect of the rhs of eq .
( [ swe2 ] ) that stems from dispersion . a snapshot of the breaking scenario obtained from the full nlse
is compared with the dispersionless limit ( swes ) in fig . [ fig1](a ) .
the dsw is delimited by two edge velocities
@xmath40 ( @xmath41 ) , where the oscillations vanish ( linear edge ) or become deepest ( soliton edge ) , respectively @xcite . according to whitham modulation theory such velocities reads as @xcite @xmath42 whereas , in the same framework , owing to the rw smoothness , one recovers eqs .
( [ rw ] ) for the rw edges .
the velocities @xmath40 in eqs . ( [ dsw ] ) and @xmath43 in eqs .
( [ rw ] ) define the wedges where the dsw and rw expand , as shown ( in dimensional units , i.e. including the multiplicative factor @xmath44 for our fiber ) in fig .
[ fig1](b ) .
importantly , the modulation theory entails a crossover between two different regimes separated by a critical condition . in the first regime ,
the dsw envelopes are monotone and the dsw power never vanishes , as shown in fig .
[ fig1](a ) . however , below a critical value of the key parameter , namely the ratio between the quiescent ( down- and up - stream ) states , that we henceforth denote as @xmath45 , the dsw exhibits a self - cavitating point ( i.e. , zero power , corresponding to a vacuum point in gas dynamics ) , which we found at @xcite @xmath46^{-1},\ ] ] where @xmath47 and @xmath48 are elliptic integrals of the first and second kind , respectively , of modulus @xmath49 .
the threshold for cavitation can be obtained by imposing @xmath50 ( cavitation on soliton edge of the dsw ) , or equivalently @xmath51 , from which we obtain @xmath52 for @xmath53 below such threshold , a vacuum point always exists , which tends towards the linear edge of the dsw in the limit @xmath54 . in this limit , however , both the oscillation amplitude and the plateau extension tends to vanish @xcite , and one recovers the limiting hydrodynamic case known as dry - bed " dam - break , characterised by a single rw extending to zero and no shock @xcite .
^{1/2}$ ] ( dashed red ) with the ideal dispersionless solution of the swes ( solid black ) .
the oscillations over the top of the rw are due to the gibbs phenomenon @xcite .
( b ) wedges in @xmath55 plane corresponding to the rw ( fable orange ) , the dsw ( green ) and the plateau in between .
the boundaries correspond to slopes @xmath56 , @xmath57 . here
@xmath58 ps@xmath13/km and @xmath14 ( w km)@xmath15 , @xmath59 w , @xmath60 mw , @xmath61 ps . , width=313 ] _ experiment. _ we performed a series of experiments to provide evidence for the decay of a photonic dam into the dsw - rw pair and for the critical transition to self - cavitation . in our experimental setup
( see @xcite ) , we make use of a continuous laser diode source emitting at @xmath62 nm , which is intensity modulated by an electro - optic modulator driven by an arbitrary waveform generator , with typical rise time @xmath63 ps ( raised tanh shape , rising from @xmath64 to @xmath65 in @xmath66 ps ) and 78 mhz repetition rate .
the signal is pre - amplified in a semiconductor optical amplifier , and then it passes through a spectral filter to remove amplified spontaneous emission in excess , and is finally amplified by means of an erbium - doped fiber amplifier .
this signal is launched in a dispersion compensating fiber ( parameters as in the caption of fig .
[ fig1 ] ) and analysed in the time domain with an optical sampling oscilloscope with 1.6 ps resolution . a typical shape of the step - like signal that we obtained is shown in fig .
[ fig2](a ) .
two elevating steps in power from zero to @xmath8 and from @xmath8 to @xmath67 , can be adjusted independently , and are followed by a descending step ( trailing edge ) back to zero .
the advantage of using this specific signal shape is that , in the same experimental run , we can compare ( i ) the dsw - rw dynamics developing around the jump from @xmath8 to @xmath67 , with ( ii ) the dry - bed " dynamics developing over the descending step .
importantly , the duration of each of the constant power states @xmath68 was adjusted to be long , up to @xmath69 ns , so that the dsw - rw pair can develop without feeling the interaction with the first step and the trailing edge of the waveform .
clearly , as can be seen in fig .
[ fig2](a ) , the main step ( @xmath8 to @xmath67 ) is not instantaneous .
however , the short rising time allows us to clearly observe the dam - break phenomenon , as we will see below .
another challenging issue faced in the experiment is the loss compensation .
indeed , the dsw - rw dynamics is very sensitive to losses and even weak losses of optical fibers ( 0.5 db / km ) are strongly detrimental .
for instance , the plateau that connects the rw and the dsw would be completely distorted , not allowing for a quantitative comparison with theory ( see figs .
s2 and s3 in @xcite ) . inspired from transparent telecommunication networks
, we counterbalanced linear losses by means of raman amplification @xcite .
to this end , a counterpropagating beam was injected in the fiber at @xmath70 nm . in this way
, we achieve significant ( close to peak ) raman gain with weak relative noise intensity transfer @xcite . ) ) and the dsw ( magenta and green lines , from eqs .
( [ dsw ] ) ) , respectively .
, width=313 ] figure [ fig2](b ) shows the output temporal trace obtained for the input step - like signal with @xmath71 w , @xmath72 mw ( @xmath73 ) .
as can be seen , the input optical dam at @xmath3 breaks into a dsw - rw pair .
the dsw is characterized by fast oscillations with @xmath74 ps average period ( the period scales proportionally to @xmath75 ) .
the rw smoothly connects the plateau with power @xmath76 w ( in agreement with the theoretical prediction @xmath77 w ) to the peak level @xmath67 .
conversely , over the trailing edge , no shock occurs and a smooth rw dropping to zero is observed , consistently with the case of dry - bed " dam - break .
overall , the data show an excellent agreement with the numerical simulations reported in fig .
[ fig2](d ) based on the nlse ( [ nls ] ) .
we emphasise that the occurrence of the dsw - rw pair is related to the non - zero background and not to the character ( ascending vs. descending ) of the step .
indeed the dsw - rw pair is observed on the trailing edge for a mirror - symmetric input . and
different background fractions @xmath78 : ( a1,a2 ) @xmath79 ; ( b1,b2 ) @xmath80 ; ( c1,c2 ) @xmath81 ; ( d1,d2 ) @xmath82 . left column : experimental results .
right column : numerical simulation based on the nlse.,width=8 ] we have then proceeded to investigate in detail the breaking dynamics of the step - like input into the dsw - rw pair for different heights of the optical dam . the results shown in fig .
[ fig3 ] , are obtained for a fixed peak power @xmath83 w ( slightly larger than that of fig . [ fig2 ] ) , and a variable background @xmath8 , i.e. a variable ratio @xmath78 . for a quantitative comparison with whitham modulation theory we also report , as vertical dashed lines superimposed on the measured data , the delays @xmath84 , @xmath85 corresponding to the edges of the rws in eq .
( [ rw ] ) and the dsw in eq .
( [ dsw ] ) , respectively .
we remark that : ( i ) the experimental traces ( left column ) show a very good agreement with both simulations based on the nlse ( right column ) , and the predicted delays @xmath86 from modulation theory ; ( ii ) the duration of the plateau connecting the dsw and the elevating rw significantly shrinks as the ratio @xmath78 decreases ( from top to bottom in fig .
[ fig3 ] ) , again in quantitative agreement with modulation theory which predicts that @xmath87 reduces when @xmath78 decreases ; ( iii ) for relatively large ratios @xmath78 the dsw never touches zero and the edge @xmath88 of the dsw ( where @xmath50 ) is constituted by a gray soliton @xcite , as shown in fig . [
fig3](a1-a2 ) .
however , by decreasing @xmath78 , we observe the onset of cavitation when the soliton at the trailing edge of the dsw becomes black ( see fig .
[ fig3 ] ( b1,b2 ) , @xmath89 ) .
the observed threshold value @xmath90 is in excellent agreement with the theoretical prediction [ eq . ( [ th ] ) ] .
while the black soliton possess a zero velocity ( with respect to its background ) , the dsw edge maintains a finite velocity due to the chirped background ( velocity @xmath91 ) .
also associated with such black soliton we expect a phase jump of @xmath92 ( see fig .
s5(b ) for more details ) which , however , can not be measured with our set - up . at even lower ratios
@xmath78 , the vacuum point shifts towards the left and the dsw envelope becomes non - monotone .
, showing a zoom around the spontaneously cavitating state ( vacuum point ) of the dsw , for @xmath59 w and different fractions @xmath93 : ( a1,a2 ) @xmath79 ; ( b1,b2 ) @xmath80 ; ( c1,c2 ) @xmath94 ; ( d1,d2 ) @xmath81 .
( e ) measured delay of the cavitating state vs. @xmath78 , compared with prediction from eq .
( [ vacuum ] ) and numerical simulations of the nlse .
the dashed vertical line stands for the threshold in eq .
( [ th ] ) .
inset : overall rw - dsw dynamics of the case @xmath78 = 0.05.,title="fig:",width=313 ] , showing a zoom around the spontaneously cavitating state ( vacuum point ) of the dsw , for @xmath59 w and different fractions @xmath93 : ( a1,a2 ) @xmath79 ; ( b1,b2 ) @xmath80 ; ( c1,c2 ) @xmath94 ; ( d1,d2 ) @xmath81 .
( e ) measured delay of the cavitating state vs. @xmath78 , compared with prediction from eq .
( [ vacuum ] ) and numerical simulations of the nlse .
the dashed vertical line stands for the threshold in eq .
( [ th ] ) .
inset : overall rw - dsw dynamics of the case @xmath78 = 0.05.,title="fig:",width=313 ] while the results in fig .
[ fig3 ] already show the crossover to the cavitating state , a detailed quantitative study of this regime requires to operate with a dsw possessing larger extension and shorter average period . to this end , we operate at the maximum available power in our set - up , @xmath59 w. this allows us to observe a dsw exhibiting several oscillations with an average period of @xmath95 ps , spanning a range that exceeds @xmath96 ps , as shown in the inset of fig .
[ fig4](e ) , for @xmath97 mw or @xmath98 .
importantly , in this regime the delay of the cavitating state ( zero power ) can be accurately identified within the dsw .
figures [ fig4](a - d ) display a zoom over the bottom part of the dsw in order to show how the cavitating state moves when the ratio @xmath78 is varied across the threshold . as shown in fig .
[ fig4](a1-a2 ) , when @xmath99 , the dsw still exhibits monotone envelopes featuring a gray soliton edge with non - vanishing dip . however , fig .
[ fig4](b1-b2 ) show very clearly the onset of cavitation ( black soliton edge ) at the threshold @xmath90 , in agreement with eq .
( [ th ] ) .
decreasing further @xmath78 ( i.e. , for higher dam heights ) leads the cavitating state to acquire increasingly negative delays , shifting progressively towards the linear edge of the dsw , as clear from fig .
[ fig4](c1-c2 ) for @xmath100 and fig .
[ fig4](d1-d2 ) for @xmath101 .
for all cases the numerical simulations are still in good agreement with the measured profiles .
we summarise in fig . [
fig4](e ) the delays of the vacuum state extracted from the measured temporal traces for different ratios @xmath78 ( see also @xcite ) in the range @xmath102 ( below @xmath103 , the residual noise makes impossible to resolve the delay of the vacuum which is quite close to the linear edge ) .
the data are contrasted with the theoretical prediction from eq .
( [ vacuum ] ) , showing a satisfactory agreement in the whole range .
we ascribe the discrepancies to the finite rise - time of the step and to the asymptotic character of whitham theory , which is expected to become more accurate as the propagation length increases and/or the dispersion decreases @xcite . in summary
, we have reported the fluid behavior of light in a dispersive dam - break experiment , revealing a transition to cavitation in close quantitative agreement with the predictions of modulation theory .
such behavior is expected to be universal for systems ruled by the nlse , while qualitatively differing from other dispersive breaking scenarios observed in fluids @xcite .
our platform could be further used to explore other critical behaviors in the general dispersive riemann problem @xcite , including the focusing case @xcite .
the present research was supported by ircica ( usr 3380 cnrs ) , by the agence nationale de la recherche in the framework of the labex cempi ( anr-11-labx-0007 - 01 ) , equipex flux ( anr-11-eqpx-0017 ) , by the projects noawe ( anr-14-achn-0014 ) , topwave ( anr-13-js04 - 0004 ) , and the ministry of higher education and research , hauts de france council and european regional development fund ( erdf ) through the contrat de projets etat - region ( cper photonics for society p4s ) .
s.t . acknowledges also the grant prin 2012bfnwz2 .
the authors are grateful to l. bigot , e. andresen and ircica - tektronix european optical and wireless innovation laboratory for technical support about the electronic devices . |
a contentious issue in x - ray astronomy , and more generally in the study of galaxy formation and evolution , is whether massive galaxies , groups , and clusters possess inhomogeneous cooling flows wherein large quantities of gas have cooled and dropped out of the hot phase resulting in a spatially extended multiphase medium ( e.g. , fabian , nulsen , & canizares 1984 ; fabian 1994 ; binney 1996 ) . in the era before _
chandra _ and _ xmm _ the instrument with the best combined spatial resolution , bandwidth , and energy resolution for studies of cooling flows was the _ rosat _ pspc . the bandwidth ( 0.1 - 2.4 kev ) of the pspc is especially critical because its sensitivity to energies both above and below the edge ( 0.532 kev ) has allowed us recently to detect intrinsic oxygen absorption in cooling flow galaxies and groups which provides the first direct evidence of a spatially extended multiphase medium in these systems ( buote 2000b , hereafter paper1 ; buote 2000d , hereafter paper3 ) .
m87 is arguably the best target for study of its hot plasma with _ rosat _ because it is the brightest nearby galaxy , group , or cluster that also possesses an ambient gas temperature ( @xmath0 kev ) that lies within the bandpass of the pspc . for these reasons
m87 has been the subject of numerous x - ray studies .
the early analysis by @xcite of the _ einstein _ fpcs spectral data of m87 provided the first evidence of a multiphase hot ism in a galaxy cluster , though this claim was disputed by @xcite using data from the other _ einstein _
detectors with lower energy resolution .
@xcite discovered intrinsic absorption in the _ einstein _ sss data of m87 which was later confirmed with _
asca _ @xcite . in these studies
the absorber is assumed to be cold gas with solar abundances similar to the intervening cold gas in the milky way .
the useful bandpasses of the _ einstein _ sss and _ asca _ sis extend down to @xmath3 kev , and thus the _ rosat _ pspc whose bandpass extends down to 0.1 kev should be even more sensitive to soft x - ray absorption from cold gas ( or dust ) . however , in the first detailed spectral analysis of the pspc data of m87 @xcite mention in passing that when the hydrogen column density is treated as a free parameter they obtain values _
less _ than galactic .
although not emphasized by @xcite this result is inconsistent with significant absorption from cold gas .
a later _ rosat _ analysis by @xcite restricted consideration only to energies above 0.4 kev and also finds no evidence for absorption from cold gas in the pspc data of m87 .
but a subsequent re - analysis by these authors now finds evidence for such absorption ( sanders , fabian , & allen 2000 ) .
the confused status of intrinsic x - ray absorption and multiphase gas in m87 warrants re - examination .
the possible connection between the cooling flow and agn in m87 has also received considerable attention recently .
it has been known for some time ( e.g. , bhringer 1997 ) that the x - ray isophotes are somewhat distorted within @xmath4 kpc of the center of m87 , and these distortions appear to correlate with the radio emission . a new deep radio image obtained by @xcite shows that the radio emission fills the central @xmath4 kpc region .
the spatial correlation between radio and x - ray emission in m87 has been interpreted as evidence that the agn has refashioned the hot ism and stifled the cooling flow in the central region @xcite .
therefore , we re - analyze the pspc data primarily to search for oxygen absorption and warm gas in m87 similar to that found for the lower s / n galaxies and groups in paper1 and paper3 .
we also investigate cooling flow models in an attempt to clarify the possible connection between the agn and cooling flow .
the paper is organized in follows . in [ obs ] we present the observations and discuss the data reduction .
the deprojection procedure is summarized in [ xdeproj ] .
we discuss the spectral fitting of single - phase models in [ single ] and of multiphase models in [ multi ] . in
[ disc ] we discuss the results and in [ conc ] we present our conclusions .
cccccccc + & & & date & exposure & & background + @xmath5 & ( @xmath6 @xmath7 ) & sequence no & ( mo / yr ) & ( ks ) & ( ct s@xmath8 ) & ( erg @xmath7 s@xmath8 ) & ( ct s@xmath8arcmin@xmath9 ) + + 0.00430 & 1.8 - 2.0 & rp800187n00 & 6/92 & 10.1/9.4 & 25.3 & @xmath10 & 1.5e-3 + & & rp800365n00 & 7/92 & 9.9/9.5 + & & rp800365a01 & 12/92 & 10.0/9.7 + + we obtained the _
pspc data of m87 from the public data archive maintained by the high energy astrophysics science archive research center ( heasarc ) .
the properties of the observations are listed in table [ tab.obs ] .
the three observations have essentially identical exposures and pointings ( i.e. , centered on m87 ) .
each observation was reduced separately , and we refer the reader to buote ( 2000c , hereafter paper2 ) for details of our implementation of the standard data reduction procedures . background spectra were obtained from source - free regions near the edge of the pspc fields ; i.e. , @xmath11 from the field centers .
the principal advantage of a local background estimate is that residual contamination from solar x - rays and any other long - term background enhancements @xcite are fully accounted for in the ensuing spectral analysis .
the disadvantage is that the extended x - ray emission of m87 and the virgo cluster fills the entire pspc field and must be disentangled from the true background .
( note that the background cluster emission is accounted for in the deprojection analysis
see 3.1 of paper2 . )
we extracted the background spectra separately from each observation and then added them together .
since each observation occurred after the october 1991 gain change the redistribution matrix file ( rmf ) , which specifies the channel probability distribution for a photon , is the same for each .
since , however , the detector location of background regions are in general slightly different for the different observations because of slight aspect differences , we averaged their respective auxiliary response files ( arfs ) which contain the information on the effective area as a function of energy and detector location . to obtain an estimate of the true background we follow the procedure outlined in paper2 and
fit the composite background spectrum with a model after subtracting the particle background .
we represent the cosmic x - ray background by a power law with photon index @xmath12 and the galactic emission by two soft thermal components following @xcite .
( actually , for m87 we find that only one of these soft thermal components is required with a temperature @xmath13 kev . ) to account for emission from the extended cluster emission we include another thermal component with variable temperature .
each component is modified by the galactic hydrogen column density listed in table [ tab.obs ] .
this composite model provides a good fit to the background spectrum ( @xmath14 for 185 dof ) , and the additional thermal component representing the cluster emission is clearly required by the data ; i.e. , @xmath15 for 181 dof without the cluster component . the temperature of the cluster component is @xmath16 kev ( 90% confidence ) which is somewhat less than the @xmath17 kev temperature inferred by @xcite over the same regions of the pspc data .
these differences are probably the result of the different models used to compute the background .
if , for example , we do not include the cosmic power law component in our fits then we obtain a temperature of @xmath18 kev similar to @xcite .
( note that the temperature of the cluster component is very insensitive to the metallicity . )
since the diffuse background is most important in the softest energy channels the stability of the background estimate to various assumptions can be assessed by considering the background flux between 0.2 and 0.3 kev .
( the need for the 0.2 kev lower limit arises from psf considerations
see below in [ src ] . )
let us compare the 0.2 - 0.3 kev flux obtained from our best - fitting background model to a similar model where instead the cluster component temperature is fixed at 3 kev .
we find that the 0.2 - 0.3 kev fluxes of the background models ( after excluding the cluster components ) differ by only 14% with the best - fitting model having the slightly larger 0.2 - 0.3 kev flux .
( we also prefer the slightly larger background estimate from the best - fitting model as it allows a more conservative assessment of any intrinsic soft emission in m87 . )
the background rate of the best - fitting model ( i.e. , after excluding the 1.5 kev cluster component ) is @xmath19 cts s@xmath8 arcmin@xmath9 ( table [ obs ] ) which is very similar to the value of @xmath20 cts s@xmath8 arcmin@xmath9 obtained for pspc observations of three other virgo galaxies in paper2 .
we subtract the background spectrum from each source annulus ( see below ) following the procedure described in 4.1.4 ( equation 1 ) in paper2 . that is , the background model ( cosmic+galactic ) is scaled to the source annulus accounting for differences in area , exposure , vignetting , and detector response between the source and background regions .
the ambient virgo emission is accounted for downstream in the deprojection analysis discussed in [ xdeproj ]
( see also 3.1 of paper2 ) . since our interest lies principally in the central regions where any intrinsic oxygen absorption is expected to be most pronounced , we confine our analysis of the x - ray emission of m87 to the region within the central ring of the pspc ( radius @xmath21 ) .
this is also the best calibrated region of the pspc and is not affected by shading by the radial spokes of the pspc support structure .
the source spectra were extracted in concentric circular annuli located at the x - ray centroid ( computed within a @xmath22 radius ) such that for each annulus the width was @xmath23 and the background - subtracted counts was larger than some value ( @xmath24 ) chosen to minimize uncertainties on the spectral parameters for each system while maintaining as many annuli as possible ( see figure [ fig.m87 ] .
data with energies @xmath25 kev were excluded to insure that the psf was @xmath26 fwhm . for our on - axis sources @xmath27 of the psf at 0.2 kev is contained within @xmath28 @xcite .
any background sources that were identified by visual examination of the image were masked out before the extraction .
to obtain the three - dimensional properties of the x - ray emitting gas we deproject the data assuming spherical symmetry using the `` onion - peeling '' technique pioneered by @xcite .
that is , one begins by determining the emission in the bounding annulus and then works inwards by subtracting off the contributions from the outer annuli .
a full account of our implementation of this procedure is given in paper2 . since we confine our analysis to the region within the central ring of the pspc there
exists significant emission from m87/virgo outside of the chosen bounding annulus .
this constitutes a `` cluster background '' component that needs to be subtracted from each source annulus in addition to the cosmic+galactic background discussed in [ bkg ] . following 3.1 of paper2 we represent the spatial distribution of this cluster background by a @xmath29-model with the same spectrum as the bounding annulus .
( x - ray emissivity of a @xmath29 model is @xmath30 at large radius . ) as in paper2 we find that the derived temperatures , abundances , and column densities are very insensitive to typical values of @xmath29 for galaxies and clusters and thus we employ @xmath31 for all results presented in [ single ] and [ multi ] .
we find that the radial profiles of the temperatures , fe abundances , and column densities are generally well behaved .
hence , unlike for the galaxies and groups analyzed in paper2 we do not perform the quasi - regularization of the temperatures and abundances described in 3.3 of paper2 . the greater stability of parameter values can be attributed to the much larger s / n of the m87 data .
( we mention that the deprojection procedure is invalid if there are gaps between annuli .
this is another reason why we confined our analysis within the central ring of the pspc . )
ccccccccc + annulus & 0nh_0fe & 1nh_0fe & 0nh_1fe & 1nh_1fe & 0nh_0fe_e & 1nh_0fe_e & 0nh_1fe_e & 1nh_1fe_e + + 1 & 1678.0/186 & 583.8/185 & 1254.1/185 & 507.1/184 & 821.4/184 & 333.2/183 & 332.2/183 & 323.4/182 + 2 & 1056.1/189 & 784.0/188 & 1019.2/188 & 469.3/187 & 292.6/187 & 288.6/186 & 287.6/186 & 278.9/185 + 3 & 1097.2/190 & 647.1/189 & 1027.1/189 & 437.6/188 & 354.2/188 & 301.6/187 & 293.3/187 & 280.4/186 + 4 & 1248.4/190 & 617.8/189 & 1088.9/189 & 473.0/188 & 341.9/188 & 302.2/187 & 268.2/187 & 246.6/186 + 5 & 1355.6/190 & 567.7/189 & 1008.7/189 & 536.2/188 & 568.2/188 & 419.4/187 & 338.5/187 & 303.9/186 + 6 & 1358.7/190 & 424.2/189 & 909.8/189 & 397.9/188 & 509.8/188 & 307.7/187 & 242.2/187 & 222.7/186 + 7 & 1429.9/190 & 514.5/189 & 966.0/189 & 498.8/188 & 512.0/188 & 388.8/187 & 280.8/187 & 237.4/186 + 8 & 1383.7/190 & 420.4/189 & 860.0/189 & 415.4/188 & 375.3/188 & 346.8/187 & 251.0/187 & 227.6/186 + 9 & 1220.5/189 & 386.5/188 & 743.4/188 & 386.2/187 & 413.5/187 & 343.3/186 & 252.5/186 & 228.7/185 + 10 & 649.0/190 & 263.6/189 & 466.2/189 & 255.0/188 & 464.6/188 & 213.9/187 & 173.5/187 & 164.6/186 + + we initially focus our analysis on `` single - phase '' models of the hot gas ; i.e. , a single value for the temperature and density describes the emission of hot plasma at each radius .
this is the simplest case and would seem _ a priori _ to be most appropriate for the low energy resolution of the _ rosat _ data .
below in [ multi ] we consider multiphase models . as in previous studies we use the mekal plasma code ( mewe , gronenschild , & van den oord 1985 ; kaastra & mewe 1993 ; liedahl , osterheld , & goldstein 1995 ) to represent the emission of hot gas at a single temperature .
we take the solar abundances in xspec to be those given by the @xcite table , because it contains the correct value of the solar fe abundance , @xmath32 by number .
the popular table of @xcite used in xspec has an old value for fe / h which is too large by a factor of 1.44 as initially pointed out by @xcite . see
4.1.3 of paper2 for further discussion .
according to the recent deconvolution analysis of _ asca _ data by @xcite the fe abundance of m87 is @xmath33 within the central arcminute , remains essentially constant at @xmath34 out to @xmath35 , and then falls to an average value of @xmath36 between @xmath37 ( all abundances scaled to meteoritic solar ) .
these results agree with the multitemperature models of the _ asca _ sis data obtained by fitting the integrated spectra within @xmath38 of m87 by @xcite .
consequently , when we fix the fe abundance in selected models we assign it the value of @xmath39 appropriate for the two - temperature model of @xcite .
following standard practice we represent the soft x - ray absorption arising from the milky way by cold material with solar abundances distributed as a foreground screen at zero redshift . in this standard absorption model
the x - ray flux is diminished according to @xmath40 , where is the hydrogen column density and @xmath41 is the energy - dependent photo - electric absorption cross section for a cold absorber with solar abundances .
@xcite have shown that the galactic column density varies over the pspc field of m87 .
inspection of their figure 2 reveals that varies almost radially outward from m87 taking a minimum value of @xmath42 @xmath7 near the center and a value of @xmath43 @xmath7 near our bounding annulus ( # 10 ) . since intermediate values of appear to prevail at intermediate radii we set @xmath44 @xmath7 in those cases where we fix in our models .
in other instances we allow to be a free parameter in our fits to indicate any excess absorption intrinsic to a galaxy or group and also to allow for any errors in the assumed galactic value and for any calibration uncertainties . note that in this standard model is measured as a function of two - dimensional radius , @xmath45 , on the sky in contrast to the emission parameters ( e.g. , gas temperature , fe abundance ) that are measured as a function of three - dimensional radius , @xmath46 .
the photo - electric absorption cross sections used in this paper are given by @xcite .
it has been pointed out by @xcite that the he cross section at 0.15 kev is in error by 13% in the @xcite compilation . recently
a new version of xspec ( v11 ) has been released which updates the @xcite table to include the more accurate he cross sections of @xcite .
we find that when using these more accurate he cross sections we obtain hydrogen column densities that are 5%-10% larger than when using the original @xcite he cross sections in xspec v10 .
where appropriate we mention these small differences below .
we begin with the model having the least number of free parameters .
this corresponds to setting @xmath47 for the standard absorber and holding the fe abundance fixed to @xmath39 as discussed above .
the values of @xmath48 for this model ( 0nh_0fe ) are listed in table [ tab.1 t ] , and the spectral fits for the inner and outer annuli are displayed in figure [ fig.1t_spec ] .
it is clear from the large reduced @xmath48 values and the large residuals in the spectral plots that this model provides a very poor fit to the _ rosat _ pspc spectra of m87 .
two regions of the spectra are primarily responsible for the poor fits . in the energy band 0.2 - 0.4
kev the data clearly lie above the model prediction .
immediately one can conclude that there is no intrinsic absorption from cold gas precisely as found for the galaxies and groups in paper3 ; see section 2.1 of paper1 and section 3.2 of paper3 for details .
since we find that the results obtained when using absorber models with partial covering are essentially identical to those obtained for the standard uniform - screen absorber , we only discuss the uniform screen . ]
for which we would have expected to see a deficiency of flux below the models with @xmath47 . instead soft emission in excess of this model is suggested .
in contrast , between 0.5 - 0.8 kev the data do lie below the model consistent with intrinsic absorption . from inspection of figure [ fig.1t_spec ]
the absorption appears to be more pronounced in the inner annulus .
these excess absorption and emission features are much too large to be explained by calibration error .
the most significant calibration problem ever reported for the gain of the pspc is by @xcite who showed that the early versions of the software to convert pha ( pulse height analyzer ) to pi ( pulse invariant ) channels did not account for a gain deficit within a few arcminutes distance of the field center .
this old problem only appeared in pi channels below 20 ( i.e. , below 0.2 kev ) and is now almost entirely removed with the most up - to - date software which we have used . in principle
such an uncorrected gain deficit would appear as an absorption feature in our analysis below 0.2 kev which is clearly unrelated to both the excess emission between 0.2 - 0.4 kev and the absorption between 0.5 - 0.8 kev observed in figure [ fig.1t_spec ] .
further testament to the reality of these features is that they are each manifested over an energy range comparable to the pspc resolution [ @xmath49 ; i.e. , in the notation of @xcite , who define resolved pspc energy bands , the 0.2 - 0.4 kev interval is essentially the r2 band while 0.5 - 0.8 kev is essentially the r4 and r5 bands .
although it might be tempting to equate the soft excess in the 0.2 - 0.4 kev band with an error in the background subtraction , this can not be the case . the emission of the diffuse background in the _ rosat _ band does peak at these low energies ( e.g. , chen et al . 1997 ) and the temperature of gas required to explain the soft excess is consistent with the diffuse background ( see below in [ 2 t ] ) .
however , the background contributes only 0.6% of the 0.2 - 0.4 flux within the inner annulus , and thus only an ( impossible ) error of over two orders of magnitude in the background level could account for the excess 0.2 - 0.4 kev flux . only for the bounding annulus ( i.e. , annulus # 10 )
is a relatively small background error required ( factor of @xmath50 ) to explain the excess soft emission .
but we have taken care to account for detector response differences between background and source positions ( [ bkg ] ) . the statistical error on
the background is @xmath51 and we estimate at most a 15% systematic error in the background level arising if we do not subtract the `` additional thermal component '' attributed to m87 ( [ bkg ] ) .
hence , errors in the background subtraction can not explain the excess 0.2 - 0.4 kev emission .
since the deviations from the 0nh_0fe model displayed in figure [ fig.1t_spec ] must be intrinsic to m87 , let us now examine whether adjusting the available model parameters can improve the fits .
if is allowed to vary in the fits while keeping the fe abundance , , fixed ( i.e. , model 1nh_0fe in table [ tab.1 t ] ) then we obtain large improvements in the fits in all annuli ; the least amount of improvement is observed for annuli # 2 - 3 . although allowing to vary removes a large portion of the deviations over 0.2 - 0.4 kev the absorption feature over 0.5 - 0.8 kev remains essentially unaffected .
the fitted values of ( r ) ( see figure [ fig.1 t ] ) are @xmath52 in most annuli with a maximum of @xmath53 in annulus # 2 .
these values are still very significantly less than the galactic value when considering the 5%-10% underestimates arising from inaccuracies in the he cross sections ( see immediately before [ std ] ) .
these sub - galactic columns are also unaffected by possible errors in the plasma emission code ( mekal ) . since the fitted temperatures are always above 1 kev inaccuracies in the plasma code near energies 0.2 - 0.4 kev are negligible because at these temperatures the continuum dominates any neglected line emission which is the principal source of error ( e.g. , liedahl 1999 ) .
if instead is allowed to vary but is held fixed at ( i.e. , model 0nh_1fe in table [ tab.1 t ] ) , then the fits are also improved though not quite so much as when is allowed to vary .
the improvement in the fits is again confined to the lower energies ( 0.2 - 0.4 kev ) leaving the absorption deviations over 0.5 - 0.8 kev uncorrected .
the sub - solar values of @xmath54 ( figure [ fig.1 t ] ) are consistent with those obtained from the previous _ asca _ studies of @xcite and @xcite .
_ measurements of @xmath55 should be more reliable since they are determined primarily by the 6.5 kev fe k@xmath56 line complex in contrast to the _ rosat _ determinations using the @xmath57 kev fe l lines .
the temperature measurements from _ asca _ and _ sax _ are also more reliable than those from _ rosat _ because of the much better energy resolution and the larger bandwidth . in figure [ fig.1 t ]
one notices that for @xmath58 the model 0nh_1fe has @xmath59 kev which is very inconsistent with the temperatures obtained from _
asca _ @xcite and _ sax _ @xcite . on the other hand ,
if both and are varied ( i.e. , model 1nh_1fe in table [ tab.1 t ] ) then @xmath60 is consistent with _
asca _ and _ sax _ but now is too large at all radii ( figure [ fig.1 t ] ) .
this latter model also has fitted column densities that deviate from the galactic value even more significantly than the 1nh_0fe model discussed above .
interestingly , when both and are varied significant improvement in the fits is also observed over 0.5 - 0.8 kev unlike when only one of these parameters is varied ( though the overall fit quality is still poor in all annuli except # 10 ) .
hence , the single - phase model with a standard absorber is demonstrably a very poor fit to the _ rosat _ data .
although improvements in the fits can be achieved by varying and the resulting fits are still poor , the column densities are significantly below the galactic value , and the fitted temperatures and abundances are not each consistent with the _ asca _ ( and _ sax _ ) data for any particular model . we are thus compelled to conclude that a single - phase gas modified by a standard galactic absorber can not describe the x - ray data of m87 .
the striking residuals of excess absorption and emission apparent from visual inspection of the spectral fits of m87 ( figure [ fig.1t_spec ] ) provide strong support for the analogous ( but less obvious ) properties deduced from the lower s / n data of the individual elliptical galaxies and galaxy groups studied in paper1 and paper3 .
an immediate result of critical importance is that since the data for energies below 0.5 kev do not lie below the 0nh_0fe model in figure [ fig.1t_spec ] , the low - energy pspc data unequivocally rule out any significant excess absorption from cold absorbing matter which has frequently been advocated in the past for cooling flows . instead
, the absorption feature between 0.5 - 0.8 kev must arise from warm material as we have discussed extensively in paper1 and paper3 .
because of the low energy resolution of _ rosat _ it is instructive first to analyze the hydrogen column density obtained for the standard cold absorber as a function of the chosen lower energy limit ( ) of the bandpass ( see 3.3 of paper3 )
. that is , we fitted the model 1nh_0fe ( defined above ) to the pspc spectrum including only data having energies between @xmath61 and 2.2 kev .
the results are displayed in figure [ fig.nhvsemin ] .
when @xmath62 kev the fitted column density is always similar to or less than . as increases to @xmath63 kev
we find that @xmath64 for @xmath65 and increases to @xmath66 for @xmath67 .
this radially decreasing profile of column density becomes even more pronounced for larger values of ; e.g. , as shown in figure [ fig.nhvsemin ] for @xmath68 kev
the outer annuli have @xmath69 whereas at smaller radii @xmath70 .
this trend continues up to @xmath71 kev after which @xmath72 for @xmath73 kev within the increasingly larger uncertainties . in order to quantify the `` warm '' absorption and yet simultaneously fit the entire _ rosat _ spectrum we follow our previous investigations of galaxies and groups ( paper1 and paper3 ) and model the intrinsic absorption with a single absorption edge represented as , @xmath74 $ ] for @xmath75 where @xmath76 is the rest - frame energy of the edge and @xmath77 is the optical depth .
similar to the standard absorber we place the edge immediately in front of the source so @xmath77 is measured as a function of 2d radius @xmath45 .
a rigorous treatment of the warm absorber would ( at least ) include a series of absorption edges from different ionization states of principally oxygen but also of carbon and nitrogen similar to the models described by @xcite that are currently not supported in xspec .
our simple parameterization of a warm absorber as a single edge is intended primarily as a phenomenological tool to establish the existence and to study the gross properties of the absorber which is appropriate for the low resolution data afforded by the pspc .
we place the edge initially at @xmath78 kev appropriate for but always allow @xmath76 to vary ; nevertheless we shall often refer to this as an `` oxygen edge '' .
the values of @xmath48 for models with an oxygen edge are listed in table [ tab.1 t ] , and in figure [ fig.1t_edge_spec ] we plot the _ rosat _ spectra for annuli # 1 and # 10 in analogy to figure [ fig.1t_spec ] with the only difference being that the oxygen edge is included in the best - fitting models ; i.e. , model 0nh_0fe_e .
although the edge greatly improves the fits of all annuli , the character of the improvement varies with radius . as seen in figure [ fig.1t_edge_spec ] for annulus # 1 the oxygen edge removes most of the residuals over 0.5 - 0.8 kev while leaving other deviations ( primarily at lower energies ) mostly unchanged .
this qualitative behavior also applies to the next few annuli ( # 2 - 4 ) .
in contrast , as seen for annulus # 10 in figure [ fig.1t_edge_spec ] the single oxygen edge itself is insufficient to remove the residuals over 0.5 - 0.8 kev and instead allows only a relatively crude improvement by cutting essentially midway through the 0.5 - 0.8 kev region ; i.e. , the model is still clearly the wrong shape in the absorbing region .
this qualitative behavior applies to the outer annuli ( # 5 - 10 ) .
when either of the standard absorber or is allowed to vary the fits are improved substantially in most annuli ; i.e. , models 1nh_0fe_e and 0nh_1fe_e in table [ tab.1 t ] . for annuli # 1 - 4
the improvement offered by variable and are very similar , and as seen in figure [ fig.1t_edge_spec_nh ] ( using model 1nh_0fe_e ) most of the large residuals displayed by the original single - phase standard absorber model ( figure [ fig.1t_spec ] ) have been removed .
the radial profiles of , @xmath79 , and for these edge models are displayed in figure [ fig.1t_edge ] .
for annuli # 1 - 4 the temperatures are consistent for both models and the fe abundance profile ( for 0nh_1fe_e ) is essentially consistent with that obtained from the _ asca _ data @xcite ( scaled to meteoritic solar fe abundance ) .
the column density profile for the standard absorber is @xmath80 below the galactic value for annuli # 1,3 - 4 .
this discrepancy , though still statistically significant , is reduced to 10%-15% when using the more accurate he cross sections of @xcite in xspec v11 ( see immediately before [ std ] ) .
note that if both and are varied ( i.e. , model 1nh_1fe_e ) the fitted parameters for annuli # 1 - 4 are consistent with those above ( though with larger errors ) and the fits are slightly improved further ( table [ tab.1 t ] ) . for the outer annuli ( # 5 - 10 )
although varying improves the fits noticeably , a much larger improvement is obtained by varying . as can be seen in figure [ fig.1t_edge_spec_nh ] the variable model ( 1nh_0fe_e )
fits the _ rosat _ spectrum for the bounding annulus ( # 10 ) quite well ; the variable - model ( 0nh_1fe_e ) fits even better ( table [ tab.1 t ] ) . for annuli # 5 - 9
the variable - model fits similarly to that shown for # 10 except that the residuals below 0.5 kev are slightly more pronounced .
if both and are varied then the fits in these outer annuli are essentially formally acceptable ( table [ tab.1 t ] ) . however , the fitted parameters obtained in these outer annuli indicate a problem with the simple single - phase , single - edge models .
if is varied and is held fixed then the inferred ( r ) ( figure [ fig.1t_edge ] ) is approximately consistent with the sub - solar values obtained by @xcite for @xmath81 with _ asca _ .
but then the temperatures of this model are always below 2 kev at large @xmath45 which are very inconsistent with the 2.5 - 3 kev temperatures inferred from _
asca _ and _ sax _
( see discussion in [ std ] ) .
on the other hand , if is varied and is held fixed the fitted temperatures are consistent with
/_sax _ but now @xmath82 at large radius .
the lower temperatures inferred for 0nh_1fe_e and the sub - galactic columns inferred for 1nh_0fe_e both imply the existence of an extra emission component having a temperature significantly less than the hot phase for annuli # 5 - 10 . for the inner radii # 1 - 4 the temperatures are also consistently 30%-50% below the _ asca _ /_sax _ values ( white 2000 ; dacri et al .
1998 ) and ( except annulus # 2 ) have sub - galactic column densities which also suggest extra emission from cooler gas .
this cooler emission component is very likely related to the absorption component that we have modeled with the single oxygen edge . in figure [ fig.1t_edge_tau ]
we plot @xmath83 and @xmath84 for the models shown in figure [ fig.1t_edge ] . for @xmath85 , @xmath86 kev
and then increases to @xmath87 kev at smaller @xmath45 . assuming collisionally ionized gas these edge energies imply a gas temperature ranging from @xmath88 k ( e.g. , krolik & kallman 1984 ; sutherland & dopita 1993 ) which apparently increases with decreasing @xmath45 .
the optical depth profile also falls with increasing @xmath45 taking values near 1 for @xmath89 and flattens out for 1nh_0fe_e or decreases steadily to values @xmath90 for 0nh_1fe_e . although @xmath84 is quite similar to those obtained for
the galaxies and groups studied in paper1 and paper3 , the values of @xmath87 kev at small radii are significantly larger and probably indicate that the warm gas is `` warmer '' in m87 . finally , we mention that attributing the 0.5 - 0.8 kev absorption feature instead to a sub - solar o / fe ratio of the hot gas always results in sub - galactic column densities for the standard absorber similar to that shown in figure [ fig.1t_spec ] for model 1nh_0fe ; i.e. , sub - solar o / fe still implies excess soft emission . and
if this soft emission originates from ionized gas it should produce absorption near 0.5 kev .
moreover , sub - solar o / fe models never fit as well as the oxygen - edge models except when is held fixed and is varied . in this case
the sub - solar o / fe models fit comparably to the oxygen - edge models at larger radius ( annuli # 5 - 10 ) but at smaller radii still have @xmath48 values that are larger by 50 - 100 .
it should also be noted that these sub - solar o / fe models all result in oxygen abundances that are zero everywhere which is in serious conflict with the theoretical expectation of approximately solar abundance ratios ( e.g. , renzini et al .
if o / fe is kept to a minimum of @xmath91 solar then no sub - solar o / fe model fits as well as a model with an oxygen edge .
hence , a sub - solar value for o / fe of the hot gas can not reasonably describe either the excess absorption over 0.5 - 0.8 kev or the excess soft emission over 0.2 - 0.4 kev .
( we also note that a very sub - solar value of o / fe is inconsistent with the modest super - solar si / fe and s / fe ratios measured by _ asca _
( e.g , buote et al 1999 ) ; i.e. , the @xmath56/fe ratios should all give a consistent picture of the supernova enrichment . )
ccccc|cccc + & & + annulus & 0nh_0fe_e & 1nh_0fe_e & 0nh_1fe_e & 1nh_1fe_e & 0nh_0fe & 0nh_1fe & 0nh_0fe_e & 0nh_1fe_e + + 1 & 329.2/183 & 324.9/182 & 325.7/182 & 320.1/181 & 529.8/184 & 389.3/183 & 288.1/182 & 256.2/181 + 2 & 299.0/186 & 295.0/185 & 289.0/185 & 276.5/184 & 746.5/188 & 370.3/186 & 288.9/185 & 229.4/184 + 3 & 312.0/187 & 308.2/186 & 295.7/186 & 275.7/185 & 597.0/188 & 375.5/187 & 283.7/186 & 244.0/185 + 4 & 290.9/187 & 294.0/186 & 270.6/186 & 238.7/185 & 564.3/188 & 404.5/187 & 261.0/186 & 236.9/185 + 5 & 424.1/187 & 372.2/186 & 335.0/186 & 298.2/185 & 527.0/188 & 489.0/187 & 321.1/186 & 295.8/185 + 6 & 365.4/187 & 286.3/186 & 247.5/186 & 230.4/185 & 384.0/188 & 353.2/187 & 240.1/186 & 227.6/185 + 7 & 422.6/187 & 330.6/186 & 267.6/186 & 232.5/185 & 473.6/188 & 452.9/187 & 257.3/186 & 242.0/185 + 8 & 415.3/187 & 299.2/186 & 242.3/186 & 223.2/185 & 384.9/188 & 377.9/187 & 231.8/186 & 215.3/185 + 9 & 413.0/186 & 303.3/185 & 247.9/185 & 226.6/184 & 360.5/187 & 359.1/186 & 240.5/185 & 225.3/184 + 10 & 259.8/187 & 180.7/186 & 160.4/186 & 156.7/185 & 248.4/188 & 241.7/187 & 203.7/186 & 150.3/185 + + we have seen that an attempt to describe the _ rosat _ x - ray emission of m87 by a single - phase hot medium fails because the intrinsic absorption feature over 0.5 - 0.8 kev and the sub - galactic hydrogen column densities ( i.e. , excess soft 0.2 - 0.4 kev emission ) imply the existence of additional gas phases and can not be explained by dust ( 5.1 of paper3 ) . in principle
we would like to fit a general emission model in order to determine the differential emission measure function which fully characterizes the temperature structure of the plasma .
unfortunately , with detectors of low ( _ rosat _ ) and moderate ( _ asca _ ) energy resolution it is impossible to distinguish between different multiphase models .
@xcite showed in particular that two - phase models can easily mimic a multiphase cooling flow spectrum of galaxies and clusters . and
in section 5.3 of @xcite we showed that even with the higher resolution of _ asca _ the following multiphase models of galaxies all produced similar spectra : cooling flow , two - phase , and integrated single - phase temperature gradient ; similar results were obtained for groups in @xcite .
( however , these very different multiphase models are each easily distinguished from isothermal models . )
since the limited energy resolution of the pspc prevents a detailed exploration of different multiphase models our primary objective in this section is to quantify the emission of the warm gas at each radius .
as in our previous studies we examine the multiphase hypothesis in terms of the simple cases of a constant - pressure cooling flow and a two - phase medium .
these models are instructive since they have been widely used to interpret the x - ray spectra of galaxies and clusters . in the inhomogeneous cooling flow scenario the hot gas
is expected to emit over a continuous range of temperatures in regions where the cooling time is less than the assumed age of the system .
we consider a simple well - studied model of a cooling flow where the hot gas cools at constant pressure from some upper temperature , @xmath92 ( e.g. , johnstone et al 1992 ) .
the differential emission measure of the cooling gas is proportional to @xmath93 , where @xmath94 is the mass deposition rate of gas cooling out of the flow , and @xmath95 is the cooling function of the gas ( in our case , the mekal plasma code ) .
since the gas is assumed to be cooling from some upper temperature @xmath92 , the cooling flow model requires that there be a reservoir of hot gas emitting at temperature @xmath92 but is not cooling out of the flow .
consequently , our cooling flow model actually consists of two components , cf+1 t , where `` cf '' is the emission from the cooling gas and `` 1 t '' is emission from the hot ambient gas .
we set @xmath92 of the cf component equal to the temperature of the 1 t component , and both components are modified by the same photoelectric absorption .
hence , the cooling flow model adds only one free parameter @xmath96 to those of the single - phase models .
if only absorption from the standard cold absorber model is considered then we obtain results identical to the single - phase models in [ std ] ; i.e. , the cf component is completely suppressed by the fits .
since the cf model includes temperature components below @xmath3 kev it has stronger ( @xmath97 kev ) and ( @xmath98 kev ) k@xmath56 lines than the single - phase models which , as shown above , already predict too much emission from these lines . therefore , in exact agreement with our _ rosat _ analysis of galaxies and groups in paper3 we find that only when an intrinsic oxygen absorption edge is included in the fits can we obtain a significant contribution from a cf component , and we shall henceforth restrict ourselves to cooling flow models with an intrinsic oxygen edge .
the @xmath48 values for selected cooling flow models with an intrinsic oxygen edge are listed in table [ tab.multi ] .
let us focus initially on the model 0nh_0fe_e for comparison to the single - phase version .
we plot 0nh_0fe_e in figure [ fig.multi_spec ] for annulus # 1 which should be compared to the single - phase version in figure [ fig.1t_edge_spec ] .
the cooling flow model in the central annulus is a much better fit because it can mostly account for the soft emission over 0.2 - 0.4 kev . to a lesser extent
the cooling flow model also provides a better fit in annuli # 3 - 7 and 10 , where the improved @xmath48 values are intermediate with those obtained when varying for the single - phase model ( 1nh_0fe_e in table [ tab.1 t ] ) .
thus , the cooling flow model is able to partially compensate for the excess soft emission .
( this is also evident from ( r ) displayed for model 1nh_0fe_e in figure [ fig.cf ] ; i.e. , for @xmath99 the fitted column densities are consistent with the galactic value especially when considering the 5%-10% underestimates arising from the old he cross sections
see immediately before [ std ] . )
if is allowed to vary then the quality of the resulting cooling flow model is very similar to the corresponding single - phase model ( 0nh_1fe_e ) , though the cooling flow gives @xmath48 values that are 5 - 10 lower in most annuli .
the temperature and fe abundance profiles of the cooling flow model are shown in figure [ fig.cf ] .
the fe abundances are consistent with the single - phase model and the _ asca _ data @xcite
. however , the temperatures of the cooling flow model at large radii are 2.5 - 3.0 kev which , unlike the single - phase model , are indeed consistent with the _ asca _ /_sax _ values ; i.e. , the cooling flow model 0nh_1fe_e provides slightly better fits and physical temperatures and fe abundances in the hot gas in contrast to the single - phase model . the absorption parameters obtained for the cooling flow models are qualitatively similar to those obtained for the single - phase models ( figure [ fig.cf_tau ] ) .
the @xmath84 values for 0nh_1fe_e are larger for the cooling flow model especially at large radii where @xmath100 compared to @xmath101 for the single - phase version .
the larger optical depths for the cooling flow models are attributed to the need to suppress the stronger and k@xmath56 emission lines arising from the cooler temperature components . finally , in figure [ fig.cf_tau ] we also plot @xmath102 for the two cooling flow models .
we see a strange profile where ( apparently ) there is significant cooling ( @xmath103 ) at the center which abruptly terminates in annuli # 2 - 4 and then rises to a nearly constant value ( @xmath104 ) at large radius . in the central bin
the value of @xmath105 is uncertain and consistent with zero at the 90% confidence level and therefore is also consistent with the original _ einstein _
value of @xmath106 ( e.g. , canizares et al .
1982 ; stewart et al .
the total mass deposition rate of all annuli is @xmath107 for model 0nh_1fe_e and @xmath108 for model 1nh_0fe_e ( @xmath109 errors ) .
these values of @xmath105 are in good agreement with the previous ( imaging ) _ rosat _ study by @xcite and are also similar to the deposition rates of the brightest galaxy groups @xcite and small galaxy clusters .
the implications of the mass deposition profile are discussed in [ feedback ] .
the constant - pressure cooling flow model is one example of a ( multiphase ) gas that emits over a continuous range of temperatures , and it has the virtue of introducing only one free parameter ( ) to the single - phase models .
a two - phase model , however , introduces two free parameters to the single - phase case : extra temperature and normalization . from our previous studies
( e.g. , buote 1999 , 2000a ; buote et al .
1999 ) we know that the two - phase model will fit as well or better than the cooling flow model , and
the temperatures and fe abundances will be similar in both cases . for our current purposes
the two - phase model is useful since it allows us ( 1 ) to assign a specific temperature to the warm phase and ( 2 ) to explicitly quantify the relative contribution of the warm gas emission to the total .
we expect that adding the `` warm '' temperature component will have a similar effect as allowing to vary in the fits of the single - phase models . for models without an intrinsic oxygen edge , it can be seen in table [ tab.multi ] that the two - phase models 0nh_0fe and 0nh_1fe have very similar ( actually somewhat better ) @xmath48 values than their single - phase counterparts with variable ( 1nh_0fe and 1nh_1fe ) in table [ tab.1 t ] .
the same behavior also applies to the corresponding two - phase models with an intrinsic oxygen edge ( 0nh_0fe_e and 0nh_1fe_e ) .
the best - fitting two - phase model 0nh_0fe_e for annulus # 1 displayed in figure [ fig.multi_spec ] looks quite similar to the corresponding single - phase version with variable ( i.e. , 1nh_0fe_e ) displayed in figure [ fig.1t_edge_spec ] .
the temperature profiles of the two - phase models are shown in figure [ fig.2 t ] .
the temperature of the warm gas ( ) is essentially constant with radius with value 0.1 kev for the model without the oxygen edge . for the model with the oxygen edge
@xmath110 kev though it can be as large as 0.4 kev at intermediate radii .
although these temperatures are similar to that of the x - ray background , we recall that error in the background subtraction can not account for the warm emission ( [ std ] ) .
note also that the value of @xmath111 in annulus # 2 is very uncertain because the warm emission there is weak , and thus we had to fix the value of @xmath111 in annulus # 2 to that of annulus # 3 .
the temperature of the hot phase ( ) in both cases is similar and is also very consistent with the 2.5 - 3 kev temperatures obtained by _
asca _ @xcite and _ sax _ @xcite at large radii . in figure [ fig.2 t ]
we also show the fraction that the warm temperature component contributes to the total emission measure .
over most radii the warm temperature component contributes 10%-20% to the emission measure .
however , in annulus # 2 the warm component is markedly suppressed and is consistent with zero for the model with an oxygen edge .
the variation of emission - measure fraction with radius is qualitatively similar to the profile of for the cooling flow model ( figure [ fig.cf ] ) ; i.e. , even though the cooling flow and two - phase models have entirely different differential emission measures they give essentially identical pictures of the variation of warm gas with radius .
the absorption parameters of the intrinsic oxygen edge for the two - phase model 0nh_0fe_e are displayed in figure [ fig.2t_edge_tau ] .
both the rest energies of the edge and the optical depths are ( nearly ) consistent with being constant with radius .
although these flat profiles are somewhat different from the cooing flow and single - phase models , the ranges of values of energies and optical depths are similar .
using a spatial - spectral deprojection analysis we have shown that a single - phase model of the hot gas modified by standard galactic absorption can not describe the _ rosat _ pspc data of m87 .
the failure of the single - phase model is dramatically evident in the spectral plot in figure [ fig.1t_spec ] : excess absorption over 0.5 - 0.8 kev and excess emission over 0.2 - 0.4 kev signal the presence of an additional warm ( @xmath1 k ) gas component .
this evidence for warm absorbing ( and emitting ) gas in m87 lends strong support to the results we have recently obtained from _ rosat _
data of several bright galaxies and groups ( paper1 and paper3 ) . only multiphase models ( e.g. , cooling flow , two - phase ) with an intrinsic oxygen edge are able to provide good fits to the _ rosat _ data and still have ( 1 ) ambient temperatures and fe abundances consistent with _
asca _ and _ sax _ studies ( [ std ] ) and ( 2 ) galactic hydrogen column densities for the known absorption arising from the intervening cold gas in the milky way .
although the multiphase models provide better fits than the single - phase models , they still do not provide perfect fits in all annuli ( figure [ fig.multi_spec ] ) .
the remaining low - level residuals can be eliminated by allowing the @xmath56-process elements to vary differently from fe ; e.g. , the most important element is si which removes most of the residual if si / fe is allowed to be several times solar .
a modest super - solar si / fe ratio is consistent with previous _
asca _ studies ( buote et al .
1999 and references therein ) . since _ rosat _ is unable to place tight constraints on the abundance ratios we do not give constraints on these quantities at this time .
it is also possible that these low - level residuals in our best models reflect calibration errors . in particular ,
the remaining feature near 0.3 kev ( figures [ fig.1t_edge_spec_nh ] and [ fig.multi_spec ] ) is very similar to that seen in the pspc spectra of some other bright clusters in the recent study by @xcite .
these authors prefer to explain these low - level residuals as a gain offset . while preparing our manuscript we became aware of the preprint by @xcite .
these authors also present a spatial - spectral deprojection analysis of the _ rosat _ data of m87 along with a large sample of clusters .
the analysis of @xcite differs from ours in two main respects .
first , they bin up each pspc spectrum into three x - ray channels ( or `` colors '' ) .
spectral models are then fitted to ratios of these channels .
second , they exclude data below 0.4 kev and ignore the sensitivity of the inferred to shown in our figure [ fig.nhvsemin ] . despite these differences between our respective analyses
, @xcite also conclude that a single - phase model is inadequate and that spatially extended intrinsic absorbing material is required . since , however , @xcite exclude data below 0.4 kev they do not detect the excess soft emission and thus are not led to conclude that the absorber must be warm .
intrinsic absorption in m87 is also inferred from two - temperature models of the integrated _ asca _ spectra within @xmath112 by @xcite which confirms the original detection with _ einstein _ @xcite .
the hydrogen column density of cold material implied by the two - temperature _ asca _ models , @xmath113 @xmath7 , is considerably larger than the value of @xmath114 @xmath7 obtained by @xcite with _ rosat _ data . the smaller value obtained by @xcite arises from their fitting _
rosat _ data with energies above 0.4 kev with a cold absorber whereas the _ asca _ analysis of @xcite includes only data above 0.6 kev .
as we have shown in figure [ fig.nhvsemin ] the inferred value of is highly sensitive to the lower energy limit of the bandpass .
our absorption models using a single oxygen edge yield optical depths of @xmath57 in the inner regions of m87 .
these optical depths translate to @xmath115 @xmath7 ( depending on the ionization state of oxygen ) which agree with the _ asca _ result .
the new _ rosat _ evidence for multiphase gas in m87 largely confirms and extends the original detection by @xcite within the central @xmath116 using the _ einstein _ fpcs . by measuring selected line blends of - @xcite
inferred the presence of plasma emission components with temperatures ranging from @xmath117 k down to @xmath118 k consistent with our results .
however , these authors also deduce from analysis of the ly-@xmath56 line that the o / fe ratio is 3 - 5 times solar in the ism ( standard galactic absorption was also assumed ) . since this is not confirmed by either the _ rosat _ or _ asca _ data ( which instead indicate that the data are absorbed over 0.5 - 0.8 kev where the ly-@xmath56 line emits ) the super - solar o / fe ratio probably can be attributed to the fpcs calibration errors discussed by @xcite for m87 .
an alternative explanation ( due to the referee ) for the fpcs measurement is as follows .
since the fpcs operated over a narrow energy band , if the absorption edges are confined to energies just below the k@xmath56 line they would not be seen by the fpcs .
consequently , the fpcs would measure the line strength with respect to the absorbed continuum level .
lower resolution detectors like _ rosat _ and _ asca _ would merge the edges and line and therefore see a smaller line strength .
we now consider how much mass is associated with the warm ionized absorber in m87 . in principle
this can be computed from the optical depths of the intrinsic oxygen edges ( figures [ fig.1t_edge_tau ] , [ fig.cf_tau ] , and [ fig.2t_edge_tau ] ) , the mass deposition rates of the cooling flow models ( figure [ fig.cf_tau ] ) , or the emission measures of the warm gas in the two - phase models ( figure [ fig.2 t ] ) .
unfortunately , our single - edge absorption model is really just a phenomenological tool to establish the existence and to study the gross properties of the absorber which is appropriate for the low resolution of the pspc ( [ edge ] ) .
as discussed in paper3 , when including additional edges the inferred optical depth for each edge decreases , and we expect several edges from different ionization states of oxygen ( also carbon and nitrogen ) to contribute . since the absorption of an edge is not linear in the energy of the edge ( i.e. , @xmath119 $ ] and @xmath120 if the edge energies @xmath121 ) , by spreading multiple edges over a large energy range one can produce the observed absorption with smaller total optical depth than can be achieved with a single edge .
consequently , the single - edge optical depths should be considered upper limits . nevertheless , to provide a convenient benchmark for comparison to paper3 and to related previous studies , we have computed the absorber masses by assuming the absorption edges correspond to ( see 5.3 of paper3 ) . using the single - phase model 1nh_0fe_e as an example , we find that within the central bin ( 5 kpc ) that the absorber mass is , @xmath122 ( @xmath109 errors ) which is about twice that of the hot gas , @xmath123 .
the total values within 98 kpc are respectively , @xmath124 and @xmath125 ; i.e. , @xmath126 similar to that found for the galaxies and groups in paper3 .
the amount of time required to accumulate @xmath127 assuming @xmath107 ( [ cf ] ) is @xmath128 ; i.e. , this absorbing material could reasonably have been accumulated by a cooling flow in agreement with the galaxies and groups in paper3 .
if , as we expect , the absorber is the same material that gives rise to the excess 0.2 - 0.4 kev emission , then the above values of should be ( nearly ) the same as @xmath129 ; i.e. , the mass of the warm gas component of the two - phase models ( [ 2 t ] ) . however , there are two key problems with translating the results of [ 2 t ] into values of @xmath129 . first , the temperatures of the warm components are 0.1 - 0.2 kev , and it has recently been emphasized that the current plasma codes ( including mekal ) can underestimate the emission around 0.2 kev by as much as a factor of 10 at these low temperatures because of neglected line emission ( liedahl 1999 ; beiersdorfer et al . 1999 ) .
second , since the warm gas absorbs photons from the hot gas the assumption of collisional ionization equilibrium is also not strictly valid nor is it clear that the warm gas is fully optically thin to its own radiation as we have assumed for convenience ( with mekal ) .
these problems are evident when we attempt to compare the emission measures ( @xmath130 of the warm gas component in the two - phase model to that implied by the values of quoted above from the oxygen edges .
we find that , e.g. , in the central radial bin the emission measure implied by the oxygen edges is @xmath4 times that inferred for the warm gas component in the two - phase model .
this discrepancy is probably exacerbated by the `` missing line '' problem in the plasma codes , but can be mitigated by consideration of the following .
first , by including multiple edges in our fits we can reduce the total optical depth by factors of a few leaving a factor of @xmath131 discrepancy .
second , if we consider the radiative transfer effects mentioned above ( i.e. , photoionization ) then the warm gas is over - ionized for its ( collisional ) temperature .
the inferred ( collisional ) emission measure is very sensitive to the temperature , and if we reduce the 0.1 - 0.2 kev temperature by a factor of @xmath17 the discrepancy of emission measures can be fully mitigated .
this exercise highlights the need for a more rigorous ( and complex ) treatment of the absorption and emission of the warm gas in order to deduce the mass of the absorber which is beyond the scope of our paper .
if our basic results are confirmed with the substantially higher quality data from _ chandra _ and _ xmm _ , then it will be appropriate to expend the effort to construct rigorous models of the warm absorber accounting for many edges and possible radiative transfer effects that are not currently available in xspec .
finally , it is also worth mentioning that ( these caveats aside ) the emission from the warm gas implied by our measurements does not violate published constraints in the fuv similar to that found for the the galaxies and groups in section 5.4 of paper3 . @xcite
have reported results in the fuv for m87 obtained with the hopkins ultraviolet telescope ( hut ) .
they obtain a @xmath132 upper limit of @xmath133 erg @xmath7 s@xmath8within @xmath28 of m87 for the flux of the ( 1034 ) line which is expected to be the strongest line associated with warm gas having temperatures between @xmath134 k. using quoted above we estimate that the strength of the line at the peak temperature of @xmath135 k ( e.g. , pistinner & sarazin 1994 ; voit & donahue 1995 ) is @xmath136 erg @xmath7 s@xmath8 .
ignoring the fact that is over - estimated as discussed above , the hut constraint implies that the must be different from @xmath135 k. from examination of figure 7 of @xcite we see that the luminosity decreases very rapidly with increasing temperature .
consequently , the emission implied by is consistent with the hut constraint provided that @xmath137 k. various conflicting reports of the detection of excess euv ( 0.07 - 0.25 kev ) emission above that expected from the hot ( x - ray ) plasma have been reported in clusters of galaxies using data from the _ extreme ultraviolet explorer _
( _ euve _ ) , but only for m87 does there appear to be a consensus among different groups of the reality of centrally concentrated excess euv emission ( lieu et al .
1996 ; berghfer et al .
if the excess _ euve _
emission is interpreted as arising from optically thin collisionally ionized gas then @xcite and @xcite infer temperatures ( @xmath138 kev ) and total masses within @xmath139 ( @xmath140 ) , which ( considering the modeling differences between our studies ) are in good agreement with those values we have inferred from the _ rosat _ data .
we mention that it has usually been reported that the _ euve _ `` soft excess '' of m87 increases with radius ( e.g , bonamente , lieu , & mittaz 2000 ) .
this result depends on the definition of the model of the hot gas that serves as the reference for the excess .
the radially increasing soft excess assumes that the reference model for the hot gas has been fitted to the pspc data using only data above 1 kev .
if instead all the pspc data are included in the fit , then there is no significant trend for the soft excess with radius in the _ euve _ data ( r. lieu 2000 , private communication ) .
this is consistent with our result for the two - phase models where the warm gas fraction is essentially constant for @xmath81 ( figure [ fig.2 t ] ) .
various problems with a gas origin of the euv emission have been discussed by @xcite , in particular the need for some as - yet - undiscovered heat source to prevent the gas from rapidly cooling to lower temperatures .
these problems have spurred many to explain the excess euv emission from other non - thermal processes ( see berghfer et al . 2000 and references therein )
. however , an independent test of the hypothesis of warm ionized gas is offered by its absorbing effect on the hot x - ray plasma .
our detection of absorption confined to energies 0.5 - 0.8 kev is precisely the signature expected from warm ( @xmath138 kev ) gas ( e.g. , krolik & kallman 1984 ) . the results from _ euve _ , _ rosat _ , and _ asca _ for m87 form a coherent picture of a multiphase medium consisting ( approximately ) of a warm and hot phase : the emission from the warm gas is detected with _ euve _ and _ rosat _ while the absorption from the warm gas is detected in _ rosat _ and _ asca _ . the warm ( @xmath1 k ) gas we have detected in both absorption and emission with
which is consistent with both the excess absorption seen in _
asca _ and the excess emission in _ euve _ , provides direct evidence for the material that may have cooled and dropped out of an inhomogeneous cooling flow ( e.g. , fabian 1994 ) .
although the warm gas is consistent with the amount of matter that could have been deposited by a cooling flow ( [ mass ] ) , it is difficult to explain the warm gas phase as an equilibrium configuration because ( 1 ) the gas at @xmath141 k can not be thermally supported in the potential of m87 , and ( 2 ) the cooling time of this gas is very short ( see section 5.5 of paper3 ) .
non - equilibrium models in the form of feedback from the central agn have been proposed to largely suppress cooling flows ( binney & tabor 1995 ; ciotti & ostriker 2000 ) . in the feedback model
whenever the black hole accretes a sufficient amount of gas to stimulate nuclear activity , the accompanying radiation stimulated by the accretion heats up the hot gas and prevents further cooling .
this is supposed to be a cyclical process such that the agn phase is sufficiently rare to be consistent with the lack of nuclear activity in most cooling flows . the spectacular radio map of m87 published by @xcite shows that radio emission pervades the hot gas of m87 out to radii of @xmath142 .
the distortions in the _ rosat _ x - ray contours present in annuli # 2 - 3 ( figure [ fig.m87 ] ) testify to the interaction of the radio emission associated with the jet and the hot gas @xcite .
the interpretation of the radio emission of m87 in terms of the agn feedback model has been discussed by @xcite .
our multiphase models of the _ rosat _ pspc data of m87 appear to support the picture of a standard inhomogeneous cooling flow that has been partially disturbed by the agn .
we focus the reader s attention on the radial profiles of both for the cooling flow model in figure [ fig.cf ] and the emission - measure fraction of warm gas in two - phase models shown in figure [ fig.2 t ] .
in particular , for the cooling flow models is consistent with a constant value for @xmath85 appropriate for a standard multiphase cooling flow .
but for annuli # 2 - 4 ( i.e. , @xmath28 to @xmath143 ) is zero . over this region
lies the asymmetrical features in the x - ray isophotes and the most prominent radio emission from the jet which would seem to support the notion that the agn has disturbed the cooling flow in this region . in the central bin ( @xmath144 )
we do see significant cooling ( especially for the two - phase models ) which is not surprising since the relaxation and cooling times are shortest there , and thus it would be the first to readjust after the disturbance from the agn .
it should be remembered that the scenario we have described has been based on the _ emission _ component of the warm gas .
the absorption given by the optical depth of the single edge is essentially constant ( or declining slowly ) with increasing radius .
consequently , although the agn may have inhibited the cooling of the warm gas in the central regions it apparently has not affected much the absorption properties of the gas . whether or not this is reasonable
requires a more complex model of the warm absorber / emitter than we have employed in this paper ( see [ mass ] ) .
if this description is correct for m87 then it very likely applies to other cooling flows . in this case
it will be critical to understand how this process evolves and how ubiquitous it is in cooling flow galaxies , groups , and clusters .
from analysis of spatially resolved , deprojected _ rosat _ pspc spectra we find the strongest evidence to date for intrinsic oxygen absorption and multiphase gas in the hot ism / igm / icm of a galaxy , group , or cluster . when attempting to describe the 0.2 - 2.2 kev _
emission of m87 by a single - phase hot plasma modified by standard galactic absorption the best - fitting model displays striking residuals in the spectrum ( figure [ fig.1t_spec ] ) : ( 1 ) excess emission above the model for 0.2 - 0.4 kev and ( 2 ) excess absorption below the model for 0.5 - 0.8 kev .
these features are apparent out to the largest radii investigated ( @xmath145 kpc ) and can not be attributed to errors in the calibration or the background subtraction ( see [ std ] ) .
the principal result is that the 0.5 - 0.8 kev absorption is consistent with that of a collisionally ionized plasma with a temperature of @xmath134 k , where the lack of evidence of absorption below 0.5 kev strongly excludes the possibility of absorption from cold material as has been assumed in essentially all previous studies of absorbing material in cooling flows . in fact , the excess _
emission _ observed between 0.2 - 0.4 kev , which is also manifested as sub - galactic column densities in models with standard ( cold ) absorbers ( [ single ] ) , has a temperature that is consistent with the emission from gas responsible for the 0.5 - 0.8 kev absorption and could not be explained if dust were responsible for both the absorption and soft emission .
( see section 5.1 of paper3 for discussion of other problems with dust . )
only multiphase models can provide good fits over the entire pspc bandpass while also yielding temperatures and fe abundances that are consistent with results from _ asca _ and _ sax _ at large radii ( [ multi ] ) .
both cooling flow and two - phase models indicate that the fraction of warm gas with respect to the total emission measure differs qualitatively for radii interior and exterior to @xmath142 ( @xmath146 kpc ) .
for @xmath58 the data are consistent with a constant ( or slowly varying ) fraction of warm gas as a function of radius .
but for @xmath147 the warm gas fraction varies from @xmath80 within the central bin @xmath148 to essentially zero within the next few bins ( out to @xmath149 ) .
this behavior is to be contrasted with the absorption optical depth profiles which are approximately constant with radius .
the oxygen absorption and soft emission from warm ( @xmath134 k ) gas in m87 we have detected using the _ rosat _ pspc ( 0.2 - 2.2 kev ) is also able to satisfy the detections of excess emission with data at lower energies ( 0.07 - 0.25 kev ) from _ euve _ ( [ euv ] ) and the detection of excess absorption at higher energies ( 0.5 - 10 kev ) from _ einstein _ and _ asca _ ( [ mass ] ) .
previous studies using only _ euve _ data ( e.g. , berghfer et al . 2000 ) could not decide between a non - thermal and thermal origin for the excess soft emission .
similarly , the previous detections of absorption with _ einstein _ @xcite and _ asca _ @xcite could not constrain the temperature of the absorber and always assumed a cold absorber with solar abundances . even previous studies with _ rosat _ that neglected data below @xmath63 kev also could not constrain the temperature of the absorber ( allen & fabian 1997 ; sanders et al .
2000 ) because of the strong dependence of the inferred absorption on the lower energy limit of the bandpass ( see our figure [ fig.1t_spec ] ) .
hence , the _ rosat _ detection of intrinsic absorption that is localized in energy ( 0.5 - 0.8 kev ) is the key piece of evidence for establishing the presence of warm ( @xmath134 k ) gas distributed throughout ( at least ) the central 100 kpc .
this evidence for a multiphase ism in m87 essentially confirms the original detection within the central @xmath2 using the _ einstein _ fpcs @xcite .
however , instead of intrinsic oxygen absorption @xcite inferred a super - solar o / fe ratio which does not agree with subsequent analyses of m87 using other instruments . the anomalous o / fe ratio
is probably attributed to either calibration error in the fpcs ( see tsai 1994 ) or to an underestimate of the continuum due to the absorption edges ( see end of [ previous ] ) .
although many previous studies have shown that the _ rosat _ data ( 0.1 - 2.4 kev ) of galaxy clusters rule out large quantities of cold absorbing gas in these systems ( e.g. , arabadjis & bregman 2000 ) , our spatial - spectral deprojection analyses of m87 in this paper and of a1795 in paper1 strongly suggest that large quantities of warm ( @xmath150 k ) gas are distributed ( at least ) throughout the central @xmath145 kpc of cluster cooling flows .
the total mass of the warm gas implied by the oxygen absorption is consistent with the amount of matter deposited by an inhomogeneous cooling flow ( [ mass ] ) . on the other hand , the mass deposition profile ( figure [ fig.cf_tau ] ) and the profile of warm emission fraction of the two - phase models ( figure [ fig.2 t ] )
indicate that the emission of the warm component is suppressed over @xmath151 where the radio emission from the agn jet clearly distorts the x - ray isophotes @xcite .
this coincidence suggests that the agn has influenced the hot ism in these central regions and may have suppressed the cooling emission of the warm component ( [ feedback ] ; see binney 1999 ) . within the central arcminute
the gas has apparently readjusted and is cooling while at large radii , @xmath58 , the cooling flow was not disturbed significantly by the agn .
a hybrid model of a standard cooling flow with agn feedback seems promising for m87 .
it is puzzling , however , why the oxygen absorption optical depth does not dip between @xmath152 as would be expected if the absorption and excess soft emission arise from the same material .
perhaps this is a result of the simplifications we have employed in this investigation ( e.g. , see [ mass ] ) , and with rigorous consideration of the radiative transfer and the absorption from several ionization states of oxygen ( and carbon and nitrogen ) a self - consistent description of the multiphase medium will be obtained .
the highly significant detection of intrinsic oxygen absorption in the _ rosat _ pspc data of m87 confirms the picture of a multiphase warm+hot medium in cooling flows deduced from the lower s / n pspc data of galaxies and groups in paper1 and paper3 . with new
chandra _ and _ xmm _ data the spatial distribution of the different phases can be mapped with novel accuracy , and thus the structure and evolution of cooling flows , and the role of agn feedback , will be elucidated .
i thank d. liedahl for discussions relating to the accuracy of plasma emission codes and the anonymous referee for helpful comments .
this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center .
support for this work was provided by nasa through chandra fellowship grant pf8 - 10001 awarded by the chandra science center , which is operated by the smithsonian astrophysical observatory for nasa under contract nas8 - 39073 .
allen , s. w. , & fabian , a. c. 1997 , , 286 , 583 allen , s. w. , di matteo , t. , & fabian , a. c. 2000 , , 311 , 493 anders e. , & grevesse n. , 1989 , geochimica et cosmochimica acta , 53 , 197 arabadjis , j. s. , & bregman , j. n. , 1999 , , 514 , 607 arabadjis , j. s. , & bregman , j. n. , 2000 , , 536 , 144 baluciska - church , m. , & mccammon , d. , 1992 , , 400 , 699 beiersdorfer , p. , lepson , j. k. , brown , g. v. , utter , s. b. , kahn , s. m. , liedahl , d. a. , mauche , c. w. 1999 , , 519 , l185 berghfer , t. w. , bowyer , s. , & korpela , e. 2000 , , 535 , 615 binney , j. j. , 1996 , in gravitational dynamics proc .
36th herstmonceux conf .
, ed o. lahav ( cambridge : cambridge university press ) , 89 binney , j. j. 1999 , in the radio galaxy m87 , ed .
rser & k. meisenheimer , ( springer - verlag ) , 136 binney , j. , & tabor , g. 1995 , , 276 , 663 bhringer , h. 1997 , in galactic and cluster cooling flows asp conf .
n. soker ( san francisco : asp ) , 11 bonamente , m. , lieu , r. , & mittaz , j.p.d .
2000 , , in press ( astro - ph/0001128 ) buote , d. a. , 1999 , , 309 , 695 buote , d. a. , 2000a , , 311 , 176 buote , d. a. , 2000b , , 532 , l113 ( paper1 ) buote , d. a. , 2000c , , 539 , 172 ( paper2 ) buote , d. a. , 2000d , , in press ( astro - ph/0001330 ) ( paper3 ) buote , d. a. , & fabian , a. c. 1998 , , 296 , 977 buote , d. a. , canizares , c. r. , & fabian , a. c. 1999 , , 310 , 483 canizares , c. r. , clark , g. w. , jernigan , j. g. , markert , t. h. 1982 , , 262 , 33 chen , l .- w .
, fabian , a. c. , & gendreau , k. c. , 1997 , , 285 , 449 ciotti , l. , & ostriker , j. p. 2000 , , submitted ( astro - ph/9912064 ) dacri , f. , de grandi , s. , & molendi , s. , 1998 , in the active x - ray sky : results from bepposax and rossi - xte , ed l. scarsi et al . , ( holland : elsevier science ) ( astro - ph/9802070 ) dixon , w. , hurwitz , m. , & ferguson , h. c. , 1996 , , 469 , l77 fabian a. c. , 1994 , , 32 , 277 fabian a. c. , 1996 , science , 271 , 1244 fabian , a. c. , nulsen , p. e. j. , canizares , c. r. 1984 , , 310 , 733 fabian , a. c. , hu , e. m. , cowie , l. l. , grindlay , j. , 1981 , , 248 , 47 feldman u. , 1992 , physica scripta , 46 , 202 harris , d. e. , owen , f. , biretta , j. a. , & junor , w. 2000 , in proceedings of the ringberg workshop : diffuse thermal and relativistic plasma in galaxy clusters , eds .
h. bhringer , l. feretti , and p. schuecker , in press ( astro - ph/0002464 ) hasinger , g. , boese , g. , predehl , p. , turner , t. j. , yusaf , r. , george , i. m. , & rohrbach , g. , 1995 mpe / ogip calibration memo cal / ros/93 - 015 ver .
1995 may 8 ishimaru , y. , & arimoto , n. , 1997 , , 49 , 1 johnstone , r. m. , fabian , a. c. , edge , a. c. , & thomas , p. a. 1992 , , 255 , 431 kaastra , j. s. , & mewe , r. , 1993 , , 97 , 443 krolik , j. h. , & kallman , t. r. , 1984 , , 286 , 366 liedahl , d. a. , 1999 , in atomic data needs for x - ray astronomy , eds .
bautista , t.r .
kallman , a.k .
pradhan , http://heasarc.gsfc.nasa.gov/docs/heasarc/atomic/ liedahl , d. a. , osterheld , a. l. , & goldstein , w. h. , 1995 , , 438 , l115 lieu , r. , mittaz , j. p. d. , bowyer , s. , lockman , f. j. , hwang , c .- y . , & schmitt , j. h. m. , 1996 , , 458 , l5 mewe r. , gronenschild e. h. b. m. , & van den oord g. h. j. , 1985 , , 62 , 197 morrison , r. , & mccammon , d. , 1983 , , 270 , 119 nulsen , p. e. j. , & bhringer , h. , 1995 , , 274 , 1093 owen , f. n. , eilek , j. a. , & kassim , n. e. 2000 , , in press ( astro - ph/0006150 ) peres , c. b. , fabian , a. c. , edge , a. c. , allen , s. w. , johnstone , r. m. , white , d. a. 1998 , , 298 , 416 pistinner , s. , sarazin , c. l. 1994 , , 433 , 577 sanders , j. s. , fabian , a. c. , & allen , s. w. 2000 , , in press ( astro - ph/0006394 ) snowden , s. l. , mccammon , d. , burrows , d. n. , & mendenhall , j. a. , 1994 , , 424 , 714 snowden , s. l. , turner , t. j. , george , i. m. , yusaf , r. , predehl , p. , & prieto a. 1995 , ogip calibration memo cal / ros/95 - 003 stewart , g. , canizares , c. r. , fabian , a. c. , nulsen , p. e. j. 1984 , , 278 , 536 sutherland , r. s. , & dopita , r. s. , 1993 , , 88 , 253 tsai , j. c. 1995 , , 429 , 119 voit , g. m. , & donahue , m. 1995 , , 452 , 164 white , d. a. , 2000 , , 312 , 663 white , d. a. , fabian a. c. , johnstone r. m. , mushotzky , r. f. , & arnaud , k. a. , 1991 , , 252 , 72 yan , m. , sadeghpour , h. r. , & dalgarno , a. , 1998 , , 496 , 1044 | we present a re - analysis of the _ rosat _ pspc data within the central 100 kpc of m87 to search for intrinsic oxygen absorption similar to that recently measured in several galaxies and groups . since m87 is the brightest nearby galaxy or cluster possessing an average temperature ( @xmath0 kev ) within the pspc bandpass ,
it is the ideal target for this study . using a spatial - spectral deprojection analysis we find the strongest evidence to date for intrinsic oxygen absorption in the hot gas of a galaxy , group , or cluster .
single - phase plasma models modified by intervening galactic absorption can not fit the 0.2 - 2.2 kev pspc data as they under - predict the 0.2 - 0.4 kev region and over - predict the 0.5 - 0.8 kev region where the emission and absorption residuals are obvious upon visual inspection of the spectral fits .
these absorption and emission features are significant out to the largest radii investigated . since the excess emission between 0.2 - 0.4 kev rules out intrinsic absorption from cold gas or dust , the most reasonable model for the excess emission and absorption features is warm , collisionally ionized gas with a temperature of @xmath1 k. simple multiphase models ( cooling flow , two phases ) modified by both intervening galactic absorption and by a single oxygen edge provide good fits and yield temperatures and fe abundances of the hot gas that agree with previous determinations by _ asca _ and _ sax _ .
the multiphase models of m87 inferred from the pspc can account for the excess euv emission observed with _
euve _ and the excess x - ray absorption inferred from _ einstein _ and _ asca _ data above 0.5 kev .
this evidence for a multiphase warm+hot ism in m87 essentially confirms the original detection by canizares et al . within the central @xmath2 using the _ einstein _ fpcs .
although the total mass of the warm gas implied by the oxygen absorption is consistent with the matter deposited by a cooling flow , the suppression of the mass deposition rate and the distortion of the x - ray isophotes in the region where the radio emission is most pronounced suggest some feedback effect from the agn on the cooling gas . |
why are we interested in elastic vector meson production ? first of all the process @xmath0 provides us with well distinguishable experimental signals in a wide range of the @xmath1 c.m .
energy @xmath2 , the virtuality of the photon @xmath3 , and the mass of the vector meson @xmath4 .
quite some data are already available for @xmath5 and @xmath6 , and even for the heavy @xmath7 first measurements were published recently . in the future the range in @xmath3 and @xmath2 and the precision of the data will increase .
this enables us to study vector meson production in detail in the very interesting regime where the transition from soft to hard qcd dynamics is expected ( and already seen ) to take place .
in addition , there is hope to make use of the high sensitivity of this process on the gluon distribution @xmath8 in the proton to constrain this quantity at small values of @xmath9 better than through other processes . in the following we first sketch the basic picture of elastic vector meson production . in section 2
we briefly discuss different theoretical models which are not based on the two gluon exchange picture which is then introduced in section 3 . there
, starting from the basic leading order result known for long time , we develop corrections which improve the leading order formula . in section 4 recent issues in pqcd calculations as off - diagonal parton distributions , the influence of the vector meson wave function and an alternative approach using parton - hadron duality
are discussed .
we mainly concentrate on diffractive @xmath10 meson electroproduction , but the presented perturbative model is also successful in the case of @xmath6 and @xmath7 .
section 4 contains our conclusions and outlook . in fig .
[ fig1 ] the basic picture for the process @xmath11 is shown : first the photon with virtuality @xmath12 fluctuates into a quark - antiquark pair .
this @xmath13 fluctuation then interacts elastically with the proton @xmath14 , where the zig - zag line represents the ( for the moment unspecified ) elastic interaction with the proton .
the @xmath1 centre - of - mass energy is denoted by @xmath2 , @xmath15 whereas @xmath16 is the four - momentum transfer squared .
( in the following we will mainly restrict ourselves to the case of small @xmath17 . ) the shaded blob at the right stands for the formation of the vector meson @xmath18 , which , to leading order , has to form from the @xmath13 pair with invariant mass squared @xmath19 . in @xmath1 collisions , scaledwidth=70.0% ]
it is important to note that at high energy @xmath2 corresponding to small values of @xmath9 , @xmath20 the timescales involved in the problem are very different , which is often called @xmath21 or @xmath22 , is common in diffractive physics and should not be confused with the ordinary bjorken-@xmath9 , @xmath23 . ] : the typical lifetime of the @xmath24 fluctuation as well as the time for the formation of the vector meson @xmath18 is much longer than the duration of the interaction with the proton , i.e. @xmath25 . therefore the basic amplitude factorizes , as sketched already in fig .
[ fig1 ] , into the @xmath13 fluctuation , the interaction amplitude @xmath26 and the wave function of the vector meson @xmath18 , @xmath27 and the process becomes calculable within various models .
formally it has been shown that for @xmath3 larger than all other mass scales in the process there is factorization into a hard scattering subprocess , non - perturbative ( and off - diagonal , as will be discussed later ) parton distributions and the meson wave function.@xcite this strict proof of factorization holds for longitudinally polarized photons , whereas meson production through transversely polarized photons is shown to be suppressed by a power of @xmath28 .
let us now turn to the discussion of different models for the @xmath1 interaction .
the following short section is far from being a review of this rich field , but is intended to give a hint at some non - perturbative models , which contrast the perturbative description of diffractive scattering , which is the main subject of this article .
+ @xmath29 we will not cover approaches based on vector meson dominance ( see e.g. @xcite ) .
+ @xmath29 for regge - phenomenology - based models of ( one or two ) pomeron exchange we refer the reader to @xcite .
+ @xmath29 _ the model of the stochastic qcd vacuum _ + dosch , gusset , kulzinger and pirner
@xcite have developped a model of the interaction with the proton , which is similar to the semi - classical model of buchmller discussed in @xcite in the context of inclusive diffractive dis .
this model , originally used for hadron - hadron scattering , leads to linear confinement and predicts a dependence of the high - energy scattering on the hadron size .
it gives a unified description of low energy and soft high - energy scattering phenomena .
dosch et al .
approximate the slowly varying infrared modes of the gluon field of the proton by a stochastic process . via a path integral
method they average over all possible field configurations . for the splitting of the photon into the @xmath13 pair and for the description of the vector meson they use light cone wavefunctions . within their model
they are able to calculate the @xmath3 dependence of the cross section , as well as the dependence on the momentum transfer @xmath17 , @xmath30 , and the ratio of the longitudinal to the transverse cross section , @xmath31 , where longitudinal and transverse refer to the polarization of the photon .
their results are in fair agreement with experimental data .
there is no prediction for the @xmath2 dependence of the cross section .
+ @xmath29 rueter has extended the model of dosch et al . to also describe the @xmath2 dependence of the cross
section.@xcite he achieves this by using a phenomenological model based on the exchange of one soft and one hard pomeron , each being a simple pole in the complex angular momentum plane , similar to the donnachie - landshoff model @xcite . for the very hard components of the photon fluctuations he treats the interaction perturbatively and achieves a good description of the experimentally observed transition from the soft to the hard regime .
to leading order in qcd the zig - zag line in fig .
[ fig1 ] , which stands for the elastic scattering via the exchange of a colourless object with the quantum numbers of the vacuum , can be described by two gluons . if the scale governing the ( transverse ) size of the photon fluctuation is large compared to the typical scale of non - perturbative strong interactions , i.e. if @xmath32 then the coupling of the two gluons to the @xmath13 fluctuation can be treated reliably within perturbative qcd ( pqcd ) .
another kinematic regime , where pqcd is applicable , is high-@xmath17 diffraction .
there the hard scale which is needed to ensure the validity of the perturbative treatment is given by the large value of the momentum transfer @xmath17 , and one expects high-@xmath17 diffractin to be a good place to search for the perturbative pomeron.@xcite it has been shown some time ago that due to the factorization property of the process the coupling of the two gluons to the proton can , in the leading logarithmic approximation , be identified with the ordinary ( diagonal ) gluon distribution in the proton.@xcite we will come back to this point later when discussing the importance of off - diagonal gluon distributions .
the basic leading order formula for diffractive vector meson production is given by @xcite @xmath33 ^ 2 \left ( 1 + \frac{q^2}{m_v^2 } \right)\ , , \label{eqleading}\ ] ] where @xmath34 is the electromagnetic coupling and the gluon distribution is sampled at the effective scale @xmath35 in eq .
( [ eqleading ] ) the non - relativistic approximation for the vector meson wave function is used and the coupling of the vector meson to the photon is encoded in the electronic width @xmath36 .
note that eq .
( [ eqleading ] ) is valid for @xmath37 . in the approach discussed in the following
there is no prediction for the @xmath17 dependence of the cross section , which is assumed to be of the exponential form @xmath38 with an experimentally measured slope - parameter @xmath39 , which may depend on the vector meson @xmath18 and on @xmath3 . on the other hand , eq .
( [ eqleading ] ) makes predictions for both the @xmath3 and the @xmath2 dependence of the cross section for longitudinally and transversely polarized photons for all sorts of vector mesons , as long as either @xmath3 or @xmath19 is large enough to act as the hard scale .
it is obvious that the @xmath2 dependence comes entirely from the gluon distribution @xmath8 , which enters quadratically in the cross section . in the following
we will discuss several improvements of the leading order formula .
+ @xmath29 eq .
( [ eqleading ] ) contains only the leading imaginary part of the positive - signature amplitude @xmath40 the real part of the amplitude can be restored using dispersion relations : @xmath41 where @xmath42 is given by the logarithmic derivative @xmath43 for the case of @xmath10 production , the contributions from the real part are roughly 15% . for @xmath6 production in the hera regime
they amount to approximateley 20% and are even bigger for @xmath7 production @xcite , where larger values of @xmath9 are probed .
+ @xmath29 in fig .
[ fig2 ] one of four leading order diagrams for the two gluon exchange model is shown with some kinematic variables which will be used below . in the general case the two gluons @xmath44 ,
@xmath45 have different @xmath9 , @xmath46 and transverse momenta @xmath47 , @xmath48 .
the leading logarithmic approximation of the @xmath49 loop integral ( indicated by the circle in fig .
[ fig2 ] ) leads to the identification with the integrated gluon distribution @xmath50 at the effective scale @xmath51 defined in eq .
( [ eqdefqbarsq ] ) . beyond leading logarithmic
accuracy one has to perform the @xmath49 integral over the unintegrated gluon distribution @xmath52 .
this can lead to numerical results which are , depending on the kinematical regime , twice as big as the result from eq .
( [ eqleading]).@xcite although here we are considering elastic production at small momentum transfer @xmath17 , the timelike vector meson with mass @xmath4 has to be produced from the spacelike ( or real ) photon with virtuality @xmath3 .
this means , that even if there is no transverse momentum transfer , @xmath53 , there has to be a difference @xmath54 in the longitudinal momentum of the two gluons @xmath44 and @xmath45 .
therefore the identification with the ordinary diagonal gluon distribution @xmath8 is only a good approximation for very small values of @xmath9 and @xmath17 , and in the general case the process @xmath55 depends on off - diagonal parton distributions.@xcite their importance for diffractive vector meson production will be discussed in the following .
off - diagonal ( also called `` skewed '' or non - forward ) parton distributions are much studied recently . in the case of small @xmath17 scattering
the skewedness comes from the difference between @xmath9 and @xmath46 of the two gluons @xmath44 and @xmath45 , and the cross section can be shown to be proportional to the square of a skewed gluon distribution , @xmath56 here @xmath57 , @xmath58 , and @xmath59 is the mass squared of the intermediate @xmath60 pair .
( taking the leading imaginary part of the amplitude corresponds to cutting the amplitude as indicated by the dashed line in fig .
[ fig2 ] and putting both @xmath61 and @xmath62 on - shell , which in turn fixes @xmath9 .
@xmath46 has to accomodate the difference between @xmath63 and @xmath4 and it not fixed due to the integration over all possible quark ( and antiquark ) momenta . at leading logarithmic order @xmath64
and we can put @xmath65 . ) for arbitrary kinematics skewed parton distributions are not connected with the diagonal ones and are unknown non - perturbative objects .
however , in the case of small @xmath9 , they are determined completely by the diagonal ones.@xcite the ratio of skewed to diagonal gluon distribution is given by @xmath66 here @xmath67 is the usual gamma - function and the effective power @xmath42 can be obtained from the logarithmic derivative of the amplitude @xmath68 for the @xmath0 cross section , @xmath69 as will be shown below , the magnitude of the resulting correction factor for the total cross section , @xmath70 , can be sizeable , especially for large @xmath3 or @xmath19 .
another important issue is the treatment of the vector meson wave function . as sketched in fig .
[ fig1 ] and eq .
( [ eqfac ] ) it enters the amplitude via a convolution with the scattered @xmath13 fluctuation . in eq .
( [ eqleading ] ) the non - relativistic approximation was adopted .
this means , that quark and antiquark equally share the longitudinal momentum of the photon , i.e. @xmath71 , and that there is no internal ( transverse ) momentum @xmath72 in the @xmath13 bound state .
therefore , in this naive approximation , @xmath73 and @xmath74 .
while this simplification may be suitable for heavy mesons like the @xmath7 , it is clear that the non - relativistic approximation has to break down for light quarks .
various groups have worked on improving this approximation by including the fermi motion of the quarks in the meson by using a nontrivial wave function.@xcite different models for the meson wave function were used which lead to quite different correction factors : whereas in gaussian models there is no strong suppression @xcite , the large @xmath72 tail typical for wave functions from non - relativistic potential models seems to lead to large corrections @xcite . on the other hand , considering that within these potential models a big part of the @xmath75 corrections comes from a regime , where @xmath72 is bigger than the quark mass itself , these large corrections may well be an artefact of the non - relativistic approximation .
another related problem is the question , which mass for the quarks should be used in the perturbative formulae .
note that eq .
( [ eqleading ] ) is written in terms of the vector meson mass @xmath4 .
however , as discussed in @xcite , the full expressions used to include higher order ( relativistic ) corrections contain the quark mass @xmath76 instead of @xmath4 .
as the ratio @xmath77 enters with a high power , this difference is not negligible and should be taken into account in the calculation of the @xmath75 corrections applied to eq .
( [ eqleading ] ) .
in addition , it is well known that there are other relativistic corrections , which in principle have to be taken into account in a consistent way .
as pointed out by hoodbhoy @xcite , gauge invariance is only preserved if higher fock states ( @xmath78 , @xmath79 ) are included in the wave function . in doing
so he arrives at the conclusion , that the relativistic corrections to the quark propagators plus the corrections from the higher fock states amount to only a few percent for @xmath6 production , in agreement with @xcite .
after all large relativistic corrections can probably be excluded , but , as different approaches lead to quite different results there remains a considerable uncertainty and the issue a hot topic . in this section we will discuss an alternative approach , which avoids the meson wave function and leads to results which are in surprisingly good agreement with available data .
it was proposed in @xcite for @xmath10 meson electroproduction , where the hard scale is provided by @xmath3 , not by @xmath80 .
dependence @xmath81 with @xmath82 gev@xmath83 for @xmath84 gev@xmath85 and the @xmath2 behaviour of the cross section @xmath86 indicate that @xmath10 meson electroproduction is not a soft , but mainly a hard process . ] due to the tiny @xmath87 and @xmath88 quark masses , in the case of the @xmath10 non - relativistic approximations can not be justified , and the wave function is not very well known .
now the crucial problem was , that all naive predictions for the ratio of the longitudinal to the transverse cross section , which are based on the perturbative formula ( [ eqleading ] ) , lead to @xmath89 this is much too steep and incompatible with experimental data ( see below ) .
the inclusion of effects from a light cone wave function for the @xmath10 does not change the picture considerably .
receives large contributions from the small @xmath72 region , which is non - perturbative . but those contributions would cause the transverse cross section to fall off even faster with increasing @xmath3 and therefore worsen the problem @xcite . ]
these observations indicate that the main effects are not coming from the @xmath10 wave function and lead to the proposal of a different model in @xcite : there the cross section for @xmath10 production is predicted via perturbative @xmath90 and @xmath91 quark pair electroproduction together with the principle of parton - hadron duality ( phd ) @xcite .
phd means that the integral of the parton ( @xmath13 ) production cross section over a mass interval @xmath92 is approximately equal to the sum over all ( corresponding ) possible hadron production cross sections in the same mass interval . in the region
@xmath93 the production of more complicated partonic configurations ( like @xmath94 , @xmath95 , @xmath96 , etc . )
is heavily suppressed . on the hadronic side the @xmath10 resonance ( plus the small admixture of the @xmath97 ) with its decay into two ( three ) pions
completely saturates the cross section .
therefore we can well approximate the @xmath10 production cross section @xmath98 by @xmath99 where @xmath100 and @xmath101 have to be chosen to embrace the @xmath10 resonance appropriately , i.e. @xmath102 gev@xmath85 .
the factor 0.9 on the right side of eq .
( [ eqphdrho ] ) corrects for the contributions from @xmath97 production . the perturbative formulae for the @xmath13 production cross section
are derived from the amplitudes depicted in fig .
[ fig2 ] and can be written in terms of the conventional spin rotation matrices @xmath103 ( see @xcite for details ) : @xmath104 where @xmath105 is the electric charge of the quark @xmath61 , @xmath34 the electromagnetic coupling and @xmath106 the polar angle of the quark @xmath61 in the @xmath13 rest frame with respect to the proton direction ( @xmath107 ) .
@xmath108 are integrals over the gluon @xmath49 and given by @xmath109 with @xmath110 being the unintegrated gluon distribution and @xmath111 in eqs .
( [ eqsigmalt ] ) the different rotation matrices appropriately reflect the different spin states of the @xmath13 produced from longitudinal and transverse photons , and the integrals @xmath112 contain the scattering off the proton via the two gluon exchange .
channel helicity conservation ( schc ) , i.e. the produced @xmath10 has the helicity of the virtual photon . however , there are small violations of schc ( e.g. @xmath113 ) which can be successfully described in a framework similar to the one discussed here . for a discussion of recent measurements of the 15 spin density matrix elements of @xmath10 production compared to different theoretical predictions
see @xcite . ]
in order to pick up only those @xmath90 , @xmath91 configurations which correspond to the quantum numbers of the @xmath10 , one has to project out the @xmath114 states .
this can be easily done on amplitude level with the same rotation matrices @xmath103 , see @xcite .
( even higher spin states like the @xmath115 can be projected out using the corresponding @xmath88 function @xcite . )
it is important to note that through the projection on amplitude level the longitudinal and transverse cross sections @xmath116 are less infrared sensitive than eqs .
( [ eqsigmalt ] ) , and therefore @xmath117 becomes calculable without a large uncertainty from the treatment of the ( non - perturbative ) infrared region . for the complete numerical predictions one also has to include the contributions from the real part of the amplitudes and the skewed gluon distribution as discussed above .
both effects are taken into account on amplitude level .
( [ eqsigmalt ] ) give the cross section differential in @xmath17 for @xmath37 .
to arrive at the @xmath17 integrated total cross section one assumes the exponential form @xmath38 .
the slope @xmath39 can be taken from experiment or theoretical models and depends in general on @xmath118 , @xmath2 and @xmath3 .
for more details we refer the reader to @xcite . to go beyond the leading order prediction in a completely consistent way would require in addition the full set of next - to - leading order gluonic corrections to the @xmath119-@xmath120 vertex .
these corrections are not known yet , but can be estimated by a @xmath121 factor @xcite .
similar to the drell - yan process , there are @xmath122 enhanced terms , which come from the @xmath123 terms in the double logarithmic sudakov form factor .
resummation of those leading corrections results in the @xmath121 factor @xmath124 which leads to a considerable enhancement of the cross section . predicted within the phd model as described in the text compared to recent h1 data @xcite .
_ continuous line _ : using the skewed gluon distribution , _ dashed line _ : without skewing , scaledwidth=70.0% ] in fig .
[ fig3 ] the complete numerical prediction for @xmath125 using the phd model in the @xmath121 factor was chosen as @xmath126 . for more details
see @xcite . ]
is shown as a function of @xmath3 together with recent h1 data.@xcite the continuous line includes all the effects discussed above , whereas the dashed line does not include the skewed gluon .
the importance of the off - diagonal gluon for the @xmath3 behaviour of the cross section is obvious and the effect seems to be required to describe the data .
of course the model prediction is not free from uncertainties like the choice of the mass interval @xmath127 in eqs .
( [ eqphdrho ] ) or the scale of @xmath128 in the @xmath121 factor . these ( and other ) uncertainties are discussed in detail in @xcite , but they affect mainly the normalization of the cross section and do not spoil the good agreement with the experimental data
. predicted within the phd model ( _ continuous line _ ) as described in the text compared to experimental data @xcite and the naive prediction ( _ dashed line _ ) from eq .
( [ eqlovertnaiv]),scaledwidth=60.0% ] in fig .
[ fig4 ] the prediction of the phd model for the ratio @xmath31 is shown as a continuous line .
it agrees fairly well with the data points , which show a very modest rise with @xmath3 in contrast to the naive prediction eq .
( [ eqlovertnaiv ] ) ( dashed line ) .
thus , in the phd picture , it is _ not _ the @xmath10 wave function , but the dynamics of the @xmath13 pair creation from longitudinal and transverse photons together with the off - diagonal two gluon interaction and the projection onto the @xmath129 state , that determines the @xmath3 dependence and the ratio @xmath31
. production as predicted in the phd model @xcite compared to recent h1 data @xcite , scaledwidth=60.0% ] it is important to note that the phd model also works in the case of massive quarks and heavy mesons .
starting from formulae for diffractive heavy quark production @xcite and modifying the projection formalism appropriately , elastic @xmath7 photoproduction was recently predicted using phd in agreement with first measurements.@xcite the same formalism can also be applied to diffractive @xmath6 production.@xcite again , as shown in fig .
[ fig5 ] , there is a surprisingly good agreement between the predicted cross section as a function of @xmath3 and the experimental data @xcite .
elastic vector meson production is a rich field , both from experimental and theoretical point of view .
different theoretical models describe the data , and more precise data in an increased kinematical range will be needed to clarify the situation .
we have briefly discussed some non - perturbative models , but mainly concentrated on perturbative approaches .
we have shown that with recent improvements pqcd - based approaches work very well and are in agreement with data .
the fairly large impact of skewed parton distributions on the predictions within these models is supported by the data .
there is good hope that in future we will be able to discriminate between the different models and to understand elastic vector meson production in more detail . by combining different observables from different processes elastic vector meson production with its high sensitivity to the gluon at small @xmath9
will finally help to constrain the gluon much better .
for this much effort will be needed also from the theoretical side in order to increase the precision of the calculations .
i would like to thank g. grindhammer , b. kniehl and g. kramer for the good organization of this stimulating and enjoyable workshop .
i also thank genya levin , alan martin and misha ryskin for pleasant collaborations . | the elastic production of vector mesons at hera is discussed from the theoretical point of view .
we briefly review different models , their successes and shortcomings .
main emphasis is put on recent issues in perturbative qcd calculations .
models including the vector meson wave function are compared with an approach based on parton - hadron duality .
we discuss several refinements of these models in some detail , including the important role of off - diagonal parton distributions .
* desy 99 - 154 * + * hep - ph/9910329 * + * september 1999 * + * theory of elastic vector meson production * + thomas teubner + _ deutsches elektronen - synchrotron desy , notkestrasse 85 , d-22607 hamburg , germany _ + |
hadron production from energetic jets in high - energy nuclear collisions is often parameterized through fragmentation functions using the universality of the process as given by qcd factorization theorems @xcite . on a more microscopic level
, one can achieve a good description of jet hadronization through a perturbative evolution of the parton spectrum inside the jet using dglap splitting kernels from the largest momentum scale to some lower cutoff @xmath2 , followed by a non - perturbative hadronization model like the lund string model applied to the parton shower .
the event generator pythia @xcite has succesfully implemented such a strategy to describe hadron production in @xmath0 collisions at high energies and in numerous other processes .
on the other hand , it has been suggested that hadron production can often be described through a process of quark recombination or coalescence @xcite .
it is an intriguing idea to combine the concepts of quark recombination and parton showers since it would easily generalize to the hadronization of jets in dense environments as in relativistic heavy ion collisions .
hwa and yang have pioneered the application of quark recombination to jet showers in ref .
however , the parton showers in their work were not obtained from first principles but fitted to measured hadron spectra .
they also ignored the smallness of the number of partons in each jet by applying recombination to averaged spectra instead of event - by - event showers .
fluctuations from shower to shower are expected to lead to significant changes to the hadron spectrum at large hadron momentum fraction @xmath3 .
using @xmath4 collisions as an example , we show that vacuum fragmentation functions can be reproduced in an event - by - event approach that uses a perturbative shower evolution in the vacuum down to a scale @xmath2 , a minimal modeling of non - perturbative effects on partons below @xmath2 , and finally recombination of the quarks and antiquarks into hadrons .
using pythia 6.3 @xcite , we generate an ensemble of parton showers from @xmath5 collisions at center - of - mass energy @xmath1 .
the light quark , strange quark and gluon spectra as functions of momentum at a cutoff scale @xmath6 are shown in the left panel of fig .
[ gluondecaygraph ] .
since gluons in the parton showers can not directly recombine into hadrons , we force them to decay into quark - antiquark pairs after the perturbative evolution is stopped .
this is a non - perturbative process , and we use constituent quark masses of @xmath7 for light quarks and @xmath8 for strange quarks with the remaining gluon virtuality randomly chosen in the interval between @xmath9 and @xmath2 .
the transverse momentum spectra of the decay light and strange quarks together with the total light and strange quark spectra are shown in the right panel of fig .
[ gluondecaygraph ] .
following the approach to quark recombination in ref .
@xcite the probability for quarks and antiquarks to form a hadron is taken to be proportional to the quark wigner function of the hadron . assuming the quark wave functions of the hadrons to be those of harmonic oscillators , the hadron wigner function is then gaussian in both the coordinate and the momentum space . here
we neglect spatial information , assuming shower partons with similar momenta to be close in space .
hence we only consider the momentum - space wigner function . for the production of a meson from a quark and an antiquark ,
the probabilities are @xmath10 respectively for @xmath11-wave and @xmath12-wave mesons @xcite . here
@xmath13 is the relative momentum between the quark and antiquark in the meson rest frame , @xmath14 is the width of the quark and antiquark relative momentum distribution in the meson , and @xmath15 is the statistical factor for the colored , spin-@xmath16 quark and antiquark to form a colorless meson . for the mesons @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 considered in our study , we have @xmath26 , @xmath27 , and @xmath28 . because of the color charge of a gluon , its decay quark and antiquark
are excluded from recombining into a color singlet meson . for @xmath11-wave baryons ,
the production probability from three quarks is @xcite @xmath29 where @xmath30 and @xmath31 are the relative momenta of the recombining quarks of momenta @xmath32 , @xmath33 , and @xmath34 in the rest frame of the baryon .
the statistical factors for the baryons @xmath35 , @xmath36 , @xmath14 , and @xmath37 considered in our study are @xmath38 and @xmath39 .
the momentum width @xmath14 is related to the oscillator constant in the quark wave function . here
we take them as parameters .
using @xmath40 , @xmath41 , and @xmath42 , we have calculated their respective transverse momentum spectra from the recombination of quarks and antiquarks on an event - by - event basis .
the spectra of directly produced pions , kaons and nucleons are shown by star symbols in the left , middle , and right panels of fig .
[ recomb ] , respectively .
they underestimate those obtained from pythia via string fragmentation shown by solid lines . after including contributions from the decay of resonances via @xmath43
we significantly improve the comparison with the pythia results as shown by the dashed lines in fig .
[ recomb ] .
we note that not all quarks and antiquarks are used in the recombination model to form hadrons . as shown in fig .
[ probremn ] , the recombination probability is close to one for low momentum quarks and decreases with increasing momentum .
the small number of non - recombined quarks thus consists mostly of leading quarks that are far removed in momentum space from the rest of the parton shower .
their contribution to hadron production can be included via independent fragmentations at the low momentum scale @xmath2 . because of their small numbers , the final hadron spectra are , however , not much affected .
our study indicates that hadron production from jets can be described by the recombination of shower partons inside the jets .
our approach provides a promising framework to study the medium modification of fragmentation functions in the presence of a quark gluon plasma ( qgp ) by allowing shower partons to recombine with the thermal partons in the qgp .
this work was supported in part by the u.s .
national science foundation under grants nos .
phy-0847538 and phy-1068572 , the jet collaboration and the us department of energy under contract no .
de - fg02 - 10er41682 , and the welch foundation under grant no .
99 sjostrand t , mrenna s , and skands p z 2006 _ jhep _ * 0605 * 026 das k p and hwa r c 1978 _ phys .
_ b * 68 * 459 fries r j , muller b , nonaka c and bass s a 2003 _ phys .
lett . _ * 90 * 202303 greco v , ko c m and levai p 2003 _ phys .
lett . _ * 90 * 202302 greco v , ko c m and levai p 2003 _ phys .
_ c * 68 * 034904 fries r j , muller b , nonaka c and bass s a 2003 _ phys .
_ c * 68 * 044902 hwa r c and yang c b 2004 _ phys .
_ c * 70 * 024904 cho s_et al .
_ [ exhic collaboration ] 2011 _ phys . rev .
_ c * 84 * 064910 oh y , ko c m , lee s h and yasui s 2009 _ phys . rev .
_ c * 79 * 044905 | we study hadron production in jets by applying quark recombination to jet shower partons . with the jet showers obtained from pythia and augmented by additional non - perturbative effects
, we compute hadron spectra in @xmath0 collisions at @xmath1 . including contributions from resonance
decays , we find that the resulting transverse momentum spectra for pions , kaons , and protons reproduce reasonably those from the string fragmentation as implemented in pythia . |
the discovery of high temperature cuprate superconductors @xcite triggered a renewed interest in quasi two - dimensional(2d ) antiferromagnetic heisenberg spin model due to the belief that superconductivity is closely associated with the antiferromagnetic ordering in copper oxide planes in undoped parent compounds@xcite . in this regard , the antiferromagnetic @xmath0 quantum spin-@xmath8 model on 2d square lattice , with @xmath1 as frustration parameter ,
has been studied extensively@xcite .
the quantum fluctuation induced by frustration destroys the semi - classical nel order , characterized by the magnetic wave - vector @xmath9 . beyond a critical value of @xmath1
, the frustration drives the spin system to a new semi - classical order called the collinear order characterized by the magnetic wave - vector @xmath10 or @xmath11 . at zero temperature ,
the quantum spin system is believed to have nel order for @xmath12 and collinear order for @xmath13 . for intermediate values @xmath14 ,
the system is believed to exist in quantum paramagnetically disordered state @xcite .
the boundary of the quantum paramagnetic@xmath15collinear order has not been conclusively pinned down and there has been work to suggest it to be at @xmath16 @xcite or more .
+ more recently , iron based superconductors have been found to have layered structure @xcite and provide further motivation to study @xmath7 model @xcite .
the role of @xmath2 on @xmath0 model has been examined through several schemes like coupled - cluster , rotation - invariant green s function @xcite , effective field theory@xcite , self consistent spin wave theory , series expansion and first order spin - wave theory@xcite . for the antiferromagnetic 2d @xmath0 model in the intermediate regime of inplane next nearest neighbor coupling @xmath17 , it has been found that the quantum paramagnetic order narrows down @xcite with increase in interlayer coupling @xmath2
. the critical value of @xmath2 at which the paramagnetic order disappears , has not been uniquely found and appears to depend strongly on the method of study employed @xcite .
+ in the present work we use exact diagonalization to study , at zero as well as finite temperatures , the layered frustrated quantum spin-@xmath8 system with model hamiltonian @xmath18 where @xmath19 and @xmath20 are inplane nearest and next nearest neighbors respectively and @xmath21 represents the interplane nearest neighbors ; the couplings @xmath22 being the isotropic exchange integrals for the respective neighbors which have been taken to be positive .
+ in order to validate our exact diagonalization code , we carried out diagonalization on @xmath23 site spin-@xmath8 chain with nearest neighbor coupling @xmath24 only and found the ground state energy per spin to be @xmath25 , in agreement with bethe ansatz results .
.the ground state and first two excited - state energies @xmath26 per spin , with total spin of the respective states given in parenthesis , on a @xmath27 two - dimensional lattice for spin-@xmath8 heisenberg antiferromagnet for several values of next nearest - neighbor exchange coupling @xmath1 . here
@xmath24 is taken to be 2 , as in reference @xcite . [
cols="<,^,^,^",options="header " , ] to further test the validity of the code on 2d lattice , we diagonalized the spin-@xmath8 hamiltonian for @xmath27 lattice with nearest and next nearest neighbor couplings @xmath24 and @xmath1 .
the ground and first two excited state energies per spin along with total spin of the respective states are given in table [ 4x4_latiice_results ] and is in agreement with the results of @xcite .
the same code has earlier been used to analyze the quasi-1d characteristics of sr@xmath28cu(po@xmath29)@xmath28 and ba@xmath28cu(po@xmath29)@xmath28 @xcite and the calculated values were found in good agreement with experimental results . + this paper is organized as follows . in section
ii , we present the physical quantities that are to be calculated in subsequent sections .
section iii describes the @xmath30site finite lattice for our model calculation . in section iv
, we present our numerical results on the finite lattice considered for @xmath7 quantum spin-@xmath8 model .
results are summarized and discussed in section v.
in order to obtain finite temperature quantities , we require to calculate the canonical partition function @xmath31 where @xmath32 is the dimensionality of the many - body hilbert space .
for the antiferromagnetic system , the order parameter , staggered magnetization , is defined as @xcite @xmath33 where * q * is the magnetic wave - vector and the sum on the right hand side is over all the lattice sites . in the thermodynamic limit , @xmath34 is non - zero in the ordered phase and zero in the disordered phase where the angular brackets @xmath35 denote the ensemble averaging . for a finite system , however , @xmath34 vanishes for all phases due to rotational symmetry in spin space and it becomes more appropriate to consider the square of the order parameter , the static spin structure - factor defined as @xmath36 where the sums @xmath37 run over all the @xmath38 lattice points .
the quantity @xmath39 represents the ssc between @xmath40 and @xmath41 spins for the @xmath42 energy eigenstate . at zero temperature , the ssc becomes @xmath43 where @xmath44 is the many - body ground state .
the spins @xmath45 and @xmath46 are the reference spin and its @xmath47 neighbor respectively .
+ another thermodynamic quantity of interest is the magnetic specific heat given as @xmath48 where @xmath49 is total spin of @xmath42 eigenstate corresponding to eigenvalue @xmath50 + the @xmath51-component of the magnetic susceptibility for an isotropic system is given as @xcite @xmath52 where @xmath53 is the total spin of the eigenstate , @xmath54 the bohr magneton and @xmath55 the electron lande g factor .
we consider a two - layered lattice containing 24 spins with 12 spins in each two - dimensional layer in ( @xmath56 ) geometry shown in fig.([24spn_lattice ] ) .
it will , henceforth , be referred to as @xmath57 lattice with @xmath24 , @xmath1 being the inplane nearest and next nearest neighbor couplings respectively and @xmath2 the interlayer coupling .
each layer is described by a @xmath58 frustrated antiferromagnetic heisenberg(fafh ) spin model . in the two - layered lattice considered here , there are four spins in the @xmath59-direction and three spins in the @xmath60-direction .
we , therefore , use periodic boundary condition ( pbc ) along @xmath59-direction and , in order to avoid geometrical frustration , open boundary condition ( obc ) along @xmath61direction .
since there are only two layers along the @xmath62direction , use of pbc leads to double counting of the interlayer coupling and hence , in our calculation , @xmath2 is taken twice its value used in a multi - layered system studied , say , in reference@xcite where several layers of @xmath0 model has been examined using coupled - cluster method . geometry . the exchange couplings @xmath24 , @xmath1 are the nearest and the next nearest neighbor inplane couplings respectively and @xmath2 the nearest neighbor interlayer coupling . periodic boundary condition is used along @xmath63direction but along @xmath61direction , open boundary condition has been used .
the value of @xmath2 in the present calculation is taken double its value in a multi - layered lattice ( see text).,scaledwidth=45.0% ] the hilbert space dimensionality for a spin-@xmath8 system on a @xmath4-site lattice is @xmath64 .
as discussed in the next subsection , it suffices to work in @xmath65 subspace , which has projections from all the total spin states @xmath66 and leads dimensionality of the hamiltonian matrix reduce to @xmath5 , which we diagonalize iteratively to obtain the ground and a few low - lying excited states using davidson algorithm for large sparse matrices .
it is readily seen that the hamiltonian in eq([hamiltonian ] ) commutes with the square of the total spin @xmath67 and one of its components @xmath68 @xmath69=0 , \ \ \
\left[{\bf s}^{z}_{tot } , h \right]=0 , \ \
\mbox { with } \ \ { \bf s}_{tot}= \sum_{i=1}^{n } { \bf s}_{i } \label{constantofmotion}\end{aligned}\ ] ] and hence are constants of motion with eigenvalues @xmath70 and @xmath71 , respectively . for a system with even number of spins , it suffices to perform the diagonalization of the hamiltonian ( [ hamiltonian ] ) in @xmath72 subspace which has projections from all the total spin subspaces @xmath73 . for an @xmath74spin system , a many - body ising basis state in a given total @xmath75 subspace
is given by @xmath76 where @xmath77 runs over the dimensionality of the @xmath38-spin hilbert space in the given @xmath78 subspace .
in the numerical results presented here , all the exchange couplings and energies have been measured in units of the inplane nearest - neighbor coupling @xmath24 .
+ the spin - gap is defined as the difference of energies of the lowest triplet state and the singlet ground state @xcite @xmath79 where @xmath80 and @xmath81 are the lowest eigenenergies in the subspace @xmath82 and @xmath81 , repspectively .
+ it has been reported that the spin - gap has non - zero values in quantum paramagnetic regime of a 2d @xmath0 model@xcite . in fig.([24spn_spin_gap ] ) , we present spin - gap @xmath83 _ vs _ inplane frustration parameter in the interval @xmath84 for different values of interlayer coupling @xmath85 .
we observe that , at a given value of @xmath85 , the spin - gap @xmath86 increases slowly , attains a maximum , with ( a diverging ) peak at @xmath87 and then decreases with the inplane frustration parameter , indicating the existance of a gapful paramagnetically disordered state @xcite in the interval @xmath88 . for small value of @xmath89
, we get @xmath90 at @xmath91 in agreement with the vlaue of @xmath92 reported in @xcite for an open rectangular cylinderical lattice . as the interlayer coupling @xmath85 is increased , the @xmath83 _ vs _ @xmath93 curve shifts downward i.e. the value of spin - gap decreases with increase in interlayer coupling .
the peak of the plots becomes narrower with increase in @xmath85 , over the frustration parameter interval @xmath94 .
this indicates that the region of gapful paramagnetically disordered regime shrinks on @xmath93 scale with increase in @xmath85 .
we further observe that the @xmath83 _ vs _ @xmath93 plots for @xmath95 and @xmath96 coincide _ i.e. _ an increase in the interlayer coupling beyond @xmath96 does not affect the spin - gap .
it is to be stated here that since our exact diagonalization calculation considers only two layers of @xmath0 model and double count the interlayer coupling , the value of @xmath85 in our case is indeed _
twice _ the value in a multi - layered lattice considered , for example , in reference@xcite .
it has been reported in reference @xcite that the paramagnetically disordered state vanishes at @xmath97 and the system goes directly from semi - classical nel order to semi - classical collinear order . in order to show the consistency of our results with the known results regarding various orders of 2d @xmath0 lattice on @xmath93 scale
, we consider the static ssc for the ground state and its variation with @xmath93 as well as @xmath85 . in fig.([24spncorr_zpt3]a )
, we present the variation of the inplane ssc with @xmath93 for the first(n=6 ) , second(n=10 ) , third(n=7 ) and fourth(n=11 ) neighbors ( fig.([24spncorr_zpt3]b ) ) .
the inplane first(n=6 ) and third(n=7 ) neighbors along @xmath59-axis along which pbc has been used , the ssc takes negative values ( anti - parallel spins ) and positive values ( parallel spins ) , respectively , for all values of @xmath93 , _
i.e. _ spin @xmath98 and spin @xmath99 remain antiparallel and parallel , respectively with respect to the reference spin 5 , irrespective of whether the system is in the semiclassical nel or the semiclassical collinear state .
however , the second(n=10 ) and the fourth ( n=11 ) neighbor ssc goes from positive to negative and negative to positive values , respectively crossing each other at @xmath100 , as @xmath101 $ ] is increased .
that is to say , for values of @xmath102 , a spin system on 2d square lattice exists in semiclassical nel ordered state whereas for values of @xmath103 , the system acquires semiclassical collinear order .
these results are consistent with known results for 2d @xmath0 fafhm spin model @xcite .
+ fig.([24spn_spn_cor_1st_nbr]a ) shows the variation of inplane nearest neighbor correlation between the @xmath104 and @xmath105 spins with @xmath93 for various values of @xmath85 .
the inplane nearest - neighbor ssc attains a ( negative ) maximum between @xmath106 indicating that the short range correlation ( disorder ) is dominant in the said interval of frustration parameter @xmath93 .
it is also seen that as the interlayer coupling @xmath85 is increased , the height of ( negative ) peak decreases , _ i.e. _ short range correlation decreases with increase in @xmath85 in the interval @xmath107 .
+ fig.([24spn_spn_cor_1st_nbr]b ) presents the interplane first - neighbor correlation between @xmath104 and @xmath108 spins _ vs _ inplane frustration parameter @xmath93 for several values of interlayer coupling .
the ssc @xmath109 increases with increase in @xmath85 for all values of inplane frustration parameter in the interval @xmath110 . for small values of @xmath85 ,
the inter - plane ssc changes very little with @xmath93 indicating that the physics is essentially the same as that of 2d @xmath0 model .
however , when the interlayer coupling is made significant _ i.e. _ @xmath111 , the inter - plane first neighbor ssc attains a peak between @xmath107 with accompanying decrease in inplane first neighbor ssc , indicating the shrinking of quantum disorder in the said interval of inplane frustration parameter , as seen in fig.([24spn_spn_cor_1st_nbr]a ) . _
@xmath93 for the layered @xmath112 lattice at two different values of @xmath113 .
the quantity @xmath114 corresponding to the nel order is dominant at small values of @xmath93 , whereas for large values of @xmath93 , @xmath115 corresponding to the inplane collinear order is dominant .
for intermediate values @xmath107 , several of the competing magnetic orders coexist leading to quantum paramagnetically disordered state.,scaledwidth=50.0% ] another quantity we examine is the zero - temperature static spin structure - factor @xmath116 defined in equation([structure_factor_0 ] ) . in fig.([24spn_sf_zpt6&pt05 ] ) , we present @xmath117 _ vs _
@xmath93 for two different values of @xmath113 , for various values of the magnetic wave - vector * q*. for small values of @xmath93 , the static spin structure - factor @xmath118 corresponding to the nel order dominates whereas for large values of @xmath93 , the spin - structure factor @xmath119 corresponding to the collinear order dominates .
it is to be noted that in our calculation @xmath120 , because pbc has been used along @xmath59-direction only , whereas obc has been used along @xmath60-direction . for intermediate values of frustration parameter in the interval @xmath121 , the spin structure factor @xmath122 corresponding to several magnetic wave - vectors @xmath123 have comparable values , implying that several of the competing magnetic orders coexist in the said interval leading to quantum paramagnetically disordered state . as the interlayer coupling
is increased to a large value , say @xmath124 , the spin structure factor @xmath118 corresponding to nel ordering takes markedly large values for @xmath125 whereas for @xmath126 , it is the spin - structure factor @xmath127 corresponding to collinear order that takes markedly large values , as seen in fig.([24spn_sf_zpt6&pt05]b ) . for intermediate values of inplane frustration parameter @xmath128 , the spin structure factors @xmath114 and @xmath119 vary very sharply and the region of quantum paramagnetically disordered state becomes narrower on @xmath129 scale . this can be seen as the paramagnetically disordered state tending to disappear with increase in interlayer coupling . at zero temperature , the quantum mechanical ground state of the frustrated antiferromaggnetic spin-@xmath8 system described by the hamiltonian in equation ( [ hamiltonian ] ) can be viewed as a statistical mixture of ising basis states , equation([isingbasis ] ) . accordingly
, the shannon entropy ( se ) , widely used in information theory , can be defined @xcite for a quantum mechanical state @xmath130 as @xmath131 where @xmath38 is the number of spins and @xmath132 is the probability of the @xmath133 ising basis state in the variational wavefunction @xmath134 obtained through exact diagonaliztion of the hamiltonian matrix .
since there are @xmath135 basis states for @xmath38 spin@xmath136 particles , @xmath137 for maximally disordered state and the maximum value of @xmath138 is @xmath139 . lattice _ vs _
@xmath93 for different values of inter layer couplings @xmath85 . at @xmath140 ,
two layers becomes separated therefore we present the , se _ vs _ @xmath93 , for a single layer @xmath141 lattice .
the inset image displays the variation of se with @xmath85 for single layer @xmath142 lattice , @xmath0 model.,scaledwidth=45.0% ] in fig.([shannon_entropy ] ) we present the variation of shanon entropy(se ) with inplane frustration parameter @xmath93 for different values of interlayer couplin @xmath85 . with the interlayer coupling taken to zero ,
the system becomes a set of two independent layers of @xmath141 lattice with pbc along @xmath63direction and obc along @xmath61direction .
the se _ vs _ @xmath93 plot for a @xmath141 lattice has been shown by red solid line and labeled as @xmath143 in fig.([shannon_entropy ] ) . for a better comparison with a more realistic two - dimensional lattice , we present in inset , se _ vs _ @xmath93
plot for a @xmath30spin @xmath144 lattice in which pbc has been imposed both along @xmath63 and @xmath61 directions . in the parameter intervals
@xmath12 and @xmath145 , we obtain large values of shanon entropy(se ) implying that the system fluctuates around the semiclassical nel state and the semiclassical collinear state , respectively .
the shanon entropy is minimum at a value of inplane frustration parameter @xmath93 a little less than 0.6 , implying that some kind of order - by - quantum - disorder@xcite takes place in spin configuration space in the parameter interval @xmath146 .
as we increase the @xmath85 , values of se increases in the regime of small and large values of @xmath93 , whereas the depth of the the plot in the intermediate regime @xmath146 corresponding to the disordered state takes further dip .
this indicates the narrowing of intermediate quantum mechanically disordered regime on @xmath93 scale , with increase in interlayer coupling @xcite .
in order to calculate a thermodynamic quantity , we require , in principle , the ground state and all the excited states .
this may be computationally prohibitive for most of the systems of interest . however , quantum fluctuation arising due to frustration is important only at low temperatures where only a few low - lying excited states are relevant to any thermodynamic quantity .
we , in the present calculation , take only the lowest six eigenstates , to examine the effect of @xmath2 on specific heat and magnetic susceptibility for the @xmath57 lattice .
we find that for the hamiltonian in equation([hamiltonian ] ) , the lowest six eigenstates belong to @xmath147 and @xmath148 subspace in the parameter regimes @xmath149 and @xmath150 whereas in the intermediate regime , @xmath146 , most of the low - lying eigenstates are singlet ( @xmath151 ) and a few are triplet ( @xmath82 ) .
the magnetic specific heat and the @xmath51-component of magnetic susceptibility , calculated with lowest six eigenstates , for the layered lattice @xmath57 are presented in the following .
it has earlier been observed @xcite that in a @xmath0 fafh model , as the frustration parameter is increased over the interval @xmath152 , the systme goes from ordered nel state to disordered paramagnetic state ; correspondingly , the peak of the specific heat curve sharpens and its position shifts to lower temperatures . with further increase in the frustraion parameter @xmath93 beyond @xmath153
, the system goes from disordered paramagnetic state to ordered collinear state and correspondingly , the peak of the specific heat broadens and its position shifts to higher temperatures .
accordingly , the above distinct behaviour of the specific heat peak _ vs _ temperature , as the system goes from disordered to ordered state or _
vice - veras _ can be extended to examine the role of @xmath85 on the inplane magnetic order .
+ in figure([24spn_sh_zeta]a ) , we present specific - heat _ vs _ temperature plot with inplane frustration parameter @xmath154 lying in the regime of semi - classical nel ordered state in a pure 2d antiferromagnetic @xmath0 model @xcite . the specific - heat peak monotonically shifts to higher temperatures as the interlayer coupling is increased from small to large values _
i.e. _ @xmath155 .
the specific heat curve is sharpest for @xmath89 , the smallest of the values considered . the specific - heat _ vs _ temperature curve flattens and the height of the peak decreases with increase in interlayer coupling . from calculations on small lattice of spins such as the one in the present work , it is difficult to conclude whether the peak of the specific - heat corresponds to any phase transition or is just an energy crossover .
similarly , in figure([24spn_sh_zeta]c ) with inplane frustration parameter @xmath156 lying in the regime of semi - classical collinear ordered state for a pure 2d antiferromagnetic @xmath0 model @xcite , the specific - heat _ vs _ temperature curve again flattens and the position of the peaks shifts to higher temperatures with increase in @xmath2 .
+ however , in figure([24spn_sh_zeta]b ) with inplane frustration parameter @xmath157 lying in the regime of quantum paramagnetic disordered state for a pure 2d antiferromagnetic @xmath0 model @xcite , the specific - heat _ vs _ temperature shows an entirely different behavior : no shifting of peak position is observed when the interlayer coupling @xmath2 is increased from @xmath158 to @xmath159 and @xmath160 . +
thus as the interlayer coupling is increased , the system is driven to an ordered state namely the nel oreder state or the collinear ordered state , depending on the value of the inplane frustration parameter @xmath93 . on the other hand , when the system is already in paramagnetic disordered state , an increase in interlayer coupling leads to no chnange in inplane magnetic order .
+ in figure([24spn_sh_alfa ] ) , we present magnetic specific heat _ vs _ temperature at given values of @xmath85 for different values of inplane frustration parameter .
we observe in figure([24spn_sh_alfa]a ) that for @xmath161 , the specific heat peaks are sharp and occur at @xmath162 in the quantum paramegnetic regime corresponding to inplane frustration parameter values @xmath163 ; the peak occures at lower temperatures compared to other values of @xmath164 lying in the nel ordered and collinear ordered state , respectively . at @xmath97 in figure([24spn_sh_alfa]b ) , the specific heat peaks for @xmath165 corresponding to paramagnetic regime are sharp and occur at @xmath166 , again lower than the temperatures at which specific heat curve peaks for @xmath167 corresponding to nel ordered state and @xmath168 corresponding to collinear ordered state . at significantly higher value of interlayer coupling @xmath169 ,
we observe in figure([24spn_sh_alfa]c ) that the shepecific heat peaks flaten and occur at higher temperatures except for @xmath170 corresponding to paramagnetic regime , for which the peaks are sharper and occur at a lower temperature @xmath171 implying the shrinkig of quantum paramagnetic rgime with increase in interlayer coupling . + for reference , we plot in figure([24spn_sh_alfa]d ) the magnetic specific heat _ vs _
temperature for different values of inplane frustration parameter @xmath93 for a @xmath56 lattice of 12 spins .
as @xmath93 is increased from @xmath172 to @xmath173 , the specific heat peaks sharpen and shift to lower temperatures .
however , as it is further increased to @xmath174 , the specific heat peak begins to flaten and shift to higher temperatures . + thus the above observations on specific heat _
vs _ temperature curve can be summarized as follows : at small value of interlayer coupling @xmath89 , there is a quantum paramagnetic order in @xmath175 regime whereas at @xmath169 , the quantum paramagnetic regime narrows to @xmath176 , implying the tendency of the paramagnetic regime to vanish with increasing interlaeyr coupling . for the finite lattice like the one in the present study , we do not expect to observe the vanishing of quantum paramagnetic regime as the interlayer coupling is incresaed due to finite size effects .
[ 24spn_chi_zta_pt1 ] [ 24spn_chi_zta_pt3 ] figure([24spn_suscept_zta ] ) presents the z - component of magnetic susceptibility _ vs _ temperature for different values of inplane frustration parameter @xmath93 at three values of interlayer coupling @xmath85 . in figure([24spn_suscept_zta]a ) , we observe that at small value of interlayer coupling @xmath177 , the system tends to shift towards magnetically disordered state leading to quantum paramagnetic order and the magnetic susceptibility reduces with increase in frustration parameter from small values @xmath178 corresponding to the nel ordered state to intermediate values @xmath179 corresponding to paramagnetically disordered state .
the magnetic susceptibilities corresponding to @xmath180 and @xmath181 almost coincide and remain lowest in the entire range of temperature considered in the figure .
+ at a higher value of interlayer coupling @xmath182 in figure([24spn_suscept_zta]b ) , the magnetic susceptibility for @xmath183 and @xmath184 corresponding to the nel and collinear ordered state respectively increases slightly but the susceptibility of the paramagneticaly disordered state corresponding to @xmath185 and @xmath186 is enhanced significantly ; the magnetic susceptibility curve corresponding to @xmath187 remains lowest in comparision to susceptibility curves for other values of @xmath93 , in the entire range of temperature considered in the figure . + with further increase in interlayer coupling with @xmath188 in fig.([24spn_suscept_zta]c ) , the peak value of magnetic susceptibility curves for the paramagnetically disordered state corresponding to @xmath189 , @xmath190 and @xmath191 is further enhanced and almost coincides with each other . furthermore , the magnetic susceptibility curve is enhanced significantly for @xmath192 corresponding to collinear ordered phase .
however , the magnetic susceptibility corresponding to @xmath178 remains large and of the same order of magnitude regardless of the values of @xmath85 in the entire range of temperatures considered in the figures([24spn_suscept_zta ] a , b , c ) + if we look at the position of the magnetic susceptibility humps for different values of inplane frustration parameter @xmath93 , it shifts towards lower temperatures with increase in @xmath93 exhibiting signature of quantum paramagnetic disorder , as lesser thermal energy is required to cause thermal disorder . after further increase in @xmath93 , the system re - orders and
the hump begins to shift towards higher temperatures .
+ [ 24spn_chi_alf_pt2 ] [ 24spn_chi_alf_pt6 ] [ 24spn_chi_alf_pt8 ] in fig.([24spn_suscept_alfa ] ) , we present magnetic susceptibility _ vs _ temperature for different values of interlayer coupling @xmath85 at three given values of inplane frustration parameter @xmath93 .
+ in fig.([24spn_suscept_alfa]a ) with @xmath178 , the system is believed to exist predominantly in @xmath118 ordered state corresponding to semiclassical nel order and consequently we do not observe significant variance in magnetic susceptibility curve with increase in interlayer coupling @xmath85 , in the entire range of temperature considered in the figure .
+ in fig.([24spn_suscept_alfa]b ) with @xmath187 , the system is believed to exist in quantum paramagnetic disordered state and hence the magnetic susceptibility for small values of interlayer coupling @xmath193 , takes small values in the entire range of temperature considered in the figure . when the interlayer coupling is increased to large value of @xmath169 , the system re - orders ( _ i.e. _ the quantum paramagnetic disorder disappears and the magnetic susceptibility increases significantly in the entire range of temperatures considered in the figure . + in fig.([24spn_suscept_alfa]c ) with @xmath192 the system
is believed to exist in @xmath127 ordered state , corresponding to semi - classical collinear order and the magnetic susceptibility is observed to be higher even at small values of interlayer coupling @xmath194 with respect to their values with @xmath195 in fig.([24spn_suscept_alfa]b ) .
however , when the interlayer coupling is increased to large value of @xmath169 , the magnetic susceptibility increases significantly in the entire range of temperatures considered in the figure . + the magnetic susceptibility corresponding to @xmath169 remains large and of the same order of magnitude regardless of the values of @xmath1 in the entire range of temperatures considered in the figures([24spn_suscept_alfa ] a , b , c ) .
if we observe the magnetic susceptibility hump , it moves towards higher temperatures implying that the system is being driven to a more ordered state as we increase the interlayer coupling , in agreement with mermin - wagner theorem@xcite .
+ the increment in magnetic susceptibilities of paramagnetically disordered state with increase in @xmath2 implies that the quantum parmagnetic ordered state tends to vanish and the spins re - order themselves as the interlayer coupling is increased .
we have presented an exact diagonalization study on @xmath7 model for a @xmath4-spin layered lattice in @xmath196 geometry to examine the role of interlayer coupling on the inplane magnetic order . for a quasi-2d @xmath0 model ,
the paramagnetic region is found to extend from @xmath197 to @xmath198 .
as we increase the interlayer coupling , the value of spin - gap in the paramagnetic region reduces since the interlayer coupling stabilizes the inplane order and thereby lowers the degree of disorder in paramagnetic regime .
we further observe that the spin - gap _ vs _ @xmath93 curves coincide for @xmath169 and @xmath199 , implying that , @xmath200 is the saturation value of the interlayer coupling beyond which the spin gap does not change with @xmath85 . for an multilayered lattice @xcite ,
the corresponding saturation value will be @xmath97 , half the value for our two layered lattice , after correcting for double counting .
+ at zero temperature , for small values of interlayer coupling , the inplane first - neighbor ssc becomes maximum in paramagnetic region _
i.e. _ short range order is dominant in paramagnetic regime .
as the interlayer coupling is increased , the inplane first - neighbor ssc reduces _
i.e. _ short range correlation decreases in quantum paramegnetic regime .
this is corroborated by our results on static spin structure factor also . in the quantum paramagnetic regime , @xmath118 corresponding to nel order decreases and @xmath127 corresponding to collinear order increases slowly with @xmath93 when @xmath85 is small .
however , when @xmath85 is increased to @xmath186 , the change in @xmath118 and @xmath127 becomes steeper indicating that in the thermodynamic limit , @xmath118 may crossover directly to @xmath127 and the intervening quantum paramagnetic regime disappears .
+ we observe that the specific heat _
vs _ temperature curve is flat in ordered states like nel or collinear state but as the system enters quantum paramagnetically disordered state , the specific heat _
vs _ temperature curve acquires a sharp peak . for small values of interlayer coupling , we observe a sharp peak in specific heat _ vs _ temperature curve for the inplane frustration parameter interval @xmath201 corresponding to paramagnetically disordered state
. however , as the interlayer coupling is increased to a significant value , say @xmath169 , a sharp peak in specific heat _
vs _ temperature curve is observed in the narrowed interval @xmath202 , indicating the shrinking of quantum paramagmetically disordered regime on @xmath93 scale .
+ from the ( z - component of ) magnetic susceptibility _ vs _ temperature plot we observe that the value of magnetic susceptibility in ordered states like nel or collinear state is higher compared to its value in quantum paramagnetically disordered state . in the paramagnetically disordered state corresponding to the inplane frustration parameter interval @xmath203 for small values of interlayer coupling @xmath85 ,
the peak height of magnetic susceptibility curve is small .
as @xmath85 is increased , the height of the magnetic susceptibility peak increases in the above interval of inplane frustration parameter , implying that the interlayer coupling drives the system to an ordered state .
+ in summary , the results presented in preceding sections show that for small values of interlayer coupling @xmath85 , the system is in ( a ) semi - classical nel ordered state for @xmath204 ( b ) semi - classical collinear ordered state for @xmath103 and ( c ) quantum paramagnetically disordered state for @xmath205 . as @xmath85
is increased @xcite , the interval of quantum paramagnetic disordered state on @xmath93 scale , narrows and long range order sets in .
it has earlier been reported @xcite that for a multilayered lattice with @xmath97 , the paramagnetically disordered state vanishes and as @xmath93 is increased , the system goes directly from semi - classical nel order to semi - classical collinear order . in our study on finite lattice , on increasing @xmath85 ,
the intervening paramagnetically disordered regime narrows on @xmath93 scale but does not vanish due to finite size effect . + * acknowledgement : * mmh and maha acknowledge financial support from university grants commission vide f. no .
41 - 891/2012(sr )
. partial financial support from martin - luther university is also acknowledged where a part of the code for the present work was developed .
9 j. g. bednorz and k. a. muller , z. phys .
b 64 , 189 ( 1986 ) . p. w. anderson , science * 235 * , 1196(1987 ) .
d. c. johnston , advances in physics , * 59 * , 803(2010 ) .
r. f. bishop , p. h. y. li and c. e. campbell , phys . rev.b * 88 * , 214418 ( 2013 ) .
p. chandra , b. boucot , phys .
b , * 38 * , 9335(1988 ) .
elbio dagotto and andriana moreo , phys .
63 * , 2148(1989 ) .
j. richter , phys .
b * 47 * 5794 ( 1993 ) .
r. f. bishop , d. j. j. farnell and j. b. parkinson , phys .
b , * 58 * , 6394(1998 ) .
luca capriotti and sandro sorella , phys .
lett . * 84 * , 3173(2000 ) .
luca capriotti , federico becca , alberto parola , and sandro sorella , phys .
lett . * 87 * 097201(2001 ) .
r. r. p. singh , w. zheng , j. oitmaa , o. p. sushkov , and c. j. hamer , phys .
lett . * 91 * , 017201 ( 2003 ) .
h. rosner , r. r. p. singh , w. h. zheng , j. oitmaa , and w. e. pickett , phys .
b * 67 * , 014416(2003 ) .
m. haque , m.a.h .
ahsan , advanced science lettes , * 21 * , 2713(2015 ) .
r f bishop , p h y li , r darradi and j richter , j. phys . : condens .
matter , * 20 * 255251 ( 2008 ) .
r. f. bishop p.
h. y. li , r. darradi , j. schulenburg , and j. richter , phys .
b , * 78 * , 054412 ( 2008 ) .
schulz h j and ziman t a l , europhys .
18 * , 355(1992 ) .
andreas voigt and johannes richter , journal of physic : cond . mat . * 8 * , 27 , ( 1996 ) .
l. isaev , g. ortiz and j. dukelsky , phys .
b * 79 * , 024409 ( 2009 ) .
yoichi kamihara , takumi watanabe , masahiro hirano and hideo hosono , j. am .
, * 130 * , 3296(2008 ) h. rosner , r. r. p. singh , w. h. zheng , j. oitmaa , s .- l .
drechsler , and w. e. pickett , phys .
* 88 * 186405 ( 2002 ) .
qimiao si and elihu abrahams , phys . rev
* 101 * , 076401(2008 ) .
michael holt , oleg p. sushkov , daniel stanek and gotz s. uhric , phys .
b , * 83 * , 144528(2011 ) .
r. melzi , p. carretta , a. lascialfari , m. mambrini , m. troyer , p. millet and f. mila , phys .
lett . * 85 * 1318(2000 ) d. schmalfu@xmath206 , r. darradi , j. richter , j. schulenburg and d. ihle , phys .
lett . * 97 * , 157201(2006 ) .
wagner a nunes , j ricardo de sousa , j roberto viana , and j richter , j. phys .
* 22 * , 146004 ( 2010 ) .
daniel stanek , oleg p. sushkov , and gotz s. uhrig , phys .
b , * 84 * 064505(2011 ) .
onofre rojas , c j hamer and j oitmaa , j. phys . : condens
. matter , * 23 * 416001(2011 ) .
e. dagotto and a. moreo , phys . rev .
b , * 39 * , 4744(1989 ) .
m. haque , m.a.h .
ahsan , journal of magnetism and magnetic materials , * 402 * , 143 - 149(2016 ) .
christopher l. henley , phys .
lett . * 62 * , 2056 ( 1989 ) .
j. t. haraldsen , t. barnes and j. l. musfeldt , pys .
b , * 71 * , 064403 ( 2005 ) j. richter and j. schulenburg , eur .
j. b , * 73 * , 117 ( 2010 ) .
johannes richter , ronald zinke , and damian j.j .
farnell , eur .
j. b , * 88 * , 2(2014 ) .
, subir sachdev .
shou - shu gong , wei zhu , d. n. sheng , olexei i. motrunich , and matthew p. a. fisher , phys , rev .
lett . * 113 * , 027201(2014 ) .
johannes reuther , peter wolfle , rachid darradi , wonfram brenig , marcelo arlego and johannes richter , phys .
rev b , * 83 * , 064416 ( 2011 ) . c. e. shannon , bell system technical journal * 27 * , 379 - 423 , 623 - 656 ( 1948 ) .
k. a. ross , y qiu , j r d copley , h a dabkowska and b d gaulin , phys .
112 , 057201 ( 2014 ) .
taner yildirim , turkish journal of physics * 23 * , 47 ( 1999 ) .
s. bacci , e. gagliano and e. dagotto , phys .
b , * 44 * , 285 - 292 ( 1991 ) .
n. d. mermin and h. wagner , phys .
* 17 * , 1133(1966 ) . | we present an exact diagonalization study on layered @xmath0 antiferromagnetic heisenberg spin model to examine the role of frustration induced by inplane next - nearest neighbor coupling @xmath1 , in presence of interlayer antiferromagnetic coupling @xmath2 .
a finite lattice of 24 spins in layered geometry of @xmath3 is considered and the resulting hamiltonian matrix diagonalized using davidson iterative algorithm to obtain the ground and a few low - lying excited states . the lattice size (
@xmath4 spins with hilbert space dimensionality of @xmath5 in @xmath6 subspace ) has been kept relatively small because of the large number of runs required to sample @xmath7 parameter space .
quantities like spin - gap , shannon entropy , spin - spin correlation(ssc ) , static spin structure - factor , magnetic specific - heat and magnetic susceptibility are calculated for various values of spin - spin coupling parameters . with increase in interlayer coupling ,
the system is driven to states with long range order and the interval of quantum paramagnetically disordered state , sandwitched between nel and collinear ordered states , narrows on the scale of inplane frustration parameter . +
pacs : 75.10.jm , 75.10.kt , 75.40.-s |
the critical temperature @xmath0 for bose - einstein condensation in an ideal gas of identical bosons was deduced in 1924 by albert einstein @xcite : @xmath7 where @xmath8 is the number density of the bosons .
an obvious next question is the shift @xmath9 in the critical temperature due to the interaction between the bosons . if the potential between two bosons is short - ranged , their scattering amplitude at sufficiently low energy is a constant proportional to the @xmath10-wave scattering length @xmath11 the bose gas with such an interaction has been studied extensively since the pioneering work @xcite of lee and yang in 1957 on bosons with a hard - sphere potential . however the shift in @xmath0 from such an interaction remained an unsolved problem until recently .
a major breakthrough was made by baym et al .
@xcite , who showed that the problem could be reduced to a nonperturbative calculation in an effective @xmath12-dimensional statistical field theory with @xmath13 symmetry .
this demonstrated conclusively that the leading order shift was linear in @xmath14 : @xmath15 where @xmath16 is a numerical constant . using a self - consistent calculation of the propagator
, they estimated the coefficient to be @xmath17 @xcite .
baym , blaizot and zinn - justin calculated the coefficient analytically in the large-@xmath4 limit and obtained @xmath18 @xcite .
the first correction in the @xmath19 expansion was calculated by arnold and tomasik and it reduces @xmath16 to 1.72 @xcite .
definitive calculations of the coefficient for the physical case @xmath2 were carried out by kashurnikov , prokofev , and svistunov @xcite and by arnold and moore @xcite by applying monte carlo methods to a lattice formulation of the effective field theory with the result @xmath20 . the second order correction to @xmath21 proportional to @xmath22 has since been calculated by arnold , moore and tomasik @xcite . before the recent breakthrough , there had been many previous attempts to calculate the leading shift in @xmath0 @xcite .
the results for the coefficient @xmath16 in ( [ c - def ] ) ranged from @xmath23 to @xmath24 .
the definitive solution of this problem provides a way to assess the reliability of the various calculational methods .
one particularly simple method that can be applied to this problem is the _
linear @xmath5 expansion _
@xcite , also known as _
optimized perturbation theory _
@xcite or _
variational perturbation theory _ @xcite . with this method ,
an arbitrary parameter @xmath25 is introduced into the theory and calculations are carried out using perturbation theory in a formal expansion parameter @xmath5 whose value is 1 .
the rate of convergence can be improved by adjusting the parameter @xmath25 at each order in @xmath5 by some criterion such as the principal of minimal sensitivity .
it was first applied to the problem of calculating @xmath26 by de souza cruz , pinto and ramos @xcite . at second ,
third , and fourth orders in @xmath5 , the predictions are @xmath27 , @xmath28 , and @xmath29 , respectively @xcite , which seem to be converging to the lattice monte carlo result . in this paper , we revisit the application of the linear @xmath5 expansion to the calculation of the shift in @xmath30 . in section
ii , we review the reduction of the calculation of @xmath26 to a problem in a 3-dimensional statistical field theory . in section iii , we introduce the linear @xmath5 expansion and review the results of refs .
we present two alternative precriptions for applying this method to the shift in @xmath30 . in section iv , we show that in the large-@xmath4 limit , these alternative prescriptions converge to the analytic result of ref .
@xcite for some range of @xmath25 .
we study the improvement in the convergence rate that is obtained by using the principal of minimal sensitivity to choose a value of @xmath25 that depends on the order in @xmath5 .
we show that there is a family of solutions to this equation that seems to improve the convergence rate from a power law in the order of the @xmath5 expansion to exponential . in section
v , we calculate the result for @xmath2 at @xmath31 and @xmath32 orders in @xmath5 . in section
vi , we compare our results with the previous calculations using the linear @xmath5 expansion of refs . @xcite .
our results are consistent with slow convergence to the monte carlo results of refs . @xcite .
we consider a system of identical bosons of mass @xmath25 at a temperature @xmath33 and number density @xmath8 close to the critical temperature @xmath0 for bose - einstein condensation .
if the effects of interactions are small , einstein s formula ( [ einstein ] ) implies that the interparticle spacing @xmath34 and the thermal wavelength @xmath35 are comparable .
we assume that they are both large compared to the range @xmath36 of the two - body potential . in this case
, the interaction between the bosons can be characterized entirely by the s - wave scattering length @xmath14 , which we assume to have magnitude of order @xmath36 .
thus we require @xmath37 .
this system can be described by a second - quantized schrdinger equation .
the corresponding imaginary - time lagrangian is @xmath38 where @xmath39 is the chemical potential associated with the particle number density . in this paper
, we generally use units such that @xmath40 . treating the system at finite temperature using the imaginary time formalism
, the field @xmath41 can be decomposed into fourier modes @xmath42 with matsubara frequencies @xmath43 the zero frequency mode @xmath44 is defined by @xmath45 if the temperature is near the critical temperature @xmath30 , the chemical potential @xmath46 is small in magnitude compared to @xmath33 . for momenta small compared to @xmath47 , the nonzero frequency modes have euclidean energies of order @xmath33 , while the zero frequency modes have euclidean energies of order @xmath48 .
thus at large distances much greater than @xmath49 , the nonzero modes decouple from the dynamics , leaving an effective theory of only the zero modes @xcite . a naive derivation of the effective lagrangian is obtained simply by substituting @xmath50 in ( [ l - it ] ) .
the action then reduces to @xmath51 \longrightarrow \int d^{3}x\text { } { \cal l}_{{\rm eff}}[\psi_{0}({\bf x})],\ ] ] where the effective lagrangian is @xmath52 it is convenient to replace the complex field @xmath53 by the 2-component real field @xmath54 defined by @xmath55 the effective lagrangian density for the dimensionally reduced theory can then be written as @xmath56 where the parameters are @xmath57 this @xmath13 field theory can be generalized to an @xmath1 field theory by replacing @xmath58 by an @xmath4-component real field @xmath59 .
this naive derivation of the effective lagrangian is supported by a more sophisticated analysis within the effective field theory framework @xcite .
this analysis reveals that there are also higher dimension operators in the effective lagrangian ( [ l - eff ] ) , such as @xmath60 and @xmath61 , but they are not needed to calculate the shift in @xmath30 to leading order in @xmath62 .
it also reveals that there are corrections to the coefficients @xmath63 and @xmath64 that can be calculated perturbatively as expansions in @xmath65 .
again , they are not needed to calculate @xmath26 to leading order in @xmath66 .
finally , since the scale @xmath33 has been removed from the problem , the effective field theory defined by the effective lagrangian ( [ l - eff ] ) requires an ultraviolet cutoff .
fortunately , the leading order shift in @xmath30 is dominated by the infrared region of momentum space and is insensitive to the ultraviolet cutoff . from einstein s formula ( [ einstein ] ) , the leading order shift @xmath67 in the critical temperature at fixed number density is related to the leading order shift @xmath68 in the critical density at fixed temperature by the simple relation @xmath69 the number density in the original theory is @xmath70 where the angular brackets indicate the expectation value in the state corresponding to a homogeneous gas . the number density in the effective theory , to the accuracy required to calculate the shift in @xmath30 to leading order in @xmath66 , is @xmath71 where @xmath72 is the short - distance contribution to the number density , which depends on the ultraviolet cutoff .
the angular brackets indicate the statistical average of the operator @xmath73 . it can be expressed as @xmath74 ^{-1 } , \label{fi2}\ ] ] where @xmath75 is the self - energy of the @xmath76 field .
the integral over @xmath77 is divergent and requires an ultraviolet cutoff .
the number density for an ideal gas is given by the same expression with @xmath78 which implies @xmath79 : @xmath80 ^{-1}. \label{na0 } \label{n0}\ ] ] again , the integral over @xmath77 is divergent and requires an ultraviolet cutoff . to the accuracy required to calculate the shift in @xmath30 to leading order in @xmath66 ,
the constant @xmath72 is the same as in ( [ n ] ) .
the critical point is determined by the condition that the correlation length associated with the propagator in ( [ fi2 ] ) is infinite : @xmath81 this condition reduces to @xmath82 if @xmath83 . the change @xmath84 in the critical density at fixed temperature @xmath33 due to interactions is the difference between ( [ n ] ) with @xmath85 and ( [ n0 ] ) with @xmath86 . the expression ( [ deltc-1 ] ) for the fractional shift in @xmath30 can therefore be written as @xmath87 where @xmath88 is a quantity that can be calculated within the effective field theory : @xmath89 ^{-1}-\left ( p^{2}\right ) ^{-1}\right ] .
\label{delta - gen}\ ] ] the integral over @xmath77 is convergent and therefore no longer requires an ultraviolet cutoff . we have replaced the prefactor 2 in ( [ delta - gen ] ) by @xmath4 to generalize it to the @xmath1 case . comparing ( [ delta - def ] ) with ( [ c - def ] ) and using ( [ u - a ] ) and ( [ einstein ] ) , we find that the coefficient @xmath16 is @xmath90 the effective theory defined by the lagrangian ( [ l - eff ] ) is superrenormalizable .
although the theory has ultraviolet divergences , @xmath91 is dominated by infrared physics and it is insensitive to the ultraviolet cutoff .
the relevant length scales are therefore those set by the parameters @xmath63 and @xmath64 of the effective field theory .
the critical point can be reached by tuning the parameter @xmath63 . at the critical point , there is only one relevant length scale and it is set by the parameter @xmath64 .
since @xmath91 has dimensions of length , it must be proportional to @xmath64 by simple dimensional analysis .
the determination of the coefficient of @xmath64 requires a nonperturbative calculation .
since @xmath64 is linear in @xmath14 as @xmath92 , this demonstrates that the leading term in the shift in @xmath30 is linear in @xmath14 .
the calculation of the first order shift @xmath9 in the critical temperature has been reduced to the calculation of the quantity @xmath91 in ( [ delta - gen ] ) for @xmath93 the first calculation of this quantity within a systematically improvable framework was a calculation to leading order in the @xmath19 expansion by baym , blaizot , and zinn - justin @xcite : @xmath94 the next - to - leading order ( nlo ) correction in the @xmath19 expansion was calculated by arnold and tomasik @xcite .
the first two terms in the @xmath19 expansion are @xmath95 for the physical case @xmath2 , @xmath91 has been calculated nonperturbatively by kashurnikov , prokofev , and svistunov @xcite and by arnold and moore @xcite using monte carlo simulations of a lattice formulation of the @xmath13 field theory .
the two results are consistent , but ref .
@xcite quotes a smaller error : @xmath96 if we set @xmath2 in the @xmath19 expansion in ( [ delta-1/n ] ) , the coefficient of @xmath97 is @xmath98 at leading order and @xmath99 at next - to - leading order .
these coefficients seem to be approaching the lattice monte carlo result @xmath100 .
any attempt to calculate @xmath9 using ordinary perturbation theory in @xmath64 is doomed to failure . in the symmetric phase where @xmath102
, perturbation theory is an expansion in powers of @xmath103 . if we naively let @xmath104 to find the critical behavior , the perturbative corrections are infrared divergent , with the degree of divergence becoming more severe at each successive order .
it is useful to think of perturbation theory as an expansion in powers of @xmath105 , where @xmath106 is the correlation length .
perturbation theory breaks down in the critical region , because @xmath105 is of order 1 and every order of perturbation theory gives a leading order contribution .
the _ linear @xmath5 expansion _
( lde ) is a particularly simple method for obtaining nonperturbative results using perturbative techniques .
calculations are carried out as in ordinary perturbation theory by evaluating a finite number of feynman diagrams .
the name `` linear @xmath5 expansion '' comes from the fact that it can be defined by a lagrangian whose coefficients are linear in a formal expansion parameter @xmath5 . if @xmath107 is the lagrangian for the system of interest , the lagrangian that generates the lde has the form @xmath108 where @xmath109 is the lagrangian for an exactly solvable theory , the lagrangian @xmath110 interpolates between @xmath109 when @xmath111 and @xmath107 when @xmath112 .
the linear @xmath5 expansion is much easier to apply than the
_ nonlinear @xmath5 expansion _ in which the lagrangian that interpolates between @xmath109 and @xmath107 involves an operator that is a nonlinear function of @xmath5 @xcite . the lagrangian @xmath109 for the exactly solvable theory typically involves an arbitrary parameter @xmath25 .
at @xmath112 , the lagrangian ( [ l - delta ] ) is independent of @xmath25 .
if a perturbation series in @xmath5 converges at @xmath112 , it must converge to a value that is independent of @xmath25 .
however at any finite order in the @xmath5 expansion , results depend on @xmath25 .
as @xmath25 varies over its physical range , the prediction for the observable often extends out to @xmath113 .
thus some prescription for @xmath25 is required to obtain a definite prediction .
a simple prescription for @xmath25 is the _ principle of minimal sensitivity _
( pms ) that the derivative with respect to @xmath25 should vanish .
the lde supplemented by the pms prescription is often referred to as _
optimized perturbation theory _
@xcite or _
variational perturbation theory _ @xcite .
a prescription such as pms may improve the convergence rate of the lde , because it allows the variable @xmath25 to change with the order @xmath8 of the lde .
the convergence properties of the lde have been studied extensively for the quantum mechanics problem of the anharmonic oscillator .
the hamiltonian is @xmath114 the lde is generated by the hamiltonian @xmath115 , which interpolates between @xmath116 at @xmath117 and @xmath118 at @xmath119 .
it can be expressed as @xmath120 , where @xmath121 the lde has been proven to converge for appropriate order - dependent choices of the variable @xmath25 that include the pms criterion as a special case .
duncan and jones showed that the finite temperature partition function converges exponentially , with the errors at @xmath122 order decreasing as @xmath123 where @xmath124 is a numerical constant @xcite .
guida , konishi , and suzuki proved that the energy eigenvalues @xmath125 converge uniformly in @xmath126 as @xmath127 @xcite .
janke and kleinert showed that the leading coefficient in the strong - coupling expansion for the ground state energy @xmath128 converges exponentially , with the errors decreasing like @xmath129 where @xmath16 is a numerical constant @xcite .
bellet , garcia , and neveu have made numerical studies of the lde optimized by the pms criterion for the anharmonic oscillator @xcite .
the pms criterion is a polynomial equation in the variable @xmath25 , so its solutions are in general complex - valued .
the authors of ref .
@xcite found that the solutions at different orders @xmath8 in the lde can be organized into distinct families .
the most accurate predictions for the ground state energy @xmath128 and for the quantity @xmath130 are obtained by using families of complex solutions .
the resulting values of @xmath128 and @xmath130 are of course complex - valued , but the imaginary parts are small .
the authors of ref .
@xcite suggested that the real parts could be taken as the predictions of the lde and the imaginary parts could be used as an indication of the error .
the anharmonic oscillator can be regarded as a euclidean field theory with a single real - valued field in 1 space dimension .
the statistical field theory defined by the lagrangian ( [ l - eff ] ) is a generalization to a multicomponent field in 3 space dimensions .
the most serious obstacles to generalizing the convergence proofs for the anharmonic oscillator to this more complicated problem come from the infrared and ultraviolet regions of momentum space .
it is reasonable to expect the convergence behavior to be similar if appropriate infrared and ultraviolet cutoffs are imposed on the field theory .
the quantity @xmath88 defined in ( [ delta - gen ] ) is insensitive to the ultraviolet region , so we do not expect any complications from taking the ultraviolet momentum cutoff to @xmath131 .
however , @xmath88 is very sensitive to the infrared region , so convergence in the presence of an infrared cutoff gives no information about the behavior of the lde in the limit when the infrared momentum cutoff goes to 0 .
we proceed to apply the lde to the @xmath1 statistical field theory defined by the lagrangian @xmath132 in ( [ l - eff ] ) .
our choice for the solvable field theory is the free field theory with mass @xmath25 .
the lagrangian @xmath110 defined by setting @xmath133 in ( [ l - delta ] ) can be written @xmath134 , where @xmath135 calculations are carried out by using @xmath5 as a formal expansion parameter , expanding to a given order in @xmath5 , and then setting @xmath112 .
thus the feynman diagrams generated by the lde will involve a propagator with mass parameter @xmath136 , a 4-point vertex proportional to @xmath137 , and a mass counterterm proportional to @xmath138 . in order to apply the lde to the shift in @xmath30
, we need a prescription for generalizing the quantity @xmath88 defined in ( [ delta - gen ] ) to the field theory whose lagrangian is @xmath139 .
the prescription must have a well - defined expansion in powers of @xmath5 , and it must reduce to ( [ delta - gen ] ) when @xmath140 .
one possibility , which we will refer to below as prescription ii , is to use the expression ( [ delta - gen ] ) , where @xmath141 is the self - energy for the field theory defined by @xmath142 .
any such prescription defines @xmath143 as a function of three variables .
the @xmath122 order approximation in the lde is then obtained by truncating the expansion in powers of @xmath5 at @xmath122 order to obtain a function @xmath144 and then setting @xmath140 : @xmath145 this sequence of approximations may not converge as @xmath127 , but if it does converge , we expect it to converge to the correct value @xmath88 . at any finite order @xmath8 , the approximation ( [ delta - n ] ) depends on @xmath25 .
we will see that as @xmath25 varies over its physical range from 0 to @xmath146 , the range of the predictions for @xmath88 extends out to @xmath146 for @xmath8 even and out to @xmath147 for @xmath8 odd .
thus some prescription for @xmath25 is required to obtain a definite prediction .
one prescription is the _ principle of minimal sensitivity _ ( pms ) : @xmath148 after setting @xmath112 , @xmath149 is a function of @xmath64 and @xmath25 only . by
dimensional analysis , a solution @xmath25 to the pms criterion must be proportional to @xmath64 and the value of @xmath149 at such a point must also be proportional to @xmath64 .
an alternative prescription is the _ fastest apparent convergence _ ( fac ) criterion : @xmath150 by allowing the variable @xmath25 to change with the order @xmath8 , prescriptions such as pms or fac may improve the convergence rate of the lde .
the lde was first applied to the problem of calculating the shift in @xmath30 by de souza cruz , pinto and ramos @xcite .
their prescription for the quantity @xmath88 defined in ( [ delta - gen ] ) is @xmath151 ^{-1}-\left ( p^{2}\right ) ^{-1}\right ] , \label{d - i}\ ] ] where @xmath141 is the self - energy for the field theory whose lagrangian is @xmath142 .
this prescription agrees with ( [ delta - gen ] ) at @xmath112 .
note that in addition to the explicit dependence on @xmath5 in ( [ d - i ] ) , there is also implicit dependence on @xmath5 through @xmath141 , which can be calculated as a power series in @xmath5 .
the subscript i on @xmath88 in ( [ d - i ] ) distinguishes this prescription from alternative prescriptions for @xmath88 that we will label ii and iii . in the feynman diagrams for evaluating @xmath141 in the lde , all the propagators have mass @xmath25 .
thus when the prescription i for @xmath88 is expanded in powers of @xmath5 , all the integrals involve only propagators of mass @xmath25 , with the exception of the subtraction term in ( [ d - i ] ) , which involves a massless propagator .
the quantity @xmath88 was calculated to @xmath152 order in @xmath3 in ref .
@xcite and to @xmath31 and @xmath32 order in ref .
after truncating the expansion in powers of @xmath5 and setting @xmath112 , they determined @xmath25 by applying the pms prescription ( [ pms ] ) . at @xmath122 order in the @xmath3 expansion ,
this reduces to a polynomial equation in @xmath25 with @xmath8 roots , most of which are complex - valued . for a particular choice of the roots at @xmath153 @xmath31 , and @xmath32 order in @xmath5 , the real parts of @xmath154 are @xmath155 @xmath156 , and @xmath157 respectively
this sequence seems to be approaching the lattice monte carlo result @xmath100 . in ref .
@xcite , the authors also applied the lde to the second order shift in @xmath0 with encouraging results .
since the result for @xmath88 is known analytically in the large @xmath4 limit , one would also like to compare the predictions of the lde to this analytic result . unfortunately , as we shall see , the prescription i of ref .
@xcite is not well - behaved in this limit .
we therefore present two alternative prescriptions for @xmath88 that are well - behaved in the large-@xmath4 limit : @xmath158 , \label{d - ii } \\ \delta_{\rm iii } & = & n\int_{p}\left [ \left ( p^{2}+m^{2}\left ( 1-\delta \right ) + \sigma(p)-\sigma(0)\right ) ^{-1}-\left ( p^{2}+m^{2}\left ( 1-\delta \right ) \right ) ^{-1}\right ] .
\label{d - iii}\end{aligned}\ ] ] these both agree with ( [ delta - gen ] ) at @xmath112 . with prescription iii ,
the lde for @xmath88 generates integrals involving only propagators of mass @xmath25 . with prescription
ii , the integrals involve both propagators of mass @xmath25 and massless propagators .
our prescription iii in ( [ d - iii ] ) is closely related to the prescription i in ref .
the difference between them is given by a simple integral that is evaluated in ( [ j0 ] ) : @xmath159 because of the additional term in @xmath160 proportional to @xmath161 , the limit @xmath162 does not commute with the large-@xmath4 limit . at @xmath122 order in the @xmath5 expansion , the terms that dominate as @xmath163 are the @xmath164 term and the highest order term in the @xmath3 expansion which is proportional to @xmath165 .
the pms criterion gives a value of @xmath25 that scales like @xmath166 and the corresponding value of @xmath91 scales like @xmath167 the correct result is that @xmath88 scales like @xmath168 in the large @xmath4 limit .
thus prescription i has the wrong behavior in the large-@xmath4 limit at any order in the lde .
in this section , we study the behavior of the linear @xmath5 expansion ( lde ) in the large-@xmath4 limit defined by @xmath170 , @xmath171 with @xmath172 fixed .
we calculate @xmath88 to all orders in @xmath5 using prescriptions ii and iii .
we show that the lde converges to the correct analytic result for @xmath173 greater than some minimum value , with errors that seem to decrease roughly like @xmath34 .
we also study the improvement in the convergence obtained by using the pms criterion for @xmath25 .
if we use the family of real solutions of the pms criterion , the errors in @xmath88 seem to decrease like @xmath174 .
if we use a particular family of complex solutions , the errors seem to decrease exponentially with @xmath8 . at the critical point .
the black blob represents the @xmath73 vertex , while the shaded blob represents the complete propagator with subtracted self - energy @xmath175.,width=113 ] at order @xmath176.,width=151 ] the quantity @xmath88 defined in ( [ delta - gen ] ) has the diagrammatic representation shown in fig .
[ fig : delta ] , where the line with the shaded blob represents the propagator @xmath177^{-1}$ ] . in the large-@xmath4 limit ,
the leading contribution to the subtracted self - energy @xmath175 comes from the series of bubble - chain diagrams whose @xmath178 and @xmath32 members are shown in figs .
[ fig:2loop ] and [ fig : largen ] .
since these diagrams are of order @xmath19 , the leading contribution at large @xmath4 is obtained by expanding the expression ( [ delta - gen ] ) for @xmath88 to first order in @xmath175 . for the prescription ii
given in ( [ d - ii ] ) , the resulting expression is @xmath179 the corresponding expression for the prescription iii is obtained by replacing @xmath180 in the denominator of ( [ delta - largen ] ) . although @xmath141 is ultraviolet divergent , the difference @xmath175 is finite . at any finite order in the lde
, @xmath175 scales as @xmath181 at small @xmath182 and as @xmath183 for large @xmath182 , so the integral in ( [ delta - largen ] ) is convergent .
in addition to the diagrams for @xmath141 generated by the interaction term @xmath184 in ( [ l - int ] ) , we must also take into account insertions of @xmath185 and insertions of @xmath186 .
the effect of the @xmath187 insertions is to replace each 1pi self - energy subdiagram in @xmath141 by the subdiagram with the zero - momentum limit subtracted .
the effect of the @xmath186 insertions is to replace @xmath136 by @xmath188 . in the series of diagrams for @xmath141 that survive in the large-@xmath4 limit.,width=188 ]
we first consider the prescription ii for @xmath88 .
the expression for the large-@xmath4 diagram for @xmath88 with @xmath189 loops is given in ( [ a21 ] ) .
the integral @xmath190 is a constant multiplied by @xmath191 . after taking into account the @xmath186 insertions ,
the sum of the large-@xmath4 diagrams is @xmath192 the integrals @xmath190 are expressed as 1-dimensional integrals in ( [ b24 ] ) . inserting these expressions into ( [ 23 ] ) ,
we obtain @xmath193 ^{n-1 } , \label{d}\ ] ] where @xmath39 is the dimensionless mass parameter @xmath194 and @xmath195 is the function @xmath196 the prediction for @xmath88 at @xmath122 order in the @xmath3 expansion is obtained by expanding the expression ( [ d ] ) as a power series in @xmath5 , truncating after order @xmath197 , and then setting @xmath112 .
it is easy to show that if the lde for ( [ d ] ) converges , it converges to the correct analytic result ( [ largen - exact ] ) . after interchanging the order of the sum and the integral ,
the sum can be evaluated and we obtain @xmath198 upon taking the limit @xmath199 all dependence on the mass parameter @xmath39 disappears : @xmath200 after evaluating the integral over @xmath201 , our final result is ( [ largen - exact ] ) .
note that no prescription for @xmath46 was required to get this result .
thus , if the @xmath3 expansion converges at some value of @xmath46 , it should converge to ( [ largen - exact ] ) . in the large-@xmath4 limit as a function of @xmath46 at @xmath122 order in the lde for @xmath202 , 4 , 5 , 7 , 11 , 19 , 35 .
the curves for @xmath203 , 11 , and 19 appear in order between those labelled @xmath204 and 35.,width=453 ] the manipulations that were used to show the convergence to ( [ largen - exact ] ) involved several interchanges of limits .
first , we summed up some terms in the @xmath3 expansion to all orders to obtain the sum over diagrams in ( [ 23 ] ) .
next , we interchanged the order of the sum over diagrams and the integral over the last momentum for each diagram .
next we took the limit @xmath205 .
finally we evaluated the last momentum integral .
it is difficult to translate the conditions for the validity of each of these steps into a statement about the convergence of the lde .
however , the convergence can be easily studied numerically . in fig .
[ fig : d - mun ] , we show @xmath88 as a function of the dimensionless variational parameter @xmath39 for several orders in the @xmath5 expansion : @xmath202 and @xmath206 @xmath207 .
the horizontal line is the analytic result ( [ largen - exact ] ) .
the results are consistent with convergence to ( [ largen - exact ] ) for all @xmath46 greater than a critical value @xmath208 which is close to 0.7 .
our best estimate for the endpoint of the convergence region is @xmath209 . if @xmath210 , @xmath91 seems to diverge to @xmath211 for @xmath8 even and @xmath212 for @xmath8 odd . for any fixed value of @xmath213 ,
the convergence with @xmath8 is very slow . in column 1 of table [ tab1 ] , we give the values of @xmath88 at @xmath214 , which is the location of the minimum of the @xmath202 curve in fig . [
fig : d - mun ] .
we define the fractional error @xmath215 by @xmath216 ( taking the real part of @xmath88 is superfluous for real values of @xmath46 , but it will be important later when we consider complex values of @xmath46 . ) in fig . [
fig : err - n ] , we show a log - log plot of @xmath215 as a function of @xmath8 . the points lie close to a straight line , indicating that the errors decrease like a power of @xmath8 .
the dotted line is @xmath217 , which goes through the points for @xmath218 and 35 .
thus the errors seems to decrease roughly like @xmath34 . as a function of the order @xmath8 in the lde .
the squares , diamonds , and circles are for @xmath214 , the real solution to the pms criterion , and the imaginary solution with maximal @xmath219 , respectively .
the lines are simple curves that pass through the last 2 points.,width=453 ] the rate of convergence can be improved by changing the value of @xmath39 at each order in the lde .
one possibility is to use the pms criterion ( [ pms ] ) to determine @xmath46 . at @xmath122 order in the @xmath5 expansion
, this condition reduces to a polynomial in @xmath39 of order @xmath220 for @xmath8 even , there are no real roots . for @xmath8 odd ,
there is always one real root .
these roots correspond to the minima of the curves in fig .
[ fig : d - mun ] .
the values of @xmath221 are given in column 2 of table i. there is slight improvement compared to column 1 in the convergence toward the correct value 1 .
the fractional errors defined in ( [ eps - n ] ) are shown as a function of @xmath8 in fig .
[ fig : err - n ] .
the points lie close to a straight line , indicating that the errors decrease like a power of @xmath8 .
the dashed line is @xmath222 , which goes through the points for @xmath218 and 35 .
thus the errors seem to decrease roughly as @xmath174 .
the pms criterion has either zero or one real solution , depending on whether @xmath8 is even or odd .
however there are always @xmath223 complex - valued solutions . in the case of the anharmonic oscillator , there are families of complex solutions with much better convergence properties than families consisting of purely real solutions @xcite . in our problem , at any odd order @xmath224 the real solution always has the largest value of re@xmath88 , which is the value farthest from the correct result ( [ largen - exact ] ) .
thus this family of solutions gives the slowest possible convergence rate .
however there is a strong anticorrelation between the errors in re@xmath88 and @xmath225 .
this is illustrated in fig .
[ fig : im - re ] , which is a scatter plot of @xmath219 vs. re@xmath88 for the solutions to the pms criterion for @xmath226 .
the solutions that give the most accurate values for re@xmath88 are those with the largest values for @xmath227 .
thus we can define a nearly optimal family of solutions by choosing those with the maximal values of @xmath227 .
the values of @xmath91 for this family of solutions are given in column 3 of table i. note that im@xmath88 for these solutions shows no sign of converging to @xmath228 , and in fact iseems to be converging to a value close to @xmath229 .
however the improvement in the convergence of re@xmath88 compared to the real values of @xmath88 in column 2 is evident . a log - log plot of the fractional errors in re@xmath88 is shown in fig .
[ fig : d - mun ] .
the downward curvature of the points indicates that the errors decrease faster than any power of @xmath8 .
the solid line is @xmath230 , which goes through the points for @xmath218 and 35 . up to the order to which we have calculated
, the errors are decreasing faster than this exponential . vs. re@xmath88 for the solutions @xmath46 of the pms criterion at @xmath231 order in the lde.,width=453 ]
we obtain similar results if we use the prescription iii for @xmath88 given in ( [ d - iii ] ) .
the sum of the large-@xmath4 diagrams is the same as in ( [ 23 ] ) , except that the integrals @xmath232 are replaced by @xmath233 .
the integrals @xmath234 are expressed as 1-dimensional integrals in ( [ b25 ] ) . interchanging the order of the sum and the integral and then evaluating the sum
, we obtain @xmath235 taking the limit @xmath199 all dependence on @xmath46 disappears and after evaluating the integral we again obtain ( [ largen - exact ] ) .
thus , if the @xmath3 expansion converges , it should converge to the correct result .
the convergence can be again studied numerically .
we find that it seems to converge to ( [ largen - exact ] ) for all @xmath213 and it seems to diverge for @xmath210 , with @xmath236 . in column 4 of table [ tab1 ] , we give the values of @xmath88 for various orders @xmath8 at @xmath237 , which is the location of the minimum of @xmath88 for @xmath202 .
the convergence at a fixed value of @xmath46 is very slow . as in the case of prescription
ii , the errors seem to decrease roughly as @xmath34 .
more rapid convergence can again be obtained by using the pms criterion for @xmath39 .
the pms criterion has no real solutions for @xmath8 even and 1 real solution for @xmath8 odd .
the values of @xmath91 for the real solution at various orders @xmath8 are given in column 5 of table i. this is the family of solutions with the slowest convergence rate for re@xmath88 .
the errors seem to decrease roughly as @xmath174 . a family of solutions with a nearly optimal convergence rate for re@xmath88 consists of those with the maximal value of @xmath227 .
the values of @xmath91 for this family are shown in column 6 of table i. the errors in re@xmath88 are numerically larger than those with prescription ii , but they also seem to decrease exponentially with @xmath8 .
in this section , we use the linear @xmath5 expansion ( lde ) to calculate @xmath88 for finite @xmath4 .
we calculate @xmath88 to @xmath32 order in @xmath5 using both of the prescriptions ii and iii . setting @xmath2 and using the pms criterion
, we obtain the predictions relevant to the shifts in @xmath30 for bose - einstein condensation . at order @xmath238.,width=453 ] the first term in the lde for @xmath88 is second order in @xmath3 .
it is obtained by calculating the 3-loop diagram involving the 2-loop self - energy diagram in fig .
[ fig:2loop ] .
the contribution to @xmath91 of order @xmath239 is obtained by calculating the 4-loop diagram involving the 3-loop propagator correction in fig .
[ fig:3loop ] .
the contribution of order @xmath240 is obtained by calculating the 5-loop diagram involving the 4-loop propagator corrections in fig .
[ fig:4loop ] .
the diagrams are reduced to momentum integrals in appendix [ app : diag ] and the momentum integrals are evaluated in appendix [ app : int ] . the complete expression for @xmath91 for the @xmath1 theory through order @xmath240 is @xmath241 .
\label{delta - ii}\end{aligned}\ ] ] inserting the values for the integrals from appendix b , truncating at successively higher orders in @xmath3 and then setting @xmath112 , we obtain the results at @xmath152 , @xmath31 and @xmath32 orders in the lde : @xmath242 \frac{1}{\mu ^{3 } } \right ) , \label{d4-ii } \end{aligned}\ ] ] where @xmath46 is the dimensionless mass parameter @xmath243 in the large-@xmath4 limit , this reduces to the mass parameter introduced in ( [ mu - largen ] ) .
the accuracy of the @xmath32 order result in ( [ d4-ii ] ) is limited by the accuracy to which we have evaluated the integrals @xmath244 and @xmath245 . in fig .
[ fig : d - mu2 ] , we show @xmath88 for @xmath2 as a function of @xmath46 for @xmath246 .
the horizontal lines are the edges of the upper and lower error bars of the lattice monte carlo result given in ( [ delta - mc ] ) . at order @xmath247.,width=680 ] at order @xmath248 the pms criterion has no solutions , because ( [ d2-ii ] ) is a monotonic function of @xmath46 . at order @xmath249
the pms criterion has a single real solution @xmath250 and the resulting value for @xmath91 is @xmath251 at 4@xmath252 order in @xmath3 , @xmath91 is a monotonic function for real - valued @xmath39 , so the pms criterion has no real solutions for @xmath39 .
it does however have a conjugate pair of complex solutions whose values at @xmath2 are @xmath253 .
the corresponding values of @xmath91 are complex , but we can take the real part as the prediction for @xmath88 at 4@xmath252 order in the lde : @xmath254 for @xmath2 , the @xmath31 and @xmath32 order approximations in ( [ delta3 ] ) and ( [ delta4:2 ] ) differ from the lattice monte carlo result of ( [ delta - mc ] ) by about 66% and 62% , respectively
. these predictions are not as accurate as in the large @xmath4 limit , where the errors in the @xmath31 and @xmath32 order approximations are about 52% and 44% , respectively .
furthermore , the percentage decrease in the error in going from @xmath31 order to @xmath32 order is only about half as large for @xmath2 as it is for large @xmath4 . for @xmath2 as a function of @xmath46 at @xmath122 order in the lde using prescriptions i ( upper 3 curves on the right ) and ii ( lower 3 curves on the right ) .
the dotted , dashed , and solid curves are for @xmath255 , 3 , and 4 , respectively.,width=453 ] with the prescription iii , the expression for @xmath88 to @xmath32 order in @xmath5 is the same as ( [ delta - ii ] ) , except that the integrals @xmath232 are replaced by @xmath233 . inserting the value for the integrals @xmath233 from appendix b ,
the results at @xmath152 , @xmath31 and @xmath32 orders in the lde are @xmath256 \frac{1}{\mu ^{3 } } \right ) , \label{d4-iii } \end{aligned}\ ] ] the accuracy of the @xmath32 order result in ( [ d4-iii ] ) is limited by the accuracy to which we have evaluated the integrals @xmath257 and @xmath258 . at order @xmath259
the pms criterion has no solution . at order @xmath249
it has one real solution for @xmath39 . at 4@xmath252 order in @xmath3
, it has a conjugate pair of complex solutions .
the corresponding values of @xmath88 are given in table [ tab2 ] .
the values of @xmath88 at @xmath31 order and of re @xmath88 at @xmath32 order are close to the corresponding values from prescription ii , but a little farther from the lattice monte carlo result .
the results of the lde for prescription i can be obtained from those for prescription iii by using ( [ i - iii ] ) : @xmath260 in fig .
[ fig : d - mu2 ] , we show @xmath149 as a function of @xmath46 for @xmath255 , 3 , and 4 . the values of @xmath149 at the solutions to the pms criterion are given in column 1 of table [ tab2 ] .
at orders 3 and 4 , the pms criterion has a conjugate pair of complex solutions . as the predictions of the lde , the authors of ref .
@xcite chose the positive value for @xmath261 and the real parts of the complex values for @xmath262 and @xmath263 .
these predictions differ from the lattice monte carlo results by about 131% , 85% , and 20% , respectively .
the calculation of the shift in @xmath30 to leading order in @xmath62 can be reduced to a nonperturbative calculation of a quantity @xmath88 in a 3-dimensional @xmath13 field theory at its critical point .
this quantity can be generalized to an @xmath1 field theory and it can be calculated systematically using the @xmath19 expansion .
we have shown that the linear @xmath5 expansion ( lde ) can be applied to the calculation of @xmath88 in such a way that the large @xmath4 and @xmath264 limits commute .
the prescriptions ii and iii defined in ( [ d - ii ] ) and ( [ d - iii ] ) have this property , but the prescription i introduced in ref .
@xcite does not . for prescription ii and iii , we calculated @xmath88 in the large-@xmath4 limit to all orders in the lde , and studied its convergence properties .
we found that it seems to converge to the correct analytic result for @xmath213 , where @xmath236 .
the convergence for fixed @xmath46 is extremely slow : the errors seem to decrease roughly like @xmath34 .
the convergence can be accelerated by using the pms criterion to choose a value of @xmath46 that depends on the order in @xmath5 .
the pms criterion is a polynomial equation with multiple roots , so there are various families of predictions for @xmath88 .
the family consisting of the real solutions of the pms criterion gives the slowest possible convergence : the errors in @xmath88 seem to decrease roughly like @xmath174 .
the family consisting of the complex solutions of the pms criterion with the maximal value of @xmath219 have nearly optimal convergence : the errors in re@xmath88 seem to decrease exponentially with @xmath8 .
unfortunately , accurate predictions still require calculations to high orders in the @xmath5 expansion . to achieve 10%
accuracy requires computing to about @xmath265 order in @xmath5 . for the case
@xmath2 relevant to bose - einstein condensation , we calculated @xmath88 to @xmath32 order in the lde using both prescriptions ii and iii .
the quantity @xmath88 was previously calculated to @xmath32 order using prescription i in ref .
the pms criterion has complex solutions beginning at @xmath31 order for prescription i and at @xmath32 order for prescriptions ii and iii .
we take the values of re@xmath149 for these solutions to be the predictions for @xmath88 .
the @xmath31 and 4@xmath252 order predictions from prescription ii are smaller than the lattice monte carlo result by about 66% and 62% , respectively .
the @xmath152 , @xmath31 and 4@xmath252 order predictions from prescription i are larger than the lattice monte carlo result by about 131% , 85% , and 20% , respectively .
we argue that the 20% error in the 4@xmath252 order prediction from prescription i is fortuitously small .
prescription i differs from prescription iii by the additive term given in ( [ i - iii ] ) .
the expansion of the factor @xmath266 truncated at @xmath122 order in @xmath5 and evaluated at @xmath140 converges to 0 as @xmath127 .
the prediction from prescription i should therefore approach the prediction from prescription iii as @xmath8 increases , which suggests that it will converge to the correct answer from below . because the @xmath255 prediction from prescription i is greater than the correct result by about a factor of 2.3 , there must be a crossover in @xmath8 where the prediction goes from values greater than the correct result to smaller values .
the high accuracy of the 4@xmath252 order result suggests that this crossover occurs at @xmath204 .
thus we expect higher order predictions from prescription i to first overshoot the correct value and then approach it from below . before the lattice monte carlo calculation of ref .
@xcite , there were many previous attempts to calculate the coefficient @xmath16 in the expression ( [ c - def ] ) for the leading order shift in @xmath30 for bose - einstein condensation .
for most of these methods @xcite , there is no known systematic way of improving the calculation to make it more accurate .
such a method is best regarded as just a model of the critical @xmath1 field theory whose accuracy can only be determined by comparison with a more reliable calculational method .
a systematically improvable method has two great advantages over such a model .
first , the accuracy of the prediction can be improved with additional effort .
second , one can make a reliable estimate of the error in the prediction .
one example of a systematically improvable method is the @xmath19 expansion @xcite , which has been used to calculate @xmath26 to leading order @xcite and to next - to - leading order @xcite .
comparison with the lattice monte carlo results show that these first two approximations have errors of 77% and 30% , respectively .
our demonstration that the lde converges in the large @xmath4 limit suggests that the lde may also be a systematically improvable method for @xmath2 .
although the convergence rate of the optimized lde appears to be exponential in the large-@xmath4 limit , it is still rather slow .
one must calculate to about @xmath265 order in @xmath5 to achieve 10% accuracy . for the case
@xmath2 relevant to bose - einstein condensation , we have calculated @xmath88 to @xmath32 order in the lde .
it may be feasible to extend the calculation to @xmath267 order , but it would be very difficult to extend it to much higher order .
the slow convergence in the large-@xmath4 limit suggests that even if the lde also converges for @xmath2 , a strict expansion in @xmath5 is not useful for quantitative calculations . it may however be possible to use order - dependent mappings to change the expansion in @xmath5 into a more rapidly converging expansion @xcite . in conclusion
, we have shown that the lde for the quantity @xmath88 that determines the leading order shift in @xmath30 converges in the large @xmath4 limit if we use an appropriate prescription .
if the pms criterion is used to optimize the convergence , the errors seem to decrease exponentially in the order of the lde .
this suggests that the lde may also be a systematically improvable approximation scheme in the case @xmath2 relevant to bose - einstein condensation . with the use of order - dependent mappings ,
it may be possible to to develop the lde into a general and powerful tool for quantitative calculations in superrenormalizable field theories , even at a critical point .
this research was supported by doe grant no .
de - fg02 - 91-er4069 .
in this appendix , we give the expressions for the diagrams that contribute to @xmath91 through order @xmath240 . the diagrams for @xmath91 are obtained by inserting propagator corrections into the diagram in fig .
[ fig : delta ] , with the propagator corrections subtracted at @xmath228 momentum . in addition , any 2-loop self - energy subdiagram like that in fig .
[ fig:2loop ] must also be subtracted at @xmath228 momentum .
we first introduce a compact notation for the massive propagator and for the functions of the momentum that correspond to subdiagrams in which two points are connected by 2 or 3 lines : @xmath268 where @xmath269 denotes the 3-dimensional integral over the momentum @xmath270 @xmath271 there is a simple analytic expression for the function @xmath272 : @xmath273 the function @xmath274 is ultraviolet divergent , but the divergence can be eliminated by a subtraction at @xmath228 momentum .
the subtracted function can be evaluated analytically : @xmath275 .\ ] ] we proceed to give the expression for the contributions to @xmath88 using the prescription ii given in ( [ d - ii ] ) .
the contribution from the 2-loop propagator correction in fig .
[ fig:2loop ] is @xmath276 .
\label{a8}\end{aligned}\ ] ] the contribution from the sum of the 3-loop propagator corrections in fig .
[ fig:3loop ] is @xmath277 d_{2}^{2}\left ( q\right ) .
\label{a10}\end{aligned}\ ] ] the contributions from the 4-loop propagator corrections in fig .
[ fig:4loop ] are @xmath278 where the momentum integrals are @xmath279 d_{2}^{3}\left ( q\right ) , \label{a16 } \\ i_{4b } & = & \int_{p}\frac{1}{\left ( p^{2}\right ) ^{2}}\int_{qr}d_{1}\left ( q\right ) d_{1}\left ( q - r\right ) d_{2}\left ( r\right ) \nonumber \\ & & \hspace{3 cm } \times \left [ d_{1}\left ( r - p\right ) d_{2}\left ( p - q\right ) -d_{1}\left
( r\right ) d_{2}\left ( q\right ) \right ] , \label{a17 } \\ i_{4c } & = & \int_{p}\frac{1}{\left ( p^{2}\right ) ^{2}}\int_{qrs}\left [ d_{1}\left ( p+q\right ) -d_{1}\left ( q\right ) \right ] d_{1}\left ( r\right ) d_{1}\left ( q+r\right ) \nonumber \\ & & \hspace{3 cm } \times d_{1}\left ( s\right ) d_{1}\left ( q+s\right ) d_{2}\left ( r - s\right ) , \label{a18 } \\ i_{4d } & = & \int_{p}\frac{1}{\left ( p^{2}\right ) ^{2}}\int_{q}\left [ d_{2}\left ( p+q\right ) -d_{2}\left ( q\right )
\right ] d_{1}^{2}\left ( q\right ) \left [ d_{3}\left ( q\right ) -d_{3}\left ( 0\right ) \right ] , \label{a19 } \\ i_{4e } & = & \int_{p}\frac{1}{\left ( p^{2}\right ) ^{3}}\left [ d_{3}\left ( p\right ) -d_{3}\left
( 0\right ) \right ] ^{2}. \label{a20}\end{aligned}\ ] ] if one uses the prescription iii for @xmath88 given in ( [ d - ii ] ) , the corresponding expressions for the diagrams are obtained by replacing each factor of @xmath280 in the integrand by @xmath281 .
we denote the integral corresponding to @xmath232 by @xmath233 . if one uses the prescription i of ref .
@xcite , there is also a diagram that is @xmath282 order in @xmath5 : @xmath283 . \label{j0}\end{aligned}\ ] ] the @xmath32 in the class of diagrams that survive in the large-@xmath4 limit is shown in fig .
[ fig : largen ] .
the expression for the first three diagrams in the series are given in ( [ a7 ] ) , ( [ a9 ] ) and ( [ a11 ] ) .
the general expression for the @xmath284-loop diagram is @xmath285 ^{n-1}\left [ d_{1}\left ( p+q\right ) -d_{1}\left ( q\right ) \right ] . \label{a22}\end{aligned}\ ] ] if we use prescription iii , the corresponding integral is @xmath286^{n-1 } \left [ d_{1}(p+q ) -d_{1}(q ) \right ] .
\label{a23}\ ] ] a closed - form expression for the @xmath287 factor associated with this class of diagrams is @xmath288 where we have chosen the overall normalization such that @xmath289 for @xmath290 and @xmath291 this expression reproduces the @xmath287 factor in ( [ a7 ] ) , ( [ a9 ] ) and ( [ a11 ] ) . note that for @xmath292 the large-@xmath4 limit of the coefficient is @xmath293 , \qquad n\geq 3.\ ] ] the corrections are suppressed by powers of @xmath294 if we use the simpler expression @xmath295 the error is of order @xmath19 for @xmath296 and of order @xmath297 for @xmath8 of order @xmath4 .
in this appendix , we evaluate the momentum integrals that appear in the expressions for the diagrams in appendix a. we first consider the integrals @xmath232 that arise if we use prescription ii for @xmath88 given in ( [ d - ii ] ) .
the 3-loop and 4-loop integrals ( [ a8 ] ) and ( [ a10 ] ) can be evaluated analytically : @xmath298 \frac{1}{m^{2}}.\end{aligned}\ ] ] the remaining integrals must be evaluated numerically .
three of the 5-loop integrals can be reduced to integrals over a single momentum .
the integral @xmath299 in ( [ a20 ] ) is already in such a form .
the integrals @xmath300 in ( a16 ) and @xmath301 in ( [ a19 ] ) can be reduced to this form by first integrating over @xmath302 @xmath303 & = & -\frac{m}{4\pi } \frac{1}{\left ( q^{2}+m^{2}\right ) ^{2 } } , \\
\int_{p}\frac{1}{\left ( p^{2}\right ) ^{2}}\left [ d_{2}\left ( p+q\right ) -d_{2}\left ( q\right ) \right ] & = & -\frac{1}{32\pi ^{2}}\frac{1}{q^{2}+4m^{2}}.\end{aligned}\ ] ] the resulting expression for the integrals are @xmath304 \frac{1}{q^{2}+4m^{2}}.\end{aligned}\ ] ]
reducing these integrals to 1-dimensional integrals and evaluating then numerically , we obtain @xmath305 the remaining two 5-loop integrals can be expressed as 3-loop integrals with the mercedes - benz topology .
the integral @xmath244 in ( [ a17 ] ) already has this form .
the integral @xmath245 in ( [ a18 ] ) can be put in this form by integrating over @xmath182 : @xmath306 these integrals can be reduced to 7-dimensional integrals that are then evaluated numerically using the monte carlo method to obtain @xmath307 the integrals @xmath233 required if @xmath88 is calculated using the prescription iii differ from the integrals @xmath190 calculated above only in the substitution @xmath308 .
the 3-loop integral can be evaluated analytically : @xmath309 the 4-loop and 5-loop integrals must be evaluated numerically .
four of them can be reduced to integrals over a single momentum .
the integrals @xmath310 and @xmath311 already have this form .
the integral @xmath312 and @xmath313 can be put into this form by first integrating over @xmath182 : @xmath314 & = & -\frac{3}{8\pi } \frac{m}{\left ( q^{2}+m^{2}\right ) \left ( q^{2}+4m^{2}\right ) } , \\
\int_{p}d_{1}^{2}\left ( p\right ) \left [ d_{2}\left ( p+q\right ) -d_{2}\left ( q\right ) \right ] & = & \frac{1}{32\pi ^{2}}\frac{1}{mq } \left [ \arctan \frac{q}{3m}-\arctan \frac{q}{2m}\right ] .\end{aligned}\ ] ] reducing the final momentum integrals to a 1-dimensional integral and then evaluating it numerically , we obtain @xmath315 the remaining two 5-loop integrals can be expressed as 3-loop integrals with the mercedes - benz topology .
the integral @xmath244 in ( [ a17])already has this form . the integral @xmath245 in ( [ a18 ] ) can be put in this form by integrating over @xmath316 these integrals can be reduced to 7-dimensional integrals that are then evaluated using the monte carlo method to obtain @xmath317 if we use the prescription i of ref .
@xcite , we must also evaluate the integral @xmath318 defined in ( [ j0 ] ) : @xmath319 the integrals @xmath190 that survive in the large-@xmath4 limit can all be reduced to integrals over a single momentum . after interchanging orders of integration in ( [ a22 ] ) and integrating over @xmath182 , the remaining integral reduces to @xmath320 ^{n-1 } , \label{b24}\ ] ] where the function @xmath195 is defined in ( [ a - y ] ) . applying the same strategy to ( [ a23 ] )
, we obtain @xmath321 ^{n-1}. \label{b25}\ ] ] p. arnold and g. d. moore , phys .
lett . * 87 * , 120401 ( 2001 ) .
p. arnold , g. d. moore and b. tomasik , phys .
a * 65 * , 013606 ( 2002 ) .
stoof , phys .
a * 45 * , 8398 ( 1992 ) ; m. bijlsma and h.t.c .
stoof , phys .
a * 54 * , 5085 ( 1998 ) .
a. duncan and h. f. jones , phys .
d * 47 * , 2560 ( 1993 ) ; c. arvanitis , h. f. jones and c. s. parker , phys . rev .
d * 52 * , 3704 ( 1995 ) .
r. guida , k. konishi and h. suzuki , ann .
phys . * 241 * , 152 ( 1995 ) ; ann .
phys . * 249 * , 109 ( 1996 ) .
w. janke and h. kleinert , quant - ph/9502019 ; h. kleinert and w. janke , phys .
* a206 * , 283 ( 1995 ) . b. bellet , p. garcia and a. neveu , int . j. mod
. phys .
a * 11 * , 5587 ( 1996 ) . | the leading correction from interactions to the transition temperature @xmath0 for bose - einstein condensation can be obtained from a nonperturbative calculation in the critical @xmath1 scalar field theory in 3 dimensions with @xmath2 .
we show that the linear @xmath3 expansion can be applied to this problem in such a way that in the large-@xmath4 limit it converges to the exact analytic result . if the principal of minimal sensitivity is used to optimize the convergence rate , the errors seem to decrease exponentially with the order in the @xmath5 expansion .
for @xmath6 we calculate the shift in @xmath0 to fourth order in @xmath3 .
the results are consistent with slow convergence to the results of recent lattice monte carlo calculations . |
jet production in neutral - current ( nc ) and charged - current ( cc ) deep inelastic @xmath0 scattering ( dis ) provides a test of perturbative qcd ( pqcd ) calculations and of the electroweak sector of the standard model ( sm ) .
jet cross sections allow the determination of one of the fundamental parameters of qcd , the strong coupling constant @xmath1 , and help to constrain the parton densities in the proton .
new particles or interactions may be observed by deviations of the measured jet cross sections with respect to the predictions . up to leading order ( lo ) in @xmath1 , jet production in nc and cc
dis proceeds via the quark - parton model ( qpm ) ( @xmath2 , where @xmath3 , @xmath4 or @xmath5 ) , boson - gluon fusion ( bgf ) ( @xmath6 ) and qcd - compton ( qcdc ) ( @xmath7 ) processes ( see figure [ one ] ) . (
18.0,5.5 ) ( 0.0,0.5 ) ( 2.0,0.3)(a ) ( 7.5,0.3)(b ) ( 14.0,0.3)(c ) the jet production cross section is given in pqcd by the convolution of the parton densities in the proton and the subprocess cross section , @xmath8 where @xmath9 is the fraction of the proton momentum taken by the interacting parton , @xmath10 are the parton distribution functions ( pdfs ) in the proton , @xmath11 is the factorisation scale , @xmath12 is the subprocess cross section and @xmath13 is the renormalisation scale .
the proton pdfs are determined from global fits to experimental data ( see e.g. @xcite ) and parametrise the long - distance structure of the target hadron .
the subprocess cross section is calculable in pqcd at any order and represents the short - distance structure of the interaction .
the determination of @xmath1 from measurements of cross sections in dis depends on the proton pdfs .
thus , to perform a qcd analysis , a functional form of the pdfs must be assumed and the determination of @xmath1 should be made from observables with a small dependence on the pdfs , e.g. ratios of observables .
for precise measurements of @xmath1 , the experimental and theoretical uncertainties should be small .
small experimental uncertainties are obtained by measuring e.g. ratios of observables and the theoretical uncertainties are reduced by including next - to - leading order ( nlo ) qcd corrections to the cross section calculations .
several programs are available to make nlo qcd calculations of jet cross sections in dis : disent @xcite , mepjet @xcite , disaster++ @xcite and nlojet @xcite . the nlo corrections include virtual corrections with internal particle loops and real corrections with a third parton in the final state .
the programs differ in the treatment of the real corrections : disent , disaster++ and nlojet use the subtraction method whereas mepjet uses the phase space slicing method .
disent and disaster++ are found to agree at the @xmath14 level , and nlojet is in good agreement with them .
mepjet agrees well with the other programs at lo , but presents differences of the order of @xmath15 at nlo .
mepjet is the only program that includes @xmath4 and @xmath16 exchange and only disaster++ allows the possibility to turn on the number of active flavours as a function of the scale .
the calculation of jet cross sections at nlo in dis depends on two scales , @xmath13 and @xmath11 .
two possible choices for these scales that have been considered in the analyses presented here are @xmath17 , where @xmath18 is the virtuality of the exchanged boson , and the jet transverse energy , @xmath19 .
since the nlo qcd calculations are for jets of partons and the measurements are done at the hadron level , the calculations need to be corrected for hadronisation effects .
these effects have been estimated using monte carlo models for parton radiation and hadronisation @xcite .
the uncertainties of the qcd calculations include that due to terms beyond nlo , which is usually estimated by varying the renormalisation scale by factors between @xmath20 and @xmath21 ; it amounts to @xmath22 and translates into @xmath23 in the determination of @xmath24 .
the uncertainties on the value of @xmath24 and of the proton pdfs amount to @xmath25 .
the size and uncertainty of the hadronisation corrections have been taken into account : the hadronisation correction factor amounts to @xmath25 with an uncertainty of the order of @xmath14 .
the use of inclusive jet cross sections in a qcd analysis presents several advantages : inclusive jet cross sections are infrared insensitive and better suited to test resummed calculations and the theoretical uncertainties are smaller than for dijet cross sections .
inclusive jet cross sections in nc interactions have been measured @xcite in the breit frame using the @xmath26 cluster algorithm @xcite in the longitudinally invariant inclusive mode for the kinematic region of @xmath27 2 , where @xmath18 is the virtuality of the exchanged boson .
these cross sections refer to jets of transverse energy measured in the breit frame , @xmath28 , above 8 gev and jet pseudorapidity also in the breit frame in the range @xmath29 .
the inclusive jet cross sections as a function of @xmath18 and @xmath28 are shown in figure [ two ] .
( 18.0,15.0 ) ( -1.0,5.0 ) ( -1.0,0.0 ) ( 8.0,5.0 ) ( 8.0,0.0 ) ( 7.0,13.5)(a ) ( 7.0,4.3)(b ) ( 16.0,13.5)(c ) ( 16.0,4.3)(d ) the measured cross section has a steep fall - off of five ( four ) orders of magnitude over the measured range of @xmath18 ( @xmath28 ) .
nlo qcd calculations using disent , the mrst99 @xcite set of proton pdfs and different choices of @xmath13 have been compared to the data .
the calculations describe reasonably well the @xmath18 ( @xmath28 ) dependence of the cross section for @xmath30 2 ( @xmath31 gev ) .
inclusive jet cross sections have been measured in the laboratory frame in nc @xcite and cc @xcite interactions to search for deviations from the sm .
the jets have been reconstructed using the @xmath26 cluster algorithm in the kinematic region of @xmath32 2 and at least one jet of @xmath33 gev and @xmath34 for nc ( cc ) interactions is required .
the inclusive jet cross sections as a function of @xmath19 for nc and cc interactions are shown in figure [ three ] .
the cross section in nc interactions displays a steep fall - off of four orders of magnitude over the measured range .
the behaviour of the @xmath19 distribution in cc interactions is very different : it is approximately constant at low @xmath19 and falls less rapidly than in nc , and it approaches the nc cross section for @xmath35 gev .
this behaviour can be interpreted as due to the presence of a massive propagator in cc events , as it has already been observed in measurements of the @xmath18 dependence of the inclusive cc dis cross section @xcite .
the cross sections for jets give an independent measurement of the different electroweak boson propagators in nc and cc dis .
the predictions of monte carlo models using either the color - dipole model ( cdm ) @xcite or first - order qcd matrix elements plus parton - showers ( meps ) @xcite have been compared to the measurements .
the calculations describe reasonably well the measured cross sections .
no significant deviation from the sm is observed in nc or cc interactions up to the highest @xmath19 measured value ( @xmath36 gev ) .
( 18.0,10.0 ) ( 2.0,-3.5 ) ( 11.7,7.5)(a ) ( 11.7,2.5)(b )
differential cross sections for dijet production in nc dis have been measured @xcite in the breit frame for @xmath37 2 . in this kinematic region ,
the experimental uncertainties on the reconstruction of both the positron and the hadronic final state are smaller than at lower @xmath18 .
in addition , the theoretical uncertainties due to the modelling of the hadronic final state , to the proton pdfs and to the higher - order contributions are minimised . for these measurements ,
the @xmath26 cluster algorithm was run in the longitudinally invariant inclusive mode in the breit frame .
two jets of @xmath38 gev , @xmath39 gev and @xmath40 in each event were required , where @xmath41 is the transverse energy of the jet in the breit frame with the highest ( second highest ) transverse energy in the event .
the use of asymmetric cuts on @xmath42 avoids infrared sensitive regions where the behaviour of the cross section as predicted by the nlo qcd programs is unphysical .
the differential dijet cross sections as functions of @xmath43 , where jet 1 ( 2 ) is the jet with highest ( lowest ) pseudorapidity in the breit frame , and @xmath18 are presented in figure [ four ] .
nlo qcd predictions calculated using disent are compared to the measurements .
the predictions , which assume @xmath44 , provide a good overall description of both the shape and magnitude of the measured cross sections . the dijet fraction ,
@xmath45 , defined as the ratio of the dijet to the inclusive cross section ( see figure [ five](a ) ) , has been measured as a function of @xmath18 .
the dijet fraction increases with increasing @xmath18 due to phase - space effects . for the cross sections as a function of the jet transverse energies and @xmath18 ,
there is agreement at the @xmath46 level between data and theory over four orders of magnitude , demonstrating the validity of the description of the dynamics of dijet production by the nlo qcd hard processes .
( 18.0,14.0 ) ( 1.0,0.0 ) ( 18.0,10.0 ) ( 1.0,0.0 ) dijet cross sections in cc interactions have been measured @xcite in the laboratory frame as a function of the transverse energy of each of the two highest @xmath19 jets ( figure [ six](a ) ) and as a function of the invariant mass , @xmath47 , of these two jets ( figure [ six](b ) ) .
the jets have been reconstructed using the @xmath26 cluster algorithm in the kinematic region of @xmath48 2 and at least two jets of @xmath49 gev and @xmath34 are required .
values as high as 60 gev for @xmath19 and @xmath47 are accessible with the luminosity used in this analysis .
the shapes of the @xmath19 distributions at low @xmath19 in inclusive jet ( see figure [ three ] ) and dijet production are very different : this is interpreted as the inclusive jet cross section being governed by electroweak interactions and the heavy mass of the boson propagator at low @xmath19 , whereas the dijet cross section is mainly dictated by qcd .
the predictions of the cdm and meps models describe reasonably well the shape of the measured dijet cross sections , but they underestimate the normalisation by @xmath50 ; this is understood since these models only include up to @xmath51 diagrams .
( 18.0,8.0 ) ( -0.5,-3.0 ) ( 8.3,-3.0 ) ( 7.5,6.1)(a ) ( 7.5,2.0)(b ) ( 16.3,6.1)(c ) ( 16.3,2.0)(d )
the investigation of the internal structure of jets gives insight into the transition between a parton produced in a hard process and the experimentally observed spray of hadrons . at sufficiently high @xmath19 , where fragmentation effects become negligible ,
the jet structure is expected to be calculable in pqcd .
the lowest non - trivial order contribution to the measurements of jet substructure is given by @xmath52 calculations .
thus , measurements of jet substructure provide a stringent test of pqcd and allow a determination of @xmath1 by comparing nlo calculations to the measurements . at present , this is only possible for jets defined in the laboratory frame since only in this frame can three partons be inside one jet ( see figure [ seven ] ) .
( 18.0,4.0 ) ( 4.0,0.0 ) the jet substructure was studied by means of the integrated jet shape in inclusive jet production in nc interactions @xcite using the @xmath26 cluster algorithm in the laboratory frame .
the integrated jet shape is defined as the average fraction of the jet s transverse energy that lies inside a cone in the @xmath53 plane of radius @xmath54 concentric with the jet axis , @xmath55 where @xmath56 is the transverse energy associated with the jet within the given cone of radius @xmath54 and @xmath57 is the total number of jets in the sample .
only the particles assigned to the jet are considered . the integrated jet shape as a function of the radius @xmath54 in different regions of @xmath19
is shown in figure [ eight ] .
( 18.0,12.5 ) ( 1.0,-0.5 ) in pqcd , @xmath58 is calculated as the fraction of the jet s transverse energy , due to parton emission , that lies outside of the cone of radius @xmath54 , @xmath59\over
\etjet\ \sigma_{\rm jet}(\etjet)},\ ] ] where @xmath60 is the cross section for inclusive jet production .
nlo qcd predictions for the integrated jet shape are derived from the above formula by computing the numerator to @xmath61 and the denominator to @xmath62 .
calculations using disent have been compared to the data .
they give a very good description of the measurements for @xmath63 ( see figure [ eight ] ) .
the jet substructure was also studied by means of the mean subjet multiplicity in nc @xcite and cc @xcite interactions .
subjets are jet - like objects within a jet which are resolved by reapplying the @xmath26 cluster algorithm until for every pair of particles the quantity @xmath64\ ] ] is above @xmath65 , where @xmath66 is the resolution parameter .
the mean subjet multiplicity as a function of the resolution parameter @xmath66 is presented in figure [ nine ] for nc interactions . in pqcd , @xmath67 is calculated as the ratio of the cross section for @xmath68 subjets over that of inclusive jet production , @xmath69 nlo qcd predictions for the mean subjet multiplicity are derived from the above formula by adding 1 and computing the numerator ( denominator ) to @xmath61 ( @xmath62 ) .
nlo calculations computed using disent and different parametrisations of the proton pdfs , which were determined assuming different values of @xmath1 , have been compared to the data .
the nlo qcd calculation using the cteq4 m proton pdfs and @xmath70 gives a good description of the measurements .
( 18.0,10.0 ) ( 2.0,-0.5 ) the measured mean subjet multiplicity as a function of @xmath66 for different regions in @xmath71 and @xmath19 is shown in figures [ ten ] and [ eleven ] , respectively , for cc interactions . the predicted mean subjet multiplicities at the hadron level using the cdm and meps models
have been compared to the measurements .
the measured mean subjet multiplicities are well described by the predictions .
( 18.0,10.0 ) ( 3.0,-2.5 ) ( 18.0,10.0 ) ( 3.0,-2.5 ) the @xmath19 dependence of the integrated jet shape at a fixed value of @xmath72 ( figure [ twelve](a ) ) and that of the mean subjet multiplicity at a fixed value of @xmath73 ( figure [ twelve](b ) ) in nc interactions have also been studied @xcite .
the jet shape increases and the subjet multiplicity decreases as @xmath19 increases : the jets become narrower as @xmath19 increases .
nlo qcd calculations using disent and different parametrisations of the pdfs have been compared to the data .
they give a good description of the measurements and display the sensitivity of these observables to @xmath1 .
( 18.0,8.0 ) ( -2.0,-1.5 ) ( 8.0,-0.5 ) ( 7.0,8.0)(a ) ( 16.2,8.0)(b ) figure [ thirteen](a ) shows the measured @xmath74 at @xmath73 as a function of @xmath71 in cc interactions .
it exhibits no significant dependence on @xmath71 , as in the case of the jet shape @xcite .
the measured @xmath74 at @xmath73 as a function of @xmath19 is presented in figure [ thirteen](b ) .
it decreases as @xmath19 increases and thus the jets become narrower as @xmath19 increases . in both figures , the predictions from the cdm model are in good agreement with the data whereas the predictions from the meps model are slightly above the data . in figure [ thirteen](b ) , the measurements from figure [ twelve](b ) are also shown .
the measured @xmath74 at @xmath73 in cc and nc are very similar .
this similarity can be attributed to a large content of final - state quark jets in these two processes and shows that the pattern of qcd radiation close to a primary quark is to a large extent independent of the hard scattering process .
( 18.0,7.0 ) ( -1.0,-9.0 ) ( 8.5,-9.0 ) ( 7.0,1.8)(a ) ( 16.5,1.8)(b )
in recent publications @xcite , a new method to determine @xmath1 has been presented .
the method consists of performing nlo calculations of an observable using the five cteq4 sets of the `` a - series '' @xcite ; the value of @xmath24 in each calculation is that of the corresponding set of pdfs .
the calculations for each @xmath24 are used to parametrise the @xmath1 dependence of the observables , and then , from the measurements , a value of @xmath24 is directly extrapolated .
the measurements of the inclusive jet cross section @xcite and the dijet fraction @xcite as a function of @xmath18 and the integrated jet shape @xcite and mean subjet multiplicity @xcite as a function of @xmath19 have been used to determine a value of @xmath24 . from each measurement in a given @xmath18 or @xmath19 region , a value of @xmath24 was determined .
the results are shown in figure [ fourteen ] separately for the inclusive jet cross section , the mean subjet multiplicity and the integrated jet shape .
( 18.0,6.0 ) ( -1.0,-0.5 ) ( 4.0,-0.5 ) ( 8.0,-2.5 ) ( 3.6,5.3)(a ) ( 7.5,5.3)(b ) ( 16.0,4.0)(c ) a combined value of @xmath24 from each observable has been obtained by performing a @xmath75 fit to the data in the following kinematic ranges : @xmath30 2 for the inclusive jet cross section , @xmath76 2 for the dijet fraction , @xmath77 gev for the integrated jet shape and @xmath78 gev for the mean subjet multiplicity .
these are the kinematic regions where the hadronisation correction factors to the nlo calculations are small for each observable .
the combined values of @xmath24 determined from each observable are : inclusive jet cross section:@xmath79 ; dijet fraction:@xmath80 ; integrated jet shape:@xmath81 ; mean subjet multiplicity:@xmath82 .
the values of @xmath24 obtained from the different observables are compatible with each other and with the world average @xcite ( see figure [ fifteen ] ) .
( 18.0,13.0 ) ( 2.5,-0.5 ) the qcd fit of the dijet fraction has been repeated in five @xmath18 bins to test the scale dependence of the renormalised strong coupling constant .
the method of the fit is the same as outlined above , but the @xmath1 dependence of the dijet fraction was parametrised in terms of @xmath83 , where @xmath84 is the mean value of @xmath85 in each bin .
the measured @xmath83 values are shown in figure [ five](b ) .
the measurements are compared with the renormalisation group predictions obtained from the pdg @xmath24 value and its associated uncertainty .
the values are in good agreement with the predicted running of the strong coupling constant over a large range in @xmath85 .
recent measurements of jet cross sections in neutral - current and charged - current deep inelastic @xmath0 scattering have been presented . from qcd analyses of these measurements , determinations of the strong coupling constant @xmath1
have been obtained .
the results are consistent with the world average value of @xmath24 .
the measurements have been performed in regions of phase space where the experimental uncertainties are small and thus the determinations are precise . however , the next - to - leading order ( nlo ) calculations used in the qcd analyses have relatively large uncertainties . for more accurate measurements of the parameters of qcd ,
improved calculations are needed .
for instance , higher - order corrections are required to reduce the largest theoretical uncertainty , namely that of the terms beyond nlo . * acknowledgments*. this report is partially based on a talk given at the _ international europhysics conference on high energy physics _ ,
budapest , hungary , july 2001 .
i would like to thank j. terrn for a critical reading of the manuscript . | recent measurements of jet cross sections in neutral - current and charged - current deep inelastic @xmath0 scattering are presented .
the results of the qcd analyses on these measurements to determine the strong coupling constant are also reported . |
quantum mechanics was originally introduced as a non commutative matrix calculus of observables by werner heisenberg ( heisenberg 1925 ) and parallel as a wave mechanics by erwin schrdinger ( schrdinger 1926 ) .
both structurally very different theories , matrix mechanics and wave mechanics could explain fruitfully the early observed quantum phenomena .
already in the same year the two theories were shown to be realizations of the same , more abstract , ket - bra formalism by dirac ( dirac 1958 ) .
only some years later , in 1934 , john von neumann put forward a rigorous mathematical framework for quantum theory in an infinite dimensional separable complex hilbert space ( von neumann 1955 ) .
matrix mechanics and wave mechanics appear as concrete realizations : the first one if the hilbert space is @xmath1 , the collection of all square summable complex numbers , and the second one if the hilbert space is @xmath2 , the collection of all square integrable complex functions .
the formulation of quantum mechanics in the abstract framework of a complex hilbert space is now usually referred to as the ` standard quantum mechanics ' .
the basic concepts - the vectors of the hilbert space representing the states of the system and the self - adjoint operators representing the observables - in this standard quantum mechanics are abstract mathematical concepts defined mathematically in and abstract mathematical space , and this is a problem for the physicists working to understand quantum mechanics .
several approaches have generalized the standard theory starting from more physically defined basic concepts .
john von neumann and garett birkhoff have initiated one of these approaches ( birkhoff and von neumann 1936 ) were they analyze the difference between quantum and classical theories by studying the ` experimental propositions '
. they could show that for a given physical system classical theories have a boolean lattice of experimental propositions while for quantum theory the lattice of experimental propositions is not boolean .
similar fundamental structural differences between the two theories have been investigated by concentrating on different basic concepts .
the collection of observables of a classical theory was shown to be a commutative algebra while this is now the case for the collection of quantum observables ( segal 1947 , emch 1984 ) .
luigi accardi and itamar pitowski obtained a analogous result by concentrating on the probability models connected to the two theories : classical theories have a kolmogorovian probability model while the probability model of a quantum theory is non kolmogorovian ( accardi 1982 , pitowski 1989 ) .
the fundamental structural differences between the two types of theories , quantum and classical , in different categories , was interpreted as indicating also a fundamental difference on the level of the nature of the reality that both theories describe : the micro world should be ` very different ' from the macro world .
the author admits that he was himself very much convinced of this state of affairs also because very concrete attempts to understand quantum mechanics in a classical way had failed as well : e.g. the many ` physical ' hidden variable theories that had been tried out ( selleri 1990 ) . in this paper
we want to concentrate on this problem : in which way the quantum world is different from the classical world .
we shall do this in the light of the approach that we have been elaborating in brussels and that we have called the ` hidden measurement formalism ' .
we concentrate also on the different paradoxes in quantum mechanics : the measurement problem , the schrdinger cat paradox , the classical limit , the einstein - podolsky - rosen paradox and the problem of non - locality .
we investigate which ones of these quantum problems are due to shortcomings of the standard formalism and which ones point out real physical differences between the quantum and classical world .
as we mentioned already in the foregoing section , the structural difference between quantum theories and classical theories ( boolean lattice versus non - boolean lattice of propositions , commutative algebra versus non commutative algebra of observables and kolmogorovian versus non kolmogorovian probability structure ) is one of the most convincing elements for the belief in a deep difference between the quantum world and the classical world . during all the years that these structural differences have been investigated ( mostly mathematically ) these
has not been much understanding of the physical meaning of these structural differences .
in which way would these structural differences be linked to some more intuitive but physically better understood differences between quantum theory and classical theory ? within the hidden measurement approach
we have been able to identify the physical aspects that are at the origin of the structural differences between quantum and classical theories .
this are two aspects that both characterize the nature of the measurements that have to be carried out to test the properties of the system under study .
let us formulate these two aspects carefully first .
_ we have a quantum - like theory describing a system under investigation if the measurements needed to test the properties of the system are such that : _ ( 1 ) : : _ the measurements are not just observations but provoke a real change of the state of the system _ ( 2 ) : : _ there exists a lack of knowledge about the reality of what happens during the measurement process _ it is the lack of knowledge * ( 2 ) * that is theoretically structured in a non kolmogorovian probability model.in a certain sense it is possible to interpret the second aspect , the presence of the lack of knowledge on the reality of the measurement process , as the presence of ` hidden measurements ' instead of ` hidden variables ' . indeed ,
if a measurement is performed with the presence of such a lack of knowledge , then this is actually the classical mixture of a set of classical hidden measurements , were for such a classical hidden measurement there would be no lack of knowledge . in an analogous way as in a hidden variable theory ,
the quantum state is a classical mixture of classical states .
this is the reason why we have called the formalism that we are elaborating in brussels and that consists in formalizing in a mathematical theory the physical situations containing the two mentioned aspects , the ` hidden measurement formalism ' .
after we had identified the two aspects * ( 1 ) * and * ( 2 ) * it was not difficult to invent a quantum machine fabricated only with macroscopic materials and producing a quantum structure isomorphic to the structure of a two dimensional complex hilbert space , describing for example the spin of a quantum particle with spin @xmath3 ( aerts 1985 , 1986 , 1987 ) . this quantum machine has been presented in different occasions meanwhile ( aerts 1988a , b , 1991a , 1995 ) and therefore we shall only , for the sake of completeness , introduce it shortly here .
the machine that we consider consists of a physical entity @xmath4 that is a point particle @xmath5 that can move on the surface of a sphere , denoted @xmath6 , with center @xmath7 and radius @xmath8 .
the unit - vector @xmath9 where the particle is located on @xmath6 represents the state @xmath10 of the particle ( see fig .
for each point @xmath11 , we introduce the following measurement @xmath12 .
we consider the diametrically opposite point @xmath13 , and install a piece of elastic of length 2 , such that it is fixed with one of its end - points in @xmath14 and the other end - point in @xmath13 .
0.7 cm 3 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 12 pt fig . 1 : a representation of the quantum machine . in ( a ) the physical entity @xmath15
is in state @xmath16 in the point @xmath17 , and the elastic corresponding to the measurement @xmath18 is installed between the two diametrically opposed points @xmath19 and @xmath20 . in
( b ) the particle @xmath15 falls orthogonally onto the elastic and stick to it . in ( c )
the elastic breaks and the particle @xmath15 is pulled towards the point @xmath19 , such that ( d ) it arrives at the point @xmath19 , and the measurement @xmath21 gets the outcome @xmath22 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ once the elastic is installed , the particle @xmath5 falls from its original place @xmath9 orthogonally onto the elastic , and sticks on it ( fig 1,b ) .
then the elastic breaks and the particle @xmath5 , attached to one of the two pieces of the elastic ( fig 1,c ) , moves to one of the two end - points @xmath14 or @xmath13 ( fig 1,d ) . depending on whether the particle @xmath5 arrives in @xmath14 ( as in fig 1 ) or in @xmath13
, we give the outcome @xmath23 or @xmath24 to @xmath12 .
we can easily calculate the probabilities corresponding to the two possible outcomes .
therefore we remark that the particle @xmath5 arrives in @xmath14 when the elastic breaks in a point of the interval @xmath25 ( which is the length of the piece of the elastic between @xmath13 and the point where the particle has arrived , or @xmath26 ) , and arrives in @xmath13 when it breaks in a point of the interval @xmath27 ( @xmath28 ) .
we make the hypothesis that the elastic breaks uniformly , which means that the probability that the particle , being in state @xmath10 , arrives in @xmath14 , is given by the length of @xmath25 divided by the length of the total elastic ( which is 2 ) .
the probability that the particle in state @xmath10 arrives in @xmath13 is the length of @xmath27 ( which is @xmath29 ) divided by the length of the total elastic .
if we denote these probabilities respectively by @xmath30 and @xmath31 we have : @xmath32 these transition probabilities are the same as the ones related to the outcomes of a stern - gerlach spin measurement on a spin @xmath3 quantum particle , of which the quantum - spin - state in direction @xmath33 @xmath34 , denoted by @xmath35 , and the measurement @xmath12 corresponding to the spin measurement in direction @xmath36 @xmath37 , is described respectively by the vector and the self adjoint operator of a two - dimensional complex hilbert space .
@xmath38 we can easily see now the two aspects in this quantum machine that we have identified in the hidden measurement approach to give rise to the quantum structure .
the state of the particle @xmath5 is effectively changed by the measuring apparatus ( @xmath10 changes to @xmath39 or to @xmath40 under the influence of the measuring process ) , which identifies the first aspect , and there is a lack of knowledge on the interaction between the measuring apparatus and the particle , namely the lack of knowledge of were exactly the elastic will break , which identifies the second aspect .
we can also easily understand now what is meant by the term ` hidden measurements ' .
each time the elastic breaks in one specific point @xmath41 , we could identify the measurement process that is carried out afterwards as a hidden measurement @xmath42 .
the measurement @xmath12 is then a classical mixture of the collection of all measurement @xmath42 : namely @xmath12 consists of choosing at random one of the @xmath42 and performing this chosen @xmath42 .
first of all we remark that we have shown in our group in brussels that such a hidden measurement model can be built for any arbitrary quantum entity ( aerts 1985 , 1986 , 1987 , coecke 1995a , b , c ) . however , the hidden measurement formalism is more general than standard quantum theory .
indeed , it is very easy to produce quantum - like structures that can not be represented in a complex hilbert space ( aerts 1986 ) . if the quantum structure can be explained by the presence of a lack of knowledge on the measurement process , we can go a step further , and wonder what types of structure arise when we consider the original models , with a lack of knowledge on the measurement process , and introduce a variation of the magnitude of this lack of knowledge .
we have studied the quantum machine under varying ` lack of knowledge ' , parameterizing this variation by a number @xmath43 $ ] , such that @xmath44 corresponds to the situation of maximal lack of knowledge , giving rise to a quantum structure , and @xmath45 corresponds to the situation of zero lack of knowledge , generating a classical structure .
other values of @xmath0 correspond to intermediate situations and give rise to a structure that is neither quantum nor classical ( aerts durt van bogaert 1992 , 1993 and aerts durt 1994 a , b ) .
we have called this model the @xmath0-model and want to introduce it again in this paper to explain in which way some of the quantum paradoxes are solved in the hidden measurement formalism . _
a ) the @xmath0-model _
we start from the quantum machine , but introduce now different types of elastic .
an @xmath46-elastic consists of three different parts : one lower part where it is unbreakable , a middle part where it breaks uniformly , and an upper part where it is again unbreakable . by means of the two parameters
@xmath47 $ ] and @xmath48 $ ] , we fix the sizes of the three parts in the following way .
suppose that we have installed the @xmath46-elastic between the points @xmath13 and @xmath14 of the sphere .
then the elastic is unbreakable in the lower part from @xmath49 to @xmath50 , it breaks uniformly in the part from @xmath51 to @xmath52 , and it is again unbreakable in the upper part from @xmath52 to @xmath14 ( see fig .
0.7 cm 3 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 12 pt fig .
2 : a representation of the measurement @xmath53 .
the elastic breaks uniformly between the points @xmath54 and @xmath55 , and is unbreakable in other points . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ an @xmath12 measurement performed by means of an @xmath46-elastic shall be denoted by @xmath56 .
we have the following cases : _ ( 1 ) _ @xmath57 .
the particle sticks to the lower part of the @xmath58-elastic , and any breaking of the elastic pulls it down to the point @xmath13 .
we have @xmath59 and @xmath60 . _ ( 2 ) _ @xmath61 .
the particle falls onto the breakable part of the @xmath46-elastic .
we can easily calculate the transition probabilities and find : @xmath62 _ ( 3 ) _
the particle falls onto the upper part of the @xmath64-elastic , and any breaking of the elastic pulls it upwards , such that it arrives in @xmath14 .
we have @xmath65 and @xmath66 .
we are now in a very interesting situation from the point of view of the structural studies of quantum mechanics .
since the @xmath0-model describes a continuous transition from quantum to classical , its mathematical structure should be able to learn us _
what are the structural shortcomings of the standard hilbert space quantum mechanics_. therefore we have studied the @xmath0-model in the existing mathematical approaches that are more general than the standard quantum mechanics : the lattice approach , the probabilistic approach and the * -algebra approach . _
b ) the lattice approach _ in this lattice approach exists a well known axiomatic scheme that reduces the approach to standard quantum mechanics if certain axioms are fulfilled . for intermediate values of @xmath0 ,
that is @xmath67 , we find that 2 of the 5 axioms needed in the lattice approach to reconstruct standard quantum mechanics are violated .
the axioms that are violated are the weak - modularity and the covering law , and it are precisely those axioms that are needed to recover the vector space structure of the state space in quantum mechanics ( aerts durt van bogaert 1993 and aerts durt 1994 a , b ) . _ c ) the probabilistic approach _ if we take the case of vanishing fluctuations ( @xmath45 ) , do we obtain the kolmogorovian theory of probability ?
this would be most interesting , since then we would have constructed a macroscopic model with an understandable structure ( i.e. , we can see how the probabilities arise ) and a quantum and a classical behavior . we ( aerts 1995 ) proposed to test the polytopes for a family of conditional probabilities .
the calculations can be found in ( aerts s. 1995 ) and the result was what we hoped for : a macroscopic model with a quantum and a kolmogorovian limit . for intermediate values of the fluctuations ( @xmath67 )
the resulting probability model is neither quantum nor kolmogorovian : we have identified here a new type of probability model , that is quantum - like , but not really isomorphic to the probability model found in a complex hilbert space . _
d ) the * -algebra approach _ the * -algebra provides a natural mathematical language for quantum mechanical operators .
we applied the concepts of this approach to the epsilon model to find that an operator corresponding to an @xmath0 measurement is linear if and only if @xmath44 ( aerts dhooghe 1996 ) .
this means that for the classical and intermediate situations the observables can not be described by linear operators .
* conclusion : * _ quantum theory and classical theories appear as special cases ( @xmath68 ) and the general intermediate case , although quantum - like , can not be described in standard hilbert space quantum mechanics .
_ _ e ) the measurement problem and the schrdingers cat paradox _
the result stated in the conclusion means that all the paradoxes of standard quantum theory that are due to the fact that quantum theory is used as a universal theory , also being applied to macroscopic system , for example the measuring apparatus , are not present in our hidden measurement formalism .
we explicitly have in mind the ` measurement problem ' and the ` schrdinger cat paradox ' .
indeed , the measurement apparatus should be described by a classical model in our approach , and the physical system eventually by a quantum model .
the problem of the presence of quantum correlations between physical system and measuring apparatus , as it presents itself in the standard theory , takes a completely different aspect .
we are working now at the elaboration of a concrete description of the measurement process within the hidden measurement formalism ( aerts durt 1994b ) .
our result also shows that it is possible in the hidden measurement formalism to formulate a ` classical limit ' , namely as a continuous transition from quantum to classical ( aerts durt van bogaert 1993 ) .
the measurement problem and paradoxes equivalent to schrdingers cat paradox disappear in the hidden measurement formalism , because standard quantum mechanics appears only as a special case , the situation of maximum fluctuations .
all quantum mysteries connected to the effect of ` non - locality ' remain ( einstein podolsky rosen paradox and the violation of the bell inequalities ) .
it is even so that non - locality unfolds itself as a fundamental aspect of the hidden measurement approach .
this is due to the fact that if we explain the quantum structures as it is done in the hidden measurement formalism a quantum measurement has two concrete physical effects : ( 1 ) it changes the state of the system and ( 2 ) it produces probability due to a lack of knowledge about the nature of this change . with the quantum machine
we have given a macroscopic model for a spin measurement of a spin @xmath69 particle .
if we apply the hidden measurement formalism to the situation of a quantum system described by a wave function @xmath70 , and to a position or a momentum measurement performed on this system , we also have the two mentioned effects .
for example in the case of a position measurement : the detection apparatus changes the state of the quantum system , in the sense that it ` localizes ' the quantum system in a specific place of space , and the probabilities that are connected with this measurement are due to the fact that we lack knowledge about the specific way in which this localization takes place .
this means that ` before the detection has taken place ' the quantum system was in general not localized : it was not present in a specific region of space . with other words ,
quantum systems are fundamentally non - local systems : the wave function @xmath70 describing such a quantum state is not interpreted as a wave that is present in space , but as indicating these regions of space were the particle can be localized , where @xmath71 is the probability that this localization will happen in region @xmath72 .
a similar interpretation must be given to the momentum measurement of a quantum entity : the quantum entity has no momentum before the measurement , but the measurement creates partly this momentum . in ( aerts durt van bogaert 1993 )
we have calculated the @xmath0-situation for a quantum system described by a wave function @xmath70 , element of the hilbert space of all square integrable complex functions , and we have found the following very simple procedure
. 0.7 cm 3 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 12 pt fig .
3 : a graphical representation of the @xmath73 situation for a quantum system described by a wave function @xmath74 . suppose that @xmath75 $ ] is given and @xmath76 is the probability distribution corresponding to @xmath77 .
we cut , by means of a constant function @xmath78 , a piece of the function @xmath79 , such that the surface contained in the cutoff piece equals @xmath73 ( step 1 of fig 3 ) .
we move this piece of function to the @xmath80-axis ( step 2 of fig 3 ) , and then renormalize by dividing by @xmath73 ( step 3 of fig 3 ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ suppose that @xmath0 is given , and the state of the quantum system is described by the wave function @xmath70 and @xmath81 is the corresponding probability distribution ( hence @xmath82 ) .
we cut , by means of a constant function @xmath83 , a piece of the function @xmath81 , such that the surface contained in the cutoff piece equals @xmath0 ( see step 1 of fig 3 ) .
we move this piece of function to the @xmath84-axis ( see step 2 of fig 3 ) .
and then we renormalize by dividing by @xmath0 ( see step 3 of fig 3 ) .
if we proceed in this way for smaller values of @xmath0 , we shall finally arrive at a delta - function for the classical limit @xmath85 , and the delta - function is located in the original maximum of the quantum probability distribution .
we want to point out that the state @xmath70 of the physical system is not changed by this @xmath0-procedure , it remains always the same state , representing the same physical reality .
it is the regime of lack of knowledge going together with the detection measurement that changes with varying @xmath0 . for @xmath86
this regime is one of maximum lack of knowledge on the process of localization , and this lack of knowledge is characterized by the spread of the probability distribution @xmath81 . for an intermediate value of @xmath0 , between 1 and 0 , the spread of the probability distribution has decreased ( see fig3 ) and for zero fluctuations the spread is 0 .
let us also try to see what becomes of the non - local behavior of quantum entities taking into account the classical limit procedure that we propose .
suppose that we consider a double slit experiment , then the state @xmath87 of a quantum entity having passed the slits can be represented by a probability function @xmath88 of the form represented in fig 4 .
we can see that the non - locality presented by this probability function gradually disappears when @xmath0 becomes smaller , and in the case where @xmath88 has only one maximum finally disappears completely .
0.7 cm 3 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 12 pt fig .
4 : a representation of the probability distribution corresponding to the state of a quantum system that has passed a double slit .
we show the three different steps of the @xmath89 procedure ( fig 3 ) in this case . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ when there are no fluctuations on the measuring apparatus used to detect the particle , it shall be detected with certainty in one of the slits , and always in the same one .
if @xmath88 has two maxima ( one behind slit 1 , and the other behind slit 2 ) that are equal , the non - locality does not disappear .
indeed , in this case the limit - function is the sum of two delta - functions ( one behind slit 1 and one behind slit 2 ) .
so in this case the non - locality remains present even in the classical limit .
if our procedure for the classical limit is a correct one , also macroscopic classical entities can be in non - local states .
how does it come that we do nt find any sign of this non - locality in the classical macroscopic world ?
this is due to the fact that the set of states , representing a situation where the probability function has more than one maximum , has measure zero , compared to the set of all possible states , and moreover these states are ` unstable ' .
the slightest perturbation will destroy the symmetry of the different maxima , and hence shall give rise to one point of localization in the classical limit .
also classical macroscopic reality is non - local , but the local model that we use to describe it gives the same statistical results , and hence can not be distinguished from the non - local model .
it is interesting to consider the violation of bell s inequalities within the hidden - measurement formalism .
the quantum machine , as presented in section 3 , delivers us a macroscopic model for the spin of a spin @xmath3 quantum entity and starting with this model it is possible to construct a macroscopic situation , using two of these models coupled by a rigid rod , that represents faithfully the situations of two entangled quantum particles of spin @xmath3 ( aerts 1991b ) .
the ` non - local ' element is introduced explicitly by means of a rod that connects the two sphere - models .
we also have studied this epr situation of entangled quantum systems by introducing the @xmath0-variation of the amount of lack of knowledge on the measurement processes and could show that one violates the bell - inequalities even more for classical but non - locally connected systems , that is , @xmath90 .
this illustrates that the violation of the bell - inequalities is due to the non - locality rather then to the indeterministic character of quantum theory . and that the quantum indeterminism ( for values of @xmath0 greater than 0 ) tempers the violation of the bell inequalities ( aerts d. , aerts s. , coecke and valckenborgh 1996 , aerts , coecke , durt and valckenborgh , 1997a , b )
this idea has been used to construct a general representation of entangled states ( hidden correlations ) within the hidden measurement formalism ( coecke 1996 , 1998 and coecke , dhooghe and valckenborgh 1997 ) .
accardi , l. , 1982 , _ nuovo cimento _ , * 34 * , 161 .
aerts , d. , 1985 , a possible explanation for the probabilities of quantum mechanics and a macroscopic situation that violates bell inequalities " , in _ recent developments in quantum logic _ , eds .
mittelstaedt , p. et al .
, in _ grundlagen der exacten naturwissenschaften _ , vol.6 , wissenschaftverlag , bibliographisches institut , mannheim , 235 .
aerts , d. , 1987 , the origin of the non - classical character of the quantum probability model " , in _ information , complexity , and control in quantum physics _ , eds .
blanquiere , a. et al . , springer - verlag .
aerts , d. , 1988a , the physical origin of the epr paradox and how to violate bell inequalities by macroscopic systems " , in _
symposium on the foundations of modern physics _ ,
lahti , p. et al .
, world scientific , singapore .
aerts , d. , 1988b , the description of separated systems and a possible explanation for the probabilities of quantum mechanics " , in _ microphysical reality and quantum formalism _ , eds . van der merwe et al . , kluwer academic publishers .
aerts , d. , 1991a , a macroscopic classical laboratory situation with only macroscopic classical entities giving rise to a quantum mechanical probability model " , in _ quantum probability and related topics , vol .
vi , _ ed .
accardi , l. , world scientific , singapore .
aerts , d. , 1991b , a mechanistic classical laboratory situation violating the bell inequalities with 2@xmath91 , exactly ` in the same way ' as its violations by the epr experiments " , _ helv .
acta _ , * 64 * , 1 .
aerts , d. , aerts , s. , coecke , b. and valckenborgh , f. , 1996 , the meaning of the violation of bell inequalities : non - local correlation or quantum behavior ? " , preprint , clea , brussels free university , krijgskundestraat 33 , brussels .
aerts , d. and durt , t. , 1994b , quantum , classical and intermediate : a measurement model " , in _ 70 years of matter - wave _ , eds .
laurikainen , k.v .
, montonen , c. and sunnarborg , k. , editions frontieres , gives sur yvettes , france .
aerts , d. , durt , t. and van bogaert , b. , 1993 , quantum probability , the classical limit and non - locality " , in _ on the foundations of modern physics 1993 _ , ed .
hyvonen , t. , world scientific , singapore , 35 .
coecke , b. , dhooghe , b. and valckenborgh , f. , 1997 , classical physical entities with a quantum description " , in _ fundamental problems in quantum physics ii _ ,
ferrero , m. and van der merwe , a. , kluwer academic , dordrecht , 103 . | in the hidden measurement formalism that we develop in brussels we explain the quantum structure as due to the presence of two effects , ( a ) a real change of state of the system under influence of the measurement and , ( b ) a lack of knowledge about a deeper deterministic reality of the measurement process .
we show that the presence of these two effects leads to the major part of the quantum mechanical structure of a theory describing a physical system where the measurements to test the properties of this physical system contain the two mentioned effects .
we present a quantum machine , where we can illustrate in a simple way how the quantum structure arises as a consequence of the two effects .
we introduce a parameter @xmath0 that measures the amount of the lack of knowledge on the measurement process , and by varying this parameter , we describe a continuous evolution from a quantum structure ( maximal lack of knowledge ) to a classical structure ( zero lack of knowledge ) .
we show that for intermediate values of @xmath0 we find a new type of structure that is neither quantum nor classical .
we analyze the quantum paradoxes in the light of these findings and show that they can be divided into two groups : ( 1 ) the group ( measurement problem and schrdingers cat paradox ) where the paradoxical aspects arise mainly from the application of standard quantum theory as a general theory ( e.g. also describing the measurement apparatus ) .
this type of paradox disappears in the hidden measurement formalism .
( 2 ) a second group collecting the paradoxes connected to the effect of non - locality ( the einstein - podolsky - rosen paradox and the violation of bell inequalities ) .
we show that these paradoxes are internally resolved because the effect of non - locality turns out to be a fundamental property of the hidden measurement formalism itself .
= 13 pt fund and clea , brussels free university , krijgskundestraat 33 , 1160 brussels , belgium , e - mail : [email protected] |
emission , the 1216 ( n = 2@xmath131 ) transition of hydrogen in emission , is a prominent tracer of ionizing photons produced by young stars
. this line can carry up to @xmath56% of the bolometric luminosity of a star forming galaxy ( partridge & peebles 1967 ) , therefore it is an easy handle for detection of both star forming galaxies and active galactic nuclei ( agn , where emission is powered by the ionizing photons produced at the accretion disk around the central massive black hole ) at redshifts @xmath14 2 . the line - search technique has been used successfully to identify samples of high redshift galaxies for over a decade , using narrowband images ( e.g. , cowie & hu 1998 ; rhoads et al .
2000 , 2003 ; malhotra & rhoads 2002 , gawiser et al . 2007 ; finkelstein et al . 2008 , 2009 ; guaita et al .
2010 ) and spectroscopic surveys ( e.g. , pirzkal et al . 2004 , deharveng et al .
2008 , martin et al .
2008 , rhoads et al .
2009 , and blanc et al .
there are thousands of photometrically selected emitters ( hereafter laes ) , with hundreds of spectroscopic confirmations ( e.g. , hu et al . 2004 , dawson et al .
2007 , wang et al . 2009 ) at redshifts ranging from @xmath15 0.3 ( deharveng et al .
2008 ) to @xmath15 7 ( iye et al .
2006 ) .
since line searches have achieved notable success in identifying high redshift star - forming galaxies , it is very important to understand the radiation transfer of the emission line .
however , interpreting the line is not trivial , because photons are resonantly scattered when they interact with the surrounding neutral hydrogen in the inter - stellar medium ( ism ) .
these radiative transfer effects can be quite complex when considering the presence of dust in a multiphase and moving interstellar medium ( neufeld 1991 , hansen & oh 2006 ; finkelstein et al .
2009c ) , and require empirical study of the fraction of photons of which can escape the ism .
_ indeed , empirical results on the escape fraction offer one of the better tools for probing the physics of escape , which is determined on spatial scales that are usually unresolved in distant or nearby galaxies .
_ on the theory side , semi - analytical models of lae populations by le delliou et al .
( 2005 ) assume a constant escape fraction of f@xmath16 = 0.02 , and hydrodynamical models of laes at z @xmath5 3 predict f@xmath16 = 0.05 - 0.1 ( nagamine et al . 2010 , shimizu et al .
2011 ) . on the observational side ,
to measure the fraction of photons escaping the ism ( hereafter the escape fraction ) , we often require a flux measurement in ultraviolet ( uv ) continuum or a non - resonant recombination line , such as h@xmath8 , and a dust extinction value .
the dust corrected uv luminosity or h@xmath8 luminosity , like the intrinsic luminosity , are mainly contributed by young stars , thus connected to the intrinsic star formation rate ( sfr ) .
recent constraints at z @xmath5 0.3 ( atek et al . 2009 ) and at z @xmath52.25 ( hayes et al . 2010 ) are based on the line ratio between h@xmath8 and . with the optical spectroscopic observation of galex selected laes , atek et al .
( 2009 ) showed that the escape fractions span a wide range ( from f@xmath16 @xmath5 0.5 to 100 % ) and decrease with increasing dust extinction .
hayes et al .
( 2010 ) estimated an average escape fraction of @xmath5 5% with a double blind survey targeting and h@xmath8 at z = 2.25 . with
the dust - corrected uv emission of 98 laes at z @xmath17 1.9 - 3.8 detected through integral - field spectroscopy survey , blanc et al .
( 2010 ) got a median escape fraction of 29% .
however , the escape fractions estimated from both of these methods are sensitive to the dust extinction law .
for example , under the case b recombination , if we ignore the resonant scattering and velocity field in the gas or ism geometry , the escape fraction is presented as f@xmath16 = 10@xmath18 ( atek et al .
2009 ) , and the extinction coefficient at wavelength goes from @xmath19 @xmath5 9.9 to 12.8 for cardelli et al . ( 1989 ) and calzetti et al .
( 2000 ) laws , respectively .
this will lead to a factor of two difference in the inferred f@xmath16 when e(b - v ) @xmath20 0.3 .
meanwhile , the escape fractions estimated here also depend on the selection methods of laes . at z @xmath5 0.3 ,
the laes are selected from a spectroscopic followup of fuv dropouts ( atek et al .
2009 ) , so the escape fractions here refer to a subsample of dropout galaxies at that redshift . at z @xmath5 2.25 ,
the average escape fraction is estimated from 18 laes and 55 h@xmath8 emitters , and only 6 have both and h@xmath8 detection ( hayes et al .
so the escape fraction at z@xmath5 2.25 estimated by hayes et al .
( 2010 ) is a global escape fraction for all galaxies at that redshift .
the narrowband surveys for high - z laes can be designed to use the same criteria , e.g. , the threshold equivalent width of ew@xmath21 @xmath20 20 and the line flux limit estimated from narrowband and broadband images , however , we should note that the escape fraction is dependent on the choice of the threshold equivalent width , and the narrowband exposure depth .
shallower narrowband surveys tend to select bright laes , which may have smaller ews ( e.g. , ando et al .
2006 ) , and a decrease in the value of threshold equivalent width will contain more objects with little - to - no fluxes as well as interlopers .
x - ray photons , especially hard x - ray photons , can give an extinction - free measurement of star formation in galaxies .
first , the x - ray emission of high - redshift star - forming galaxies are mainly from supernovae ( sne ) , hot interstellar gas ( i.e. , @xmath22 k ) , and high - mass x - ray binaries ( hmxbs ) , all of which trace star formation ongoing in the last few million years .
nandra et al .
2002 , grimm et al .
2003 , ranalli et al .
2003 , and gilfanov , grimm & sunyaev ( 2004 ) have shown a linear correlation between l@xmath0 and sfr .
calibration with the total sfr from far - infrared ( fir ) and ultraviolet ( uv ) bands for local and high - z ( z @xmath5 1 and z @xmath5 3 ) star - forming galaxies ( in the chandra deep field north and south , cdf - n and cdf - s , where the deepest x - ray observations exist ) show relations consistent with ranalli et al .
( 2003 ) @xmath9 5 m@xmath23 and salpeter s initial mass function ( imf ) ) following condon ( 1992 , radio luminosities ) and kennicutt ( 1998 , fir luminosities ) .
] : @xmath24 second , x - rays penetrate a typically dusty interstellar medium much more easily than uv light . for example
, a column density of @xmath25@xmath26 is required to attenuate 2 kev x - rays by a factor of 2 .
the corresponding reddening would be @xmath27 mag , corresponding to @xmath28 mag and @xmath29 mag , or a factor of @xmath30 , based on ratios of dust to gas from either our galaxy ( @xmath31 mag @xmath32 atoms@xmath4 ) or from a set of three gravitationally lensed galaxies at high redshift ( @xmath33 = ( 1.4 @xmath34 0.5 ) @xmath35 10@xmath36 mag @xmath32 atoms@xmath4 ; dai et al 2006 ) . in this paper , we present an independent analysis of the escape fraction at @xmath12 @xmath5 0.3 , @xmath12 @xmath5 1 , and 2 @xmath2 @xmath12 @xmath2 6.5 , using x - ray emission as a tracer of the intrinsic sfr .
we use the new 4 msec _
chandra _ x - ray image in the cdf - s , which include laes at z @xmath5 0.3 and 1.0 selected by galex , and ground - based narrowband selected laes at z = 2.1 , 3.1 , 3.15 , 4.5 , 5.7 , and 6.5 . by comparing the ratio of derived sfrs from observed line flux and x - ray , we will perform an independent measurement of the escape fraction for laes .
the optical samples and x - ray data are presented in 2 , and the x - ray detection and stacking results on laes are presented in 3 .
the results and discussion on x - ray constrained escape fraction are presented in 4 and 5 , respectively .
in the chandra deep field south region , samples of emitters have been observed at various redshifts , including emitters at z = 0.195 - 0.44 and z = 0.65 - 1.25 through galex grism spectroscopy ( deharveng et al .
2008 , cowie et al .
2010 , cowie et al . 2011 ) , and ground - based narrowband imaging selected emitters at z = 2.1 ( guaita et al .
2010 ) , z = 3.1 ( gronwall et al .
2007 , ciardullo et al .
2011 in prep . ) , z = 3.15 ( nilsson et al .
2007 ) , z = 4.5 ( finkelstein et al .
2009 ) , z = 5.7 ( wang et al 2005 ) and z = 6.5 ( rhoads et al in prep . ) .
the galex laes were confirmed by ground based optical spectroscopy ( cowie et al .
2010 , 2011 ) . at higher redshift
( z @xmath9 2 ) , all the narrowband selected samples typically have a spectroscopic confirmation fraction greater than 70% ( dawson et al .
2007 , gawiser et al .
2007 , wang et al . 2009 ) . in this paper
we focus on laes covered in the cdf - s proper region .
the number of laes at each redshift and their stacked x - ray properties are presented in tables 1 and 2 .
the 4 ms _ chandra _ advanced ccd imaging spectrometer ( acis ) exposure of the cdf - s is composed of 52 individual acis - i observations , 9 of which were obtained in 2000 , 12 from september to november 2007 , and 31 from march to july 2010 .
the event-2 file and exposure - map files are available at the _ chandra _ website ( http://cxc.harvard.edu/cdo/cdfs.html ) .
to search for potential x - ray counterparts for our laes , we use the 4-ms cdf - s catalog published by xue et al .
( 2011 ) for the following source cross - match and source - masking processes .
we search for x - ray counterparts for individual laes within a 2@xmath37 radius . among laes selected in the 4 msec cdf - s region using the galex fuv channel ( which covers at @xmath38 ) , 3 of 6 ( found at @xmath39 )
are detected in x - rays .
these three have relatively modest luminosities l@xmath40 @xmath2 10@xmath41 ergs s@xmath4 , and have been classified as galaxies from their optical spectra ( cowie et al .
2010 , 2011 ) .
their uv and x - ray properties are presented in table [ z03lae ] .
note that all three detected galaxies show a soft - band x - ray detection , while their hard - band luminosities are upper limits , but remain consistent with their soft band luminosities under the @xmath42 = 2 assumption . for galaxies selected in the galex nuv channel ( which covers at @xmath43 ) ,
the corresponding number is 4 x - ray detections out of 5 objects in the 4 msec chandra coverage .
these four objects have x - ray luminosities l@xmath40 @xmath9 6@xmath3510@xmath44 ergs s@xmath4 , and can be classified as agns .
x - ray counterparts have previously been detected for a small fraction ( @xmath2 5% ) of the laes at higher redshifts .
guaita et al .
( 2010 ) detected x - rays from 10 of their 216 @xmath12 = 2.1 laes in the ecdf - s region , using the 2msec _ chandra _ image for those laes in the central cdf - s ( luo et al .
2008 ) , and the 250 ksec _
_ image ( lehmer et al . 2005 ) for those in the wider ecdfs .
gronwall et al . ( 2007 ) and ciardullo et al ( in prep . )
selected 278 lae candidates at z = 3.1 , and only 5 were x - ray detected . at z = 4.5 ,
only 1 of 113 lae candidates show x - ray detection in ecdf - s field ( zheng et al .
2010 ) and is classified as an agn at the same redshift , and no x - ray detections were found for 24 lae candidates at z = 3.15 ( nilsson et al .
2007 ) , 25 lae candidates at z = 5.7 ( wang et al . 2005 ) , or 11 lae candidates at z = 6.5 ( rhoads et al . in prep . ) . in this paper , these previous x - ray detections are excluded , and we find no new x - ray detection with the 4 ms chandra exposure .
we calculate the x - ray signal - to - noise ratio for the x - ray non - detected laes , all of which show s
/ n @xmath2 3 .
the s / n ratios were calculated as @xmath45 ( gehrels 1986 ) , where @xmath46 and @xmath47 are the net counts and total counts extracted from their 50% chandra acis psf circles ) were estimated from an annulus with 1.2r@xmath48 @xmath2 r @xmath2 2.4r@xmath48 ( after masking out nearby x - ray sources ) . ] in 0.5 - 2 , 2 - 7 and 0.5 - 7 kev band respectively . to convert from x - ray counts to fluxes ,
we have assumed a powerlaw spectrum with photon index of @xmath42 = 2 ( except where explicitly stated otherwise ) , which generally represents the x - ray spectra of both starburst galaxies ( e.g. , lehmer et al .
2008 ) and type 1 agns .
when converted from psf - corrected count - rate to flux , the full and hard bands were extrapolated to the standard upper limit of 10 kev .
all x - ray fluxes and luminosities presented in this paper have been corrected for galactic absorption ( dickey & lockman 1990 ) .
the 4 ms cdf - s data reaches on - axis sensitivity limits of x - ray luminosity lg@xmath49 @xmath17 [ 39.6 , 40.9 , 41.5 , 41.9 , 41.9 , 42.3 , 42.5 , 42.6 ] at z = [ 0.3 , 1.0 , 2.1 , 3.1 , 3.2 , 4.5 , 5.7 , 6.5 ] ( with the on - axis soft band flux limit of @xmath50 @xmath17 9.1 @xmath35 10@xmath51 erg @xmath26 s@xmath4 , xue et al . 2011 ) , which means that we may find all the agns ( including type 2 agns ) hidden in the [email protected] and z@xmath51 laes located in the cdf - s region .
we stacked the x - ray data at the positions of the non - detections at each redshift , following zheng et al .
( 2010 ) . after masking out
the x - ray detected sources with 95% psf circles , we co - added the net and background counts of the x - ray non - detected laes in each sample .
the stacked net and background counts in the soft , hard and 0.5 - 7 kev band , as well as the summed effective chandra exposure time are presented in table [ stkdata ] .
we used the 95% psf region for bright source extraction , and 50% psf region for upper limit estimate .
the results are about 0.2 dex lower than upper limit from 95% psf extraction .
in figure [ sntheta ] we plot the 1 @xmath1 upper limits ( upper plot ) on the mean x - ray luminosity for each studied redshift in the selection region ( parameterized by the maximum off - axis angle @xmath52 for inclusion in the sample ) , and the s / n ratios of the stacked signals are also plotted in the lower part . from figure [ sntheta ]
we can see than the mean luminosity can be better constrained by excluding laes with larger off - axis angles .
this is mainly because the chandra acis has much larger psf and much lower collection area at larger off - axis angles , so that including those laes with large off - axis angles would bring strong fluctuations to the signal without necessarily increasing the s / n . in the following study
, we exclude those laes with acis off - axis angle above 6@xmath53 , and the number of laes used for stacking and their stacked results at relative redshifts are presented in table [ stkdata ] .
we adopted the relations established at z = 0 to convert luminosities to sfrs for our lae samples : kennicutt ( 1998 ) for conversion of the luminosity ( assuming case - b conditions and salpeter s imf with mass limits 0.1 and 100 @xmath54 ) , and ranalli et al .
( 2003 ) for the x - ray luminosities : @xmath55 note that these relations might give uncertainties as well as the factor of @xmath52 dispersion ( e.g. , dust extinction , burst age and imf ) when applied to individual sources
. however , previous x - ray stacking works ( e.g. , nandra et al . 2002 , seibert et al . 2002 , lehmer et al . 2005 , and reddy & steidel 2004 ) have proved reasonable l@xmath0-sfr@xmath0 connection on local starburst galaxies and 1 @xmath56 z @xmath56 3 lbgs .
table [ z03lae ] shows the sfr estimates based on the ( `` sfr@xmath6 '' ) luminosity and 2 - 10 kev ( `` sfr@xmath0 '' ) luminosity for three [email protected] galex selected laes .
table [ stkdata ] shows the sfr@xmath0 estimates from the stacking signal of the x - ray undetected laes at z @xmath5 0.3 , 1.0 , 2.1 , 3.1 , 3.2 , 4.5 , 5.7 and 6.5 .
it can be seen that at z @xmath20 4.5 , the sfr@xmath0 limits are weaker due to a larger luminosity distance , and comparatively smaller sample size . note that low - mass x - ray binaries ( lmxbs ) can also contribute to x - ray emission from galaxies , but they have longer evolutionary time scales ( on the order of the hubble time ) , and therefore track the integrated star - formation history of galaxies ( i.e. , the total stellar mass ) . where lmxbs are important , e.g. for nearby normal galaxies ( colbert et al .
2004 ) and luminous infrared galaxies ( lehmer et al .
2010 ) , the x - ray luminosity can be fitted as a function of both galaxy galaxy stellar mass and sfr : @xmath57 here lx , m@xmath58 , and sfr@xmath0 have units of ergs s@xmath4 , m@xmath3 , and m@xmath3 yr@xmath4 , respectively , and constants @xmath59 ergs s@xmath4 m@xmath60 = 1.51 and 0.91 , and @xmath61 ergs s@xmath4 ( m@xmath3 yr@xmath4)@xmath62 = 0.81 and 1.62 for colbert et al .
( 2004 ) and lehmer et al .
( 2010 ) , respectively .
the differences in the parameters might be introduced by the high obscuration in luminous infrared galaxies .
we can see from these relations , the contribution of x - ray emission from lmxbs can be ignored in actively star - forming galaxies , which often have low dust extinction ( 0@xmath63 e(b - v ) @xmath63 3 ) , lower stellar mass ( m@xmath58 @xmath63 10@xmath64 m@xmath3 ) , and more active star formation ( sfr@xmath0 @xmath65 1 m@xmath3 yr@xmath4 ) than local galaxies ( e.g. , guaita et al .
2011 ) . in terms of the specific star formation rate ( ssfr ) , which is defined as the ratio of star formation rate to stellar mass
, we can neglect the lmxb term whenever @xmath66 , so that @xmath67 , i.e. , whenever the typical stellar ages are much below 1 gyr .
the sfr@xmath0 of high redshift star - forming galaxies was measured by nandra et al .
( 2002 ) , who detected a stacked x - ray signal of @xmath15 3 lyman break galaxies ( lbgs ) in the hubble deep field north .
the average x - ray luminosity of @xmath15 3 lbgs is l@xmath68 = 3.4 @xmath35 10@xmath69 ergs s@xmath4 ( 6-@xmath1 significance ) , implying a sfr@xmath0 = 64@xmath3413 m@xmath23 @xmath70 , in excellent agreement with the extinction - corrected uv estimates . in the same field ,
laird et al . ( 2006 ) found the average sfr@xmath0 of [email protected] m@xmath3 yr@xmath4 for z @xmath5 3 lbgs .
additionally , lehmer et al .
( 2005 ) reported the average sfr@xmath0 of @xmath530 m@xmath3 yr@xmath4 for z @xmath5 3 lbgs in cdf - s ( old 1 ms _ chandra _ exposure ) .
lehmer et al .
also stacked lbgs in cdf - s at z@xmath54 , 5 , and 6 , and did not obtain significant detections ( @xmath23 @xmath1 ) , deriving rest - frame 2.0 - 8.0 kev luminosity upper limits ( 3 @xmath1 ) of 0.9 , 2.8 , and 7.1 @xmath35 10@xmath69 ergs s@xmath4 , corresponding to sfr@xmath0 upper limits of 18 , 56 and 142 m@xmath3 yr@xmath4 , respectively .
note also that a @xmath53 @xmath1 stacking signal of the optically bright subset ( brightest 25% ) of lbgs at z@xmath54 was detected , corresponding to an average sfr@xmath0 of @xmath528 m@xmath3 yr@xmath4 .
reddy & steidel ( 2004 ) examined the stacked radio and x - ray emission from uv - selected spectroscopically confirmed galaxies in the redshift range 1.5 @xmath71 3.0 .
their sample showed sfr@xmath0 @xmath5 50 m@xmath3 yr@xmath4 , and found a consistent sfr@xmath0/sfr@xmath72 @xmath5 4.5 - 5.0 for galaxies over the redshift range 1.5 @xmath71 3.0 .
these studies demonstrate the value of stacking the deepest x - ray observations to measure star formation activity , with little sensitivity to dust .
the laes x - ray emission should be directly connected to the intrinsic sfr , since laes are thought to be less massive and much younger than lbgs at high - redshift ( e.g. , venemans et al 2005 ; pirzkal et al 2007 ; finkelstein et al 2008 , 2009c ) , and the agn fraction in laes is also very small ( @xmath2 5% , e.g. , zheng et al .
an x - ray detection could give us a more accurate unbiased sfr estimate , or more properly an upper limit , since faint agn may contribute to the x - ray flux .
the first x - ray observations of high redshift laes were presented in malhotra et al .
( 2003 ) and wang et al .
( 2004 ) at [email protected] with two 170 ks _
chandra _ exposures .
no individual laes were detected , and a 3-@xmath1 upper limit on the x ray luminosity ( l@xmath73 @xmath2 2.8 @xmath35 10@xmath41 ergs s@xmath4 ) was derived by an x - ray stacking method ( wang et al .
2004 ) . from a stacking analysis of the non - detected laes in the much deeper 2 ms cdf - s field and a larger 250 ks ecdfs field , guaita et al .
( 2010 ) , gronwall et al .
( 2007 ) and zheng et al .
( 2010 ) found a smaller 3-@xmath1 upper limit on the luminosity of @xmath51.9 @xmath35 10@xmath69 ergs s@xmath4 , 3.1 @xmath35 10@xmath69 ergs s@xmath4 , 2.4 @xmath35 10@xmath41 ergs s@xmath4 at z = 2.1 , z = 3.1 and z = 4.5 , respectively .
these imply upper limits of unobscured average sfr@xmath0 @xmath2 43 , 70 and 290 m@xmath3 yr@xmath4 , respectively .
the above results are not surprising , since we would expect that laes have lower sfr rates than lbgs .
the average calculated using sfr@xmath6 and sfr@xmath0 for laes at different redshifts are presented in table [ xsfr ] .
the sample selection limits are also presented in table [ xsfr ] .
we should note that the laes at z@xmath5 0.3 and 1 are selected among the galex fuv and nuv band drop - out galaxies with @xmath75 @xmath9 20from their follow - up galex spectra .
so the lae samples at z @xmath63 1 are quite different compared to the lae samples at z@xmath9 2 , which are selected from narrowband excess over broadband with nearly same criteria on the equivalent width @xmath75 ( estimated from the broad band to narrowband ratio ) . in the following discussion , we treat them separately .
we only take the soft band upper limits for z@xmath9 2 laes , because soft band flux are more sensitive than the total band and hard band , and at z @xmath9 2 , the observed soft band x - ray photons are closer to rest - frame hard photons , and therefore more robust to a change in photon index @xmath76 ( see figure 2 of wang et al .
2007 ) assumed when converting x - ray count - rate to flux . at @xmath77 0.3 ,
we only have l@xmath78 for the three selected galaxies .
although they are not detected in the hard x - ray band , their 3 @xmath1 upper limit on l@xmath40 show good consistency with their l@xmath78 ( see table 1 ) . here
we use the sfr@xmath0-l@xmath78 relation from ranalli et al .
( 2003 ) , since our detections are in the soft band .
we find sfr@xmath0 @xmath5 1 , 11 , and 22 m@xmath23 yr@xmath4 , implying escape fractions of 22 , 3 , and 9% , respectively .
finkelstein et al . (
2009b ) also derived the dust extinction of a@xmath79 = 0.9@xmath52 from the sed fitting , corresponding to @xmath80 = sfr@xmath6/(sfr@xmath81 ) = 2 - 5% ( see table 3 ) , which is systematically lower than the from x - ray .
x - ray stacking of the undetected z @xmath5 0.3 laes shows a 3 @xmath1 upper limit of sfr@xmath82 @xmath5 3.2 m@xmath23 yr@xmath4 , implying an average escape fraction of @xmath928% at 3 @xmath1 significance level .
these x - ray constraints may tell us that at z@xmath5 0.3 , the escape fraction is substantially larger than we thought from the uv or h@xmath8 view .
there are only 2 star - formation dominated lae galaxies at z@xmath5 1 in the ecdfs region ( excluding the four agn with individual x - ray detections ) , so at z@xmath5 1 the escape fraction is poorly constrained ( with ( @xmath83 ) @xmath9 7% ) . at @xmath12 @xmath9 2 , we do not obtain an x - ray detection after stacking the 69 star - formation dominated laes in the central cdf - s field , or after stacking the 351 laes in the ecdfs field ( see table [ stkdata ] ) .
we get the 1-@xmath1 upper limits ( same as below ) on the intrinsic ( dust - free ) sfrs as sfr@xmath0 @xmath2 [ 14 , 28 , 28 , 139 , 440 , 876 ] @xmath84 yr@xmath4 for laes at z = [ 2.1 , 3.1 , 3.2 , 4.5 , 5.7 , 6.5 ] .
the observed sfrs from the line are about 1.9 , 2.6 and 2.3 m@xmath23 yr@xmath4 for z = 2.1 , 3.1 and 3.2 laes on average , and @xmath5 5 m@xmath23 yr@xmath4 for z @xmath9 4 laes .
the ratio sfr@xmath6
/ sfr@xmath0 measures the escape fraction : @xmath85= sfr@xmath6 / sfr@xmath0 ( see table 3 ) .
this is plotted in figure [ lyafrac ] , showing the constraints on as a function of redshift . at z = 2.1 ,
the escape fraction of laes is above 14% . at z = 3.1 and 3.2 , the f@xmath74 is greater than 8% and 9% , separately .
combining the two samples at z@xmath5 3 , we increase the upper limit to f@xmath74 @xmath9 14% , the same as the upper limit at z = 2.1 . at higher redshift ,
the limits are weaker due to a larger luminosity distance , and comparatively smaller sample sizes .
if the emitters do nt evolve from z = 3.2 to z = 2.1 , then we can combine the samples in this redshift range to obtain a more robust limit on the escape fraction .
a 1-@xmath1 upper limit on soft band flux is derived as f@xmath78 @xmath2 8.5@xmath35 10@xmath86 2s .
this implies an sfr@xmath87 14%@xmath528% ( due to the different sfr@xmath6 average value ) during the redshift range 2.1 @xmath88 @xmath12 @xmath88 3.2 .
this value is consistent with the median value of @xmath89 @xmath5 29% from blank fields ifu spectroscopically - selected laes at 1.9 @xmath90 3.8 ( blanc et al . 2011 ) , but larger than the value of @xmath91 2 , 5 - 10% in some theoretical models of laes ( e.g. , le delliou et al .
2005 , shimizu et al .
2011 ) .
there is no x - ray detection for emitters coadded at any individual redshift bin ( z @xmath9 2 ) . by coadding the 53 emitters at 2.1 @xmath92 3.2
, we reach an x - ray flux limit of 8.5@xmath3510@xmath86 ergs @xmath26 s@xmath4 ( 1 @xmath1 ) , but still do not detect x - ray photons from these galaxies .
this means that the star - formation rate is truly low , as indicated by the line strength . to increase our sensitivity to detect the star formation rate of laes , which is relatively low when compared to continuum - selected star - forming galaxies at comparable redshifts , we coadd all the undetected emitters between redshifts 2 and 6.5 .
the average x - ray emission is still undetected at a 1-sigma flux level of 7.6@xmath3510@xmath86 ergs @xmath26 s@xmath4 . to convert the flux into luminosity and hence sfr , we need to model the redshift distribution of the sources .
let us , instead , predict the expected x - ray counts on this average image of galaxies .
assuming the escape fraction for laes is @xmath7 , then sfr@xmath0 = sfr@xmath6/(@xmath7)= ( l@xmath93 , we get @xmath94 if we assume that all laes have an effective x - ray photon index of @xmath42 = 2 , then with the x - ray count - rate to flux conversion of 6.64 @xmath35 10@xmath95 erg/@xmath32 at soft band and sfr@xmath6 from the table [ xsfr ] , we estimate that the expected number of x - ray photons at @xmath54 ms cdf - s for laes should be @xmath96)$ ] for laes at @xmath97 $ ] .
this means that we should observe @xmath5 3 soft x - ray photons per z = 2.1 lae galaxy in the 4 ms cdf - s field when @xmath7 @xmath5 5% , and the observed expected soft x - ray photons decrease to @xmath5 1 when @xmath98 increases to 15% .
since the background x - ray photons per point source on the acis - i ccds varies from position to position , we take the background value extracted from each lae candidate , and add the estimated x - ray photons from their corresponding sfr@xmath6 to analyse the probability on @xmath7 . by coadding all laes between redshift z = 2 and 6.5 ,
the estimated counts are plotted in figure [ estcts ] as a function of @xmath7 .
the signal should be s / n @xmath99 3 when @xmath7 @xmath2 5.7% , and s / n @xmath100 when @xmath7 @xmath2 17% .
so we can reject the value of 5% at 99.87% confidence level , and report that the real value of @xmath7 @xmath9 17% in 84% confidence level .
the escape fraction of laes has also been discussed using other tracers of the total sfr .
atek et al .
( 2009 ) studied z = 0.3 laes by using the extinction - corrected h@xmath8 to ratio , which has a ranges of @xmath101@xmath5 0.5% to 100% and a median value of @xmath5 20% .
blanc et al .
( 2011 ) studied [email protected] laes through extinction - corrected uv to ratio and found an average @xmath89 @xmath5 29% .
the f@xmath74 for laes estimated from the ratio of sfr@xmath6 and dust - corrected sfr@xmath103 , @xmath89 = sfr@xmath6/(sfr@xmath81 ) , also exist for our sample .
guaita et al .
( 2011 ) , gawiser et al .
( 2007 ) and nilsson et al .
( 2007 ) did the stacked sed fitting on the samples at z = 2.1 @xmath5 3.2 . at z@xmath5 0.3 and 4.5 , finkelstein et al .
( 2008 , 2010 ) did individual sed fitting for laes with existing _
hubble _ and _ spitzer _ observations .
the dust properties of the sed fitting results at different redshifts are converted to a@xmath79 ( see in table 3 ) using calzetti et al .
the @xmath89 from dust - corrected uv to ratio for our sample show no evolution from z@xmath52 to [email protected] , as @xmath89 @xmath5 26% , and is consistent with @xmath5 29% of blanc et al .
2011 . at z @xmath5 4.5 ,
the sed fitting results might be affected by the poor spatial resolution of _ spitzer_. as an independent estimate on from x - ray , our at z = 0.3 are located at the low end of atek et al .
( 2009 ) , while consistent at 1.9 @xmath90 3.8 with blanc et al .
2011 ( see figure [ lyafrac ] ) , and the @xmath89 from the sed fitting results at at z = 2.1 , 3.1 , 3.2 and 4.5 .
the laes at @xmath77 0.3 seems different compared to high - redshift laes , as they are more agn contaminated ( agn fraction of @xmath515%40% , e.g. , scarlata et al .
2009 ; cowie et al . 2010 ; finkelstein et al .
2009a ) and more massive ( finkelstein et al .
at high - redshift , the agn fraction in laes is very low ( agn fraction @xmath63 5% , zheng et al . 2010 and references there in ) , and agns are relatively easy to detect in this deepest _ chandra _ field , so the lower limit of f@xmath74 should be very robust .
our x - ray constraints on the @xmath7 , as well as the @xmath89 estimated from sed fitting , show that the escape fraction need not evolve during the redshift 2.1 @xmath56 @xmath12 @xmath56 3.2 .
hayes et al . ( 2011 ) and blanc et al .
( 2011 ) also reported the global evolution of f@xmath74 , which is defined as the ratio of integrated luminosity functions from laes and the global extinction - corrected sfr density at different redshift .
the global extinction - corrected sfr densities were integrated from h@xmath8 or uv luminosity .
however , we should point out that at z@xmath5 1 , 3 , 4 , and 5 , the sfr@xmath0/sfr@xmath72 @xmath5 4.5 - 5 for lbgs , consistent with the dust extinction values from other methods ( nandra et al . 2002 , lehmer et al .
2005 , reddy & steidel 2004 ) .
note that the luminosity functions of laes do not evolve during 3@xmath56 z @xmath56 6 ( e.g. , see ouchi et al .
2008 ) , while the relative numbers of laes to lbgs increased with redshift from z@xmath5 3 to 6.5 ( e.g. , clement et al .
2011 ) , the evolution trend of global f@xmath74 could be explained as the evolution of relative numbers of laes to lbgs at different redshift , with little implication on the escape mechanism for laes only .
x - ray constraints on the of laes are independent of dust - extinction law , and we have found the same lower limits of for laes at z@xmath5 0.3 and 1 , and laes at z = 2.1 and 3.2 with different average sfr@xmath6 values . however , a larger sample of laes with deep x - ray observation is needed to give a constraint on the evolution of in lae only .
we should note that the x - ray radiation from galaxies is predicted to be relatively low for the youngest stellar populations .
mas - hesse et al .
( 2008 ) predict the soft x - ray to far infrared luminosities ratio in star - forming galaxies from synthesis models .
they find that the ratio is dependent on the age of the star formation episode for ages @xmath2 myr .
after 30 myr , the correlation becomes stable and is consistent with sfr@xmath0
l@xmath104 relation of ranalli et al .
2003 . in the hard x - ray band
, mas - hesse & cervino ( 1999 ) predicted that a few hmxb should be active in starbursts that are older than 5 - 6 myr , contributing a few times 10@xmath105 erg s@xmath4 to the total x - ray luminosity .
recent _ chandra _ studies in the interacting galaxy pair ngc 4038/4039 ( the antennae ; rangelov et al . in prep ) found that 22 of 82 x - ray binaries are coincident or nearly coincident with star clusters .
the ages of these clusters were estimated by comparing their multi - band colors with predictions from stellar evolutionary models .
they found 14 of the 22 coincident sources are hosted by star clusters with ages of @xmath56 myr or less .
so the hmxbs in star - forming galaxies might form earlier than suggested in the mas - hesse & cervino ( 1999 ) model .
the estimates of lae ages are based on sed fitting and are quite uncertain .
such results have been reported at z = 2.1 ( guaita et al . 2011 ) , 2.25 ( nilsson et al . 2009 ) , 3.1 ( gawiser et al . 2007 ) , 4.5 ( finkelstein et al . 2008 ) and z=4.0 - 5.5 ( pirzkal et al . 2007 ) .
all the results show that the best fit age parameter is 20@xmath540 myr , but extends to @xmath50.1 - 1 gyr at the 68% confidence level . given this range of best fit ages
, we expect that a minority of the laes in our sample will be younger than 15 myr .
this would imply a modest upward correction to the sfr inferred from x - rays , and a corresponding correction downwards in the ; both corrections might reasonably be factors of @xmath106@xmath107 .
from the 4 ms x - ray chandra image of cdf - s , we find that the from x - rays for laes at z@xmath5 0.3 are about @xmath52 - 7 times larger than that from dust corrected uv or h@xmath8 .
we coadded @xmath108 emitting galaxies between redshifts @xmath109 .
none of these galaxies were individually detected .
the absence of signal in the coadded image implies an average flux of less than 7.6@xmath3510@xmath86 ergs @xmath26 s@xmath4 ( 1-sigma ) .
this implies that the sfrs in these galaxies are quite modest , as indicated by the line emission . and
the ratio of the average line intensity to the upper limits of x - ray flux constrain the escape fraction of @xmath9 17% at 84% confidence level .
+ + _ acknowledgements : _ we would like to thank the anonymous referee whose comments improved this manuscript significantly .
zyz would like to thank the support from the china scholarship council ( csc no .
2009634062 ) and arizona state university .
sm & jer are supported by the united states national science foundation grant ast-0808165 .
the work of jxw is supported by national basic research program of china ( 973 program , grant no .
2007cb815404 ) , and chinese national science foundation ( grant no .
10825312 ) . cccccccccc name & r.a . & dec .
& redshift & @xmath110l@xmath6 & @xmath111ew(ly@xmath8 ) & sfr@xmath112 & @xmath110l@xmath113 & @xmath110l@xmath68 ( 3@xmath1 ) & @xmath114sfr@xmath0 + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) & ( 10 ) + galex0332 - 2744 & 53.08000 & -27.74589 & 0.217 & 41.38 & 17 & 0.22 & 39.6 & @xmath240.4 & 1.0 + galex0333 - 2753 & 53.23600 & -27.88797 & 0.365 & 42.32 & 23 & 2.0 & 41.0 & @xmath241.3 & 21.7 + galex0333 - 2744 & 53.28029 & -27.74242 & 0.220 & 41.55 & 12 & 0.3 & 40.7 & @xmath241.1 & 10.9 + [ z03lae ] redshift & number@xmath115 & time & & & lg@xmath116(l@xmath0(@xmath117))@xmath118 & sfr@xmath119 + & & ms & tot@xmath120 & net@xmath120 & tot@xmath121 & net@xmath121 & tot@xmath122 & net@xmath122 & f@xmath104 & f@xmath123 & f@xmath124 & l@xmath40 & ( m@xmath23 yr@xmath4 ) + + 0.3 & 3 & 9.8 & 7 . & 1.3 & 13 . &
-0.8 & 20 . & 0.5 & 0.69 & 2.28 & 1.33 & 39.8 & 1.3 + 2.1 & 12 & 39.1 & 20 . & -4.3 & 71 . &
11.5 & 91 .
& 7.1 & 0.19 & 2.54 & 0.98 & 40.8 & 14 + 3.1 & 18 & 59.2 & 50 . & -4.9 & 120 . &
-8.2 & 170 . & -13.0 & 0.18 & 0.96 & 0.51 & 41.2 & 32 + 3.2 & 23 & 73.9 & 54 . & -9.4 & 145 . & -3.5 & 199 . & -12.9 & 0.15 & 0.84 & 0.44 & 41.1 & 28 + 4.5 & 12 & 39.3 & 30 . &
2.5 & 59 . &
-6.8 & 89 . & -4.3 & 0.31 & 1.05 & 0.58 & 41.8 & 139 + 5.7 & 3 & 9.8 & 12 . & -0.6 & 34 . &
& 5.4 & 0.62 & 6.19 & 2.90 & 42.3 & 439 + 6.5 & 1 & 3.3 & 3 . &
-3.6 & 16 . & 0.6 & 19 . & -2.9 & 1.17 & 8.16 & 3.53 & 42.7 & 1002 + + 0.3 & 12 & 11.6 & 19 & 6.4 & 31 & -0.1 & 50 & 6.3 & 1.3 & 2.7 & 2.7 & 39.9 & 1.6 + 1.0 & 2 & 0.4 & 1 & -1.2 & 9 & 3.4 & 10 & 2.2 & 7.3 & 83.9 & 32.9 & 41.6 & 84 + 2.1 & 122 & 72.2 & 49 . &
1.9 & 125 .
& 11.7 & 169 .
& 9.1 & 0.18 & 1.54 & 0.68 & 40.8 & 14 + 3.1 & 118 & 85.8 & 71 . & -6.2 & 169 .
& -3.4 & 234 . &
-12.4 & 0.14 & 0.76 & 0.40 & 41.1 & 28 + 3.2 & 24 & 74.9 & 55 . & -9.1 & 159 . &
5.9 & 216 . &
-2.7 & 0.15 & 1.24 & 0.45 & 41.1 & 28 + 4.5 & 64 & 49.8 & 44 . &
& 6.8 & 136 .
& 13.7 & 0.38 & 1.65 & 1.13 & 41.9 & 160 + 5.7 & 17 & 12.6 & 14 .
& -0.0 & 37 .
& 5.6 & 51 .
& 5.6 & 0.50 & 4.73 & 2.33 & 42.3 & 440 + 6.5 & 6 & 4.5 & 4 . &
-3.0 & 19 . & 2.8 & 23 . & -0.2 & 0.93 & 8.55 & 2.79 & 42.6 & 876 + [ stkdata ] ccc|ccc|c|cc redshift & lg l@xmath125@xmath114 & ew(@xmath126)@xmath127@xmath114 & sfr@xmath128 & sfr@xmath129 & sfr@xmath130 & a@xmath131 & f@xmath132 & f@xmath133 + & & & & ( mag ) & ( % ) & ( % ) + 0.217@xmath111 & & & 1.0 & 0.22 & 2.9 & [email protected] & 3.3@xmath134 & 22.0 + 0.220@xmath111 & & & 10.9 & 0.3 & .. & .. & .. & 2.8 + 0.374@xmath111 & & & 21.7 & 2 & 6.2 & [email protected] & 5.1 & 9.2 + @xmath50.3 & 41.2 & 20 & @xmath21.6 & 0.9 & .. & .. & .. & @xmath956.3 + @xmath51.0 & 42.6 & 20 & @xmath2 84 & 5.7 & .. & .. & .. & @xmath9 6.8 + 2.1 & 41.8 & 20 & @xmath214 & 1.9 & 2.4 & 1.2@xmath135 & 26@xmath136 & @xmath914 + 3.1 & 42.1 & 20 & @xmath228 & 2.6 & 5.3 & @xmath2 0.6 & 28@xmath137 & @xmath99.4 + 3.2 & 42.9 & 22 & @xmath2 28 & 2.3 & 4.2 & 0.8@xmath138 & 26@xmath139 & @xmath98.3 + 4.5 & 42.6 & 16 & @xmath2 139 & 6.0 & 24 & 1.5@xmath140 & 6@xmath141 & @xmath94.3 + 5.7 & 42.8 & 11 & @xmath2 439 & 5.5 & ... & ... & ... & @xmath91.3 + 6.5 & ... & ... &
@xmath2876 & 5 & ... & ... & ... & @xmath90.57 + [ xsfr ] ando , m. et al .
2006 , apjl , 645 , 9 atek , h. , kunth , d. , et al .
2009 , a&a , 506 , l1 blanc , g. a. , adams , j. et al .
2011 , apj , 736 , 31 burgarella , d. , et al .
2007 , mnras , 380 , 986 calzetti , d. , et al .
apj , 533 , 682 colbert , e. j. m. , heckman , t. m. , ptak , a. f. , et al .
2004 , apj , 602 , 231 condon , j. j. 1992 , ara&a , 30 , 575 cowie , l. l. , & hu , e. m. 1998 , aj , 115 , 1319 cowie , l. l. , barger , a.j . , &
hu , e. m. 2010 , apj , 711 , 928 dawson , s. , rhoads , j. e. , malhotra , s. , et al .
2007 , apj , 671 , 1227 deharveng , j. m. , small , t. , et al .
2008 , apj , 680 , 1072 dickey , j. & lockman , f. , 1990 , ara&a , 28 , 215 finkelstein , s. l. , rhoads , j. e. , malhotra , s. , grogin , n. , & wang , j. x. 2008 , apj , 678 , 655 finkelstein , s. l. , cohen , s. h. , malhotra , s. , rhoads , j. e. , et al .
2009a , apj , 703 , 162 finkelstein , s. l. , cohen , s. h. , malhotra , s. , & rhoads , j. e. 2009b , apj , 700 , 276 finkelstein , s. l. , rhoads , j. e. , malhotra , s. , & grogin , n. 2009c , apj , 691 , 465 finkelstein , s. l. , cohen , s. h. , moustakas , j. , et al .
, 2010 , astro - ph/1012.4010 finkelstein , s. l. , hill , g. j. , et al .
2011 , apj , 729 , 140 gawiser , e. , francke , h. , lai , k. , et al . , 2007 ,
apj , 671 , 278 gehrels , n. 1986 , apj , 303 , 336 gilfanov , m. , grimm , h. j. , sunyaev , r. , 2004 , nuphs , 132 , 369 grimm , h.j .
, gilfanov , m. , & sunyaev , r. , 2003 , mnras , 339 , 793 gronwall , c. et al .
2007 , apj , 667 , 79 guaita , l. , gawiser , e. , et al .
2010 , apj , 714 , 255 guaita , l. , gawiser , e. , et al .
2011 , astro - ph/1101.3017 hansen , m. & oh , s. p. 2006 , mnras , 367 , 979 hayes , m. , ostlin , g. , et al .
2010 , nature , 464 , 562 hayes , m. , schaerer , d. , et al .
2011 , apj , 730 , 8 hu , e. m. , cowie , l. l. , et al .
2004 , aj , 127 , 563 iye , m. , ota , k , kashikawa , n. , et al .
2006 , nature , 443 , 14 kennicutt , r. c. , jr .
1998 , ara&a , 36 , 189 laird , e.s . , nandra , k. , et al .
2005 , mnras , 359 , 47 laird , e.s . ,
nandra , k. , hobbs , a. , & steidel , c.c .
2006 , mnras , 373 , 217 lehmer , b.d . , brandt , w. n. , et al .
2005 , apjs , 161 , 21 lehmer , b.d . , brandt , w. n. , et al .
2008 , apj , 681 , 1163 lehmer , b.d . ,
alexander , d.m .
, et al . 2010 , apj , 724 , 559 luo , b. , bauer , f. e. , brandt , w . n. , et al .
2008 , apjs , 179 , 19 malhotra , s. & rhoads , j.e .
2002 , apjl , 565 , 71 malhotra , s. , wang , j , rhoads , j. e. , et al .
2003 , apj , 585 , l25 martin , c. l. , et al .
2008 , apj , 679 , 942 mas - hesse , j. m. & cervino , m. 1999 , iaus , 193 , 550 mas - hesse , j. m. , oti - floranes , h. , & cervino , m. 2008 , a&a , 483 , 71 nagamine , k . , ouchi , m. , springel , v. , hernquist , l. 2010 , pasj , 62 , 1455 nandra , k. , mushotzky , r. f. , arnaud , k. , et al .
2002 , apj , 576 , 625 nilsson , k. k. , moller , p. , moller , o. , et al .
2007 , a&a , 471 , 71 nilsson , k. k. , tapken , c. , et al .
2009 , a&a , 498 , 13 neufeld , d. a. 1991 , apjl , 370 , 85 partridge , r. b. & peebles , p. j. e. 1967 , apj , 148 , 377 persic , m. , rephaeli , y. et al .
2004 , a&a , 419 , 849 pirzkal , n. , et al .
2004 , apjs , 154 , 501 pirzkal , n. , malhotra , s. , rhoads , j. e. , xu , c. 2007 , apj , 667 , 49 ranalli , p. , comastri , a. , & setti , g. 2003 , a&a , 399 , 39 rhoads , j. e. , malhotra , s. , et al . 2000 , apjl , 545 , 85 rhoads , j. e. , dey , a. , malhotra , s. , et al .
2003 , aj , 125 , 1006 rhoads , j. e. , 2009 , apj , 697 , 942 scarlata , c. et al .
2009 , apjl , 704 , 98 tullmann , r. , et al .
2006 , a&a , 457 , 779 venemans , b. p. , et al .
2005 , a&a , 431 , 793 wang , j. x. , rhoads , j. e. , malhotra , s. , et al .
2004 , apjl , 608 , 21 wang , j. x. , zheng , z. y. , malhotra , s. , et al .
2007 , apj , 669 , 765 wang , j. x. , malhotra , s. , rhoads , j. e. , et al .
2009 , apj , 706 , 762 zheng , z. y. , wang , j. x. , finkelstein , s. l. , malhotra , s. , rhoads , j. e. , & finkelstein , k. d. 2010 , apj , 718 , 52 | we have coadded the x - ray flux of all known emitters in the 4 msec chandra deep field south ( cdf - s ) region , achieving the tightest upper limits yet on the x - ray to ratio .
we use the x - ray data to place sensitive upper limits on the average unobscured star - formation rate ( sfr@xmath0 ) in these galaxies . a very small fraction of galaxies in the field are individually detected in the x - rays , implying a low fraction of agn activity . after excluding the few x - ray detected lyman alpha emitters ( laes ) , we stack the undetected laes located in the 4 ms cdf - s data and 250 ks extended cdf - s ( ecdfs ) data , and compute a 1-@xmath1 upper limit on sfr@xmath0 @xmath2 1.6 , 14 , 28 , 28 , 140 , 440 , 880 m@xmath3 yr@xmath4 for laes located at z @xmath5 0.3 and z = 2.1 , 3.1 , 3.2 , 4.5 , 5.7 and 6.5 , respectively .
the upper limit of sfr@xmath0 in laes can be then be compared to sfr@xmath6 derived from line and thus can constrain on the escape fraction ( @xmath7 ) .
the from x - ray at z @xmath5 0.3 is substantially larger than that from uv or h@xmath8 .
three x - ray detected lae galaxies at z@xmath5 0.3 show @xmath5 3 - 22% , and the average escape fraction from stacking the x - ray undetected laes show @xmath9 28% at 3 @xmath1 significance level at the same redshift . we derive a lower limit on @xmath9 14% ( 84 % confidence level , 1 @xmath1 lower limit ) for laes at redshift z @xmath102.1 and z @xmath5 3.1 - 3.2 . at z @xmath9 4 ,
the current lae samples are not of sufficient size to constrain sfr@xmath0 well . by averaging all the laes at z@xmath11 2
, the x - ray non - detection constrains @xmath11 17% ( 84 % confidence level , 1 @xmath1 lower limit ) , and rejects @xmath2 5.7% at the 99.87% confidence level from 2.1 @xmath2 @xmath12 @xmath2 6.5 . |
the isogal project is a multi - wavelength survey at high spatial resolution of the inner galaxy .
the general scientific aims are to quantify the spatial distributions of the various stellar populations in the inner galaxy , together with the distribution of the warm interstellar medium ( ism ) .
optical , near - ir and mid - ir ( to 15 @xmath0 m ) data with near arcsec spatial resolution are being obtained covering the central galactic bulge , and sampling the obscured disk within the solar circle .
complementary data at other wavelengths are being obtained in regions of specific interest .
to date , most isogal effort has focussed on a large area broad - band imaging survey at 7 @xmath0 m and 15 @xmath0 m with isocam on the iso satellite , and on complementary denis ijk@xmath1 observations of the central galaxy .
a full description of the isogal project will be published elsewhere ( omont et al , in preparation ) .
first imaging results are available ( prault et al @xcite ) , as is a complementary paper to this , discussing late type giants at somewhat higher galactic latitudes ( glass et al @xcite ) . in this paper
we discuss first results on the luminous agb stellar population in the previously unobserved inner bulge ( see also frogel et al @xcite ) , from observations of a field whose line of sight projects some 140pc above the galactic centre , on the minor axis .
the agb stage is one of the most complex phases of stellar evolution , while at the same time being of crucial importance for nucleosynthesis and galactic chemical evolution .
although it is clear that mass - loss dominates the final evolution of agb stars , the detailed physics of this process remains rather uncertain .
theoretical progress in the past years has emphasized the relationship between mass - loss and luminosity and hence thermal pulses , radiation pressure on dust and stellar pulsations of long period variables ( lpv ) .
the status of recent modelling and the observational data are fully reviewed by habing ( @xcite ) [ see also the proceedings of iau symposium 191 edited by le bertre et al .
( @xcite ) ] .
observationally , rates of mass loss are relatively well documented , especially from millimeter co studies , far infrared iras results and near infrared studies .
however , this direct knowledge of agb mass - loss is mostly limited to the solar neigbourhood .
in particular , in the galactic bulge and central disk , iras was able to detect only the relatively few agb stars with the largest mass - loss rates @xmath11 10@xmath12 m@xmath13/yr .
we are indeed still lacking the observational information to characterize the influence of metallicity and initial mass on the properties of mass - loss on the agb .
near infrared observations , and in particular the denis ( epchtein et al .
@xcite ) and 2mass ( skrutskie et al .
@xcite , cutri @xcite ) surveys , can detect practically all the agb stars in the galaxy .
however , it is extremely difficult to both identify an agb star , and to distinguish between small to moderate mass - loss rates and patchy interstellar extinction , solely from near - infrared data in regions of high extinction .
data at longer wavelengths , which are more sensitive to the infrared excess that is a consequence of mass loss , and less sensitive to interstellar reddening , are required .
mid - infrared post - iras space surveys , such as the present surveys with isocam and msx ( price et al .
@xcite , egan et al .
@xcite ) are thus uniquely suited for carrying out a census of mass - losing agb stars in the inner galaxy and for quantifying the distribution function of mass loss - rates .
our isogal survey of selected regions of the inner galaxy , at 15 and 7 @xmath0 m ( prault et al .
@xcite , omont et al .
@xcite,@xcite ) has a sensitivity two orders of magnitude better than iras , and their spatial resolution is better by one order of magnitude .
the main purpose of this paper is to show that the isogal data , which combine isocam mid - infrared observations with near - infrared denis data , are ideal to identify bulge mass - losing agb stars , even with dust mass - loss rates as small as 10@xmath14 m@xmath13/yr .
we analyse here the isogal / denis data of a small field ( area 0.035 deg@xmath5 ) centered at @xmath2 = 0.0@xmath3 , @xmath4 = 1.0@xmath3 in the inner bulge .
this field is approximately one - half a bulge scale height down the minor axis , in a previously poorly studied region mid - way ( in @xmath15 ) between the innermost optical `` windows '' ( glass et al .
@xcite ) and the galactic centre .
the very high stellar density in this field leads to a near confusion - limited survey , thus providing a maximum number of detected sources , the majority of them in the bulge and central disk .
the relatively low and well behaved extinction allows reliable identification of mass - losing agb stars down to the rgb tip ( k@xmath16 @xmath6 8.2 for d = 8.0 kpc , tiede et al @xcite ) .
it permits also easier comparisons with earlier works on ( apparently fainter ) stars of the red giant branch ( rgb ) , lpvs and iras sources in the more outer bulge ( see e.g. frogel & whitford @xcite , frogel et al .
@xcite , tiede et al .
@xcite , glass et al .
@xcite , van der veen & habing @xcite and references therein ) .
we can also compare the data with those of the companion paper ( glass et al .
@xcite ) on isogal observations in two fields of the baade windows .
the analysis of the stellar sources is made easier in the latter by the smaller extinction , the greater distance from the galactic disk and previous identifications of lpvs from optical surveys .
.journal of isocam and denis observations in the @xmath2 = 0.0@xmath3 , @xmath4 = 1.0@xmath3 field [ cols="<,^,^,^,^",options="header " , ] + cross - identifications of lw3 and lw2 sources , between themselves and with denis sources , provide the multi - colour data which allow discussion of the nature and properties of individual sources which makes up the bulk of this paper ( figures [ k_jk_den ] to [ 7_k7 ] discussed below ) .
the cross - identification process is also useful for determining data quality .
we have now routine standard procedures for isogal - isogal and denis - isogal cross - identifications ( see appendix b )
. a substantial fraction of the iso sources have thus been identified with denis sources ( 93% and 84% for lw2 and lw3 sources respectively ) .
cross - identifications are essential to establish the reliability of the isogal detections . indeed , because of possible residual artifacts , mainly due
to pixel memory or of noise peaks simulating sources , we consider that the reality of a weak isogal source is not yet well warranted here if it is not confirmed by another detection , either in the other isogal band , or in the k@xmath1 band .
only 9% of lw3 sources are not associated with an lw2 or a k@xmath1 source .
the proportion of unassociated lw2 sources is slightly smaller , 5% . in order to check the completeness of lw3 sources
, we can use the more sensitive lw2 observations ; denis k@xmath1 detections can be used in a similar way to estimate the completeness of lw2 sources ( see appendix b and figure [ iso_cmpl ] ) .
the completeness is probably close to 80% at least , for @xmath17~<~9.5 $ ] and for @xmath18~<~8.5 $ ] .
a more detailed analysis of the source surface density , and its implications for the structure of the galactic bulge , will appear elsewhere , following more sophisticated modelling of source incompleteness as a function of position in this and other fields .
for present purposes such an incompleteness is not a limiting factor .
the quality of isogal photometry has been checked in this field and others by repeated observations ( see appendix b and ganesh et al . in preparation ) .
the uncertainty thus proved to be better than @xmath19 mag rms above @xmath20 mjy in both bands .
however , there is not yet a good standard procedure to fully correct for the detector time behaviour effects for fields such as isogal ones with strong sources and background .
because of that and of confusion , our photometry is thus still uncertain by a few tenths of a magnitude systematically . in conclusion , we consider the reliability of the existence of most of the sources discussed to be well established .
the completeness is also well characterised .
however , the photometric accuracy can still be improved .
table [ table2 ] gives a catalogue of bright isogal sources ( @xmath17~<~7.5 $ ] ) , with three - band denis associations and identification of foreground sources and of candidate variable stars .
a complete catalogue of all isogal sources will be available at cds by october 1999 , when the data reduction is improved .
the data at five wavelengths available for most of the sources allow in most cases a good characterisation of the isogal sources , with some redundancy , as well as of their interstellar reddening .
the very large stellar density in the inner bulge and central disk brings a considerable simplification by ensuring that the majority of the sources are located within it .
in addition , it happens that the interstellar extinction is relatively small on this line of sight and nearly constant for this whole small field .
the discussion of the nature of the sources and of their properties , such as mass - loss , is thus much easier .
the multi - dimensional analysis of magnitude - colour space defined by these data allows one to visualise and to determine the source properties . deferring detailed discussions to the next sections , we limit this section to a general presentation of these diagrams and of the general information they provide about interstellar extinction , circumstellar dust emission and circumstellar absorption .
the k@xmath1/j - k@xmath1 magnitude - colour diagram of all denis sources in our field ( figure [ k_jk_den ] ) shows a remarkably well - defined bulge red giant sequence shifted by fairly uniform extinction of a@xmath21=5.8@xmath221 mag . with respect to the reference k@xmath1@xmath23 of bertelli et al .
( @xcite ) values have been computed .
the differences are very small ( on average about 0.04 mag for the m giants ) , so concerning the internal dispersion in k we can neglect it .
this result is in agreement with persson et al .
( @xcite ) who presented a new grid of infrared standard stars in j , h , k and k@xmath1 . ] with z = 0.02 and a distance modulus of 14.5 ( distance to galactic centre 8 kpc ; we have assumed that [ a@xmath24-a@xmath25/a@xmath26 = 0.167 ) .
most of the extinction should thus be associated with interstellar matter outside of the bulge .
we have checked from the individual values of a@xmath26 and the denis source counts that there is apparently no strong spatial variation of the extinction in this field .
the isogal sources with anomalously low values of a@xmath21 are probably foreground .
they are visible in figure [ k_jk_iso ] , which shows the subset of the k@xmath1/j - k@xmath1 sources of figure [ k_jk_den ] which were also detected at longer wavelengths .
there is of course some uncertainty for the marginal cases : for bright sources with good photometry ( k@xmath1 @xmath27 10 ) , those located left of line a ( a@xmath28 ) are almost certainly foreground , while those to the right of line b ( a@xmath29 ) are very probably in the `` bulge '' , and the case is uncertain for those between lines a and b. the case of foreground isogal sources will be further discussed below .
those with a@xmath28 and with consistent data in the other diagrams are identified by special symbols in the various diagrams .
the distribution of sources about line c is dominated by photometric errors , and residual extinction variations .
there is no evidence for a significant background population , more highly reddened , in the disk beyond the galactic centre , though a few such sources may be present .
similarly , it is difficult from just these data to identify any intrinsic red j - k@xmath1 excess in sources below the rgb - tip , which is near k@xmath30 ( tiede et al @xcite ) .
the j - k@xmath1 excess of the six bright sources much redder than line d is very probably related to an intrinsic j - k@xmath1 excess generated by a relatively thick dusty circumstellar shell , as confirmed by the very large value of k@xmath1-[15 ] for these sources ( see section 6 ) .
a few other sources , just redder than line d in figure [ k_jk_iso ] , might also have an intrinsic j - k@xmath1 excess .
only in the situation of relatively small , foreground , and uniform interstellar extinction , could even a fraction of the agb stars with high mass - loss be identified , and their mass - loss characterised , from the near - infrared denis ( or 2mass ) data alone .
in general , longer wavelength data are critical .
the i band data , when they exist , can provide additional interesting constraints .
however , the much larger spread of intrinsic i - j values in the bulk of the distribution , compared to j - k@xmath1 complicates the identification of foreground sources from i - j data alone . while ( j - k@xmath1)@xmath31 is confined to a very small range ( @xmath320.5 mag ) for most sources , it is well known ( see , e.g. , frogel & whitford @xcite , appendix a ) that there is a large spread in the i magnitudes of bulge agb giants and hence in ( i - j)@xmath33 , that we find ranging over more than 2 magnitudes .
such a spread is certainly related to the behaviour of the tio absorption bands , and hence probably to the metallicity
. however , there is as yet no very detailed modelling of this behaviour . the average value of i - j increases by about 1.5 mag . along the intermediate - agb sequence defined in section 5 .
as discussed below , the most numerous classes of sources detected both at 7 & 15 @xmath0 m in the isogal survey are probably `` bulge '' intermediate agb stars or rgb tip stars with low mass - loss , and high mass - loss rate agb stars ( @xmath10/yr ) .
these are discussed in detail below . in this section
we consider first minor populations in the survey .
there are practically no good young star candidates among isogal sources in this field .
they should be found among sources with large 15@xmath0 m excess that are too faint to be agb stars with large mass - loss .
there are no really convincing cases in the diagrams of figures [ 15_k15 ] & [ 15_715 ] ( however , see section 5 ) .
this is consistent with the relatively small value of a@xmath34 , indicating that there is no very thick molecular cloud on the line of sight .
some 40 stars ( @xmath35 of isogal sources with denis counterparts ) which lie to the left of line a ( a@xmath26@xmath64 ) in figure [ k_jk_iso ] are probably foreground stars , in front of the main line of sight extinction .
this is independently confirmed by another colour for most of them .
the brightest 19 stars , with k@xmath1@xmath36 9 , are detected in lw3 with colours 0.1@xmath37k@xmath1-[15]@xmath370.5 and apparent magnitudes consistent with their being foreground disk giants .
the fainter sources are consistent with being either m giants with very low reddening , or moderately reddened disk k giants
. one should add to these probable foreground stars , a few very bright sources saturated in the denis data .
seven such stars are thus tentatively identified in the iso data .
let us stress that the identification of foreground stars is more difficult for bright sources ( k @xmath38 8) because the intrinsic colour ( j - k@xmath1)@xmath16 is more uncertain .
altogether , we have thus identified about 8% of the isogal sources as foreground stars .
they are distinguished by special symbols in figures [ k_jk_iso ] to [ 7_k7 ] .
as described above , the isogal sensitivity is much greater in the lw2 band than in the lw3 band for red giants with no or little dust .
the number of sources detected in lw2 is more than twice that in lw3 .
most sources detected with lw2 and not with lw3 are fainter in the lw2 and k@xmath1 bands than are the detected lw3 sources ( see figure [ 7_k7 ] ) .
we are able to define the nature of these sources reliably , since the interstellar extinction is well characterised on this line of sight .
few , if any , of the sources detected only at 7 @xmath0 m can have intrinsic infrared excess , from circumstellar dust , or they would have been readily detected at 15 @xmath0 m .
the foreground sources are identifiable from combination of the denis and 7 @xmath0 m flux .
thus , one may isolate those sources which are predominately bulge rgb sources , below the rgb tip ( see figure [ 7_k7 ] ) .
an analysis of the bulge density distribution of both agb and rgb stars , based on their surface density distribution , will be provided elsewhere .
this analysis however requires very careful modelling , as the faint source counts are strongly affected by incompleteness ( see e.g. unavane et al @xcite ) .
the main additional information in the iso mid - infrared bands with respect to the shorter wavelength data solely from denis , is the much increased sensitivity to emission from cold circumstellar dust .
this is well exemplified by the j - k@xmath1/k@xmath1-[15 ] colour - colour diagram of figure [ jk_k15 ] . while the range of j - k@xmath1 values is restricted to @xmath60.5 mag for most sources ( with another 0.5 mag for a few sources ) , k@xmath1-[15 ] ranges from 0 to 2.2 for the bulk of the sources ( with an extension up to 4 magnitudes for a few sources ) .
the colours [ 7]-[15 ] and k@xmath1-[7 ] ( figures [ 15_715 ] & [ 7_k7 ] ) also display large ranges of excess , although somewhat smaller than for k@xmath1-[15 ] . as we discuss now
, only the presence of circumstellar dust can explain such a large excess ; only a portion of it can be attributed to the very cold photosphere .
[ table3 ] lccccccccc + + & & & & & & & & & + + + tip & 2.2 & 1.4 & 5.8 & 8.0 & 1.8 & 7.5 & -7.0 & -4.0 & 3200 + base & 0.4 & 0.0 & 8.5 & 8.9 & 0 & 8.4 & -6.1 & -3.1 & 1400 + + _ notes . _ @xmath39 with a distance modulus of 14.5 ( d = 8 kpc ) + @xmath40 with the k bolometric correction m@xmath41 - m@xmath42 = 3.0 ( groenewegen @xcite ) , which yields m@xmath41@xmath6k@xmath1 - 12 in this field . in the magnitude - colour diagrams [ 15]/k@xmath1-[15 ] and [ 15]/[7]-[15 ] ( figures [ 15_k15 ] & [ 15_715 ] , respectively ) ,
the majority of the sources follow a loose linear sequence .
characteristic values of the colours and magnitudes corresponding to the two ends of this sequence are given in table 3 .
the magnitude of the lower end of the sequence accidently coincides with the isogal sensitivity at 15@xmath0 m .
it is almost exactly that of the tip of the bulge rgb ( @xmath43 8.2 , tiede et al @xcite , @xmath43 8.0 , frogel et al @xcite ) . 0.1mag per scale length . in their study of sgr
i miras , glass et al 1995 have found a dispersion sigma of 0.35 mag from the regression line in the k logp diagram .
this gives an upper limit for the dispersion caused by front - to - back spread , at least in the sgr i region .
thus , line of sight spreads in the photometry , for fields near the minor axis where systematic bar - induced effects are unimportant , are probably small compared to other uncertainties .
of course , the magnitude spread is much larger for the minor , but non negligible , number of sources of the central disk : @xmath60.9 mag per 2.7 kpc scale length ] since there is some uncertainty about the position of the base of the sequence with respect to the rgb tip , it is difficult to know whether the 15@xmath0 m sources with the smaller infrared excess ( k@xmath1-[15]@xmath44 ) are intermediate - agb or rgb - tip stars .
however , most of the sequence with larger 15 @xmath0 m excess ( k@xmath1-[15]@xmath45 ) seems to correspond to agb stars up to @xmath61 mag in k@xmath1 brighter than the rgb tip .
this k@xmath46 magnitude range , 7.5 - 8.5 , corresponds to m spectral types from m6 to m9 ( see table 3a of frogel & whitford @xcite and figs 10 and 12 of glass et al .
since these stars are fainter by one or two magnitudes in k@xmath1@xmath31 or m@xmath41 than the few very luminous agb stars discussed in section 6 , we propose to describe this sequence as `` the intermediate - agb mass - loss sequence '' .
there are also some stars with large 15@xmath0 m excess ( k@xmath1-[15]@xmath47 ) apparently significantly below the rgb tip ( see figure [ k_k15 ] ) .
however , we have checked that the majority ( 8 out of 13 ) have very poor photometry because of blends .
the nature of the few remaining cases is unclear : photometry or association problems , background agbs , young stars or red giants below the rgb tip ( agb or rgb ) with mass - loss ? in order to explore the amount of circumstellar dust involved and its properties , we have used the models developed by one of us ( mg ) which calculate absolute magnitudes within the relevant isocam and denis filters .
the most robust conclusion from the models is confirmation of the need for circumstellar dust to achieve such large infrared excess with respect to photospheric emission . without dust the k@xmath1-[15 ] colours for giant spectral types m5 , m8 and m10 are only 0.15 , 0.55 and 1.07 , respectively . as concerns the specific dust model , in the absence of detailed information ,
the simplest assumption is a time - independent dust mass - loss rate @xmath48 .
of course the value of @xmath48 inferred from the denis - isogal colours strongly depends on the assumed intrinsic dust properties .
depending on these properties , the dust mass - loss rate of the tip of the intermediate agb sequence ranges from @xmath48 = @xmath610@xmath49/yr to @xmath65 10@xmath49/yr .
the infrared colours of the beginning of the intermediate agb sequence imply mass - loss rates 10 - 30 times smaller than for the tip .
an appropriate value of the dust - to - gas ratio during mass loss remains an open question .
the range of spectral type and of mass - loss rate discussed above is typically the domain of validity of the reimers formula ( 1975 ) for the total mass - loss rate @xmath50 . for m@xmath41
= @xmath514 , this formula gives @xmath52m@xmath13/yr .
if one assumes that this result , derived in the solar neighbourhood , can still be applied to bulge stars with the same luminosity , it yields a gas - to - dust ratio in the range @xmath6100 - 500 , depending on the dust properties .
however , the gas - to - dust ratio should be significantly smaller at the base than at the tip of the `` intermediate - agb sequence '' of figure [ 15_k15 ] .
the nature of the dust of bulge stars with weak mass - loss will be discussed in a forthcoming paper ( blommaert et al . in preparation ) from the isocam - cvf spectral observation ( 5 - 16.5 @xmath0 m ) of a few @xmath53 isocam fields in the bulge .
let us stress that the intermediate agb stars in the central galactic bulge with low mass - loss rates ( several 10@xmath54 to close to 10@xmath55 m@xmath13/yr ) seem rather similar in their properties to solar neighbourhood iras agb stars with typical [ 12]-[25 ] colours in the range 0.2 - 1 ( guglielmo 1993 , unpublished phd thesis , hacking et al .
the apparently brightest stars detected by iso at mid - infrared wavelengths form a loose group of @xmath6 30 sources ( see figure [ 15_715 ] ) taking into account those saturated in denis bands , in the various magnitude - colour diagrams .
they are brighter than the tip of the sequence of `` intermediate - agb '' sources , which we have shown are agb stars with low mass - loss rates .
these stars have [ 15]@xmath36 6 , [ 7]@xmath36 7 , and k@xmath1@xmath36 8 in the various figures .
note that with conversion between k@xmath1 and m@xmath41 ( table 3 ) , this limit corresponds to m@xmath56@xmath36 -4 .
a large proportion ( @xmath6@xmath57 ) have at least two very red colours among k@xmath1-[15]@xmath582.0 , [ 7]-[15]@xmath581.2 , k@xmath1-[7]@xmath581.0 and j - k@xmath1 @xmath582.8 , characteristic of larger mass - loss than for the intermediate - agb sequence .
it is tempting to identify them with the onset of the agb `` large mass loss '' at @xmath59@xmath6010@xmath61/yr .
it is known that such a strong wind is classically associated with long period variability ( lpv , see , e.g. habing @xcite and references therein ) . from the simbad data base ,
we have found two lpv stars previously identified in this field , and ( terzan & gosset @xcite ) .
iras17382 - 2830 is an oh / ir star with s@xmath62/s@xmath63 = 2.12 .
it is denoted by `` i '' in figure [ 15_715 ] .
remarkably , this source ( together with two other very bright 15@xmath0 m sources ) is not detected in our 1996 denis observations .
its derived colours are extremely red : k@xmath1-[7]@xmath587 , k@xmath1-[15]@xmath589 .
such very red colours are confirmed by the 1998 denis observations ( see table 1 ) where the source is detected , giving k@xmath1 = 11.7 , k@xmath1-[7]@xmath647.56 and k@xmath1-[15]@xmath649.43 ( schultheis et al , in preparation ) .
these extreme near - ir colours are consistent with its very cold 12/25@xmath0 m iras colour ( see , e.g. , blommaert et al .
@xcite ) . in order to investigate variability in this field
we compared the observations performed with @xmath65 pixels , at two different dates ( see table 1 ) with both lw2 and lw3 filters .
we consider that a bright source is a candidate to be considered for variability , when there is a 3@xmath66 difference in one band , or consistent weaker indications in both bands .
the sources selected in this way are displayed with special symbols in figures [ k_jk_iso ] to [ 7_k7 ] .
we emphasise that this is just a positive indication in favor of variability , but without any rigorous statistical meaning .
in particular , the significance of the candidate variables on the intermediate agb sequence ( figure [ 15_k15 ] ) is unclear , since they are not confirmed by variability in denis data ( schultheis et al . in preparation ) , and it is known that there is no strong variables on this sequence in baade s window ( glass et al @xcite ) . on the other hand , with observations at only three epochs , we can miss a few variables .
this is the case for example for the known variable terzan v 3126 .
a first striking feature in the distribution of these suspected variables is their high proportion in the regions of the magnitude - colour diagrams corresponding to the high mass - loss agb stars defined above ( @xmath50@xmath6710@xmath68/yr ) and delimited in figures [ k_jk_iso ] , [ jk_k15 ] , [ 15_k15 ] , [ 15_715 ] and [ 7_k7 ] . in all cases these candidate variables are at least 50@xmath69 of the stars found there .
their proportion exceeds 80@xmath69 in figure [ 7_k7 ] ( [ 7]@xmath377.0 , [ 7]-[15]@xmath701.0 .
this region of the [ 7 ] @xmath71 [ 7]@xmath72[15 ] plane has been shown from analysis of isogal data by glass et al .
( @xcite ) to be best correlated with the lpv phenomenon in baade window fields .
this identification of luminous variable stars using iso photometry has been confirmed by preliminary comparisons of denis data at two epochs ( schultheis et al , in preparation ) , in particular for most individual stars of this field .
altogether , we have 16 candidate lpvs among the 33 brightest stars ( [ 7 ] @xmath37 7.0 ) .
most of the sources with [ 7 ] @xmath37 7 which are not candidate lpvs are grouped in the region 6.4 @xmath37 [ 7 ] @xmath37 7 , 0.4 @xmath37 k@xmath1 - [ 7 ] @xmath37 1 . out of 12 sources there ,
only two are candidate variables .
of course , we have not enough data to claim with any certainty that any given star among the 10 other is not a variable . however , it is clear , from comparison with the similar group with k@xmath1 - [ 7 ] @xmath58 1 where almost all stars are candidate variables , that the majority of these 10 stars are not strong variables .
the relationship of this apparently non - variable group with agb stars of similar k magnitude which are lpvs , and the relationship with less bright agbs with however similar colours is still unclear ( see also glass et al @xcite )
. it will be interesting to check in particular whether the difference between variable and non variable agb stars in apparently similar evolutionary states is related to differences in metallicity or initial mass , or to a difference in mass loss history on the agb .
another possibility is that most of the `` non variable '' bright stars are in the inner disk with @xmath73 , since their values of j - k@xmath1 are smaller than the average value ( curve c in figure [ k_jk_iso ] ) for most of them .
isogal , combining isocam and denis data , is providing the first detailed and systematic mid - infrared study of an inner bulge field with sufficient sensitivity to isolate the entire agb population .
the isogal performance allows a breakthrough in the analysis of infrared stellar populations , with an improvement of two orders of magnitude with respect to iras .
the density of detected sources is close to the confusion limit with 3@xmath74 pixels at 7 @xmath0 m .
as expected , most of the sources are m giants of the bulge or central disk . in the low reddening fields analysed to date practically all bulge giants have near - infrared denis counterparts .
this high proportion of associations is possible only because of the very good denis astrometry and the relatively good astrometry with isocam .
sources detected above our completeness limit at 15 @xmath0 m are mainly just above the rgb tip ; they are thus mostly agb stars .
sources detected only at 7 @xmath0 m are mainly normal bulge red giants just below the rgb tip .
the completeness and reliability of detection at 7 @xmath0 m and 15 @xmath0 m of point sources are high and well quantified down to @xmath3215 mjy and @xmath3210 mjy respectively .
the photometric accuracy is reduced by a variety of effects : by the complex time- and illumination history - dependant behaviour of isocam pixels , especially in regions with a very high density of bright sources , and also by the integration time available for a wide - area survey .
however , the photometric accuracy we have achieved is good enough to be able to take advantage of the rich information provided by the combination of the five wavelength data of isogal+denis .
of particular importance is our ability to detect reliably quite small reddening - corrected 15 @xmath0 m excesses .
a detailed analysis of the stellar populations is also much eased by the overwhelming preponderance of bulge or central disk stars with a well defined distance , and , in the present field , by the relatively low and constant interstellar reddening in front of the whole field studied .
we have shown here that our isogal survey is ideal ( though not unique ) for analysis of agb stars with large mass - loss rates ( @xmath10/yr ) , by providing a complete census in the field , independent of large amplitude variability .
the most important conclusion for future analyses from the present analysis is our demonstration that the combination of near - ir ( denis ) and mid - ir ( 7 @xmath0 m and 15 @xmath0 m ) isogal data allows reliable detection of circumstellar dust and low rates of mass - loss in bulge agb stars not deducible from near - infrared data alone .
isogal is uniquely suitable for systematic studies of agb stars with low rates of mass - loss in the whole inner galaxy , especially in the bulge .
the very small amounts of dust associated with low rates of mass loss are undetectable in the near - infrared , both in absorption and in emission , while being readily detectable at 15 @xmath75 m .
stars with low rates of mass loss are too faint to have been detectable by iras , except in the solar neighbourhood ; most of them will escape detection by msx because of the confusion limit arising from the 18@xmath74 msx pixels , which have areas almost an order of magnitude larger than those used for isogal .
although mid - infrared evidence for dust emission corresponding to low rates of mass loss is seen in the iras data for agb stars in the solar neighbourhood , inevitable distance uncertainties make the analysis of such iras data for mass - loss rates and agb evolution much less clear - cut .
the most important immediate scientific conclusion of this paper is our detection of low rates of mass loss which are ubiquitous for red giants with luminosities just above or possibly close to the rgb tip , and thus still in relatively early stages .
the most luminous of these stars , that we defined as `` intermediate '' agb ( in the early agb thermal pulse phase ) , form a well defined sequence in the [ 15]/k@xmath1-[15 ] and [ 15]/[7]-[15 ] magnitude - colour planes . in order to explain the colours the presence of dust is required , with ( model - dependant ) dust mass - loss rates of a few 10@xmath76/yr .
it is thus well established that dust formation is already associated with weak mass - loss during the early tp - agb phase , this obviously puts important constraints on the physics of dust formation .
these first results can obviously be improved and exploited in several ways . for the data presented for this specific field
, we still hope to improve the photometry and the reliability of the isogal data , in particular by including the verification observations not yet fully exploited .
the study of mass - losing red giants will be extended to the other 200 isogal fields : i ) in a straightforward way to the other isogal bulge fields with @xmath77 , as already done for the isogal observations of two baade window fields ( glass et al .
@xcite ) ; ii ) to the bulk of the isogal fields closer to the galactic plane , where there is large and variable extinction , and where the uncertainty on the distance and the mixing with young stars somewhat complicate the analysis .
we are currently analysing the surface density of the various classes of agb stars to characterize the structure and stellar populations of the bulge , to determine to what extent it is meaningful to consider the various structures : bulge , central disk , bar , central cluster , and so on .
as concerns the theoretical interpretation , much work is still needed in order to : \i ) improve the models of red giants with weak mass - loss both for photospheric and for dust emission .
many questions can be addressed : what is the chemical nature of the dust , considering silicates and possible other components ; is the base of the large spread we have observed in i - j colours at fixed luminosity a metallicity effect ; and more generally , how can we disentangle the effects due to initial mass , age and metallicity ?
\ii ) use our isogal results to further constrain the models of dust formation and of tp - agb evolution , and in particular : determine whether dust formation can begin in rgb stars close to the rgb tip , or whether it is specific of agb stars ; and explain dust formation in the context of very weak mass - loss .
_ acknowledgements _ + this work was carried out in the context of eara , the european association for research in astronomy .
we would like to thank m. prault , p. hennebelle , s. ott , r. gastaud , h. aussel , and f. viallefond for useful discussions during the course of the reduction of the iso data .
we thank the referees , especially , j.a .
frogel , for their very constructive and useful suggestions .
sg was supported by a fellowship from the ministere des affaires etrangres , france .
ms acknowledges the receipt of an esa fellowship .
the denis project is partially funded by european commission through science and human capital and mobility plan grants .
it is also supported , in france by the institut national des sciences de lunivers , the education ministry and the centre national de la recherche scientifique , in germany by the state of baden - wrtemberg , in spain by the dg1cyt , in italy by the consiglio nazionale delle ricerche , in austria by the fonds zur frderung der wissenschaftlichen forschung und bundesministerium fr wissenshaft und forschung , in brazil by the foundation for the development of scientific research of the state of sao paulo ( fapesp ) , and in hungary by an otka grant and an esoc&ee grant .
the fluxes and magnitudes used were defined in the following way ( blommaert @xcite ) .
the isocam units of adu / gain / sec were first converted into mjy / pixel units within cia , with the conversion factors : @xmath78 for lw2 and @xmath79 for lw3 .
these are correct for a f@xmath80 @xmath6 @xmath81@xmath82 power spectrum at wavelengths 6.7 @xmath0 m and 14.3 @xmath0 m respectively .
magnitudes are defined by @xmath83 = 12.39 - 2.5 \times log[f_{lw2}(mjy ) ] \label{eq : cnv3}\ ] ] @xmath84 = 10.74 - 2.5 \times log[f_{lw3}(mjy ) ] \label{eq : cnv4}\ ] ] where the zero point has been chosen to provide zero magnitude for a vega model flux ( a0v star , not including the infrared excess emission of the circumstellar disk ) at the respective wavelengths mentioned earlier .
the special procedure for isogal source extraction , developed by alard et al ( in preparation ) , uses several `` cia '' procedures not yet implemented in the standard ( auto - analysis ) treatment ( corrections for distortion of the isocam field , and for time behaviour [ `` vision '' and `` inversion '' methods ] ) .
it also includes a sophisticated source extraction , after a regularisation of the point - spread - function ( psf ) . indeed , there is not yet a good standard procedure to fully correct for the time behaviour effects for fields such as isogal ones with strong sources and background .
here we have used the fluxes given by the `` inversion '' method , which provides better photometry , but we have used the `` vision '' method to identify and drop the false sources generated by the detector `` memory '' of bright sources previously observed at a different location on the sky , but in the same pixel .
after the elimination of the false replication sources , the source counts in this field are 599 and 282 respectively in lw2 ( [ 7 ] @xmath37 9.75 ) and lw3 ( [ 15 ] @xmath37 8.5 ) .
this is indeed close to the confusion limit for lw2 sources ( 85 pixels [ @xmath85 per source , density 1.7 10@xmath86 deg@xmath87 ) , but farther from this limit for lw3 ones ( 8.1 10@xmath88 deg@xmath87 ) . for denis sources ,
the sensitivity is mostly limited by confusion in the k@xmath1 and j bands ( @xmath89 [ for k@xmath1 @xmath27 12 and j @xmath27 14 ] , giving @xmath619 pixels @xmath90 per source ) .
the completeness limit is thus probably close to 11.5 in the k@xmath1 band and 13.5 in the j band .
the density at the sensitivity limit in the i band , @xmath91 , is farther from confusion ( density @xmath92 with @xmath93 pixels ) .
independent magnitudes are available for many denis sources in the overlap region between adjacent observations .
analysis of these repeated observations shows that the internal dispersion in the photometry , in this crowded region , is less than 0.1 mag for k@xmath1 @xmath27 11 , j @xmath27 13.5 and i @xmath27 16.5 ( it rises to 0.18 mag for k@xmath1 @xmath27 13 , 0.13 mag for j @xmath27 14.5 and 0.2 for i @xmath27 17.5 ) . for the determination of the zero point
all standard stars observed in this night have been used .
we derived the following zero points : @xmath94 , @xmath95 and @xmath96 , respectively .
the internal rms in the zero - points is found to be 0.03 , 0.07 and 0.04 mag in the k@xmath1 , j and i bands respectively .
we have now routine standard procedures for isogal - isogal and denis - isogal cross - identifications ( copet et al . in preparation ) . the good quality of the pointing of iso and of the correction of the isocam field distortions permits , after optimisation of a small rotation - translation of the fields , a reduction of the rms of the nominal offsets of matched sources to @xmath97 and @xmath98 for lw3/lw2 and isogal / denis respectively . however , the search radius was fixed at a large value , 2.7@xmath74 , for lw3/lw2 associations in order not to miss associations .
the chance of spurious association with an lw2 source is then @xmath99 . because of the very high density of denis sources ,
the search radius was reduced to 2.1@xmath74 for the denis / isogal associations .
nevertheless , the density of the denis sources is so high that the chance of spurious associations remains @xmath100 for k@xmath1 sources with k@xmath1 @xmath27 12 .
the chance of spurious association is reduced to @xmath101 when one limits the associations to k@xmath1 = 11 . a substantial fraction of the iso sources
have thus been identified with denis sources . out of a total number of 599 lw2 sources , 557 ( @xmath102 )
are matched with a k@xmath1 @xmath27 12 source , 552 with a jk@xmath1 source and 522 with an ijk@xmath1 source .
out of 282 lw3 sources , 248 ( @xmath103 ) are matched with an lw2 source , 237 ( @xmath104 ) with a jk@xmath1/lw2 , and 221 ( 78@xmath69 ) with an ijk@xmath1/lw2 source .
the number of lw2/lw3 sources without k@xmath1 or lw3/k@xmath1 sources without lw2 is very small , 10 in both cases . in order to check the completeness of lw3 sources
, we can use the more sensitive lw2 observations .
one can check , for instance , that only 13 lw2 sources with [ 7 ] @xmath27 8.3 , among 183 in total , are missing in lw3 . from the known range of values of [ 7 ] - [ 15 ]
, we can conclude from figure [ iso_cmpl ] that the completeness of lw3 sources is close to @xmath105 for [ 15 ] @xmath27 7.5 ( @xmath106 mjy ) and @xmath107 in the range 7.5 @xmath27 [ 15 ] @xmath27 8.5 ( @xmath108 mjy ) .
denis k@xmath1 detections can be used in a similar way to estimate the completeness of lw2 sources .
figure [ iso_cmpl ] compares the number of lw2 sources detected per half magnitude bin with an approximate estimate of the number expected from k@xmath1 sources with typical colours .
it is seen that the detections are practically complete for [ 7 ] @xmath27 8.5 ( @xmath109 mjy ) and that they remain more than 80@xmath69 complete for 8.5 @xmath27 [ 7 ] @xmath27 9.5 ( @xmath110 mjy ) ; however , the completeness rapidly decreases below @xmath111 mjy .
this incompleteness is mainly due to confusion .
the quality of isogal photometry has been checked in this field ( table 1 ) and others by repeated observations both with 6@xmath74 pixels ( ganesh et al . in preparation ) .
the uncertainty thus proved to be better than @xmath19 mag rms above @xmath20 mjy in both bands .
it is poorer for weaker sources , especially in the lw3 band .
one can expect a similar repeatibility accuracy with 3@xmath74 pixels .
however , this does not take into account systematic errors .
in particular , the comparison of the 3@xmath74 and 6@xmath74 pixels measurements which were performed on this field shows a small systematic difference in the fluxes , with average differences up to 0.1 - 0.2 mag rms .
further work is in progress to understand these details , but the effect may be explained by the source confusion as is discussed in depoy et al ( @xcite ) .
our photometry is thus still uncertain by a few tenths of a magnitude systematically .
the photometry is expected to be poorer on the edges of the isogal 7 pointings ) , @xmath112 the image is observed with a single exposure instead of the double exposure on average for the points of the central part . in addition , the source extraction is not able to recover the full intensity of sources very close to the edges within a few pixels .
bertelli , g. , et al . , 1994 a&as , 106 , 275 + binney , j.j . , gerhard , o.e . , and spergel d.n . , 1997 , mnras , 288 , 365 + blommaert , j.a.d.l , 1998 : isocam photometry report , http://www.iso.vilspa.esa.es/users/expl_lib/cam_list.html + blommaert , j.a.d.l .
, van der ween , w.e.c.j . ,
van langevelde , h.j . ,
habing , h.j . ,
sjouwerman , l.o . , 1998 ,
a&a 329 , 991 + cesarsky , c. , et al .
1996 a&a , 315l , 32 + cutri , r. m. 1998 , baas , 30 , no .
+ depoy , d.l . ,
terndrup , d.m . ,
frogel , j.a . ,
atwood , b. , blum , r. , 1993,aj 105,2121 + egan , m.p . ,
shipman , r.f . , price , s.d . ,
carey , s.j . ,
clark , f.o .
, cohen , m. , 1998 , apj 494 , 199 + epchtein n. , et al . , 1997 ,
messenger 87 , 27 + frogel , j.a . ,
whitford , a.e . , 1987 , apj 320 , 199 ( fw ) + frogel , j.a . ,
terndrup , d.m . ,
blanco , v.m . ,
whitford , a.e .
, 1990 , apj 353 , 494 + frogel , j.a . ,
tiede , g.p . ,
kuchinski , l.e . , 1999 ,
aj in press ( astro - ph/9901328 ) + glass , i.s . ,
whitelock , p.a . ,
catchpole , r.m . , feast , m.w . , 1995 ,
mnras 273 , 383 + glass , i.s . ,
ganesh , s. , alard , c. , blommaert , j.a.d.l . , gilmore , g. , lloyd evans , t. , omont , a. , schultheis , s. , simon , g. , 1999 , mnras in press ( astro - ph/9904010 ) + groenewegen , m.a.t . , 1997 , in `` the impact of large scale near - ir sky surveys '' , f. garzon et al .
page 165 + guglielmo , f. , 1993 , thesis , universite paris vii + habing , h.j .
, 1996 , astron .
rev . 7 , 97 + hacking , p. , et al . , 1985 ,
pasp 97 , 616 + le bertre , t. , lebre , a. , waelkens , c. 1999 , eds , proceedings of iau symposium 191 + omont a. & the isogal collaboration 1999 , in `` astrophysics with infrared surveys : a prelude to sirtf '' , pasadena , asp conference series , in press + omont a. et al . 1999b , to appear in the universe as seen by iso , eds .
p. cox and m.f .
kessler , esa special publications series ( sp-427 ) + prault m. , omont a. , simon g. , sguin p. , ojha d. , blommaert j. , felli m. , gilmore g. , guglielmo f. , habing h. , price s. , robin a. , de batz b. , cesarsky c. , elbaz d. , epchtein n. , fouqu p. , guest s. , levine d. , pollock a. , prusti t. , siebenmorgen r. , testi l. , tiphne d. , 1996 , a&a 315 , l165 + persson , s.e . ,
murphy , d.c . ,
krzeminski , w. , roth , m. , rieke , m.j .
, 1998 , aj 116 , 2475 + price , s.d . ,
tedesco , e.f .
, cohen , m. , walker , r.g . ,
henry , r.c . , moshir , m. , paxton , l.j . ,
witterborn , f.c . , 1997 , iau symp .
179 , p115 + reimers , d. , 1975 , in `` problems in stellar atmospheres and envelopes '' b. baschek , w.h .
kegel , g. traving ( eds ) , springer , berlin , p 229 + skrutskie , m.f .
1997 , in `` the impact of large scale near - ir sky surveys , '' p187 - 195 , f. garzon et al .
( eds . ) , kluwer ( netherlands ) .
+ terzan , a. , gosset e. , 1991 , a&as 90 , 451 + tiede , g.p . ,
frogel , j.a . ,
terndrup , d.m . , 1996 , aj 110 , 2788 +
unavane , m. , gilmore , g. , epchtein , n. , simon , g. , tiphene , d. , de batz , b. , 1998 , mnras 295 , 119 + van der veen , w.e.c.j . ,
habing , h.j . , 1990 , a&a 231 , 404 + | the isogal project is a survey of the stellar populations , structure , and recent star formation history of the inner disk and bulge of the galaxy .
isogal combines 15 @xmath0 m and 7 @xmath0 m isocam observations with denis ijk@xmath1 data to determine the nature of a source and the interstellar extinction . in this paper
we report an isogal study of a small field in the inner galactic bulge ( @xmath2 = 0.0@xmath3 , @xmath4 = 1.0@xmath3 , area = 0.035 deg@xmath5 ) as a prototype of the larger area isogal survey of the inner galaxy .
the isocam data are two orders of magnitude more sensitive than iras ones , and its spatial resolution is better by one order of magnitude , allowing nearly complete and reliable point - source detection down to @xmath6 10 mjy with the lw3 filter ( 12 - 18 @xmath0 m ) and @xmath6 15 mjy with the lw2 filter ( 5.5 - 8 @xmath0 m ) .
more than 90% of the isocam sources are matched with a near - infrared source of the denis survey .
the five wavelengths of isogal+denis , together with the relatively low and constant extinction in front of this specific field , allow reliable determination of the nature of the sources .
while most sources detected only with the deeper 7 @xmath0 m observation are probably rgb stars , the primary scientific result of this paper is evidence that the most numerous class of isogal 15 @xmath0 m sources are red giants in the galactic bulge and central disk , with luminosities just above or close to the rgb tip and weak mass - loss rates
. they form loose sequences in the magnitude - colour diagrams [ 15]/@xmath7-[15 ] and [ 15]/[7]-[15 ] .
their large excesses at 15 @xmath0 m with respect to 2 @xmath0 m and 7 @xmath0 m is due to circumstellar dust produced by mass - loss at low rate ( @xmath8a few @xmath9/yr ) .
these isogal results are the first systematic evidence and study of dust emission at this early stage ( intermediate agb and possibly rgb - tip ) , before the onset of the large mass - loss phase ( @xmath10/yr ) .
it is thus well established that efficient dust formation is already associated with such low mass - loss rates during this early phase .
about twenty more luminous stars are also detected with larger excess at 7 and 15 @xmath0 m .
repeated isogal observations suggest that the majority of these are long period variables with large amplitude , probably in the large mass - loss stage with @xmath10/yr .
0.3true cm |
in this paper , which is a sequel to appendix b in , we shall consider the computation of the disc partition function for the superstring z= e^- , = _ 0 + _ 1 + ... for the euclidean minimal surface of ending on @xmath1-wound circle at the boundary . here
the expansion is in inverse powers of the string tension @xmath4 and we will be interested in first subleading ( 1-loop ) correction @xmath5 .
the string partition function should be representing the expectation value @xmath6 of the circular wilson loop in @xmath1-fundamental @xmath7 representation in dual @xmath8 sym theory taken in the limit @xmath9 and then expanded in large t hooft coupling @xmath10 for fixed @xmath1 .
the well - known exact expression for @xmath6 in the planar approximation gives & & w = 2 k i_1 ( k ) = e^- , = _ 0 + _ 1 + ... , + & & _ 0 = - k , _ 1 = + 3 2 k + 3 2 . assuming that the planar @xmath1-fundamental circular wilson loop is indeed represented by the open string path integral in with the boundary condition that the disc - like world sheet ends on @xmath1-wound circle at the boundary the corresponding minimal surface in , e.g. , @xmath11 with the metric @xmath12 is described by = k , z= k , r= ^2 k , z= , where @xmath13 .
then the induced metric is ds^2 = k^2 ^2 k ( d^2 + d ^2 ) = d ^2 + k^2 ^2 d ^2 , = . for @xmath14
this is the usual @xmath2 metric but for @xmath15 it has a conical singularity at @xmath16 with negative deficit @xmath17 , i.e. this is a cone of @xmath2 .
the corresponding classical action is proportional to the regularized @xmath2 volume .
to recall , the euclidean @xmath18 space in poincare coordinates @xmath19 with boundary being a line @xmath20 is not equivalent to @xmath21 in angular coordinates @xmath22 with boundary @xmath23 ; in particular , their regularized volumes are different v__2 = 0 , v_ads_2 = 2 ( 1 - 1 ) - 2 . here
we set the @xmath2 scale to 1 , @xmath24 is the length of the boundary of @xmath25 and @xmath26 is a radial ir cutoff . as a result , the renormalized classical string action for ( [ 3 ] ) is @xmath27 which matches the value of @xmath28 in ( [ b3 ] ) .
this suggests that the prescription to use the regularized volume of @xmath2 is a required part of definition of string theory partition function in the context of ads / cft .
the starting point to compute @xmath5 is the general form of the superstring one - loop correction @xcite _ 1= .
[ b1 ] in the simplest cases of minimal surfaces with symmetric - space induced metric ( like @xmath2 spaces of straight line and circular loop ) it is natural to preserve this symmetry in the loop computation till very end , i.e. not to introduce radial cutoff directly on contributions of individual modes .
then , as standard for determinants on a symmetric spaces they can be computed using heat kernel technique ( utilizing in an essential way the unbroken global symmetry ) and the final expression will be proportional to the space volume factor for which we should use its renormalized value in ( [ o ] ) .
for example , in the straight line case the classical action and loop corrections will be proportional to the volume of @xmath29 with boundary @xmath20 which has zero renormalized value in ( [ o ] ) and should thus be assumed to vanish , in agreement with @xmath30 on the gauge theory side .
such heat kernel based computation of ( [ b1 ] ) in the @xmath14 circular loop case gives ( using that the renormalized @xmath31 ) _ 1 ( 1 ) = ( 2 ) .
the same expression was found in using a different ( gelfand - yaglom ) method to compute the determinants with the radial ir cutoff implemented explicitly on modes and all power divergences dropped in the final expression .
the same method applied to the @xmath32 case led to the following generalization of ( [ 4 ] ) _ 1 kt ( k ) = ( 2 ) + ( 2 k + ) k - k ! .
this expression does not match the gauge theory result @xmath33 in ( [ b3 ] ) .
the third @xmath34 term in @xmath33 may be attributed to the presence of the string tension normalization factor for the three ( mobius - symmetry ) ghost zero modes on the disc @xcite , which was not included in ( [ b1 ] ) and thus should be absent in ( [ 4]),([b2 ] ) .
the remaining difference may be coming from a numerical factor in normalization of the disc zero modes or from the ratio of the ghost and the two longitudinal mode determinants ( assumed to be equal to one in ( [ b1 ] ) ) once they are computed with proper boundary conditions ( cf . ) .
below we shall ignore these open issues and aim at just deriving the analog of ( [ b2 ] ) in the heat kernel approach of .
explicitly , we will find that _ 1 ( k ) = k ( 2 ) + ( k ) , ( 1 ) = 0 , + ( k ) = - _ 0^ dyy y here the terms with @xmath35 and @xmath36 are the contributions of the scalar factors under log in ( [ b1 ] ) and the term with @xmath37 the contribution of the spinor factor . for large @xmath1 the integral @xmath38 is well approximated by a straight line , @xmath39 .
this result for @xmath5 with @xmath40 is thus different from ( [ b2 ] ) of and also different from the simple gauge - theory expression @xmath41 in ( [ b3 ] ) . here we will not attempt to understand the difference with ( [ b2 ] ) , concentrating on developing technical tools required for derivation of ( [ 5 ] ) that may have other potential applications .
the main issue is to find scalar and fermion heat kernels on the cone of in ( [ 3 ] ) .
let us mention also that another possible application of the above partition function is to computation of rnyi entropy on @xmath2 ( see ) .
it will be useful to replace @xmath1 in ( [ 3 ] ) by an arbitrary real number @xmath42 and set @xmath43 only at the end
. then if we take @xmath42 to be equal to @xmath44 with an integer @xmath45 the metric ( [ 3 ] ) becomes that of an orbifold @xmath46 ( with a conical singularity of positive deficit ) and the corresponding scalar or fermion heat kernel can be found as a sum over images
. this special case of @xmath47 will thus allow us to check the general @xmath42 expressions for heat kernels . in the case of a cone of a 2-plane @xmath48 with a general @xmath42
the scalar laplacian heat kernel can be found using re - periodisation " trick ( or sommerfeld formula ) and the same idea applies also to the construction of heat kernel on the cone of @xmath2 with generic @xmath42 .
we shall present a detailed derivation of the expression of for the scalar heat kernel in section 2 .
the analogous construction for the spinor heat kernel can be given using the result for the regular @xmath2 and generalizing the examples of cones of 2-plane and 2-sphere in .
we shall present it in section 3 .
this will be our main new technical result . as a check
, we shall match the small @xmath49 expansion of the spinor kernel with the general form of asymptotic expansion of spinor heat kernel on a conical singularity in and also show that in the special case of @xmath50 corresponding to @xmath46 orbifold it agrees with the result in . in section 4 we shall use the results for the scalar and spinor heat kernels to compute the determinants in ( [ b1 ] ) deriving ( [ 5 ] ) and comparing it to ( [ b2 ] ) . in appendix
a we shall review the derivation of the sommerfeld formula for heat kernels of 2d scalar and spinor laplacians on a manifold with a conical singularity . in appendix b we shall present the derivation of the phase factor matrix @xmath51 in the spinor heat kernel in ( [ 31 ] ) .
in order to compute the @xmath52-loop partition function ( [ 0]),([b1 ] ) we need to know the traces of heat kernels of the scalar and spinor laplacians on the cone of with an arbitrary deficit @xmath53 ( we shall denote this space as @xmath54 ) . in the regular case
we have = - _
^ ( t ) , ( t ) = d^2 x g k(x , x;t ) , k(x , x;t ) = x| e^-t | x + ( t ) = v_ads_2
k(t ) , k(t ) = k(x , x ; t ) , v_ads_2 = - 2 .
here @xmath55 is the trace of heat kernel which on symmetric space is a product of ( regularized ) volume and heat kernel at coinciding points . in this section
we will review the computation of the heat kernel @xmath55 of the scalar laplacian @xmath56 on @xmath54 using the sommerfeld formula , rederiving the expression for it given in .
we shall check the consistency of the result in the special case when @xmath47 with integer @xmath45 by comparing it with the scalar heat kernel on the orbifold @xmath57 .
the generalization of the scalar heat kernel on regular @xmath58 to the case of can be constructed using the sommerfeld formula as reviewed in appendix a. the starting point is the expression for the untraced heat kernel of the scalar laplacian on the regular @xmath58 @xmath59 here @xmath60 is the geodesic distance between the two points @xmath61 and @xmath62 of in polar coordinates & & ds^2 = d ^2 + ^2 d ^2 , + & & d(x , x ) = - ( - ) . according to the sommerfeld formula , the traced heat kernel in presence of a conical singularity when @xmath63 is given by & & [ s]_ads_2(t ) _
( t)= k(0 ; t ) + _ dz k(z;t ) , + & & k(z;t)_0 ^ 2d_0^d k(=^,^=+z;t ) , + & & k(=^,^=+z;t)= _
d(,z)^ , + & & k(0;t)= _ 0 ^ 2d_0^d k(=^,^=;t)= _ ads_2 ( t ) , + & & _ ads_2 ( t ) = v_ads_2 k_0(t ) , k_0(t ) = _ 0^ dy y 1 e^- .
here @xmath64 is the integrated heat kernel with points separated by @xmath65 along angular direction only .
the contour @xmath66 on the complex plane is made of two vertical lines going from @xmath67 to @xmath68 and from @xmath69 to @xmath70 ( see appendix a ) . the geodesic distance @xmath71 between the points @xmath72 and @xmath73 is given by @xmath74 . in ( [ sis ] ) we used that for regular symmetric @xmath2 space the heat kernel does not depend on coordinates at coinciding points .
note that the ir divergence regularized by @xmath75 will be present only in the first
@xmath2 " term @xmath76 in ( [ s ] ) , while the second integral term will be finite . to compute @xmath64 we introduce @xmath77 to get k(z;t ) = _
0^ _ d(v , z)^ dy y e^- ^2(y , z)-v^2 , ( y , z ) .
now we interchange the integration limits @xmath78 and integrate over @xmath79 using @xmath80 . then integrating by parts one finds that @xmath81 changing the variable @xmath82 we finally get @xmath83 according to ( [ s ] ) , the scalar heat kernel on cone of @xmath58 is thus given by & & _
( t)= _ ads_2(t ) + _ 0^ dy y e^-y^2 t _ dz f(z;y , ) , + & & f(z;y,)= . for @xmath84
the function @xmath85 becomes @xmath86 periodic in @xmath65 , and due to the structure of the contour @xmath66 , this leads to the vanishing of the integral contribution . to evaluate the contour integral we use the residue theorem .
the two vertical lines @xmath87 and @xmath88 surround one pole of @xmath89 at @xmath90 with the residue @xmath91 . setting @xmath92
we get @xmath93 with two poles on the imaginary axis at @xmath94 . using the limit @xmath95
@xmath96 therefore @xmath97 finally , the scalar heat kernel on @xmath54 is given by @xmath98 which is the same expression as given in .
the integral vanishes for @xmath84 , i.e. @xmath99 .
it is obviously convergent at @xmath100 and also at @xmath101 since @xmath102 .
the special case of @xmath47 corresponds to the @xmath103 orbifold of @xmath58 which is obtained by identification @xmath104 .
the analog of the sommerfeld formula for orbifolds is equivalent to the summation over images expression discussed in @xcite .
there it was found that the small @xmath49 expansion of the heat kernel for a massless scalar laplacian on @xmath105 is given by @xmath106 to compare to the general expression ( [ 28 ] ) we need first to set there @xmath107 and expand for small @xmath49 . changing the variable @xmath108 we get for @xmath109 @xmath110 + o(t^{2})\\ = & \k\ , \kk_{ads_{2}}(t)+ \te \frac{1}{12}\big(\k^{-1 } - \k \big ) - \frac{1 + 10\k^{2}-11\k^{4}}{360 \k^{3}}\ , t + o(t^{2})\ . \end{split}\ ] ] setting here @xmath111 we indeed match the orbifold expansion ( [ 30 ] ) .
let us now turn to the spinor case .
the fermions in ( [ b1 ] ) are real ( majorana ) so that log of their determinant or trace of heat kernel should be understood with extra @xmath112 included ( so that on a trivial background the contribution of one 2d majorana fermion is minus that of one real scalar ) .
the starting point is the heat kernel for the ( majorana ) spinor laplacian @xmath113 on @xmath58 @xmath114 here @xmath115 is a @xmath116 matrix acting on the @xmath58 spinor bundle and satisfying the equation of the parallel transport along the geodesic connecting the two points @xmath117 : @xmath118 here the derivatives are taken with respect to the point @xmath119 and @xmath120 is the tangent vector to the geodesic .
the general structure of @xmath121 on hyperbolic spaces was not spelled out in the literature .
as we show in appendix b in the present case of @xmath2 in polar coordinates ( [ aa ] ) it is given by ( @xmath122 is the pauli matrix ) @xmath123 we can now apply the fermionic version of the sommerfeld formula ( see appendix a ) to obtain the heat kernel of the spinor laplacian on the cone of @xmath58 ( cf .
( [ s]),([ss ] ) ) & & _ ( t)= k(0;t ) + _ dz k(z;t ) , + & & k(z;t)_0 ^ 2d_0^d k(=^,^=+z;t ) , + & & k(=^,^=+z;t)= u(,z ) _
d(,z)^ + & & u(,z)= , u(,z)= , + & & k(0;t)= k(x , x;t ) = _ ads_2 ( t ) , _ ads_2 ( t ) = v_ads_2 ( t ) ,
+ & & ( t ) = _ 0^ dy y e^- . here
@xmath124 is the fermionic kernel in the regular case in ( [ 31 ] ) .
the contour @xmath66 is the same as in the scalar case in ( [ s ] ) ( see appendix a ) . to compute @xmath125 in ( [ ff ] ) we integrate over the angle and after a change of variables ( @xmath126 ) we find @xmath127 interchanging the integration limits as explained below ( [ 23 ] ) we get @xmath128 where @xmath129 as in ( [ 23 ] ) . as a result @xmath130 integrating by parts in the remaining integral over @xmath131 in ( [ 355 ] )
one can put it into the form k(z;t ) = _ 0^ dy e^-y^24 t . changing the variable @xmath82 gives @xmath132 so that ( [ f ] ) takes the form ( cf .
( [ 26 ] ) ) & & _ ( t)= _ ads_2(t ) + _ 0^ dy e^-y^2 t _ dz ( z;y , ) , + & & ( z;y,)= .
for @xmath84 the function @xmath133 becomes the same as @xmath134 in ( [ 27 ] ) in the scalar case , so that the second integral term vanishes because of the structure of the contour @xmath66 .
it remains to evaluate the contour integral using the residue theorem .
as in the scalar case , the contour @xmath66 contains one pole of @xmath135 at @xmath90 and two imaginary poles @xmath136 of @xmath137 , i.e. ( cf .
( [ 277 ] ) ) _ dz ( z;y,)= - ( - ) .
finally , the spinor heat kernel the @xmath54 is given by @xmath138 the integral term here vanishes at @xmath84 and is convergent at both @xmath139 and @xmath140 where @xmath141 = \frac{1}{6 } \big ( 1-\frac{1}{\k^{2}}\big ) .
$ ] to check ( [ 42 ] ) we may compare its small @xmath49 expansion with the asymptotic expansion of spinor heat kernel in the case of a general curved manifold with a conical singularity derived in . setting @xmath107 in ( [ 42 ] ) and changing @xmath142 we get @xmath143 \ , \ ] ] so that expanding in small @xmath49 gives @xmath144 in agreement with general expressions in . as in the scalar case , we may also match ( [ 45 ] ) with the small @xmath49 expansion of the the massless spinor heat kernel on @xmath105 in corresponding to the special case of @xmath145 ( cf .
( [ 30 ] ) ) @xmath146
let us now apply the above results for heat kernels ( [ 28 ] ) and ( [ 42 ] ) to compute the string partition function ( [ 0]),([b1 ] ) on the cone @xmath54 . using the explicit values of masses ( @xmath147 and @xmath148 ) in ( [ b1 ] ) and ( [ ab1 ] ) we get & & _ 1 ( ) = _ 1(1 ) + ( ) , _
1(1 ) = ( 2 ) , + & & ( ) = - ^ _ _ 0^ dy e^- ( 4t)^1/2 h(t , y ; ) , + & & h(t , y ; ) = 1y . here
we isolated the contribution @xmath149 of the first ( @xmath2 ) terms in ( [ 28 ] ) and ( [ 42 ] ) where @xmath150 given by ( [ 4 ] ) was already computed in .
the first term in the bracket in @xmath151 is the combined contribution of 5 + 3 scalars and the second term is the contribution of 8 fermions ( taken with the minus sign as required in ( [ b1 ] ) ) .
both vanish for @xmath84 , i.e. @xmath152 . the proper time integral is actually convergent at @xmath109 as one can see from the small @xmath49 expansion of the heat kernels in ( [ 31a ] ) and ( [ 45 ] ) or directly from ( [ 522]),([52 ] ) after setting @xmath153 to get @xmath154 : h(t , t u ; ) |_t0 = - ( - ^-3 ) u^2 t + o(t^2 ) .
the resulting cancellation of 2d uv divergences in ( [ 522 ] ) is an important check of the consistency of this result as the ( non - trivial part of the ) 1-loop correction in superstring should be finite on generic 2d background .
we can thus remove the proper - time ( uv ) cutoff @xmath155 in ( [ 522 ] ) and compute the finite integral over @xmath49 ( keeping @xmath49 and @xmath131 as integration variables and using that @xmath156 ) ( ) = - _ 0^ dyy y .
this integral is convergent at both @xmath139 and @xmath157 as for small @xmath131 the integrand goes as @xmath158 .
it is not clear how to compute the integral ( [ 54 ] ) analytically .
for some integer values of @xmath42 it is given by a combination of values of the riemann @xmath159-function and polygamma functions , but we could not find an interpolating expression .
it is of course straightforward to evaluate it numerically and conclude that for @xmath160 it is different from the earlier result ( [ b2 ] ) of and also from the gauge - theory @xmath161 behaviour in ( [ b3 ] ) .
we illustrate the difference between @xmath162 in ( [ 51 ] ) and @xmath163 in ( [ b2 ] ) in figure 1 . for large @xmath42
we find that @xmath164 grows linearly , @xmath165 , while for small @xmath42 we get @xmath166 . in ( [ 51 ] ) ( black dots ) and @xmath167 in ( [ b2 ] ) ( blue line ) ]
a.a.t . thanks e. buchbinder for a collaboration at an initial stage and useful discussions .
he also acknowledges discussions with v. forini , d. fursaev , m. kruczenski , s. solodukhin and a. tirziu . the work of a.a.t .
is supported by the erc advanced grant no.290456 , the stfc grant st / j0003533/1 and by the russian science foundation grant 14 - 42 - 00047 associated with lebedev institute .
let us consider a @xmath168-dimensional space with a radial coordinate @xmath169 and a polar angle @xmath170 of period @xmath171 with metric near @xmath16 being @xmath172 so that there is a conical singularity with angular deficit @xmath173 .
given the expression for the heat kernel of a scalar or spinor laplacian on a regular manifold ( @xmath84 ) it is possible to derive its counterpart on the space with @xmath174 using so called sommerfeld formula .
we shall review this construction below .
let us consider first the case of flat 2d cone @xmath175 to construct the heat kernel for @xmath176 on this cone it is sufficient to consider the two points @xmath177 and @xmath178 separated only in the angular direction by angle @xmath170 .
the starting point is the heat kernel for @xmath84 , i.e. on a regular @xmath179 plane , where we have for @xmath180 @xmath181 since this is a smooth function of @xmath170 , we can rewrite it using the cauchy representation : @xmath182 where @xmath183 is a contour in the complex plane which encircles the point @xmath184 .
now let us deform @xmath183 to a family of contours @xmath185 , where @xmath186 goes from @xmath187 to @xmath188 in the higher half plane and from @xmath189 to @xmath190 in the lower half plane one .
then the heat kernel can be expressed as the sum over images " k(x(),x(0);t ) = _ n=-^ _ a_nk(x(z),x(0);t ) + = _ n=-^ _ a_0k(x(z),x(0);t ) = _ a_0 dz k(x(z),x(0);t ) .
here we used the periodicity of the kernel in ( [ a2 ] ) and the identity @xmath191 we can now repeat the same procedure changing the period of the angle from @xmath192 to @xmath193 to get the heat kernel on the cone ( [ a1 ] ) @xmath194 or after shifting @xmath195 @xmath196 the contour @xmath197 is obtained from @xmath198 by a shift by @xmath170 and consists of two pieces : one in the upper half plane running from @xmath199 to @xmath200 and one in the lower half plane going from @xmath201 to @xmath202 .
we may then deform @xmath197 into a small circle around @xmath90 and another contour @xmath203 made of two vertical lines : one going from @xmath204 to @xmath205 and one from @xmath206 to @xmath207 .
near @xmath90 the integrand has a pole of @xmath208 with residue being the regular plane kernel @xmath209 .
then denoting @xmath210 we finish with @xmath211 this is the sommerfeld formula for the heat kernel of the scalar laplacian on the cone ( [ a1 ] ) in terms of the heat kernel on the regular 2-plane .
the same procedure can be applied to the case of an arbitrary space @xmath212 where the heat kernel depends only on the angular distance between two points .
denoting @xmath61 , @xmath213 the coordinates of two points on the corresponding cone @xmath214 of @xmath212 , the sommerfeld formula for the heat kernel of the scalar laplacian reads @xmath215 here the angular distance is @xmath216 and the contour @xmath66 is formed by two vertical lines from @xmath217 to @xmath218 and from @xmath219 to @xmath220 .
taking the coincident points limit we get @xmath221 defining the integrated heat kernel @xmath222 we finally get the expression for the trace of heat kernel _
( t)=_m _ d^2 x g x| e^- t | x = k(0 ; t ) + _ dz k(z;t ) .
since in ( [ a9 ] ) @xmath223 , here the contour @xmath66 reduces to two vertical lines going from @xmath224 to @xmath225 and from @xmath226 to @xmath70 .
thus in addition to the expected term @xmath227 inherited " from the regular space case there is a correction given by a contour integral on the complex plane . for @xmath84
this second term vanishes as expected : the integrand becomes @xmath228 periodic in @xmath65 and the contributions coming from the integration along the two vertical lines at @xmath229 and @xmath230 cancel each other .
let us start again with a flat space cone .
on @xmath179 in cartesian coordinates the spin connection is trivial and the fermionic laplace operator @xmath231 reduces to the scalar laplacian @xmath232 .
thus solution to the heat kernel equation is the same as for the scalar ( [ a2 ] ) up to @xmath116 identity matrix @xmath233 : @xmath234 in polar coordinates @xmath235 the spinor covariant derivative is @xmath236 with @xmath237 .
thus @xmath238 then the solution to the spinor heat kernel equation in polar coordinates may be written as @xmath239 with @xmath240 given by ( [ a10 ] ) . due to
the presence of the spin factor @xmath241 the spinor heat kernel is antiperiodic in @xmath170 . to get spinor heat kernel on a cone of @xmath179 we follow the same route as in the scalar case : we choose @xmath242 , @xmath243 and first rewrite @xmath244 in ( [ a12 ] ) using the method of the images as in ( [ a22]),([a3 ] ) : & & k ( x(),x(0);t)= _ c ( z ) k(x(z),x(0);t ) + & & =
_ n=-^ _ a_n ( z ) k(x(z),x(0);t ) + & & = _
n=-^ _ a_0 ( z ) k(x(z),x(0);t ) + & & = _ a_0 dz ( z ) k(x(z),x(0);t ) .
here we used the antiperiodicity of @xmath241 and @xmath245 .
we can now change the periodicity of @xmath170 from @xmath192 to @xmath246 and to get the expression for the spinor heat kernel on the cone of @xmath179 k _
( x(),x(0);t)&= & _ a_0 dz k(x(z),x(0);t ) + & = & _ a_0^ dz k(x(z+),x(0);t ) , where @xmath197 is again given by two lines in the upper and lower half plane : one running from @xmath199 to @xmath200 and the other from @xmath201 to @xmath202 .
next , we deform @xmath197 into a small circle around @xmath90 and a contour @xmath203 made of two vertical lines that run from @xmath204 to @xmath205 and from @xmath206 to @xmath207 . setting @xmath247 we obtain @xmath248 this expression
can be generalized to the case of a cone of curved 2d space where the heat kernel depends on the angular distance between two points @xmath61 and @xmath213 @xmath249 here , as in the scalar case in ( [ a7 ] ) , @xmath66 is modified by @xmath250 . defining the integrated kernel as in ( [ a88 ] )
we then get the final expression for the traced spinor heat kernel : @xmath251 here @xmath252 is over the spinor indices and the integration contour reduces for @xmath253 to the two vertical lines going from @xmath224 to @xmath225 and from @xmath226 to @xmath70 . for @xmath84
the integrand becomes periodic ( as the product of two antiperiodic functions in @xmath65 ) and thus the integral contribution vanishes .
here we solve the equation ( [ 32 ] ) for the matrix @xmath51 in the @xmath2 metric ( [ aa ] ) written in polar coordinates .
the corresponding zweibein and tangent space and spinor connections are ( @xmath254 ) & & e^1_=(1,0 ) , e^2_=(0 , ) , _ ^=0 , _ ^12=-_^21=- , + & & _ = 1 2 _ ^ _ = ( 0,- _ 3 ) , _ = 14 [ _ , _ ] , ^= ( _ 1,_2 ) . in @xcite
it was shown that ( [ 32 ] ) is equivalent to the requirement that the spinor covariant derivative acts on @xmath255 in the following way @xmath256 where the geodesic distance between the two points @xmath257 is given in ( [ aaa ] ) .
this equation for @xmath51 can be rewritten as _
_ 3u , = ( - _ d ) _ 3 u .
the first equation in ( [ 35b ] ) has solution [ 36b ] & & u(,^,,^)= c(,^ ) , + & & b(,^,,^)= _ . using ( [ aaa ] ) we find b&= & + & = & 2 - .
the matrix @xmath258 can be determined from the second equation in ( [ 35b ] ) _ c= _ 3 ( - _ d-_b ) c = 0 , where we used ( [ aaa ] ) and ( [ 3888 ] ) .
thus @xmath183 is constant and is fixed by the initial condition @xmath259 @xmath260 as a result , @xmath261 j. k. erickson , g. w. semenoff and k. zarembo , `` wilson loops in n=4 supersymmetric yang - mills theory , '' nucl .
b * 582 * , 155 ( 2000 ) [ hep - th/0003055 ] .
g. w. semenoff and k. zarembo , `` wilson loops in sym theory : from weak to strong coupling , '' nucl .
suppl . * 108 * , 106 ( 2002 ) [ hep - th/0202156 ] .
n. drukker , d.j .
gross and a.a .
tseytlin , `` green - schwarz string in ads(5 ) x s5 : semiclassical partition function , '' jhep * 0004 * ( 2000 ) 021 [ hep - th/0001204 ] .
s. forste , d. ghoshal and s. theisen , `` stringy corrections to the wilson loop in n=4 superyang - mills theory , '' jhep * 9908 * , 013 ( 1999 ) [ hep - th/9903042 ] .
i. chavel ,
heat kernels and spectral theory " , cambridge u.p . ,
r. camporesi , `` harmonic analysis and propagators on homogeneous spaces , '' phys .
* 196 * ( 1990 ) 1 .
d. fursaev and d. vassilevich , `` operators , geometry and quanta : methods of spectral geometry in quantum field theory , '' d. v. vassilevich , `` heat kernel expansion : user s manual , '' phys .
rept . * 388 * , 279 ( 2003 ) [ hep - th/0306138 ] . c. kristjansen and y. makeenko , `` more about one - loop effective action of open superstring in @xmath262 , '' jhep * 1209 * ( 2012 )
053 [ arxiv:1206.5660 ] . s. n. solodukhin , `` entanglement entropy , conformal invariance and extrinsic geometry , '' phys .
b * 665 * , 305 ( 2008 ) [ arxiv:0802.3117 ] .
h. casini and m. huerta , `` entanglement entropy for the n - sphere , '' phys .
b * 694 * , 167 ( 2010 ) [ arxiv:1007.1813 [ hep - th ] ] .
i. r. klebanov , s. s. pufu , s. sachdev and b. r. safdi , `` renyi entropies for free field theories , '' jhep * 1204 * , 074 ( 2012 ) [ arxiv:1111.6290 ] .
r. aros , f. bugini and d. e. diaz , `` on renyi entropy for free conformal fields : holographic and q - analog recipes , '' j. phys .
a * 48 * , 105401 ( 2015 ) [ arxiv:1408.1931 ] .
r. k. gupta , s. lal and s. thakur , `` heat kernels on the @xmath2 cone and logarithmic corrections to extremal black hole entropy , '' jhep * 1403 * ( 2014 ) 043 [ arxiv:1311.6286 ] .
`` logarithmic corrections to extremal black hole entropy in n = 2 , 4 and 8 supergravity , '' arxiv:1402.2441 . j. s. dowker , `` quantum field theory on a cone , '' j. phys . a * 10 * , 115 ( 1977 ) .
`` vacuum averages for arbitrary spin around a cosmic string , '' phys .
d * 36 * , 3742 ( 1987 ) .
d. v. fursaev , `` the heat kernel expansion on a cone and quantum fields near cosmic strings , '' class .
grav . *
11 * , 1431 ( 1994 ) [ hep - th/9309050 ] .
j. s. dowker , `` effective actions with fixed points , '' phys . rev .
d * 50 * , 6369 ( 1994 ) [ hep - th/9406144 ] .
r. camporesi , `` the spinor heat kernel in maximally symmetric spaces , '' commun .
* 148 * ( 1992 ) 283 .
r. camporesi and a. higuchi , `` spectral functions and zeta functions in hyperbolic spaces , '' j. math .
* 35 * ( 1994 ) 4217 .
d. v. fursaev and g. miele , `` cones , spins and heat kernels , '' nucl .
b * 484 * , 697 ( 1997 ) [ hep - th/9605153 ] .
d. v. fursaev , `` euclidean and canonical formulations of statistical mechanics in the presence of killing horizons , '' nucl .
b * 524 * , 447 ( 1998 ) [ hep - th/9709213 ] . | we compute the one - loop correction to partition function of @xmath0 superstring that should be representing @xmath1-fundamental circular wilson loop in planar limit .
the 2d metric of the minimal surface ending on @xmath1-wound circle at the boundary is that of a cone of @xmath2 with deficit @xmath3 .
we compute the determinants of 2d fluctuation operators by first constructing heat kernels of scalar and spinor laplacians on the cone using sommerfeld formula .
the final expression for the @xmath1-dependent part of the one - loop correction has simple integral representation but is different from earlier results .
submission data authors : r. bergamin and a. a. tseytlin report no .
: imperial - tp - at-2015 - 07 title : comments : pages .
abstract : tr a s |
during the last decade the heavy hadron spectroscopy has been a field of permanent and growing interest .
this interest dictated by the exciting experimental results obtained in regard to this subject . at present all ground state baryons containing a single charm quark have already been observed and their masses are also measured . many of the spin-1/2 hadrons with a single bottom quark , such as @xmath3 , @xmath4 , @xmath5 and @xmath6 , and spin-3/2 hadrons such as @xmath7 have also been observed @xcite . recently
, spin-3/2 @xmath8 baryon has been discovered @xcite , and latest measurement of the life - time of @xmath3 baryon has been announced @xcite . on the other hand
the experimental situation with two and three heavy quarks is not reach yet .
only selex collaboration announced the observation of the spin-1/2 @xmath9 baryon @xcite , which is not confirmed yet by the other collaborations . experimental observation and investigation of the properties of doubly and triply heavy baryons constitute one of the most promising research area in particle physics .
one of the main static quantities which could give valuable information about the internal structure of baryons is their magnetic moments .
magnetic moment of heavy baryons have extensively been investigated within the framework of different approaches like , naive quark model , chiral quark model , non - relativistic quark model and qcd sum rules method .
the main advantage of the qcd sum rules method compared to the other non - perturbative approaches is that , it is based on the fundamental qcd lagrangian , and it takes into account the non - perturbative nature of the qcd vacuum . in order to solve the problems which are inherit in the traditional qcd sum rules method , the light cone version of the qcd sum rules is proposed ( for more about this method , see for example @xcite ) . in this version of the qcd sum rules ,
the operator product expansion is performed over twists of the operators and non - perturbative effects are all encoded into the matrix elements of the non - local operators between the vacuum and one - particle states .
the light cone qcd sum rules ( lcsr ) has so far been applied to many problems of the hadron physics ( for the very recent applications of the lcsr , see for example @xcite ) . in the present work
we calculate the @xmath0-@xmath1 transition magnetic moment within the qcd sum rules method .
note that the @xmath10-@xmath11 transition magnetic moment was studied within the same framework in @xcite .
the paper is organized as follows . in section 2 the sum rules for the @xmath0-@xmath1 transition magnetic moment are obtained .
the following section contains numerical calculations , discussion and comparison with the predictions of the other theoretical methods existing in literature .
we start this section by summarizing the useful information on @xmath12 classification of the heavy hadrons with single heavy quarks .
these baryons belong to either @xmath12 antisymmetric @xmath13 or symmetric @xmath14 flavor representations .
it is well known that , in @xmath14 representation the ground state baryons must have total spin 1 , while their total spin is 0 in @xmath13 representation .
therefore , baryons in @xmath14 representation can have spin-1/2 or -3/2 , but hadrons in @xmath13 representation can have only spin-1/2 . in the present work we consider only spin-1/2 baryons from both representations ,
i.e. , we consider @xmath0 from @xmath14 and @xmath1 from @xmath13 representations , respectively . following this brief information , we can proceed now to calculate the @xmath0-@xmath1 transition magnetic moment within lcsr .
for this aim we consider the following correlation function : [ etrn01 ] = id^x e^ipx 0 \ { _ _ q ( x ) |__q^ ( 0 ) } 0 _ , where @xmath15 is the external electromagnetic field ; @xmath16 and @xmath17 are the interpolating currents of @xmath1 and @xmath0 baryons .
the general forms of the interpolating currents of @xmath1 and @xmath0 baryons are as follows : [ etrn02 ] _ _ q ^abc \ { 2 ( s^at c q^b ) _ 5 q^c + ( s^at c q^b ) _ 5 q^c - ( q^at c q^b ) _ 5 s^c .
+ .2 t ( s^at c _ 5 q^b ) q^c + t ( s^at c _ 5 q^b ) q^c - t ( q^at c _ 5 q^b ) s^c } , + + _ _
q^ ^abc \ { ( s^at c q^b ) _ 5 q^c + ( q^at c q^b ) _ 5 s^c .
t ( s^at c _ 5 q^b ) q^c + t ( q^at c _ 5 q^b ) s^c } , where @xmath18 are the color indices ; @xmath19 or @xmath20 , and @xmath21 or @xmath22 quark ; @xmath23 is the charge conjugation operator ; and @xmath24 is the arbitrary parameter . it should be noted here that , @xmath1 and @xmath0 baryons do have the same quantum numbers and in principle there can be mixing between them .
this mixing can be implemented by modifying the interpolating currents for physical states as follows : [ etrn03 ] _ _ q^phys _ q _ _ q + _ q _ _
q^ , + _ _ q^^phys - _ q _ _ q + _ q _ _ q^ .
it is shown in @xcite that the mixing angle between @xmath25 and @xmath26 is equal to @xmath27 ( @xmath28 ) , which is quite small , and hence we can safely neglect it . according to sum rules method philosophy the correlation function is calculated in terms of hadrons , and in terms of quarks and hadrons in two different kinematical regions .
equating then the obtained results to each other one can get the sum rules for the appropriate physical quantity .
we start calculating the correlation function from the hadronic side . inserting complete set of hadrons with @xmath16 and @xmath17 state quantum numbers , and isolating the ground state contributions ,
we get : [ etrn04 ] _
q(p_1 ) . _
q^(p_2 ) _ + , where @xmath29 , @xmath30 and @xmath31 is the photon momentum ; and dots represent the contributions of higher states and continuum .
the matrix element of the interpolating current between one baryon and vacuum states is defined in the following way : [ etrn05 ] 0 _ _ q _
q(p ) _ _ q u__q(p ) , + _
q^(p)|__q^ 0__q^ |u__q^(p ) .
the second matrix element in ( [ etrn04 ] ) is parametrized in terms of two form factors @xmath32 and @xmath33 as : [ etrn06 ] _
q(p_1 ) . _
|u(p_1 ) u(p_2 ) ^ , + using eqs . ( [ etrn05 ] ) and ( [ etrn06 ] ) in hadronic side of the correlation function
, we get the following expression : [ etrn07 ] = _ _ q _ _
q^ m__q^2 - p_1 ^ 2 ( /p_1+m__q)m__q^^2-p_2 ^ 2 \{(f_1+f_2 ) _ - f_2 ( p_1+p_2 ) _ } ( /p_2+m__q^)^ .
we see from this expression that there are many structures appearing in the phenomenological part of the correlation function in calculation of the transition magnetic moment . in the present work we chose the structure @xmath34 in determining the transition magnetic form factor @xmath35 . at @xmath36
this combination gives the transition magnetic moments in natural units , .i.e .
, @xmath37 .
the coefficient of the structure @xmath38 gives the expression of the invariant function for the transition magnetic moment @xmath39 in the following form : [ etrn08 ] = _ _ q _ _ q^ ( m__q^2 - p_1 ^ 2 ) ( m__q^^2 - p_2 ^ 2 ) _ _ q _
theoretical part of the correlation function is calculated in the following way . at the first step
the correlation function is written in terms the quark operators using the wick theorem .
the correlation function contains two parts , namely , photon interacting with quarks perturbatively , and photon interacting with quarks non - perturbatively . in order to calculate the perturbative contribution
it is enough replacing one of the three propagators with [ etrn09 ] s -12 dy s^free ( x - y ) _ s^free ( y ) y _ _ , and the remaining two quark propagators are taken as free quark propagators . in eq .
( [ etrn09 ] ) the fixed point gauge ( the so - called fock - schwinger gauge ) is used , i.e. , @xmath40 , and @xmath41 is the free quark operator .
free quark operator for the light and heavy quarks is given as : [ etrn10 ] s^free(x ) - m_q4 ^2 x^2 + s_q^free(x ) -
i m_q^2 /x 4 ^2 x^2 k_2(m_q ) , respectively , where @xmath42 are the modified bessel function of the second kind ; and @xmath43 and @xmath44 are the masses of the light and heavy quarks , respectively . following the above - explained steps , the perturbative part of the correlation function can be calculated straightforwardly .
the non - perturbative contribution to the correlation function can be calculated by replacing one of the light quark propagators with [ etrn11 ] s_^ab = - 14 |q^a _
i q^b ( _ i ) _ , where @xmath45 is the full set of the dirac matrices , i.e. , @xmath46 ; and the remaining two propagators are taken as full quark propagators .
so , for calculation of the perturbative and non - perturbative parts of the correlation function we need the expressions of the light and heavy propagators in external field .
the light cone expansion of the propagator in external field is performed in @xcite and it receives contributions from non - local four - quark @xmath47 , three - particle @xmath48 , four - particle @xmath49 , etc .
operators , where @xmath50 is the gluon field strength tensor . in the present work
we take into account contributions coming only from non - local operators with one gluon .
the contributions of four - quark and two - gluon - two - quark operators are neglected due to their small contributions . under these approximations
the heavy and light quark operators , in presence of the external field , have the following forms : [ etrn12 ] i s_light(x ) i s_light^free(x ) + ( 1 - m_0 ^ 2 x^2 16 ) - i g_s _ 0 ^ 1 du , + + i s_heavy(x ) i s_heavy^free(x ) + ig_s e^-ikx _ 0 ^ 1 du .
it follows from eqs .
( [ etrn11 ] ) and ( [ etrn12 ] ) that , in order to calculate the non - perturbative contribution to the correlation function we need to know the following matrix elements @xmath51 and @xmath52 .
these matrix elements are defined in terms of the photon distribution amplitudes ( das ) , whose expressions all can be found in @xcite . using eq .
( [ etrn12 ] ) and definitions of the photon das one can calculate the theoretical part of the correlation function . the sum rules for the magnetic moment of @xmath0-@xmath1 transition is obtained by equating the coefficients of the structure @xmath34 from both sides of the correlation function .
the final step in deriving the sum rules for the magnetic moment of the @xmath0-@xmath1 transition is performing double borel transformation over the variables @xmath53 and @xmath54 in order to suppress higher state and continuum contributions . finally we get : [ etrn13 ] _ _ q _ _ q^ _ _ q _ q^ = e^(m__q^2/m_1 ^ 2 + m__q^^2/m_2 ^ 2 ) ^theor .
the expression for @xmath55 is rather lengthy , therefore we do not present it here . it should be noted here that , since the masses of @xmath1 and @xmath0 are very close to each other , we can set @xmath56 and @xmath57 .
it follows from eq .
( [ etrn13 ] ) that in calculation of the transition magnetic moment @xmath39 we need to know the residues of the @xmath1 and @xmath0 baryons .
the residues of sextet and anti - triplet heavy baryons have already been calculated in @xcite , and for this reason we do not present them in this work .
in this section we perform numerical analysis of the sum rules for the transition magnetic moment obtained in the previous section .
the main input parameters of the sum rules are photon das , whose explicit expressions are given @xcite .
the remaining input parameters which we use in the numerical analysis of the sum rules are : @xmath58 , @xmath59 , @xmath60 @xcite , the magnetic susceptibility @xmath61 @xcite .
the sum rules contain the following auxiliary parameters : borel mass square @xmath62 , continuum threshold @xmath63 , and the arbitrary parameter @xmath64 in the interpolating current .
obviously one expects the transition magnetic moment be independent of these parameters .
for this reason in further numerical analysis we shall look for the regions of these parameters where magnetic moment is practically independent of them .
the domain of @xmath62 is determined from the following two conditions : \(i ) the ground state contribution should be larger compared to higher states and the continuum contributions \(ii ) the highest terms in @xmath65 should constitute , say about 30% - 40% of the higher power @xmath62 terms .
our numerical analysis shows that both conditions are satisfied when @xmath66 for the @xmath5 baryon , and @xmath67 for the @xmath68 baryon , respectively .
the working regions of @xmath63 and @xmath64 are also determined from the consideration that the magnetic moment should not change appreciably in the respective regions of these parameters . the dependence of the magnetic moments for the @xmath69-@xmath70 , @xmath71-@xmath72 , @xmath73-@xmath74 and @xmath75-@xmath76 transitions on @xmath77 , where @xmath78 , at fixed values of @xmath62 and @xmath63 are presented in figs .
( 1)-(4 ) , respectively .
we see from these figures that the working region for @xmath64 in the @xmath79-baryon case is @xmath80 , and for the baryons containing @xmath22-quark it is @xmath81 .
we deduce from these figures the following values for the @xmath82-@xmath83 transition magnetic moments : @xmath84 , @xmath85 , @xmath86 and @xmath87 , where @xmath88 is the nuclear magneton . in the case of @xmath79-baryons
the change in the values of the magnetic moments is about 10% if @xmath63 varies in the domain @xmath89 .
hence we can safely be say that the results for the magnetic moments are stable with respect to the variation of @xmath63 in the determined domain . in table 1
we present our results on the transition magnetic moments together with the ones predicted by different approaches , such as effective quark mass and screened quark charge scheme ( for @xmath79-hadron sector see @xcite , for @xmath22-hadron sector see @xcite ) , bag model @xcite , non - relativistic quark model @xcite , and chiral constituent quark model @xcite .
@xmath90 from a comparison of our results with the predictions of the above - mentioned approaches we see that for the @xmath91 transition all results are very close to each other . in the case of @xmath92 transition , except the bag model the results of all other approaches are in good agreement .
our prediction for the @xmath93 transition is very close to the results of @xcite and @xcite , but considerably different from that of @xcite . as far as @xmath94 transition
is concerned , our result confirms the prediction of @xcite , it is approximately 1.5 times smaller than that of @xcite , two times larger than that of @xcite and 10 times larger than the results predicted in @xcite .
finally we express our concluding remarks about the present work , in which we calculate the magnetic moments of the @xmath95 ( @xmath21 or @xmath22 ) transitions within the framework of light cone qcd sum rules .
we also perform a comparison of our results with the predictions of various approaches existing in literature .
we observe that , for the @xmath96 and @xmath97 transitions the predictions of all approaches do practically agree with each other .
similar situation takes place for the @xmath93 transition , except the bag model prediction . as it comes to @xmath98 transition
, our results coincide with the ones predicted in @xcite , but differ substantially with the predictions of all other approaches . at present
, these magnetic moments have not yet been measured in experiments , and we hope that they all can be measured in future planned experiments .
we believe that the measurement of @xmath95 transition magnetic moments will also be very useful in determination of the mixing angle among heavy baryons
. 99 j. beringer _
( particle data group ) , phys . rev .
d * 86 * , 010001 ( 2012 ) .
t. aaltonen _
( cdf collaboration ) , phys .
* 107 * , 102001 ( 2011 ) ; ( cdf collaboration ) , phys .
d * 85 * , 092011 ( 2012 ) .
et al . _ ( belle collaboration ) , phys .
lett . * 94 * , 122002 ( 2005 ) ; b. aubert _
( babar collaboration ) , phys .
lett . * 98 * , 012001 ( 2007 ) .
s. chatrchyan _
( cms collaboration ) , phys .
lett . * 108 * , 252002 ( 2012 ) .
s. chatrchyan _
( cms collaboration ) , j. high energy phys .
07 ( 2013 ) 163 .
a. ocherashvili _
( selex collaboration ) , phys .
b * 628 * , 18 ( 2005 ) ; j. engelfried _ et al . _
( selex collaboration ) , nucl .
a752 * , 121 ( 2005 ) .
v. m. braun , arxiv:9801222 .
t. m. aliev , k. azizi , m. savci , phys .
b * 723 * , 145 ( 2013 ) .
t. m. aliev , y. ktem , m. savci , phys .
d * 88 * , 036002 ( 2013 ) .
t. m. aliev , k. azizi , m. savci , phys .
d * 87 * , 096013 ( 2013 ) .
t. m. aliev , a. zpineci , m. savci , phys .
d * 65 * , 096004 ( 2002 ) .
t. m. aliev , a. zpineci , v. s. zamiralov , phys .
d * 83 * , 016008 ( 2011 ) .
i. i. balitsky , v .
m. braun , nucl .
b311 * , 541 ( 1989 ) . p. ball , v. m. braun , and n. kivel , nucl .
b * 649 * , 263 ( 2003 ) .
t. m. aliev , k. azizi , a. zpineci , phys .
d * 79 * , 056005 ( 2009 ) . v. m. belyaev , b. l. ioffe , sov .
jetp * 56 * , 493 ( 1982 ) .
j. rohrwild , j. high energy phys . 09 ( 2007 ) 073 .
r. dhir , c. s. kim , and r. c. verma , phys .
d * 88 * , 094002 ( 2013 ) .
s. kumar , r. dhir , and r.c .
verma , j. phys .
g * 31 * , 141 ( 2005 ) .
a. bernotas , v. simonis , arxiv:1209.2900 .
j. franklin , d. b. lichtenberg , w. namgung , and d. carydas , phys .
d * 24 * , 2910 ( 1981 ) .
n. sharma , h. dahiya , p. k. chatley , and m. gupta , phys .
d * 81 * , 073001 ( 2010 ) .
1 * the dependence of the transition magnetic moment @xmath99 on @xmath77 at several fixed values of the borel mass - square @xmath62 , and at fixed value of the continuum threshold @xmath100 . + + * fig .
2 * the same as in fig . 1 , but for the @xmath101 transition . + + * fig .
3 * the dependence of the transition magnetic moment @xmath102 on @xmath77 at several fixed values of the borel mass - square @xmath62 , and at fixed value of the continuum threshold @xmath103 . + + * fig .
4 * the same as in fig . 1 , but for the @xmath104 transition . | the @xmath0-@xmath1 transition magnetic moments are calculated in framework of the light cone qcd sum rules method ( lcsr ) .
the values of the transition magnetic moments obtained are compared with the predictions of the other theoretical approaches .
# 1#2#3 @xmath2 # 1#2#3 @xmath2 0^*0 5_5 q 5_5 o _ ^0 _ pacs numbers : 11.55.hx , 13.25.hw , 13.25.jx |
the problem of laser light compton scattering on high - energy electrons are now considered in the projects relating to the creation of @xmath1 colliders , laser - synchrotron sources , laser cooling , diagnostic of sub - picosecond electron bunches and others @xcite .
it is supposed that intensity of laser flash in these problems is so high that an electron can undergo several successive collisions passing through a photon bunch @xcite .
the distributions over the electrons energy and collisions number , the moments of these distributions , and the spectra of scattered photons have been studied in this paper using corresponding kinetic equations and by statistical simulation methods .
it is seen from this analysis that multiple scattering of electrons influences on the spectra both electrons and photons even for the photon target of small thickness .
penetration of an electron through a photon target is a stochastic process , where both the number of collisions of each electron with photons and the energy loss in individual collision are random .
let us consider an electron with energy @xmath2 traveling through a bunch of photons with energy @xmath3 .
the typical value of the scattering electron angle in the compton back - scattering process is determined by the laser photon energy @xmath3 and does nt depend on the electron energy @xmath2 : @xmath4 @xmath5 being the rest energy of electron .
we shall consider the head - on collisions of laser photons with energy @xmath6 ev and electrons with energy @xmath7 gev .
in this case the electron deflection angle is much less than characteristic radiation angle @xmath8 and one can neglect the electron angular deflection ( the straight - ahead approximation ) . in this approximation
the probability to undergo @xmath9 collision along the pass @xmath10 , @xmath11 , obeys the adjoint balance equation ( the kolmogorov - chapman equation ) @xcite : @xmath12 @xmath13 and @xmath14 being the total and differential macroscopic cross - sections of the compton scattering , @xmath15 is a small part of @xmath10 , and @xmath16 is the energy of scattered photon @xmath17 @xmath18 is the maximum value of the scattered photon energy , @xmath19 @xmath20 @xmath21 @xmath22 and @xmath23 are the total and differential cross - sections , @xmath24 is the concentration of laser photons in a bunch .
note that @xmath25 is the mean number of collisions of an electron per unit path length and @xmath14 is the mean number of an electron collisions with energy loss in unit interval about @xmath16 per unit path length .
the first term in the right side of eq . ( [ sec : pn ] ) corresponds to electrons which pass the path @xmath26 without collisions and @xmath27 is corresponding probability .
these electrons have to undergo @xmath28 collisions along the rest path @xmath29 the second term corresponds to the electrons which undergo the first scattering passing the path @xmath26 and @xmath30 is corresponding probability .
these electrons have to undergo @xmath31 collisions after that but the energy of electron after first scattering equals @xmath32 , where @xmath33 is random energy of the scattered photon and @xmath34 is the probability density function of @xmath16 . in the limit @xmath35 eq .
( [ sec : pn ] ) gives the integro - differential equation for @xmath36 @xcite : @xmath37 with boundary condition @xmath38 in a similar way one can obtain the kinetic equation for the probability density function @xmath39 describing the energy distribution of electrons after travelling the path @xmath10 : @xmath40 with boundary condition @xmath41 @xmath42 being the dirac @xmath43-function .
( [ dpn ] ) , ( [ pq ] ) can be transformed into the equations for the moments of distributions @xmath44 : @xmath45 @xmath46 the equation for @xmath0 and @xmath47 has a form @xmath48 @xmath49 the boundary conditions for the moments are @xmath50 @xmath51 if the relative energy loss of an electron in one collision is small the integro - differential equations ( [ bar_n ] ) and ( [ q1 ] ) can be transformed by the taylor expansion of integrands : @xmath52 @xmath53 this gives the partial differential equations : @xmath54 @xmath55 the quantities @xmath25 , @xmath56 , and @xmath57 in ( [ n_csd ] ) , ( [ q1csd ] ) are the moments of the macroscopic differential cross - section : @xmath58 @xmath59 @xmath60 the eq .
( [ q1csd ] ) for the second moment @xmath61 can be transformed into the equation for the variance @xmath62 this equation is @xmath63
the linear compton scattering differential cross - section is @xmath64 where @xmath65 , and @xmath66 is the classical radius of electron . in the energy region of our interest
the invariant dimensionless parameter @xmath67 is small and the integral interaction coefficients @xmath25 , @xmath56 , and @xmath57 are described by the approximate formulas @xmath68 @xmath69 @xmath70 where @xmath71 the quantities @xmath72 and @xmath73 are the mean energy loss and the mean squared energy loss of an electron per unit path length .
the partial differential equations ( [ n_csd ] ) , ( [ q1csd ] ) , and ( [ delta ] ) can be solved by the iteration method . in the first approximation , where the terms with the second derivation are neglected , @xmath74 @xmath75 @xmath76 the second term in eq .
( [ n - sol ] ) is due to increasing of the interaction cross - section because of electron energy loss in collisions .
but it is seen from the equation that this effect can be neglected in the energy region under consideration and the problem can be solved in one - velocity @xmath77 approximation .
it follows from eq .
( [ var ] ) that the variance @xmath78 has a maximum at the point , where @xmath79 and @xmath80 using eqs .
( [ var ] ) and ( [ eps ] ) one can derive the formula @xmath81 which was earlier obtained in @xcite .
it can be shown that in the one - velocity approximation the solution of eq .
( [ pq ] ) is the sum of the terms corresponding to nonscattered and scattered electrons : @xmath82 @xmath83 where @xmath84 is the poisson distribution with the mean value @xmath85 and the function @xmath86 obeys the recurrent convolution formula @xmath87 where @xmath88 is the probability density function of @xmath16 and @xmath89 it should be pointed out that @xmath86 is the energy distribution of n - scattered electrons .
decreasing of electrons energy due to their repeated collisions changes the energy spectrum of scattered photons .
the photons produced in the secondary compton scatterings are softer and the resulting spectrum can be written as the sum of terms corresponding to individual collisions : @xmath90 where @xmath91 is the spectrum of photons resulting from the @xmath92th electron collision and can be written in the form @xmath93
the results of analytical calculations above agree with the data of our statistical simulation for 10 gev electrons and 1 ev photon head - on collisions .
it was supposed in this simulation that the number of electron collisions with laser photons is random .
this number was selected from the poisson distribution with fixed @xmath94 .
the simulation of individual collisions was carried out in the electron rest frame using the klein and nishina formula with the lorentz transformation to the lab system . in the same way as in analytical calculations above we neglected the angular deflection of electrons but accounted for the energy decreasing after each collision .
all results were obtained with statistics more than @xmath95 trajectories .
the energy spectra of electrons @xmath86 after fixed number of collisions are given in fig .
( [ fig : un ] ) .
the data on the electrons energy distributions @xmath96 are given in fig .
( [ fig : pq ] ) for several values of @xmath97 it should be pointed out the discontinuous of the spectra for small @xmath0 at the point @xmath99 due to single scattered electrons and contribution of multiple scattered electrons at energies below the point of discontinuous .
this contribution exists even for small @xmath0 .
the energy spectra of photons @xmath91 from the @xmath92th scattering of electron and the resulting spectra for several @xmath0 are shown in fig .
( [ fig : phi - n ] ) and fig .
( [ fig : phi ] ) .
it is seen from fig .
( [ fig : phi ] ) that the contribution of multiple scattering should be taken into account even for the photon target of small thickness . fig .
( [ fig : collimator ] ) shows that the electrons multiple collisions influence on the spectra of photons with emission angle @xmath100 .
it is seen from the figure that multiple compton scattering results in broadening of the photon spectra with increasing of the photon target thickness . 8 a.d.angelo at al .
instr.meth._ 455 ( 2000 ) , 1 .
i.gizburg , g.kotkin , v.serbo , and v.telnov .
_ nucl.instr .
_ 285 ( 1983 ) 47 .
_ nucl.instr .
_ a 355 ( 1995 ) 3 .
_ nucl.instr .
meth . _ a 472 ( 2001 ) 43 .
w.feller , _ an introduction to probability theory and its application _ , vols 12 , wiley , new york , 1971 .
a.m.kolchuzhkin , v.v.uchaikin , _ introduction to the theory of particles penetration through a matter _ , ( book in russian ) .
moscow , atomizdat , 1978 .
a.kolchuzhkin , a.potylitsyn , a.bogdanov , i.tropin , physics let .
a 264 ( 1999 ) 202 .
_ , 78 ( 1997 ) 4757 . | the problem of stochastis in multiple electron - photon head - on collisions has been considered in this paper .
the kinetic equations for the distributions over the electron energy and collisions number along with the equations for these distributions moments have been obtained .
the equations for the first moments have been solved by the iteration method .
it has been shown that the variance of the energy distribution as a function of the mean number of collisions has a maximum at some value of @xmath0 .
it is seen from this analysis that multiple scattering of electrons influences on the spectra both electrons and photons even for the photon target of small thickness .
the data of approximate analytical calculations agree with the results of the monte carlo simulation .
pacs : 07.85.fv ; 13.60.fz ; 24.10.lx ; 41.75.ht , electron - photon head - on collision ; multiple energy loss ; kinetic equation ; monte carlo simulation . |
fusion categories over a field @xmath1 generalize categories of finite group representations ; they are tensor categories with finitely many classes of simple objects , all having duals .
they arise in representation theory@xcite , operator algebras@xcite , topological quantum field theories@xcite , and quantum invariants of @xmath2-manifolds@xcite .
they also play an important role physically in the study of topological quantum computing@xcite and topological phases of matter@xcite . as such
, a classification of fusion categories up to various notions of equivalence is desirable .
one such notion of equivalence is _ grothendieck equivalence _ , wherein categories have isomorphic decategorifications , which is a based ring in the sense of @xcite .
another notion of equivalence is _ monoidal equivalence _ , wherein categories are related via an invertible monoidal functor .
a natural line of inquiry is then : given a based ring , does it admit categorification ?
if so , given its grothendieck class of fusion categories , how many monoidal classes are there and how can they be distinguished ?
this is an incredibly difficult question to answer in general , especially at the level of fusion categories , with no further assumptions .
most results along these lines are either obtained for small examples which can be easily computed@xcite , but there are some families@xcite for which classifications are known .
in this paper we obtain an answer to these questions for a family of categories we will call metaplectic fusion categories .
these are fusion categories underlying metaplectic modular categories .
recall that a metaplectic modular category@xcite is a modular category grothendieck equivalent to @xmath0 , the category of affine @xmath3 representations with highest integer weight @xmath4@xcite .
other common constructions for such categories come from @xmath5@xcite for @xmath6 some @xmath7 root of unity , or from @xmath8 equivariantizations of @xmath9@xcite where @xmath10 . in (
* theorem 3.2 ) it was shown that all metaplectic modular categories arise from gauging the particle - hole symmetry in @xmath11 modular categories .
our main result is obtained by classifying solutions to the pentagon equations and hexagon equations coming from the @xmath0 based ring .
we stop short of demonstrating that these are all of the categories as this is a much harder problem .
however , the modular data for @xmath0 categories is known and the modular data computed for our solutions coincides . as such , we refine the classification of @xcite one step further . for fixed @xmath12
, our categories are parameterized by pairs @xmath13 where @xmath14 and @xmath15 a positive odd integer less than and co - prime to @xmath16 .
let @xmath17 be the set of these @xmath15 .
define @xmath18 with @xmath19 the quotient map .
elements of @xmath20 can be represented as positive integers less than and co - prime to @xmath16 .
likewise , elements of @xmath21 can be represented as positive integers less than or equal to @xmath12 and co - prime to @xmath16 .
particularly , each @xmath22 represents an element of @xmath20 and the action of @xmath21 on @xmath17 via @xmath23 makes sense .
similarly , the evaluation @xmath24 makes sense .
our main result can then be stated as follows for fixed @xmath12 the following are true : 1 .
the monoidal classes of fusion categories constructed from @xmath13 , @xmath14 and @xmath22 are the fusion categories underlying metaplectic modular categories .
2 . let @xmath13 and @xmath25 parameterize two different solutions to the pentagon equations .
then the fusion categories constructed from these solutions are monoidally equivalent if and only if @xmath26 and there exists @xmath27 such that @xmath28 .
3 . for @xmath29 , there are exactly @xmath30 monoidally inequivalent metaplectic modular categories if @xmath31 , otherwise there are exactly @xmath32 . as a sanity check on this , following ( * ? ? ? * theorem 1.1 ) , we note that the number of metaplectic fusion categories admitting the structure of a modular category is always less than or equal to the number of metaplectic modular categories .
the structure of this paper is as follows . in [ s : fusion ] we review the basic information for fusion and modular categories and their arithmetic descriptions , fusion and modular systems . in [ s : invs ] we review the gauge invariants of @xcite which allow us to extend deductions about our fusion systems to entire gauge and monoidal classes of fusion categories . in [ s : sost ] we present our solutions to the pentagon and hexagon equations , and construct the modular data . in [ s : equivs ] we determine the monoidal equivalence classes of the fusion categories . in [ s : examples ] we explicitly compute the equivalence classes and modular structures for several examples .
appendix [ a : proof ] demonstrates that the explicit @xmath33- and @xmath17-symbols we present are indeed solutions to the pentagon and hexagon equations .
_ acknowledgements .
_ e. ardonne would like to thank zhenghan wang for stimulating discussions .
m. titsworth would like to thank the organizers of the 2014 ams mathematics research communities program on fusion categories and quantum information : zhenghan wang , eric rowell , and richard ng .
e. ardonne was supported , in part , by the swedish research council .
p. e. finch was supported by the deutsche forschungsgemeinschaft under grant no .
fr 737/7 - 1 .
in this section we review relevant facts about fusion / modular categories and their arithmetic descriptions .
we do not provide proofs , but most can be found in one of @xcite , or now @xcite . in the sequel
we will always work over @xmath34 for simplicity .
[ def : basedring ] a _ unital based ring _
@xmath35 is a @xmath36- ring @xmath17 together with a set @xmath37 and identity @xmath38 such that * ( structure constants ) there exist non - negative integers @xmath39 for @xmath40 such that @xmath41 * ( duality ) a bijection @xmath42 such that @xmath43 which extends to an anti - involution on @xmath35 , i.e. @xmath44 . by associativity the structure constants must satisfy @xmath45 for all @xmath46 .
the structure constants define a map @xmath47 which we can extend recursively to arbitrary @xmath48 via @xmath49 for all @xmath50 .
those structure constants which are non - zero will play an important role and so we define @xmath51 with @xmath52 the notation for @xmath53 .
we will say that @xmath35 is _ multiplicity free _ if @xmath54 . for based rings @xmath35 and
@xmath55 , every bijection @xmath56 satisfying @xmath57 defines a unique based ring isomorphism @xmath58 and vice versa .
let @xmath59 to be the group of based ring automorphisms of @xmath35 . since all based rings in this paper are all multiplicity free , unless otherwise noted all future statements will be made under this assumption .
let @xmath35 be a based ring and define @xmath60 be the matrix with entry @xmath61 .
the _ frobenius - perron dimension _
@xmath62 of @xmath63 is the largest positive real eigenvalue of @xmath60 .
the frobenius - perron dimension of @xmath35 is @xmath64 .
we now define fusion categories .
[ def : fusioncategory ] a _ fusion category _ @xmath65 over @xmath66 is a monoidal semi - simple abelian category by @xmath67 and direct sum of objects by @xmath68 with identity @xmath69 such that 1 .
( @xmath66-linearity ) @xmath65 is enriched over @xmath70 .
this is to say that @xmath71 is a finite dimensional vector space over @xmath1 for all objects @xmath72 .
2 . ( finiteness )
there are finitely many isomorphism classes of simple objects in @xmath73 and @xmath74 for all simple objects @xmath75 .
3 . ( rigidity ) for every object @xmath75
, there is an object @xmath76 and evaluation and co - evaluation maps @xmath77 such that @xmath78 we denote by @xmath79 the grothendieck ring of @xmath65 .
this is a based ring as defined in [ def : basedring ] with basis elements corresponding to equivalence classes of simple objects @xmath80 and multiplication induced from @xmath67 via @xmath81 .
we say that @xmath65 is multiplicity free if @xmath79 is multiplicity free .
given two fusion categories @xmath65 and @xmath82 we say that they are _ grothendieck equivalent _ if and only if @xmath83 .
we say that a based ring @xmath35 _ admits categorification _ if there exists a fusion category @xmath65 such that @xmath84 .
let @xmath65 be a fusion category and @xmath85 be a natural isomorphism @xmath86 of the double dual and identity functors .
for all equivalence classes of simple objects @xmath87 of @xmath65 this gives us a morphism @xmath88 . from this
we can define the left and right _ quantum trace _ of a morphism @xmath89 on @xmath87 , @xmath90 where @xmath91 and @xmath92 are the evaluation and coevaluation maps .
a natural isomorphism @xmath85 such that @xmath93 is called a _
pivotal isomorphism_. a
_ pivotal fusion category _ is a fusion category equipped with a pivotal isomorphism . for every isomorphism class of simple objects @xmath87 in a pivotal fusion category @xmath65 , one defines the left quantum dimension @xmath94 and the right quantum dimension @xmath95 as the quantum traces of the identity morphism on @xmath87 .
spherical fusion category _ is a pivotal fusion category such that the left and right quantum dimensions of all objects coincide . in this case
we refer simply to the quantum dimension of an object @xmath87 as @xmath96 . for a spherical fusion category
@xmath65 we define the matrix @xmath97 to be the diagonal matrix with entries @xmath96 for all equivalence classes @xmath87 .
we also define the _ categorical quantum dimension _
. a spherical fusion category is called _ pseudo - unitary _
iff @xmath99 and there exists a canonical choice of spherical structure such that @xmath100 for all @xmath87 .
there is also the stronger notion of a _ unitary fusion category_. a _ conjugation _ on a fusion category is a family of conjugate linear maps @xmath101 , @xmath102 satisfying @xmath103 a fusion category is _ unitary _ if @xmath104 whenever @xmath105@xcite . in this case
then , all of the @xmath106 spaces are hilbert spaces @xcite .
unitary implies @xmath107 and so then pseudo - unitary and spherical .
let @xmath65 be a fusion category .
a braiding on @xmath65 is a family of natural isomorphisms @xmath108 satisfying the hexagon equations as in @xcite . a _ braided fusion category _ is a fusion category equipped with a braiding . for braided fusion categories
there is a canonical natural isomorphism @xmath109 a _ balanced fusion category _ is a braided pivotal fusion category with balancing @xmath110 . the trace of this morphism , by abuse of notation ,
will also be denoted @xmath111 and the diagonal matrix with entries @xmath111 will be called the @xmath112-matrix .
a _ ribbon fusion category _ is a balanced fusion category that is also spherical .
there also exist a family of invariants @xmath113 corresponding to the evaluations of hopf links colored by each pair of equivalence classes @xmath114 .
the matrix with entries @xmath113 is called the @xmath115-matrix and a _ modular category _ is a ribbon fusion category with non - degenerate @xmath115-matrix . in the case that @xmath65 is modular than all @xmath116 are roots of unity .
the matrices @xmath117 and @xmath118 give a representation of the modular group @xmath119 .
it is believed that the pair @xmath120 uniquely determine the modular category and typically what one does in classifying modular categories ( e.g. @xcite,@xcite ) is enumerate the admissible pairs @xmath120 .
an important question is how one goes about constructing fusion categories .
one way to do so@xcite is through a collection of numbers called a fusion system . a _ fusion system _ @xmath121 over @xmath1 consists of 1 . a set of labels @xmath122 containing an element called @xmath69
an involution @xmath123 such that @xmath124 .
a set map @xmath125 ( written @xmath126 for @xmath127 ) satisfying @xmath128 the entries of matrix @xmath129 are referred to as @xmath130-symbols .
define @xmath131 to be the group of permutations on @xmath122 such that for all @xmath132 in @xmath131 @xmath133 this then extends to an action on @xmath33 via @xmath134 central to our analysis is the ability to interpolate between fusion categories over @xmath66 and fusion systems over @xmath66 . given a fusion system @xmath135 over @xmath65 , one can construct a fusion category @xmath136 . given a fusion category @xmath65 over @xmath66 , a fusion system over @xmath66 , denoted @xmath121 ,
can be extracted such that @xmath137 as fusion categories . for our construction , this is proven in @xcite as proposition 3.7 .
a similar theorem also appears in @xcite .
it is by analyzing a family of fusion systems for @xmath0 grothendieck rings that we develop our classification . to do this
, we will utilize the following : two fusion categories @xmath65 and @xmath82 are grothendieck equivalent if and only if given any two fusion systems @xmath135 and @xmath138 extracted from them there exists a bijection @xmath139 such that for all @xmath140 there exists @xmath141 such that @xmath142 . if @xmath143 there exists a based ring isomorphism in the sense of @xcite 2.1.iv , and this is determined by its action on basis elements .
this allows us to specify a common basis for @xmath144 from which to work .
there also exists arithmetic data for pivotal structures .
let @xmath135 be a fusion system extracted from a fusion category @xmath65 and define the set @xmath145 .
pivotal structures on @xmath65 are in 1 - 1 correspondence with solutions to @xmath146 this is proven in ( * ? ? ?
* proposition 3.12 ) . given a fusion category @xmath65 with pivotal structure @xmath85 ,
if it is a spherical structure then there exists a fusion system @xmath121 such that all @xmath147 . by skeletalizing the rigidity conditions and using the choice of basis as in (
* lemma 3.4 ) , the quantum dimensions can be computed from @xmath33 and @xmath85 as @xmath148 if @xmath65 admits the structure of a unitary fusion category , then there exists a fusion system @xmath121 such that all @xmath149 matrices are unitary .
conversely ( see @xcite , section 4 ) , if given @xmath150 there exists a solution to - such that all @xmath149 matrices are unitary , then @xmath136 admits a unitary structure . a _ modular system _
@xmath151 is a fusion system @xmath135 such that @xmath152 , @xmath153 is a solution to , and @xmath154 is a collection of numbers satisfying @xmath155 such that the matrix with entries @xmath156 is invertible .
similar to [ eq : autaction ] , we can define the action of @xmath140 on @xmath17 via @xmath157 the matrix @xmath158 is related to the matrix @xmath115 via @xmath159 .
additionally , we can compute the twists as @xmath160 obtained by taking the trace of the morphism above@xcite .
note that through teasing out definitions , all of the information above can be written in terms of @xmath33 , @xmath17 , and @xmath85 .
since in our case we have all of the @xmath33 s , @xmath17 s , and @xmath85 s , so we can simply compute @xmath120 for the given the arithmetic data .
given two fusion ( modular ) categories @xmath65 and @xmath82 , a central question is one of whether or not they are equivalent in some suitable sense . the strongest of these is _
monoidal equivalence_.
@xmath65 and @xmath82 are said to be _ monoidally equivalent _ iff there exists a pair of monoidal functors(see @xcite ) @xmath161 and @xmath162 such that @xmath163 and @xmath164 are naturally monoidal isomorphic to the identity functors on @xmath65 and @xmath82 respectively . if @xmath65 and @xmath82 are braided monoidal , then they are _ braided monoidally equivalent _ if @xmath165 and @xmath166 are braided monoidal functors . a family of categories @xmath167 which are monoidally equivalent are automatically grothendieck equivalent .
a difficult and important question goes the other direction : given a grothendieck equivalence class of categories , how many monoidal equivalence classes are there ? that there are finitely many
is known as _
ocneanu rigidity_@xcite .
given that we can describe fusion categories arithmetically , there is the natural question of whether or not an equivalence between them can be described arithmetically . the action @xmath131 on a fusion system @xmath151 is given by and it was shown in @xcite that this gives an object permuting monoidal functor .
we say that two fusion systems @xmath168 are _ permutation equivalent _ if there exists @xmath169 such that @xmath170 .
the other operation one can perform on @xmath135 to obtain equivalent categories is given by a _ gauge transformation_. this corresponds to a change of basis on the @xmath171 homspaces and can be represented arithmetically as follows : define @xmath172 to be the set of functions @xmath173 fixing the notation @xmath174 and @xmath175 .
then given a solution @xmath33 to - and @xmath176 , one obtains another solution @xmath177 via the equations @xmath178 and two solutions @xmath33 and @xmath179 are gauge equivalent if and only if there exists @xmath176 such that @xmath180 .
can be used to construct an object fixing monoidal functor .
given @xmath65 , it is clear from the definition that any two fusion systems extracted from @xmath65 are gauge equivalent .
monoidal equivalence of fusion categories @xmath65 and @xmath82 is then determined by the existence of a permutation equivalence together with a gauge equivalence .
the above extends readily to braided monoidal equivalence between modular categories . given @xmath17 and @xmath181
, permutation equivalence requires the imposition of the extra condition @xmath182 as defined in .
gauge equivalence requires the imposition that @xmath183 as given by @xmath184 given a set of fusion(modular ) systems , determining their monoidal equivalence is not an easy task in general .
however , in @xcite a construction is given for invariants which can distinguish gauge classes of fusion categories ( with or without multiplicity ) .
the key to this is noting that for a given @xmath150 solutions to the equations - define an algebraic scheme @xmath185 in variables @xmath186 .
@xmath33 can then be interpreted as an algebra homomorphism @xmath187\rightarrow \c$ ] with @xmath188 for @xmath189 . from here , on an open subset @xmath190 of @xmath185 , it is straightforward to extend @xmath33 to @xmath191 , the ring of regular functions on @xmath63 defined at @xmath190 .
gauge equivalence classes correspond to @xmath172-orbits in @xmath185 and monoidal equivalence classes correspond to @xmath192-orbits in @xmath185 . given this
, one can leverage the geometric invariant theory of @xcite to obtain the following result : [ thm : invs ] fix @xmath150 .
let @xmath193 be the number of gauge equivalence classes of fusion categories and @xmath194 be the number monoidal equivalence classes of fusion categories .
then there exist @xmath195 @xmath172-invariant rational monomials with @xmath196 such that for fusion systems @xmath33 and @xmath179 , @xmath33 and @xmath179 are gauge equivalent if and only if @xmath197 similarly , there exist @xmath198 @xmath131-linear combinations @xmath199 of @xmath172-invariant monomials as in such that @xmath33 and @xmath179 are monoidally equivalent if and only if @xmath200 @xmath131-linear combinations of @xmath172-invariants will be called @xmath192-invariants for obvious reasons .
this is theorem 1.2 of @xcite .
it is straightforward to extend this construction to classifying modular systems @xmath151 .
one goes through the same arguments of using the scheme defined from -,- , and noting that they have the same essential properties ( @xmath172 is reductive , orbits are closed ) . in the fusion case
, the choice of @xmath192-invariants is not necessarily apparent a priori .
however , as we will show in section [ s : equivs ] , there is an easy choice which works for our @xmath0 categories . for modular categories
there are invariants which immediately present themselves since it was previously mentioned that the pair @xmath120 are ( conjecturally ) strong enough for classification .
the quantum dimensions as defined in [ eq : qdims ] are categorical invariants sufficient to distinguish between spherical structures , but the first row and column of the matrix @xmath201 are essentially these numbers .
it has also been previously mentioned that the @xmath202 , @xmath203 , and @xmath204 can all be written using @xmath205 , @xmath206 , and @xmath207 .
thus they all define @xmath172-invariant regular functions on @xmath63 .
in the following we present arithmetic data for the @xmath33 and @xmath17 matrices which makes much of the structure for our categories apparent . we also provide explicit computations of the relevant categorical quantities .
there are several contexts in which the grothendieck rings for our categories naturally arise , though two primary sources both come from the study of lie algebras .
as should be apparent from our notation , one of these is as grothendieck rings for the categories of representations of the ( untwisted ) affine kac - moody algebras @xmath208 ( i.e. @xmath3 ) with highest integral weight 2 . as determined by @xcite , these are equivalent ( as modular categories ) to the semi - simplification ( la @xcite ) of the representation category for @xmath209 with @xmath210 .
the affine kac - moody and quantum group constructions for the grothendieck rings can be found in @xcite and @xcite respectively .
another source for our fusion rules comes from tambara - yamagami categories as follows .
let @xmath12 a positive integer , @xmath211 a symmetric bicharacter on @xmath11 and @xmath212 .
in @xcite it was shown that @xmath8-de - equivariantizations of the tambara - yamagami category @xmath213 gives rise to an @xmath0 category .
for some @xmath0 category @xmath65 , we label the basis elements of @xmath144 ( i.e. equivalence classes of simple objects ) by elements of the set @xmath214 with @xmath215 .
these have frobenius - perron dimensions @xmath216 respectively .
@xmath144 is commutative with non - trivial products given by : @xmath217 where @xmath218 with @xmath219 is given by @xmath220 let @xmath144 be the grothendieck ring of a @xmath0 category
. then 1 .
the automorphisms which permute the @xmath221 are given by @xmath222 , 2 .
the automorphisms which permute the @xmath223 are given by @xmath8 , 3 .
and the automorphism group of @xmath144 is @xmath224 . that the fusion rules are invariant under exchange of @xmath223 is straight forward .
the automorphisms which permute the @xmath221 follow restricting @xmath225 to @xmath20 and promoting it to a group homomorphism . @xmath20 has even order where @xmath226 is euler s totient function . for power @xmath1 of prime @xmath15 @xmath227 , which is even .
that the evaluation of @xmath226 is even for every @xmath16 follows from prime factorization and the multiplicative property of @xmath226 . ] and can be represented by integers in @xmath228 .
if @xmath229 is in @xmath20 so is @xmath230 , thus the quotient @xmath231 is well defined .
@xmath225 is precisely this quotient map . for all @xmath232 . to see this consider @xmath233 , which has order six rather than eight , since three has no multiplicative inverse in @xmath234 . ]
@xmath21 acts on @xmath221 in the obvious way and since @xmath225 is a group homomorphism , this preserves the fusion rules .
there are no automorphisms on pointed objects or between objects of different frobenius - perron dimension , and so the automorphism group of @xmath144 must be the direct product .
this gives a separate proof from that in @xcite .
our solutions to - are indexed by pairs @xmath13 where @xmath15 is an odd integer between @xmath235 and @xmath16 , such that @xmath236 and @xmath14 .
fix @xmath210 . to write the general @xmath33-symbols , we first introduce the following matrices .
@xmath238 where the @xmath239 take values @xmath240 and satisfy @xmath241 such that the associated @xmath33-matrices are orthogonal .
in addition , we define the matrices @xmath242 , @xmath243 and @xmath244 from the following function : @xmath245 where @xmath246 .
@xmath242 is a @xmath247 matrix with entries @xmath248 whose indices run over @xmath249 . @xmath243 and @xmath244 are @xmath250 matrices with entries @xmath251 whose indices run over @xmath252 .
we note that the matrices @xmath253 and @xmath254 are orthogonal provided that @xmath15 is an odd integer relatively prime to @xmath16 .
before we give the specific @xmath130 symbols , we make a few general remarks .
first , the order of the entries in the @xmath33-matrices respects the order we specified above : @xmath255 , where @xmath256 .
second , we will actually not give the values @xmath129 individually , but instead give the @xmath33-matrices @xmath149 .
in the basis we use to give the @xmath33-matrices , we have the following property @xmath257 in addition , it is implicitly assumed that the label @xmath258 of @xmath259 is in the tensor decomposition of @xmath260 .
this allows us to only specify a reduced set of @xmath33-matrices , while the others can be deduced by using this relation .
we note that often , but not not always , the @xmath33-matrices are symmetric . in addition , all the @xmath33-symbols are real in our basis .
we first observe that from @xcite that for fixed @xmath12 the fusion rules for @xmath261 give the tensor structure for @xmath262 .
@xmath144 is then a @xmath8-extension of @xmath263 .
we will proceed by building subcategories grothendieck equivalent to @xmath8 and @xmath262 before finally specifying @xmath33-matrices which correspond to the @xmath264 objects .
first , if one or more of the labels @xmath265 equals @xmath266 , @xmath267 .
next we have @xmath268 equivalent to @xmath8 and the action of @xmath85 on the objects @xmath221 is determined by @xmath269 @xmath270 the rest of the @xmath33-matrices for the category @xmath262 are given by : @xmath271 there are no other solutions to the pentagon equations for @xmath262 which extend to solutions for @xmath0 . by @xcite
all fusion categories @xmath65 for @xmath262 are group theoretical and thus morita equivalent to a pointed fusion category @xmath82 of the form @xmath272 where @xmath273 , the group of @xmath16 roots of unity .
by @xcite we know that ( left ) module categories over @xmath274 are parameterized by @xmath275 where @xmath276 such that @xmath277 and @xmath278 .
in our case we have that since the universal grading of @xmath0 is @xmath8 , we have that @xmath279 are indecomposable @xmath8 module categories over @xmath262
. we can count which @xmath262 categories have @xmath8 module categories by looking at which @xmath274 categories have @xmath8 module categories . since @xmath280 the only @xmath281 which fixes the @xmath8 subgroup of @xmath274 is @xmath282 and its cohomology is @xmath8 .
thus only @xmath262 has indecomposable @xmath8 module categories and there are two of them .
we now specify the @xmath33-matrices that are labeled by @xmath223 and start with the @xmath33-matrices whose labels consist of only @xmath223 .
@xmath283 there are three classes of @xmath33-matrices involving two labels that are @xmath223 .
the first class of this type is @xmath284 where @xmath285 may or may not be equal .
the second class of @xmath33-matrices with two labels equal to @xmath223 and @xmath286 is @xmath287 finally , the third class of @xmath33-matrices with two labels equal to @xmath288 or @xmath289 is @xmath290 with these symbols , and the general rules described above , we have exhausted all the @xmath33-matrices . given our solutions to the pentagon equations , one would anticipate it is not too complicated to construct solutions to the hexagon equations as well .
given a solution to the hexagon equations one always has a second solution given by inverses .
this simply corresponds to a choice of @xmath17 and @xmath291 in and .
we may also obtain another solution by replacing all @xmath17-symbols @xmath292 , such that @xmath87 and @xmath293 are both one of @xmath223 by @xmath294 .
however , this is monoidally equivalent to our original solution via the automorphism @xmath295 .
this gives us that for any @xmath12 , solutions to the pentagon and hexagon equations can be uniquely identified by the tuple @xmath296 , where @xmath297 indicates whether or not one is referring to the r - symbols given below or their inverses .
we then provide only one such solution .
we specify the @xmath17-symbols using a form inspired by conformal field theory .
namely , for each simple object type , we introduce the scaling dimensions @xmath298 which determine the @xmath17-symbols up to a sign @xmath299 where @xmath300 and @xmath301 .
below we specify the scaling dimensions @xmath302 for all the simple objects , as well as the signs @xmath303 and @xmath304 . the scaling dimensions are given by @xmath305 the notation @xmath306 denotes the jacobi symbol , which is defined for @xmath307 an integer , and @xmath308 a positive , odd integer . to completely specify the @xmath17-symbols we still need to specify the signs @xmath303 and @xmath304 .
we start with @xmath303 , and note that for the @xmath17-symbols presented here we have the property @xmath309 , which implies that we only need to specify a limited set of these symbols .
the non - trivial symbols come in two classes .
the first class is @xmath310 the second class of symbols is @xmath311 the symbols @xmath303 that are not specified by the rules above , are all equal to @xmath312 .
finally , the signs @xmath304 also satisfy the property @xmath313 and there is only one class of non - trivial signs of type , namely @xmath314 the symbols @xmath304 that are not specified by the rules above , are all equal to @xmath312 . with this
, we have completely specified one of the solutions of the hexagon equations , from which the other one follows by taking the inverse .
we close by commenting briefly on the appearance of the jacobi symbols in the expression for the @xmath17-symbols .
one can think of the list of hexagon equations as labeled by the different @xmath33-symbols , namely the one appearing on the left hand side of the symbolic form @xmath315 . to derive the form of @xmath316 above , consider the hexagon equation associated to @xmath317 .
the sum on the right hand side gives rise to a quadratic gauss sum ( because one needs to sum over all @xmath221 ( as well as the identity ) ) , which are closely related to the jacobi symbols .
this leads to their presence in the scaling dimensions @xmath316 .
see appendix [ a : proof ] for more details . as stated above , the jacobi symbols @xmath318 are defined for positive odd integers @xmath308 and arbitrary integers @xmath307 .
interestingly , we observed that the these jacobi symbols can be written in terms of the matrices @xmath243 and @xmath242 ( whose sizes depend on @xmath12 ) .
in particular , one can show @xmath319 this gives an analytic function of @xmath15 , that goes through all the jacobi symbols , defined for @xmath15 integer .
we note that eisenstein also constructed such a function , @xmath320 , see for instance @xcite . the difference with the function that we found , is that the values @xmath240 are the extrema of the function while this is certainly not the case for eisenstein s function .
all of our f - matrices are manifestly unitary , thus all of our categories admit at least one spherical structure . since they are also braided , they must be ribbon and so we can use the formula .
our formula is defined for any braided category , however one can obtain the more common form by taking @xmath321 .
we present the latter to more easily demonstrate that we obtain the same modular result as @xcite .
however , we also provide an explicit classification at the fusion level .
pivotal structures on modular categories are in bijective correspondence with the group of invertible objects and spherical structures with the maximal abelian 2-subgroup ( * ? ? ?
* lemma 2.4 ) .
since the group of invertible objects is @xmath8 , there are exactly two pivotal structures for each category , both of which are spherical .
it is straightforward to compute that for @xmath69 , @xmath85 , and @xmath221 , all pivotal coefficients must be @xmath235 .
we also have @xmath322 , the choice of which switches the sign of @xmath323 .
below , we present the modular data for solutions with positive quantum dimensions .
it s worth noting that for all @xmath12 and @xmath15 the pivotal coefficients which yield positive quantum dimensions for @xmath324 are not those that yield positive quantum dimensions for @xmath325 .
this can be seen from noting the effect of @xmath326 on . in particular , @xmath327 if we take @xmath328 .
we then have that , for modular categories corresponding to the solution indexed by @xmath296 , @xmath329 for the twists we have @xmath330
in this section we introduce those gauge invariants which we will use to classify monoidally inequivalent solutions for a given @xmath12 using the method of section [ s : invs ] . at the level of gauge equivalence it is easy to see that two solutions @xmath13 and @xmath25 are gauge equivalent if and only if @xmath331 and @xmath332 .
our goal then is to find a set @xmath333 of @xmath131-invariant linear combination of @xmath172-invariant monomials as defined in theorem [ thm : invs ] which determine the monoidal classes .
we then determine the number of monoidal equivalence classes for a given @xmath12 .
for all pairs @xmath13 as defined in section [ ss : fs ] let @xmath334 be the evaluation map as defined for theorem [ thm : invs ] .
we first note that the adjoint category is basically useless to us .
[ prop : adjsub ] the monoidal classes of solutions specified in [ ss : ad ] can not be distinguished by @xmath192-invariants coming from @xmath262 subcategories . for fixed @xmath12 , the only @xmath33-matrix entries which differ are those in which @xmath15 arises , i.e. those specified by the @xmath97 , @xmath335 , @xmath336 , and @xmath172 matrices .
these come from @xmath33-matrices involving at least one @xmath223 and so do not come from the @xmath262 subcategory .
so if @xmath337 is an @xmath192-invariant which comes from @xmath262 , i.e. only involves the objects @xmath235 , @xmath85 , and @xmath338 then @xmath339 from here , we can narrow our search even further . from - , the entries of @xmath340 can all be written in some fixed way as scalar multiples of entries in @xmath341 or @xmath342 .
these relationships are preserved under permutation .
the entries of @xmath341 and @xmath342 are given by the matrices @xmath243 and @xmath343 respectively .
let @xmath344 be solutions to - parameterized by @xmath13 and @xmath25 .
then @xmath345 if and only if @xmath346 .
the direction @xmath347 is obvious , so assume @xmath348 .
@xmath243 determines @xmath242 up to a sign and @xmath349 this corresponds to a choice @xmath223 , which is irrelevant to @xmath13 , so this implies @xmath345 .
[ cor : hdiag ] @xmath345 if and only if @xmath350 @xmath351 determines @xmath326 and @xmath352 narrows @xmath15 to one of two choices . taken modulo @xmath16 only one of @xmath15 and @xmath353 is odd , and so @xmath354 determines @xmath243 . by inspection
, we see that @xmath355 [ cor : hinvs ] @xmath356 uniquely determines the values for all @xmath172-invariant monomials and thus all @xmath192-invariant monomials .
the permutation @xmath357 acts as a gauge transformation .
there is a natural action of @xmath358 on @xmath359 given by @xmath360 [ lem : trace ] fix @xmath12 and let @xmath13 and @xmath361 be solutions as defined in [ ss : fs ] .
the following are equivalent : 1 .
@xmath13 and @xmath361 are monoidally equivalent , 2 .
there exists @xmath27 such that @xmath362 , and 3 .
there exists @xmath27 such that @xmath363 .
first assume that @xmath13 and @xmath361 are monoidally equivalent via some @xmath27 .
@xmath364 acts on @xmath365 by @xmath366 this implies @xmath367 and @xmath368 are congruent modulo @xmath16 and so @xmath28 .
assume @xmath369 such that @xmath363 .
then , modulo @xmath16 , @xmath370 or @xmath371 , but regardless @xmath372 . by corollary [ cor : hinvs ] , for any @xmath373-invariant @xmath337
, @xmath374 thus @xmath361 and @xmath13 are monoidally equivalent .
it is not an easy task to calculate how many monoidally inequivalent sets of f - moves there are .
nevertheless we can construct a framework in which the question is seemingly simple . from the previous section we showed that the f - moves labeled by @xmath13 and @xmath25 are monoidally equivalent if and only if @xmath332 and there exists a @xmath27 such that @xmath375 .
thus we define @xmath376 each set contains elements such that the f - moves labeled by @xmath13 and @xmath361 are monoidally equivalent .
furthermore , @xmath13 and @xmath361 are monoidally equivalent if and only if @xmath377 .
we realize that @xmath378 is a normal subgroup and that @xmath379 are conjugacy classes .
thus the number of monoidally inequivalent f - moves must be twice ( due to @xmath326 ) @xmath380 .
+ it is possible to simplify this further by considering the map @xmath381 defined by @xmath382 .
this map is clearly onto , i.e. the image of @xmath89 is @xmath378 . as we are dealing with the finite groups we can use the first isomorphism theorem and deduce that @xmath383 where @xmath384 is the kernel of @xmath89 and is the subgroup @xmath385 thus the number of monoidally inequivalent f - moves is twice the number of elements of @xmath386 that square to the identity ( because of @xmath326 ) .
we know that @xmath387 for some @xmath388 .
if @xmath389 satisfies @xmath390 or @xmath391 then @xmath392 satisfies @xmath390 .
we can then deduce that @xmath393 firstly , consider the prime decomposition @xmath394 where the @xmath395 are prime and @xmath396 .
we define sets @xmath397 we note that as @xmath398 is also a group and that its order is precisely @xmath32 .
next , suppose that @xmath399 it follows that @xmath400 , similarly if @xmath401 and @xmath402 then @xmath403 .
thus we have that the cardinality of @xmath404 is zero or @xmath405 .
for @xmath406 to be non - zero we must have that there exists @xmath407 such that @xmath408 for all @xmath229 .
+ from this we can deduce the number of monoidal classes @xmath409 where @xmath337 is the number of primes that divide @xmath410 .
here we have used that @xmath411
in this section we explicitly compute the classifying invariants for a number of `` small '' categories . the case of @xmath412
is known to be grothendieck equivalent to @xmath413 and both the fusion and modular structures are fully classified ( see @xcite ) . beyond this , there are several guideposts : * the classification of weakly integral modular categories of dimension @xmath414 is given in @xcite .
this contains those @xmath0 of our family for which @xmath16 is square free . *
the classification of integral modular categories of dimension @xmath415 is given in @xcite where @xmath6 is prime .
* explicit formulae for the modular data of @xmath8-equivariantizations of tambara - yamagami categories is given in @xcite to which our categories are grothendieck equivalent . for @xmath412 , @xmath416 .
thus we have only one two dimensional simple object for which the automorphism group is trivial .
we have two solutions and two fusion categories . up to permutation there are four modular categories , one for each of the four possible t - matrices coming from : @xmath417 corresponding to solutions @xmath418 , @xmath419 , @xmath420 and @xmath421 respectively .
this is in agreement with @xcite . in this case
the two braidings for each fusion category are inequivalent .
there are two two dimensional objects and @xmath423 .
both elements square to the identity and so there are four fusion categories from . this can be seen from the sets @xmath424 from @xcite there are also four modular categories , corresponding to solutions @xmath425 , @xmath426 , @xmath427 , and @xmath428 with @xmath112-matrices @xmath429 in this case the braidings corresponding to @xmath430 are equivalent .
there are three two dimensional objects and @xmath432 .
we have that @xmath433 these are all clearly related via permutation . computing the modular data for our solutions , we find that there are four modular categories distinguished by their @xmath112-matrices @xmath434 corresponding to solutions @xmath435 , @xmath436 , @xmath437 , and @xmath438 respectively . for @xmath439
there are four two dimensional objects , but we have that @xmath440 since there are three odd integers less than and coprime to @xmath441 , @xmath235 , @xmath442 , and @xmath443 .
these correspond to the @xmath444 subgroup generated by the cycle @xmath445 .
the sets @xmath446 are then : @xmath447 from this it s easy to see that @xmath445 sends @xmath448 to @xmath449 and @xmath450 sends @xmath451 to @xmath449 .
thus there is only one equivalence class .
finally , as before there are four modular structures distinguished by their t - matrices : @xmath452 corresponding to solutions @xmath453 , @xmath454 , @xmath455 , and @xmath456 respectively .
in this appendix , we briefly explain how we obtained the @xmath33- and @xmath17-symbols that we present in the main text , and how we verified that they indeed are solutions to the pentagon and hexagon equations , for arbitrary tuples @xmath296 with @xmath12 a positive integer , @xmath15 an odd integer such that @xmath457 and @xmath236 , @xmath458 and @xmath430 .
we obtained the @xmath33-symbols , by numerically calculating them from the quantum group based on @xmath3 at the appropriate roots of unity for small @xmath12 . from this data
, we were able to extract the general pattern , presented in section [ ss : fs ] .
the main difficulty in verifying that the @xmath33- and @xmath17-symbols we obtained do indeed satisfy the pentagon and hexagon equations lies not so much in the actual verification of the equations , but rather in convincing oneself that all the pentagon and hexagon equations one needs to verify are indeed covered .
the reason for this is the structure of the fusion rules of the dimension two objects , which are such that one has to consider many cases separately .
this pertains both to all the possible labelings of the equations , as well as to the sum present in these equations . in order to complete this task
, we used the algebraic manipulation program _
mathematica_. we first carefully constructed all the pentagon equations that can occur algebraically , that is , for arbitrary tuples @xmath459 . after having convinced ourselves that we indeed covered all the possible cases ,
we then verified this , for low , explicit values of @xmath12 , by using the algebraic result to explicitly construct the labels of all the pentagon equations for these explicit values of @xmath12 . for explicit @xmath12
, one can also generate all possible cases directly . as expected , we indeed covered all the cases . after we generated all the cases of the pentagon equations for the tuples
@xmath459 , we did the actual verification . in all but one class of equations ,
the equations are straightforwardly verified , and again , we used mathematica for this .
the most complicated cases occur when the sum over @xmath460 in runs over all dimension two objects ( and possibly @xmath69 or @xmath85 ) . to check that the @xmath33-symbols satisfy these equations , one rewrites the ( maximally triple ) products of sines and cosines that occur as a sum of sines and cosines , and one uses lagrange equations to perform the sum .
one can actually use the same procedure to verify that the matrices @xmath172 and @xmath336 , are orthogonal . to verify that the solutions of the @xmath33- and @xmath17-symbols presented also satisfy the hexagon equations
, we used the same strategy as we used to verify the pentagon equations .
that is , we explicitly constructed all different cases of the hexagon equations using mathematica , which we also used to do the actual checking of the equations , except for the one class of equations , that is non - trivial , and was checked by hand . the class of equations that we explicitly verified has all the labels @xmath265 of , equal to @xmath223 .
the sum over @xmath461 runs over all dimension two objects , and possibly @xmath462 or @xmath85 .
this class of equations falls apart into four distinct cases , of which we discuss the simplest case for ease of presentation .
the other three cases can be shown to hold using similar arguments .
we first give the four classes of equations ( without simplification ) explicitly .
the first class is @xmath463 the second class is @xmath464 the third class of equations reads @xmath465 finally , the fourth class of equations takes the form @xmath466 these equations hold if @xmath12 is a positive integer , @xmath15 is an odd integer @xmath457 with @xmath236 , @xmath458 and @xmath467 are integer . in the actual hexagon equations one has @xmath468 .
we note that ( and ) is obtained from by setting @xmath469 ( and @xmath470 ) .
we also note that in the right hand side of is a sum of only @xmath12 terms , in comparison to , and , for which the right hand side is a sum of @xmath471 terms .
to verify the validity of , we consider the rhs of this equation first .
replacing @xmath307 by @xmath472 in the summand leaves the summand the same ( if @xmath473 ) , which means we can extend the sum from @xmath474 to @xmath475 at the cost of a factor of two .
the sum can then be split in two pieces , namely in the even @xmath307
s @xmath476 , with @xmath477 and odd @xmath307 s , @xmath478 , with @xmath479 . in both cases , one checks that the summand takes the form @xmath480 , for both @xmath12 even and odd .
hence , the rhs becomes a quadratic gauss sum ( see , for instance , @xcite ) @xmath481 where @xmath482 for @xmath483 ( i.e. , @xmath12 even ) and @xmath484 for @xmath485 ( i.e. , @xmath12 odd ) , and we used that @xmath486 .
standard manipulations of the jacobi - symbol gives + @xmath487 , so that the rhs takes the form @xmath488 we can now compare this result for the rhs of with the lhs , by explicitly checking the lhs for the four different cases of @xmath12 modulo 4 .
one finds that the lhs is indeed independent of @xmath458 , and that the lhs equals the rhs , as we wanted to show .
paul bruillard , csar galindo , seung - moon hong , yevgenia kashina , deepak naidu , sonia natale , julia yael plavnik , and eric c. rowell , _ classification of integral modular categories of frobenius - perron dimension @xmath489 and @xmath490 _ , canad .
math . bull .
* 57 * ( 2014 ) , no . 4 , 721734 , mr3270794 , doi:10.4153/cmb-2013 - 042 - 6 , arxiv : .
mr 3270794 paul bruillard , csar galindo , siu - hung ng , julia y. plavnik , eric c. rowell , and zhenghan wang , _ on the classification of weakly integral modular categories _ , j. pure appl .
algebra * 220 * ( 2016 ) , no .
6 , 23642388 , mr3448800 , doi:10.1016/j.jpaa.2015.11.010 , arxiv : .
mr 3448800 bojko bakalov and alexander kirillov , jr .
, _ lectures on tensor categories and modular functors _ , university lecture series , vol . 21 , american mathematical society , providence , ri , 2001 , mr1797619 .
mr 1797619 ( 2002d:18003 ) paul bruillard , siu - hung ng , eric c. rowell , and zhenghan wang , _ rank - finiteness for modular categories _ , j. amer . math .
* 29 * ( 2016 ) , no . 3 , 857881 , mr3486174 , doi:10.1090/jams/842 , arxiv : .
mr 3486174 david e. evans and yasuyuki kawahigashi , _ quantum symmetries on operator algebras _ , oxford mathematical monographs , the clarendon press , oxford university press , new york , 1998 , oxford science publications .
mr 1642584 ( 99m:46148 ) pavel etingof , dmitri nikshych , and viktor ostrik , _ on fusion categories _ , ann . of math .
( 2 ) * 162 * ( 2005 ) , no . 2 , 581642 , mr2183279 , doi:10.4007/annals.2005.162.581 , = math/0203060 .
mr 2183279 ( 2006m:16051 ) shlomo gelaki , deepak naidu , and dmitri nikshych , _ centers of graded fusion categories _ , algebra number theory * 3 * ( 2009 ) , no . 8 , 959990 ,
mr2587410 , doi:10.2140/ant.2009.3.959 , arxiv : .
mr 2587410 ( 2011b:18011 ) tobias j. hagge and seung - moon hong , _ some non - braided fusion categories of rank three _ , commun .
* 11 * ( 2009 ) , no . 4 , 615637 , mr2559711 , doi:10.1142/s0219199709003521 .
mr 2559711 ( 2011b:18012 ) david kazhdan and hans wenzl , _ reconstructing monoidal categories _ , i. m. gelfand seminar , adv .
soviet math .
, vol . 16 , amer . math . soc . ,
providence , ri , 1993 , mr1237835 , pp .
mr 1237835 ( 95e:18007 ) michael mger , _ from subfactors to categories and topology .
i. frobenius algebras in and morita equivalence of tensor categories _ , j. pure appl .
algebra * 180 * ( 2003 ) , no . 1 - 2 , 81157 , mr1966524 , doi:10.1007/s00022 - 4049(02)00247 - 5 .
mr 1966524 ( 2004f:18013 ) david mumford , _ the red book of varieties and schemes _ , expanded ed . , lecture notes in mathematics , vol .
1358 , springer - verlag , berlin , 1999 , includes the michigan lectures ( 1974 ) on curves and their jacobians , with contributions by enrico arbarello .
mr 1748380 ( 2001b:14001 ) sonia natale and julia yael plavnik , _ solvability of a class of braided fusion categories _ , appl .
structures * 22 * ( 2014 ) , no . 1 , 229240 , mr3163515 , doi:10.1007/s10485 - 012 - 9299-y , arxiv : .
mr 3163515 eric c. rowell , _ from quantum groups to unitary modular tensor categories _ , representations of algebraic groups , quantum groups , and lie algebras , contemp .
413 , amer .
soc . , providence , ri , 2006 , mr2253097 , doi:10.1007/conm/413/07848 , arxiv : , pp . 215230 .
mr 2263097 ( 2007i:18007 ) eric rowell , richard stong , and zhenghan wang , _ on classification of modular tensor categories _ , comm .
* 292 * ( 2009 ) , no . 2 , 343389 , mr2544735 , doi:10.1007/s00220 - 009 - 0908-z , arxiv : .
mr 2544735 ( 2011b:18013 ) daisuke tambara and shigeru yamagami , _ tensor categories with fusion rules of self - duality for finite abelian groups _ , j. algebra * 209 * ( 1998 ) , no . 2 , 692707 , mr1659954 , doi:10.1006/jabr.1998.7558 .
mr 1659954 ( 2000b:18013 ) zhenghan wang , _ topological quantum computation _ , cbms regional conference series in mathematics , vol .
112 , published for the conference board of the mathematical sciences , washington , dc ; by the american mathematical society , providence , ri , 2010 , mr2640343 . mr 2640343 ( 2011f:81054 ) | we study a family of fusion and modular systems realizing fusion categories grothendieck equivalent to the representation category for @xmath0 .
we conjecture a classification for their monoidal equivalence classes from an analysis of their gauge invariants and define a function which gives us the number of classes . |
this work was supported by the deutsche forschungsgemeinschaft through research unit 945 .
computing time at the leibniz computing center in garching is gratefully acknowledged .
we thank martin hck for fruitful discussions . | the characterization and manipulation of deposited magnetic clusters or molecules on surfaces is a prerequisite for future utilization . for this purpose spin - flip inelastic electron tunneling spectroscopy using a scanning tunneling microscope proved to be very precise in determining e.g. exchange constants in deposited finite spin chains in the mev range . in this letter
we provide numerical justification for the underlying assumptions made in such investigations . to this end
we study the impurity magnetization of antiferromagnetic chains for varying couplings to a conduction electron band of a metal substrate .
we show under which circumstances the screening of a part of the system enables one to deduce molecular parameters of the remainder from level crossings in an applied field . _
introduction_the question whether and how deposited magnetic clusters or molecules change their magnetic properties when deposited on a metallic substrate is of fundamental importance especially in view of possible applications as next generation storage devices or magnetic logic circuits @xcite .
one experimental method to investigate local magnetic properties is spin - flip inelastic electron tunneling spectroscopy with a scanning tunneling microscope @xcite . in this method
jumps of the differential conductivity signal transitions between magnetic levels of the deposited entity , of course under the assumption of a weak coupling to the substrate . following this philosophy the authors of ref .
@xcite conjectured the strength of superexchange interactions in deposited molecular nanomagnets .
the latter were constructed by depositing layers of cobalt phthalocyanine molecules which self - organized in a stacked manner so that the cobalt ions formed finite spin chains on the substrate .
this arrangement is scetched in fig .
[ molsurf - f - a ] ( a ) . , the lowest moment
is coupled to the metal surface with strength @xmath0 .
@xmath1 and @xmath0 are unknown , @xmath1 shall be derived from experimental data .
( b ) shows a triangular configuration discussed later in this letter.,width=207 ] when interpreting the observed structure of magnetic levels in such measurements one has to conjecture whether and how much of the deposited magnetic moment is screened by the conduction electrons , since this has a great influence on the magnetic properties of the remainder . in ref . @xcite it was deduced that the co@xmath2 spin with @xmath3 closest to the surface was completely kondo - screened and therefore the observed magnetic levels should result from a practically uncoupled remainder of @xmath4 spins . in this letter
we investigate by means of the numerical renormalization group method ( nrg ) @xcite numerically exactly how the magnetic properties of deposited structures as those shown in fig .
[ molsurf - f - a ] depend on the exchange coupling @xmath0 to the substrate s conduction electrons , the internal exchange coupling @xmath1 , and the applied magnetic field @xmath5 at low temperature @xmath6 .
depending on @xmath0 the screening can vary between full screening and partial screening of the spin-1/2 closest to the surface . in the case of full screening the magnetic properties of the deposited entity are indeed identical to those of the remaining spin system without the screened spin . for our investigations
we employ a single - channel single - impurity kondo model as a minimal model to describe the correlations induced by the exchange interaction between the conduction electrons of the non - magnetic metal and one spin of the magnetic molecule @xcite . as already demonstrated in ref .
@xcite the magnetization steps due to ground state level crossings in an applied magnetic field can be used as fingerprints of the underlying spin model .
this will be shown by investigating different sizes of chains as well as triangular arrangements of magnetic moments .
_ reminder on nrg_in order to model a molecule and its coupling to a surface we use the following hamiltonian which consists of three parts @xcite : @xmath7 the first part @xmath8 represents non - interacting electrons on a lattice .
the hopping parameter @xmath9 is non - zero only if the lattice sites @xmath10 , @xmath11 are nearest neighbours . @xmath12 and @xmath13 are fermionic creation and annihilation operators for electrons with spin direction @xmath14 . the interaction with an external magnetic field @xmath5
is given by the zeeman term with @xmath15 representing the effective electron spin at lattice site @xmath10 , @xmath16 the g - factor and @xmath17 the bohr magneton .
the second part @xmath18 models the impurity , i.e. , the molecule or chain via an effective heisenberg model for all connected spins @xmath19 and a zeeman term .
@xmath20 is the interaction between spins @xmath10 and @xmath11 and antiferromagnetic for @xmath21 , @xmath22 the last part @xmath23 describes the kondo - like interaction of the molecule with the surface .
the coupling constant @xmath0 is positive for antiferromagnetic coupling @xmath24 to calculate thermodynamic values within this modell we use wilson s numerical renormalization group ( nrg ) with the discretization scheme proposed by itko and pruschke @xcite and a @xmath25-averaging for 2 values
. a constant density of states is assumed .
it is possible that in reality molecular orbitals of phthalocyanine molecules hybridize with surface states , compare e.g. @xcite .
our approach , like others @xcite , simplifies the situation to a point where the deposited molecule is reduced to its spin degrees of freedom which interact with the metal s conduction electrons .
_ results and interpretation_before we discuss our numerical results of a finite spin chain interacting with a metal substrate we would like to shortly look at a free three - site chain of spins @xmath3 .
its levels split under the influence of an applied magnetic field as depicted on the l.h.s . of fig .
[ molsurf - f - c ] . at a certain magnetic field value @xmath26 the lowest @xmath27 and
@xmath28 levels cross .
if the lowest spin of the trimer would be completely screened , the remaining dimer would possess a different level scheme with a different crossing field @xmath29 as depicted on the r.h.s . of fig .
[ molsurf - f - c ] .
therefore , infering @xmath1 from spectroscopic data or equivalently from crossing fields strongly depends on the degree of screening .
( l.h.s . ) and of a dimer of two spins @xmath3 ( r.h.s.).,title="fig:",width=151 ] ( l.h.s . ) and of a dimer of two spins @xmath3 ( r.h.s.).,title="fig:",width=151 ] in the following we discuss the so - called impurity magnetization , i.e. the thermal expectation value of molecular magnetic moment @xmath30 as a function of the applied field for various couplings to the substrate . in order to work with reasonable numbers we set the intramolecular coupling to @xmath31 mev and the half - bandwidth of the metal to @xmath32 ev . as the temperature
we choose @xmath33 k , which is lower than the intramolecular coupling , but not too low so that thermodynamic functions are still smooth . after looking at the lowest zeeman levels of a free trimer we now consider the coupling of such a trimer to a surface as described by eq .
( [ e-2 - 4 ] ) and shown in fig .
[ molsurf - f - a ] ( a )
. the strength of the coupling is parametrized by @xmath0 .
the impurity magnetization curves depicted in fig .
[ trimer ] cover the whole range from the free ( @xmath34 ) or weakly coupled case to the strongly coupled case , which is reached for @xmath35 ev . while the case @xmath0 = 0 coincides with the discussed free trimer by construction , the strongly coupled case coincides with the magnetization curve of a free dimer .
( a ) ; vertical lines mark the crossing fields for various scenarios ; @xmath36 k.,width=321 ] additionally to the magnetization curves , fig .
[ trimer ] includes the crossing fields for various scenarios which can be derived from the magnetization curves .
the vertical lines marking these fields shift from the analytical solution of a free trimer towards the analytical solution of a free dimer for increasing coupling to the substrate .
the analytical solutions thus are boundaries for the crossing field independent of the coupling @xmath0 .
focusing on the strong coupling regime , fig .
[ trimer15 ] shows variations of the couplings @xmath1 and @xmath0 . within this regime ,
i.e. for strong enough @xmath0 , the magnetization curves are independent of @xmath0 and coincide with the solution for a free dimer and thus depend on the coupling @xmath1 in the same way the analytical solution does .
in particular the crossing field is given by @xmath37 . given a strong enough coupling to reach the strong coupling regime it is therefore possible to derive @xmath1 from the crossing field .
our investigation also shows that the maximum uncertainty in the determination of @xmath1 , in the case of unknown @xmath0 , is given by the difference between the ( analytical ) solutions for the crossing field of the unscreened system and the fully screened one .
k. the magnetization curve depends only on @xmath1 , not on @xmath0 in the strong coupling case.,width=321 ] similiar results are obtained if a triangular structure is used , where all three spins couple to each other , but still only one couples to the substrate , compare fig .
[ molsurf - f - a ] ( b ) .
for the case of @xmath34 analytical results can be obtained for both trimer chain and triangle @xcite .
these show that the magnetization curves and thus the crossing fields coincide for t @xmath38 0 .
figure [ trimertriangle ] shows the magnetization curves of triangle and trimer chain for finite temperature where the curves for @xmath34 feature small differences .
for intermediate couplings to the substrate , @xmath39 ev , the differences are more pronounced .
k.,width=321 ] figure [ trimertriangle ] furthermore shows the case of strong coupling to the substrate , again @xmath35 ev . in this case trimer chain and triangle
are indistinguishable on the basis of their magnetization curves and thus their crossing fields , compare @xmath40-symbols and upright triangles in fig .
[ trimertriangle ] . after considering the triangle , and thus a slightly different structure of the impurity
, we now come back to chain - like impurities and extend our original trimer chain to four and five spins analogous to the experiment in ref .
@xcite .
figure [ tetramer ] shows the magnetization curves of a tetramer chain for various couplings to the substrate .
k.,width=321 ] the curves are affected by the coupling to the substrate in a similar way to those of the trimer chain with the crossing fields decreasing for stronger couplings . in the strong coupling regime
the tetramer chain then shows the same magnetization curve and crossing field as the free trimer .
analogously the magnetization curve and crossing fields of the pentamer chain , as depicted in fig . [ pentamer ] , coincide with those of the free tetramer in the strong coupling limit .
k.,width=321 ] unlike the other figures , figures [ tetramer ] and [ pentamer ] include vertical lines which confine areas of ( practically ) constant magnetization for the free and the strongly coupled chains at a now very low temperature of @xmath41 k. for those areas an arrow indicates whether the magnetization is increased ( green arrow ) or decreased ( red arrow ) due to the coupling to the substrate .
for the tetramer chain with its ground state spin of @xmath42 the screening leads to an increased magnetization at ( and thus a responce to ) small magnetic fields .
in contrast the magnetization for the pentamer chain with its ground state spin of @xmath43 is decreased at small magnetic fields and thus will not respond to small magnetic fields in the strong coupling regime . summarizing ,
nrg calculations of the impurity magnetization provide a very valuable tool in order to rationalize experimental assumptions and results as for instance those of ref .
@xcite .
future investigations of more complicated impurities as for instance magnetic molecules with non - heisenberg terms in the hamiltonian pose no problem , but nrg is technically limited to single - channel ( or with massive super computing time to two - channel ) contacts .
this is due to the rapidly growing hilbert space dimension which is already for two channels the square of the dimension for one channel . |
first , we briefly introduce the essential background to topic of this paper .
about one century ago , @xcite published an interesting paper titled `` in what way does it become manifest in the fundamental law of physics that space has three dimensions ? '' .
this paper remains actively discussed even today ( see a paper `` dimensionality '' by @xcite ) .
about 40 years ago , in quantum field theory , a method of dimensional regularization , permitting a distortion of space - time dimension @xmath17 , was proposed by @xcite , @xcite and @xcite .
this method is equivalent to assigning an infinitesimal fictitious mass @xmath18 to a photon or introducing a large momentum cut - off @xmath19 in the momentum integration of the feynman diagram ( @xmath20 corresponds to @xmath21 or @xmath22 , where @xmath23 is the mass of an electron ) .
these methods are important in calculations of physical quantities such as the self - energy and the anomalous magnetic moment of electron ( and muon ) , suggesting that non - integer values of @xmath24 may have a feasible physical interpretation . shortly after t hooft and veltman , and their contemporaries proposed their method in 1972 , @xcite conceptualized the fractal dimension of space - time .
inspired by his book , several researchers have investigated fractal dimensions in the natural sciences @xcite .
a range of estimated physical quantities is summarized in table [ tab-1 ] . among these
, the phase space of fractal @xmath25-dimension ( named the hausdorff dimension ) is given by @xcite based on @xcite @xmath26^{d_h}}{\gamma ( 1+d_h/2 ) } , \label{eq-1}\end{aligned}\ ] ] where @xmath27 is the gamma function , and eq .
( [ eq-1 ] ) can be multiplied by a factor @xmath28 in actual calculations .
( [ eq2 ] ) below . ) [ cols="<,<,<",options="header " , ] here , we focus on a paper by @xcite . the planck distribution , i.e. , the black - body radiation law in @xmath29 dimensions ( where @xmath6 and @xmath2 denote the space - dimension and its distortion based on eq .
( [ eq-1 ] ) , respectively ) is given by @xmath30 where @xmath31 , @xmath32 $ ] .
the quantities @xmath33 , @xmath34 , @xmath35 , @xmath36 , and @xmath3 are planck s constant , the light frequency , boltzmann s constant , speed of light , and temperature , respectively .
@xcite suggested that upper limit may be placed on the distortion of space - dimension , such that @xmath37 ( note that their investigation was published eight years earlier than the nasa cobe data @xcite ) .
measurements by nasa s cobe satellite revealed that our universe is almost homogenously filled with cosmic microwave radiation at 2.725 k. @xcite analyzed the @xcite data ( where firas denotes far infrared absolute spectrophotometer ) in terms of the following formula : @xmath38 where @xmath39 is assumed constant .
the results of their analysis are summarized in table [ tab1 ] .
our recent analyses of the same data using eq .
( [ eq2 ] ) @xcite are also depicted in table [ tab1 ] . as evident from the table ,
the analyses yield significantly different results , including the signs of @xmath2 .
these differences are contradicting in spite fact that both studies used the same data , a similar formulation and the cern minuit program . after recalculating our results ( listed in the third row of table [ tab1 ] ) , we are able to attribute these discrepancies to differences in the numerical values of @xmath40 @xcite .
second , we note that the cobe data contain two effects , namely , the chemical potential @xmath0 , and the sunyaev zeldovich ( sz ) effect @xmath1 @xcite in the black - body radiation spectra . as a next step ,
it is worthwhile to analyze the cobe data in terms of these effects , in addition to the distortion of the space - dimension @xmath2 .
as is well known , the cobe monopole data contain the following residual spectrum , given by @xmath41\nonumber\\ & \!\!\!\!\!\!&\!\!\!\!\!\ ! \equiv\ [ { \rm monopole\ data}]\ - [ u_{\rm planck}\ ( t=2.725\ { \rm k } ) ] \label{eq3}\end{aligned}\ ] ] the monopole spectrum is analyzed by means of the bose
einstein distribution , which involves the dimensionless chemical potential @xmath0 , @xcite @xmath42 where @xmath43 .
moreover , it is necessary to include for the effect of inverse compton scattering @xmath44 , described by @xmath45 here , the sz effect @xmath1 @xcite is the parameter for inverse compton scattering , defined by @xmath46 where @xmath47 , @xmath48 , @xmath49 , and @xmath50 denote the size of the high - temperature region in the universe , number density of electrons , the cross section of thomson scattering , and the temperature of electrons , respectively . in this paper , we focus on the positivity of the sz effect @xmath1 . to account for the distortion of the spatial dimension in the cobe monopole data , we adopt the following formula , in which @xmath51 in the above equation is replaced by @xmath52 , @xmath53 \label{eq7}\end{aligned}\ ] ] equation ( [ eq7 ] ) forms the basis of our investigation on the distortion of the space - dimension in the cobe monopole data .
@cccccc@ & @xmath39 or @xmath54 & @xmath3 & @xmath2 & @xmath55 & inputs for @xmath40 + & ( mjy @xmath56 sr@xmath57 ) & ( k ) & ( @xmath58 ) & & ( cm k@xmath57 ) + eq .
( [ eq2 ] ) & @xmath59 & @xmath60 & @xmath61 & @xmath62 & @xmath63 + eq . ( [ eq-3 ] ) & @xmath64 & @xmath65 & @xmath66 & @xmath67 & 1.43939 + + eq .
( [ eq-3 ] ) & @xmath68 & @xmath69 & @xmath70 & @xmath71 & 1.43878 + this paper is divided into several sections . in section [ sec2 ] , we apply eq .
( [ eq7 ] ) , including and excluding @xmath2 , to the cobe monopole data and estimate the physical quantities ( @xmath3 , @xmath2 , @xmath0 , @xmath1 ) and @xmath55 . assuming these estimates , the cobe data are analyzed at fixed temperatures by the monte carlo ( mc ) method .
average parameter values are estimated for allowed combinations of parameter ensembles .
section [ sec3 ] is denoted to the cobe residual spectrum for these combinations of parameter ensembles .
concluding remarks and discussion , including our analyses of relativistic formula of eq .
( [ eq5 ] ) in @xcite and an analysis of the dipole spectrum derived from eq .
( [ eq2 ] ) , are presented in section [ sec4 ] .
applying eq .
( [ eq7 ] ) to the cobe monopole data @xcite , we obtain the results listed in the upper sections of table [ tab2 ] . from these results
, we calculate two parameter sets ( @xmath72 , @xmath0 , @xmath1 ) and ( @xmath73 , @xmath0 , @xmath1 ) at fixed temperatures .
the @xmath55 widths ( @xmath74 ) of the allowed parameter sets are then estimated from the center - column parameters ( listed in the central columns of table [ tab2 ] ) and fig .
[ fig1 ] . from the widths @xmath74 and the standard deviations in the parameters ( @xmath75 , @xmath76 , @xmath77 ) , we can estimate the distributions of parameters ( @xmath2 , @xmath0 , @xmath1 ) at fixed temperatures using the mc method .
more details are provided in the following sub - section .
@crrrr@ @xmath3 ( k ) & @xmath78 & @xmath79 & @xmath80 & @xmath55 + @xmath81 & & @xmath82 & & @xmath67 + @xmath83 & @xmath84 & @xmath85 & & @xmath86 + @xmath87 & & @xmath88 & @xmath89 & @xmath90 + @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath86 + @xmath95 fixed + @xmath96 & & @xmath97 & @xmath98 & @xmath99 + @xmath100 & & @xmath101 & @xmath102 & @xmath103 + @xmath104 & & @xmath105 & @xmath106 & @xmath107 + @xmath108 & & @xmath109 & @xmath110 & @xmath111 + @xmath112 & & @xmath113 & @xmath114 & @xmath115 + @xmath95 fixed + @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 + @xmath104 & @xmath121 & @xmath122 & @xmath123 & @xmath124 + @xmath125 & @xmath126 & @xmath127 & @xmath128 & @xmath129 + @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath139 + @xmath55 as a function of fixed temperature @xmath3 .
( a ) @xmath72 and ( b ) @xmath73,title="fig:",height=207 ] @xmath55 as a function of fixed temperature @xmath3 .
( a ) @xmath72 and ( b ) @xmath73,title="fig:",height=207 ] the random variables are introduced as follows : @xmath140 where @xmath141 are uniform random numbers in @xmath142 ( c.f .
the resulting @xmath55 values are presented in fig .
the parameter distributions obtained from our analysis are shown in fig .
[ fig2 ] . assuming that the sz effect @xmath1 is positive ( as defined in the caption of fig .
[ fig2 ] ) , we can determine the intervals of the temperature @xmath3 and the chemical potential @xmath0 .
the results are presented in table [ tab3 ] .
parameter ensemble ( @xmath143 ) with @xmath144 .
number of events generated is @xmath145 ( 30k ) .
( a ) @xmath146 .
( b ) @xmath147.,title="fig:",height=219 ] parameter ensemble ( @xmath143 ) with @xmath144 .
number of events generated is @xmath145 ( 30k ) .
( a ) @xmath146 .
( b ) @xmath147.,title="fig:",height=219 ] + parameter ensemble ( @xmath143 ) with @xmath144 .
number of events generated is @xmath145 ( 30k ) .
( a ) @xmath146 .
( b ) @xmath147.,title="fig:",height=219 ] parameter ensemble ( @xmath143 ) with @xmath144 .
number of events generated is @xmath145 ( 30k ) .
( a ) @xmath146 .
( b ) @xmath147.,title="fig:",height=219 ] inserting the space distortion @xmath4 into the above calculations yields the distributions shown in fig .
numbers of events satisfying @xmath148 for the two calculations ( assuming isotropic and distorted space ) are listed in table [ tab3 ] .
@lccccc@ calc .
& @xmath149 & @xmath3 ( k ) & @xmath79 & @xmath150 ) & @xmath151 + i ( @xmath1521529 ) & @xmath153 & @xmath154 & @xmath155 & & @xmath156 + i ( @xmath1571142 ) & @xmath158 & @xmath154 & @xmath159 & & @xmath160 + ii ( @xmath1522699 ) & @xmath161 & @xmath162 & @xmath163 & @xmath164 & @xmath165 + ii ( @xmath157597 ) & @xmath166 & @xmath87 & @xmath167 & @xmath168 & @xmath169 + +
if @xmath170 , eq . ( [ eq7 ] ) is simplified , and the residual spectrum becomes @xmath171\nonumber\\ & \!\!\!\!\ !
\cong & \!\!\!\!\ !
n_c(d)\cdot\left(\frac{\nu}c\right)^d\nonumber\\ & \!\!\!\!\!&\!\!\!\!\ !
+ \frac{yxe^x}{(e^x-1)^2}\left(x\coth \frac x2 - 4\right)\right ] .
\label{eq9}\end{aligned}\ ] ] computing the residual spectrum by means of eq .
( [ eq9 ] ) yields fig .
[ fig4 ] , in which the allowed combinations of parameters ensembles are evaluated for @xmath72 and @xmath73 .
this figure reveals slight difference between the residual spectra for @xmath72 and @xmath172 .
analysis of cobe residual spectrum for allowed combinations of parameter ensembles @xmath173 .
( a ) @xmath174 k , @xmath175 and @xmath176 ( see table [ tab3 ] ) , and @xmath177 .
( b ) @xmath178 k , @xmath179 , @xmath180 and @xmath181 ( see table [ tab3 ] ) , and @xmath182.,height=219 ]
* ( c1 ) * we have computed the distortion of the space - dimension @xmath2 in the nasa cobe monopole data @xcite using eqs .
( [ eq-3 ] ) and ( [ eq2 ] ) .
difference between the two results is attributed to numerical difference in the @xmath40 values .
the magnitude of our estimation of @xmath2 is @xmath183 , should be compared with that of @xcite ( namely , @xmath184 ) . as seen in third row of table [ tab1 ] , when we adopt the same numerical value for @xmath40 as @xcite in eq .
( [ eq-3 ] ) , we obtained @xmath185 @xmath186 ( and @xmath187 @xmath188 ) .
it seems to be inadequate to tune the physical constant factor @xmath40 as @xcite .
+ * ( c2 ) * to extend our investigation of the distortion of the space - dimension in the cobe data , we propose that chemical potential @xmath0 and the sz effect @xmath1 , in addition to @xmath2 , be included in eq .
( [ eq7 ] ) . applying eq .
( [ eq7 ] ) to the cobe monopole data and the mc method to parameters sets ( @xmath3 , @xmath2 , @xmath0 , and @xmath1 ) and ( @xmath3 , @xmath72 , @xmath0 , and @xmath1 ) , we have obtained the allowed parameter ensembles , assuming the positivity of the sz effect @xmath1 .
the following limits of the parameters have been obtained at @xmath189 k ; @xmath190 assuming no distortion of the space - dimension , i.e. , @xmath191 , the following limits are obtained at @xmath192 k , @xmath193 the above limit of @xmath194 is almost identical to that reported by @xcite ( @xmath195 )
. however , our estimated limit of @xmath196 is smaller than that of these authors ( @xmath197 ) , probably because sz effect @xmath196 was assumed as positive in our study .
moreover , it is interesting that limits @xmath194 and @xmath196 slightly depend on the magnitude of distortion @xmath198 .
allowing for the distortion of the space - dimension , among three limits we observe the following order : @xmath199 * ( c3 ) * according to @xcite , as the universe is re - ionized at some redshift @xmath200 , during the ionization process , the electrons gain kinetic energy ( estimated at around @xmath201 ev ) . denoting the electron temperature at the re - ionization by @xmath202
, we obtain the following expression for the sz effect : @xmath203 where @xmath204 , @xmath205 , and @xmath33 are the matter density , the baryon density , and the scale parameter , respectively .
given @xmath206 , @xmath207 , the re - ionization redshift for the limit @xmath208 is @xmath209 .
the limit @xmath210 implies that @xmath211 this set of parameters ( @xmath212 ev , and @xmath16 ) yields @xmath213 , which is consistent with the limits in eqs .
( [ eq10 ] ) and ( [ eq11 ] ) . + * ( d1 ) * for the sake of reference , we analyze the same data by means of the relativistic formula denoted by @xmath214 @xcite @xmath215 , \nonumber\\ \label{eq15}\end{aligned}\ ] ] where @xmath216 , @xmath217 , @xmath218 , and @xmath219 @xmath220 , @xmath221 and @xmath222 are explicitly given in @xcite . according to descriptions on temperature of electron @xmath50 s in @xcite , the temperatures of the ionized plasma in cluster of galaxies is approximately on @xmath223 k. our results are presented in table [ tab4 ] .
it is shown that estimated values are almost the same in table [ tab2 ] .
( notice that `` @xmath224 '' in table [ tab4 ] corresponds to `` @xmath1 '' in table [ tab2 ] . ) @ccccccc@ @xmath50 ( kev ) & @xmath3 ( k ) & @xmath79 & @xmath225 & @xmath226 & @xmath78 & @xmath55 + @xmath227 & @xmath87 & @xmath88 & @xmath228 & @xmath89 & & @xmath90 + @xmath229 & @xmath87 & @xmath88 & @xmath230 & @xmath231 & & @xmath90 + @xmath232 & @xmath87 & @xmath88 & @xmath233 & @xmath231 & & @xmath90 + @xmath227 & @xmath234 & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath86 + @xmath229 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath86 + @xmath232 & @xmath244 & @xmath235 & @xmath245 & @xmath246 & @xmath247 & @xmath86 + + * ( d2 ) * the dipole spectrum can be analyzed by the derivative of the planck distribution @xcite , given as : @xmath248 the results of this analysis are provided in table [ tab5 ] and fig . [ fig5 ] . as seen in table [ tab5 ] ,
the @xmath55 is improved by imposing a finite @xmath249 on the system . from the dipole amplitude @xmath250 mk , the velocity of our solar system ( with the universe fixed )
is estimated as @xmath251 km s@xmath57 , about @xmath252 % higher than the estimate of @xcite .
the anomalously large value of @xmath2 obtained in this analysis requires further investigation .
+ * ( d3 ) * from eq .
( [ eq2 ] ) , a modified stefan - boltzmann law is derived as , @xmath261 , \label{eq-18}\end{aligned}\ ] ] where @xmath262 is the riemann s zeta - function and the second term @xmath263 . through eq .
( [ eq-18 ] ) , we are able to discuss the problem of the distortion of the space - dimension in a future in more detail , provided that the error bar of @xmath2 becomes smaller .
the authors sincerely thank prof .
j. c. mather and prof .
d. j. fixsen for their kindness in supplying nasa cobe data and discussing analysis methods during the early stages of this investigation .
subsequently , one of the authors ( m. b. ) would like to thank prof .
n. sugiyama for insightful conversations .
moreover , it is our pleasure to acknowledge dr . f. caruso and dr .
v. oguri for our communications with them .
ehrenfest , p. : proc . of the amsterdam academy
* 20 * , 200 ( 1917 ) ; annalen phys . *
366 * , 440 ( 1920 ) ; _ paul ehrenfest : collected scientific papers hardcover _ , ed . by kelin ,
m.j . , north - holland , reprint ( 1959 ) | a concise explanation of studies on distortion of space - time dimension is briefly introduced .
second we obtain the limits ( i.e. , bounded values ) of the dimensionless chemical potential @xmath0 , the sunyaev zeldovich ( sz ) effect @xmath1 and distortion of the space - dimension @xmath2 by monte carlo ( mc ) analysis of the parameter set ( @xmath3 , @xmath4 , @xmath0 , and @xmath1 ) in cosmic microwave data assuming that the sz effect is positive @xmath5 . in this analysis ,
the magnitude of the space - dimension @xmath6 with distortion of the space - dimension @xmath2 is defined by @xmath4 .
the limits of @xmath0 and @xmath1 are determined as @xmath7 ( @xmath8 ) ( @xmath9 ( @xmath10 ) ) , @xmath11 ( @xmath8 ) ( @xmath12 ( @xmath10 ) ) , while the distortion of the space - dimension is @xmath13 ( @xmath8 ) ( @xmath14 ( @xmath10 ) ) .
the magnitudes of these three estimated limits are ordered as @xmath15 .
the estimated limit of @xmath11 appears to be related to re - ionization processes occurring at redshift @xmath16 .
we also present data analysis assuming a relativistic sz effect . |
lorsquon dveloppe la gomtrie relle algbrique selon le modle de la gomtrie algbrique complexe , des obstacles techniques apparaissent .
lun des exemples le plus emblmatique en tant que les versions algbriques complexes des thormes a et b de cartan ( cf .
@xcite ) nont pas leurs analogues en gomtrie relle algbrique .
dans la thorie rgulire telle que dveloppe dans ( * ? ? ? * chapitre 3 ) , aucun des thormes a et b de cartan nest valable . de mme travers lapproche nash de la gomtrie relle algbrique , les thormes a et b de cartan ne sont pas vrifis .
dans cet article , nous proposons une approche nouvelle de la gomtrie relle algbrique , base sur les fonction rationnelles continues , et plus gnralement les fonctions rationnelles de classe @xmath1 , quon appellera @xmath2-rgulues .
pour ces fonctions , qui constituent un assouplissement des fonctions rgulires lorsque @xmath3 , les analogues des thormes a et b de cartan deviennent valables ( cf .
[ thax ] et [ thbx ] ) .
avant de donner un aperu des principaux rsultats de cet article , il convient de rappeler certaines notions de gomtrie relle algbrique .
suivant un abus courant en gomtrie algbrique , nous dirons quune _ fonction rationnelle _
sur @xmath0 ( cf .
[ dfn.ration ] pour une dfinition formelle ) est une fonction valeurs relles @xmath4 dfinie sur un ouvert de zariski non vide @xmath5 de @xmath0 telle quil existe des polynmes @xmath6 $ ] pour lesquels @xmath7 pour chaque @xmath8 .
cette criture fractionnaire sous - entend que la fonction polynomiale @xmath9 ne sannule pas sur @xmath5 .
deux fonctions rationnelles sur @xmath0 sont alors gales si elles concident sur un ouvert de zariski non vide contenu dans leurs domaines de dfinition .
une fonction rationnelle @xmath4 sur @xmath0 possde un plus grand ouvert de zariski sur lequel elle est dfinie ( cf .
* proposition 3.2.3 ) ) .
cet ouvert est le _ domaine _ de @xmath4 , il est not @xmath10 .
nous noterons @xmath11 le complmentaire du domaine de @xmath4 , cest le _
_ de @xmath4 . a titre dexemple , le domaine de la fonction rationnelle @xmath4 sur @xmath12 dfinie par @xmath13 est gal @xmath14 et @xmath15 . lensemble des fonctions rationnelles sur @xmath0 est un corps et sidentifie au corps des fractions rationnelles @xmath16 , il peut aussi tre not @xmath17 .
une fonction rationnelle @xmath4 sur @xmath0 est une _ fonction rgulire _
sur @xmath0 si elle est dfinie sur @xmath0 tout entier ,
i.e. , si @xmath18 ( * ? ? ?
* dfinition 3.2.1 ) .
a titre dexemple , la fonction rationnelle @xmath19 est rgulire sur @xmath12 .
lensemble des fonctions rgulires sur @xmath0 est un sous - anneau du corps des fonctions rationnelles sur @xmath0 , que nous noterons @xmath20 .
nous justifierons plus loin ce choix de notation .
dans ce travail , nous nous proposons donc dtudier les fonctions rationnelles sur @xmath0 qui stendent par continuit @xmath0 tout entier .
nous entendons ici la continuit par rapport la topologie euclidienne .
remarquons tout de suite quune fonction rationnelle sur @xmath0 tant continue sur son domaine de dfinition , il sagit de fonctions qui stendent par continuit leur lieu dindtermination . plus prcisment , une _ fonction rgulue _
sur @xmath0 est une fonction valeurs relles dfinie en tout point de @xmath0 , qui est continue pour la topologie euclidienne et qui est rationnelle sur @xmath0 . a titre dexemple , la fonction rgulire @xmath21 sur @xmath22 stend par continuit en lorigine et dfinit donc une fonction rgulue sur @xmath23 .
son graphe est la toile du clbre parapluie de cartan ( voir [ exem.umbrellas ] ) .
lensemble des fonctions rgulues sur @xmath0 est un sous - anneau du corps @xmath17 des fonctions
rationnelles sur @xmath0 , que nous notons @xmath24 .
une fonction rgulire sur @xmath0 tant videmment rgulue , nous obtenons une chane de sous - anneaux @xmath25 plus gnralement , une fonction sur @xmath0 est _ @xmath2-rgulue _ , si elle est la fois rgulire sur un ouvert de zariski non vide , et de classe @xmath1 sur @xmath0 .
ici , @xmath2 dsigne un _ entier surnaturel _ , i.e. , @xmath2 est ou bien un entier naturel , ou bien @xmath2 est gal @xmath26 . a titre dexemple , la fonction rgulire @xmath27 sur @xmath22 stend par continuit en lorigine et dfinit une fonction @xmath2-rgulue sur @xmath23 , si @xmath2 est un entier naturel .
nous dmontrons ( cf .
thorme [ thinfregestreg ] ) quune fonction @xmath26-rgulue sur @xmath0 est ncessairement rgulire .
pour @xmath2 un entier surnaturel , lensemble des fonctions @xmath2-rgulues est un sous - anneau du corps @xmath17 des fonctions rationnelles sur @xmath0 , qui sera not @xmath28
. remarquons quil ny a pas de conflit de notation ni avec lanneau des fonctions
rgulues @xmath29 , ni avec lanneau de fonctions rgulires @xmath20 introduits ci - dessus .
nous obtenons finalement une chane de sous - anneaux @xmath30 le plus petit de ses sous - anneaux est gal lintersection de tous les autres sous - anneaux de la chane , i.e. , @xmath31 revenons au contenu de cet article .
dans un premier temps nous dterminons les proprits algbriques de lanneau @xmath28 des fonctions @xmath2-rgulues sur @xmath0 , o @xmath2 est un entier naturel .
ces anneaux ont t assez peu tudis ; les seules rfrences qui nous sont connues tant @xcite .
lanneau @xmath20 des fonctions
rgulires sur @xmath0 , en revanche , a attir beaucoup dattention @xcite .
nous montrons que @xmath28 est un anneau non - noethrien pour lequel le nullstellensatz est valable ( [ nullstellensatz ] ) .
cela est dautant plus remarquable que lintersection de tous ces anneaux , savoir @xmath20 , est un anneau noethrien pour lequel le nullstellensatz nest pas valable !
une version affaiblie du nullstellensatz est nanmoins vraie pour @xmath20 , cette version fait intervenir le radical rel dun idal de fonctions rgulires @xcite .
ce nullstellensatz rel , bien quintressant en lui - mme , ne rpare en rien le dfaut accablant de lanneau des fonctions rgulires qui est de possder trop didaux premiers , ce qui ne manque pas de poser des problmes en gomtrie relle rgulire .
nous verrons , en revanche , que la gomtrie relle @xmath2-rgulue , pour @xmath2 fini , ne pose pas ces problmes , et , de ce fait , se rapproche plus de la gomtrie algbrique sur un corps algbriquement clos que la gomtrie relle rgulire . malgr le caractre non noethrien de lanneau @xmath28 , son usage reste raisonnable en gomtrie algbrique car nous dmontrons que son spectre de zariski @xmath32 est un espace topologique noethrien .
de manire quivalente ( grce la validit du nullstellensatz justement ) lensemble @xmath0 muni de la topologie @xmath2-rgulue est un espace topologique noethrien , lorsque @xmath2 est fini ( cf .
[ topregnoeth ] ) .
ici et dans toute la suite de cet article , la topologie @xmath2-rgulue est la topologie do nt une base de ferms est la collection des sous - ensembles de la forme @xmath33 o @xmath4 est @xmath2-rgulue sur @xmath0 .
cette topologie est strictement plus fine que la topologie de zariski sur @xmath0 , lorsque @xmath34 .
cette dernire pouvant galement tre dnomme topologie rgulire ou @xmath26-rgulue .
dans un deuxime temps , nous posons les bases de ltude des varits rgulues abstraites .
nous revenons sur un fibr en droites rgulier pathologique sur @xmath23 bien connu
qui nest pas engendr par ses sections rgulires globales ( * ? ? ?
* example 12.1.5 ) , et montrons que cette pathologie disparait si ce fibr est interprt comme fibr en droites rgulu .
nous en dduisons alors la validit des thormes a et b de cartan dans le cadre rgulu ( [ thax ] et [ thbx ] ) .
nous concluons par un chapitre consacr la caractrisation gomtrique des ferms rgulus de @xmath0 et montrons un rsultat fondamental : les ferms rgulument irrductibles de @xmath0 concident avec les sous - ensembles algbriquement constructibles ferms de @xmath0 ( thorme . [ const ] ) .
dans le cas des courbes et des surfaces , nous poussons notre tude et proposons en particulier une relecture _ rgulue _ des fameux parapluies de cartan , de whitney , et dun parapluie de kollr .
nous introduisons aussi un nouveau parapluie _ cornu _ ( [ exem.umbrellas ] ) . a notre connaissance , les fonctions rgulues ont t tudies de faon systmatique pour la premire fois par kucharz dans @xcite ( o elles sont appeles _ continuous rational _ ) .
dans son article , kucharz montre que ce sont les bonnes fonctions pour approcher le plus algbriquement possible les fonctions continues .
il dmontre notamment que toute classe dhomotopie dune application continue entre deux sphres de dimensions quelconques contient une application rgulue , l o les applications polynomiales et rgulires font dfaut !
dans @xcite , kollr tudie les problmes de restriction et dextension de fonctions continues rationnelles dfinies sur une varit relle algbrique affine .
lorsque la varit est lisse , notre notion de fonction rgulue ( cf .
dfinition [ dfn.regulue ] ) concide avec la notion de fonction continue rationnelle de kollr et de kucharz .
lorsque la varit est singulire , notre notion de fonction rgulue concide avec ce que kollr appelle `` hrditairement rationnelles continues '' ( cf .
remarque [ rem.here ] ) .
mentionnons quune nouvelle version @xcite de larticle @xcite , en collaboration avec k. nowak simplifie certains arguments .
objets naturels , les fonctions rgulues apparaissent aussi dans des rsultats antrieurs .
en 1978 , voir @xcite , kreisel remarque ( sans employer le terme `` fonction rgulue '' bien sr ) que le positivstellensatz de stengle @xcite permet de reprsenter tout polynme @xmath35 $ ] positif sur @xmath0 comme une somme de carrs de fonctions rgulues sur @xmath0 .
il est noter que ltude des isomorphismes @xmath26-rgulus sur les surfaces a connu rcemment des progrs importants , cf .
e.g. @xcite .
pour tre complets , rappelons que sur @xmath36 , toute fonction rationnelle continue sur une varit _
normale _ est rgulire .
dans le cas dune varit singulire gnrale , ltude des fonctions rationnelles continues amne au concepts de semi - normalit et de semi - normalisation , cf .
@xcite .
nous remercions j. kollr pour nous avoir transmis une version prliminaire de son article , ainsi que s. cantat , m. coste , l. evain , w. kucharz , k. kurdyka , d. naie et a. parusiski pour lintrt prcoce quils ont port nos travaux et pour leurs suggestions qui ont contribu amliorer cet article .
merci aussi f. broglia et f. acquistapace pour nous avoir signal les rfrences @xcite et @xcite .
ce travail a bnfici dun support partiel provenant du contrat anr `` birpol '' anr-11-js01 - 004 - 01 .
soit @xmath37 un entier naturel . on dfinit , de manire gnrale , lensemble des zros et lensemble des non zros dune fonction relle sur @xmath0 , ou dun ensemble de fonctions relles : soit @xmath38 une fonction relle sur @xmath0 . on note @xmath39 lensemble des zros de @xmath4 dans @xmath0 , i.e. , @xmath40 lensemble des non zros de @xmath4 est @xmath41 soit @xmath42 un ensemble de fonctions relles sur @xmath0 . on note
@xmath43 lensemble des zros communs des fonctions dans @xmath42 , i.e. , @xmath44 on a les proprits habituelles suivantes : soit @xmath37 un entier naturel . 1 . @xmath45 et @xmath46 .
soit @xmath47 , @xmath48 , une collection densembles de fonctions relles sur @xmath0 .
alors @xmath49 3 .
soient @xmath50 un nombre fini densembles de fonctions relles sur @xmath0 , o @xmath51 est un entier naturel .
alors @xmath52 4 .
soit @xmath42 un sous - ensemble dun anneau @xmath53 de fonctions relles sur @xmath0 .
alors @xmath54 , o @xmath55 est lidal de @xmath53 engendr par @xmath42 .
un proprit moins habituelle , mais cruciale en gomtrie relle , est la suivante : [ prop.princip ] soit @xmath37 un entier naturel .
soit @xmath53 un anneau de fonctions relles sur @xmath0 , et soient @xmath56 un nombre
fini dlments de @xmath53 .
alors , il existe une fonction relle @xmath57 telle que @xmath58 plus prcisment , la fonction @xmath59 convient .
rappelons quun sous - ensemble @xmath60 de @xmath0 est un _ ferm de zariski _ , sil existe un sous - ensemble @xmath42 de lanneau des fonctions polynomiales relles sur @xmath0 tel que @xmath61 comme ce dernier anneau est isomorphe lanneau noethrien des polynmes rels @xmath62 $ ] , un sous - ensemble @xmath60 de @xmath0 est un ferm de zariski si et seulement sil existe un nombre fini de fonctions polynomiales @xmath56 sur @xmath0 telles que @xmath63 en appliquant la proposition prcdente [ prop.princip ] , on pourra encore dire quun sous - ensemble @xmath60 de @xmath0 est un ferm de zariski si et seulement sil existe une fonction polynomiale @xmath4 sur @xmath0 telle que @xmath64 les ferms de zariski de @xmath0 constituent la collection des ferms dune topologie sur @xmath0 , la _ topologie de zariski _ sur @xmath0 .
soit @xmath37 un entier naturel , et soit @xmath5 un ouvert de zariski de @xmath0 .
soit @xmath4 une fonction relle sur @xmath5 , et @xmath8 .
la fonction @xmath4 est _ rgulire en @xmath65 _ sil existe un voisinage zariski ouvert @xmath66 de @xmath65 dans
@xmath5 et des fonctions polynomiales @xmath67 et @xmath9 sur @xmath0 tel que @xmath68 pour tout @xmath69 , o il est sous - entendu que @xmath9 ne sannule pas sur @xmath66 .
la fonction @xmath4 est rgulire sur @xmath5 si elle est rgulire en tout point de @xmath5 . on note @xmath70 lensemble des fonctions rgulires sur @xmath5 .
cet ensemble est une algbre relle de manire vidente .
soit @xmath66 un ouvert de zariski contenu dans @xmath5 , la restriction @xmath66 dune fonction rgulire sur @xmath5 est rgulire sur @xmath66 .
il sensuit que @xmath71 est un prfaisceau dalgbres relles sur @xmath0 .
compte tenu de la dfinition locale dune fonction rgulire , il est vident que @xmath71 est un faisceau sur @xmath0 .
soit @xmath65 un point de @xmath0 .
la fibre @xmath72 sidentifie avec lalgbre relle locale des fonctions rationnelles @xmath73 dfinies en @xmath65 , i.e. , @xmath74 .
le faisceau @xmath71 sur @xmath0 est donc un faisceau en algbres relles locales .
il est remarquable que les sections du faisceau @xmath71 au - dessus dun ouvert de zariski @xmath5 de @xmath0 admettent une description globale ( cf .
* proposition 3.2.3 ) ) : [ prq(u ) ] soit @xmath37 un entier naturel , et soit @xmath5 un ouvert de zariski de @xmath0 .
soit @xmath4 une fonction relle sur @xmath5 .
alors @xmath4 est rgulire sur @xmath5 si et seulement sil existe des fonctions polynomiales @xmath67 et @xmath9 sur @xmath0 telles que @xmath75 , avec @xmath76 , pour tout @xmath8 .
voir proposition [ prqf(u ) ] ci - dessous o on dmontre lnonc dans un cadre plus gnral .
soit @xmath37 un entier naturel ,
et soit @xmath60 un ferm de zariski de @xmath0 .
on considre @xmath60 muni de la topologie induite par la topologie de zariski , quon appelle _ topologie de zariski _ sur @xmath60 .
notons que cette topologie est encore noethrienne .
soit @xmath77 le faisceau didaux de @xmath71 des fonctions rgulires sannulant sur @xmath60 . le faisceau quotient @xmath78 sera not @xmath79 .
son support est gal @xmath60 . de ce fait , on considre @xmath79 comme un faisceau sur @xmath60 .
cest un faisceau en algbres relles locales de corps rsiduel @xmath12 . on peut
donc le considrer comme un sous - faisceau des fonctions relles sur @xmath60 , et on lappelle le faisceau des fonctions rgulires sur @xmath60 . on a nouveau une description globale
des sections de @xmath79 ( cf .
* proposition 3.2.3 ) ) : [ prqf(u ) ] soit @xmath37 un entier naturel , et soit @xmath60 un ferm de zariski de @xmath0 .
soit @xmath5 un ouvert de zariski de @xmath60 et soit @xmath4 une fonction relle sur @xmath5 .
alors @xmath4 est rgulire sur @xmath5 si et seulement sil existe des fonctions polynomiales @xmath67 et @xmath9 sur @xmath0 telles que @xmath75 , avec @xmath76 , pour tout @xmath8 . pour la commodit du lecteur , on reproduit la dmonstration de ( * ? ? ?
* proposition 3.2.3 ) .
il suffit de dmontrer limplication directe .
supposons donc que @xmath4 est rgulire sur @xmath5 .
comme la topologie de zariski sur @xmath60 est noethrienne , il existe un recouvrement ouvert fini @xmath80 de @xmath5 et des fonctions polynomiales @xmath81 sur @xmath0 telles que @xmath82 pour tout @xmath83 , pour @xmath84 .
soit @xmath85 une fonction polynomiale sur @xmath0 avec @xmath86 , pour tout @xmath87 .
montrons que @xmath88 sur @xmath5 .
montrons dabord que le second membre , quon notera @xmath89 dans la suite , est bien une fonction rgulire sur @xmath5 . soit @xmath8 .
il existe @xmath87 avec @xmath83 . on a donc @xmath90 .
comme @xmath91 , pour @xmath92 , le dnominateur de @xmath89 ne sannule pas en @xmath65 .
le second membre est donc bien une fonction rgulire sur @xmath5 .
il reste montrer que @xmath93 sur @xmath5 .
soit @xmath8 .
lensembles des indices @xmath87 pour lesquels @xmath83 .
lorsque @xmath94 , on a @xmath95 , et donc aussi @xmath96 observons que cette dernire formule est galement valable lorsque @xmath97 puisque @xmath98 dans ce cas .
il sensuit que @xmath99 et donc aussi que @xmath100 .
une _ varit relle algbrique affine
_ est un espace localement annel @xmath101 , o @xmath102 est un faisceau en algbres relles locales ayant la proprit suivante .
il existe un entier naturel @xmath37 et un ferm de zariski @xmath60 de @xmath0 tels que @xmath103 soit isomorphe @xmath101 .
si @xmath101 est une varit relle algbrique affine , on appellera encore sa topologie la topologie de zariski sur @xmath104 , et une section de @xmath102 sur un ouvert @xmath5 une fonction rgulire sur @xmath5 . on pourrait
penser que cette dfinition induit une notion naturelle de varit relle algbrique abstraite , mais lintrt dune telle notion est limit par le fait que tout sous - ensemble localement
ferm de @xmath0 ( voire de @xmath105 ) muni de son faisceau naturel des fonctions rgulires est une varit relle algbrique affine ( voir ( * ? ? ?
* proposition 3.2.10 ) , ( * ? ? ?
* thorme 3.4.4 ) et la remarque qui suit ) .
[ dfn.ration ] soit @xmath104 une varit relle algbrique affine .
une _ fonction rationnelle _
sur @xmath104 est une classe dquivalence de paires @xmath106 , o @xmath5 est un ouvert dense ( i.e. non - vide ) de @xmath104 et @xmath4 est une fonction rgulire sur @xmath5 .
deux telles paires @xmath106 et @xmath107 tant quivalentes sil existe un ouvert dense @xmath108 de @xmath104 contenu dans @xmath109 tel que @xmath110 . notons que cette dernire condition est quivalente la condition @xmath111 soit @xmath4 une fonction rationnelle sur la varit relle algbrique affine @xmath104 .
considrons lensemble @xmath112 de toutes les paires @xmath107 qui sont quivalentes @xmath106 .
on introduit une relation dordre partiel sur @xmath112 en dfinissant @xmath113 lorsque @xmath114 .
comme @xmath102 est un faisceau , lensemble @xmath112 contient un plus grand lment @xmath107 .
comme cet lment est unique , on appelle @xmath66 le domaine de dfinition de @xmath4 et on le note @xmath10 .
du coup , on pourra galement considrer @xmath4 comme fonction rgulire dfinie sur @xmath10 , son extension rationnelle tant unique .
[ prop.pq ] soit @xmath104 une varit relle algbrique affine et soit @xmath4 une fonction rationnelle sur @xmath104 .
alors il existe des fonctions rgulires @xmath67 et @xmath9 sur @xmath104 , avec @xmath115 sur @xmath10 , telles que @xmath116 sur @xmath10 .
consquence immdiate de la proposition [ prqf(u ) ] . on aurait
pu sattendre ce que les fonctions @xmath67 et @xmath9 ci - dessus soient des fonctions polynomiales sur @xmath104 .
la notion de fonction polynomiale
sur une varit relle algbrique affine nas pas de sens intrinsque .
cette notion dpent du choix dun plongement de @xmath104 dans un espace affine @xmath0 .
cela dit , si @xmath104 est un ferm de zariski de @xmath0 , pour un certain entier naturel @xmath37 , on peut supposer que @xmath67 et @xmath9 sont des fonctions polynomiales avec les proprits voulues , comme le montre la dmonstration de [ prqf(u ) ] .
soit @xmath104 une varit relle algbrique affine .
lensemble des fonctions rationnelles sur @xmath104 est un anneau de manire vidente , _
lanneau total des fonctions rationnelles _
sur @xmath104 . on le note @xmath117 .
soit @xmath104 une varit relle algbrique affine irrductible .
alors @xmath117 est un corps .
soit @xmath104 une varit relle algbrique affine et notons @xmath118 ses composantes irrductibles . les morphismes de restriction @xmath119 induisent un isomorphisme dalgbres relles @xmath120 en particulier , lanneau @xmath117 est un produit de @xmath51 corps .
rappelons la dfinition dune varit relle algbrique lisse ( cf .
* definition 3.3.9 ) ) : soit @xmath104 une varit relle algbrique affine .
la varit @xmath104 est _
lisse _ en un point @xmath65 de @xmath104 si lanneau local @xmath121 est rgulier .
la varit @xmath104 est _ lisse _
si elle lest en chacun de ses points . on rappelle le rsultat suivant ( cf .
* proposition 3.3.10 ) ) : soit @xmath104 une varit relle algbrique affine et notons @xmath118 ses composantes irrductibles .
alors , @xmath104 est lisse si et seulement si 1 .
@xmath104 est runion disjointe des @xmath122 , et 2 .
chaque varit relle algbrique affine @xmath122 est lisse . on va tendre la notion de fonction
@xmath2-rgulue de lintroduction aux fonctions relles dfinies sur une varit relle algbrique affine lisse .
remarquons que si @xmath104 est une varit relle algbrique affine lisse , alors @xmath104 possde une structure sous - jacente de varit diffrentiable de classe @xmath1 , pour tout entier surnaturel @xmath2 ( cf .
* proposition 3.3.6 ) ) .
la dfinition suivante est donc naturelle .
[ dfn.regulue ] soit @xmath2 un entier surnaturel .
soit @xmath104 une varit relle algbrique affine lisse , et soit @xmath4 une fonction relle sur @xmath104 .
la fonction @xmath4 est _
sur @xmath104 si 1 .
@xmath4 est de classe @xmath1 sur @xmath104 , et 2 .
il existe un ouvert de zariski @xmath5 dense dans @xmath104 tel que @xmath123 est rgulire . on dira quune fonction relle sur @xmath104 est _ rgulue _
lorsquelle est @xmath124-rgulue sur @xmath104 . on note @xmath125 lensemble des fonctions @xmath2-rgulues sur @xmath104 .
cet ensemble est de toute vidence un anneau .
si @xmath126 , la dfinition ci - dessus a mme un sens lorsque la variet relle algbrique affine est singulire .
pourtant , elle ne donne pas lieu la bonne notion de fonction rgulue sur une telle varit .
en effet , kollr montre quune fonction rationnelle sur une varit relle algbrique affine singulire qui stend par continuit toute la varit peut avoir des proprits non souhaitables ( * ? ? ?
* ex . 2 ) .
voil pourquoi on introduit , pour linstant , uniquement la notion de fonction @xmath2-rgulue sur une varit relle algbrique affine lisse .
pour une varit relle algbrique affine lisse @xmath104 donne on dispose dune chane croissante @xmath127 le but du paragraphe suivant est de montrer que @xmath128 .
soit @xmath4 une fonction relle dfinie sur @xmath0 .
le but de cette section est de dmontrer que si @xmath4 est @xmath26-rgulue sur @xmath0 , alors elle est rgulire sur @xmath0 .
cet nonc est connu ( voir par exemple ( * ? ? ?
* proposition 2.1 ) ) , mais nous allons en donner une dmonstration un peu plus dtaille . on a bien linclusion @xmath129 .
le but de cette section est de montrer que cette inclusion est une galit .
dans le reste de larticle , comme dans lintroduction , il sera ainsi justifi dutiliser la notation @xmath20 pour lanneau des fonctions rgulires sur @xmath0 .
soit @xmath4 une fonction relle dfinie sur @xmath0 .
rappellons que @xmath4 est _ semi - algbrique _
si son graphe est un sous - ensemble semi - algbrique de @xmath130 ( * ? ? ?
* dfinition 2.2.5 ) .
rappelons galement que @xmath4 est _ nash _
si @xmath4 est la fois semi - algbrique et de classe @xmath131 ( * ? ?
* dfinition 2.9.3 ) .
de manire quivalente , @xmath4 est nash si @xmath4 est une fonction analytique relle et elle est algbrique sur lanneau des fonctions rgulires sur @xmath0 , i.e. , il existe des fonctions polynomiales @xmath132 sur @xmath0 telles que @xmath133 sur @xmath0 , o @xmath134 est non identiquement nulle ( * ? ? ?
* proposition 8.1.8 ) .
[ regulsa ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
une fonction @xmath2-rgulue sur @xmath0 est semi - algbrique .
en particulier , une fonction @xmath26-rgulue sur @xmath0 est nash .
soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 .
notons @xmath5 le domaine de @xmath4 comme fonction rationnelle , et soient @xmath67 et @xmath9 deux fonctions polynomiales sur @xmath0 telles que @xmath75 pour tout @xmath135 .
clairement , le graphe de la restriction de @xmath4 @xmath5 est un sous - ensemble semi - algbrique de @xmath136 .
comme @xmath4 est continue , le graphe de la fonction @xmath4 sur @xmath0 est ladhrence de lensemble prcdent pour la topologie euclidienne dans @xmath137 . or , ladhrence dun ensemble semi - algbrique est semi - algbrique ( * ? ? ?
il sensuit que @xmath4 est une fonction semi - algbrique .
[ zregulsa ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
un sous - ensemble ferm @xmath2-rgulu de @xmath0 est semi - algbriquement ferm .
[ thinfregestreg ] soit @xmath37 un entier naturel . on a @xmath138
i.e. , une fonction relle sur @xmath0 est @xmath26-rgulue si et seulement si elle est rgulire .
cet nonc se gnralise certainement au cas o @xmath104 est une varit relle algbrique affine lisse mais la preuve en devient plus technique et ne nous semble pas apporter dide nouvelle .
comme une fonction rgulire est trivialement @xmath26-rgulue , il suffit de dmontrer la rciproque .
soit @xmath5 le domaine de la fonction rationnelle @xmath4 .
ecrivons @xmath116 , o @xmath67 et @xmath9 sont des fonctions polynomiales sur @xmath0 , @xmath9 ne sannulant pas sur @xmath5 .
on montre que la fraction rationnelle @xmath73 est dfinie en tout point de @xmath0 .
autrement dit , on montre que , pour tout @xmath139 , la fraction rationnelle @xmath73 appartient lanneau local @xmath72 des germes des fonctions rgulires en @xmath65 .
il suffit de le montrer pour @xmath65 lorigine de @xmath0 .
daprs la proposition prcdente , la fonction @xmath4 sur @xmath0 est semi - algbrique .
comme @xmath4 est de classe @xmath131 , elle est nash .
une fonction nash sur @xmath0 est , en particulier , analytique relle ( * ? ? ?
* proposition 8.1.6 ) . du coup , @xmath4 stend un ouvert de @xmath140 contenant @xmath0 comme fonction analytique complexe , encore note @xmath4 .
en particulier , @xmath4 dfinit un germe dune fonction analytique complexe en @xmath124 ayant la proprit que @xmath141 .
la fonction polynomiale @xmath9 divise donc @xmath67 dans lanneau local @xmath142 des germes des fonctions analytiques complexes en @xmath124 .
cela implique que @xmath9 divise @xmath67 dans la compltion de @xmath142 , savoir lanneau des sries formelles complexes @xmath143 $ ] .
ce dernier est aussi la compltion de lanneau local @xmath144 des fonctions rationnelles complexes sur @xmath140 dfinies en @xmath124 .
comme ce dernier anneau est noethrien , @xmath9 divise @xmath67 dans @xmath144 ( cf .
* corollary 2 , p. 257 ) .
comme @xmath145 la fonction rationnelle relle @xmath73 appartient @xmath146 . on a donc dmontr que @xmath4 est une section globale du faisceau @xmath71 , i.e. , @xmath4 est une fonction rgulire sur @xmath0 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
un sous - ensemble @xmath60 de @xmath0 est un _ ferm @xmath2-rgulu _ sil existe un sous - ensemble @xmath42 de @xmath28 tel que @xmath61 remarquons que @xmath60 nest pas a priori le lieu des zros communs dun nombre fini de fonctions @xmath2-rgulues .
un sous - ensemble @xmath5 de @xmath0 est un _ ouvert @xmath2-rgulu
_ si son complmentaire est un ferm @xmath2-rgulu .
les ouverts @xmath2-rgulus de @xmath0 constituent une topologie sur @xmath0 , la _
topologie @xmath2-rgulue_. soit @xmath104 un espace topologique .
on rappelle que @xmath104 est _ noethrien _ si et seulement si toute suite dcroissante dnombrable @xmath147 , @xmath148 , de sous - ensembles ferms de @xmath104 est stationnaire
. de manire quivalente , tout recouvrement ouvert dun ouvert de @xmath104 contient un sous - recouvrement fini .
compte tenu du thorme [ thinfregestreg ] , un ouvert @xmath26-rgulu de @xmath0 est un ouvert de zariski , et rciproquement .
comme lanneau des fonctions polynomiales sur @xmath0 est noethrien , la topologie de zariski sur @xmath0 est noethrienne .
il sensuit que la topologie @xmath26-rgulue sur @xmath0 est noethrienne . on verra
ci - dessous quil en est de mme pour la topologie @xmath2-rgulue lorsque @xmath2 est fini ( voir corollaire [ topregnoeth ] ) .
si @xmath149 , la topologie @xmath2-rgulue est plus fine que la topologie @xmath150-rgulue .
en particulier , la topologie @xmath2-rgulue est plus fine que la topologie de zariski sur @xmath0 .
lexemple suivant montre que la topologie @xmath2-rgulue est strictement plus fine que la topologie de zariski , lorsque @xmath2 est un entier naturel et @xmath34 .
[ cubique ] soit @xmath151 la courbe cubique dquation @xmath152 dans @xmath23 .
comme le polynme dfinissant la courbe @xmath151 est irrductible dans @xmath153 $ ] , le sous - ensemble @xmath151 de @xmath23 est un ferm de zariski irrductible .
en particulier , la clture de zariski de @xmath154 est gale @xmath151 .
soit @xmath4 la fonction rationnelle sur @xmath23 dfinie par @xmath155 il est clair que @xmath4 stend par continuit en une fonction rgulue sur @xmath23 . on a @xmath156 et @xmath157 .
remarquons que lorigine @xmath158 est un point isol de la courbe @xmath151 .
par consquent , @xmath154 est rgulument ferm dans @xmath23 , mais non ferm au sens de zariski . en particulier ,
le ferm @xmath26-rgulu irrductible @xmath151 nest pas irrductible pour la topologie rgulue sur @xmath23 et scrit comme runion de deux sous - ferms rgulus stricts .
@xmath159 pour montrer que @xmath154 est un ferm @xmath2-rgulu , pour tout entier naturel @xmath2 , on peut supposer que @xmath160 est impair .
posons @xmath161 la fonction @xmath162 est bien @xmath2-rgulue sur @xmath23 , et , comme @xmath160 est impair , on a @xmath163 .
cela montre bien que ce dernier ensemble est un ferm @xmath2-rgulu , pour tout entier naturel @xmath2 .
cet exemple reprsente en fait le cas gnral .
en effet , on verra dans le corollaire [ cor - const ] que les topologies @xmath2-rgulue et @xmath150-rgulue coincident pour @xmath2 et @xmath150 des entiers naturels quelconques .
soit @xmath37 un entier naturel et @xmath2 un entier surnaturel .
dans ce paragraphe nous allons tudier les fonctions @xmath2-rgulues sur @xmath0 , et les comparer avec dautres classes de fonctions sur @xmath0 .
dans lnonc suivant , et dans le reste de larticle dailleurs , on utilise librement la notion de dimension , ou plutt de codimension , dun ferm de zariski de @xmath0 , et plus gnralement dun ensemble semi - algbrique de @xmath0 @xcite .
[ codim ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath164 .
soient @xmath67 et @xmath9 des fonctions polynomiales sur @xmath0 telles que @xmath75 pour chaque @xmath165 .
si @xmath166 sont premiers entre eux , alors @xmath167 et @xmath168 .
il suffit de montrer lnonc lorsque @xmath126 .
montrons dabord linclusion @xmath169 .
soit @xmath170 .
nest pas identiquement nulle , lensemble de ses zros @xmath171 est nulle part dense dans @xmath0 .
il existe donc une suite @xmath172 dans @xmath0 convergeant vers @xmath65 telle que @xmath173 , pour tout @xmath2 . on a alors
@xmath174 i.e. , @xmath175 .
montrons ensuite que @xmath176 .
par labsurde , supposons que @xmath177 . comme @xmath9
nest pas identiquement nulle , on a @xmath178 .
il existe donc un diviseur irrductible @xmath179 de @xmath9 dans @xmath62 $ ] avec @xmath180 .
daprs ce qui prcde , @xmath181 .
comme @xmath180 , on en dduit que @xmath179 divise @xmath67 ( * ? ? ?
4.5.1 , p.85 ) .
cela contredit lhypothse que @xmath67 et @xmath9 sont premiers entre eux .
comme on a vu , lnonc ci - dessus peut tre considrablement renforc lorsque @xmath182 .
en effet , dans ce cas la fonction @xmath4 est rgulire et est de la forme @xmath73 o @xmath9 ne sannule pas sur @xmath0 , i.e. , @xmath183 .
soient @xmath184 et @xmath2 un entier surnaturel .
toute fonction @xmath2-rgulue sur @xmath0 est rgulire .
soit @xmath2 un entier surnaturel .
une fonction @xmath2-rgulue sur @xmath23 est rgulire en dehors dun ensemble fini .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath164 .
le lieu @xmath11 o @xmath4 nest pas une fonction rgulire est un ferm de zariski de @xmath0 de codimension @xmath185 .
une application de @xmath0 dans @xmath186 est _
@xmath2-rgulue _ lorsque toutes ses fonctions coordonnes le sont .
il nest pas clair que la composition de deux applications @xmath2-rgulues soit encore @xmath2-rgulue .
un cas o cest vident est le suivant : soient @xmath187 des entiers naturels et @xmath2 un entier surnaturel .
soient @xmath188 deux applications @xmath2-rgulues .
si limage de @xmath4 est de codimension @xmath189 , alors la composition @xmath190 est @xmath2-rgulue .
on verra ci - dessous que la conclusion est valable mme si la codimension de limage de @xmath4 est plus grand que @xmath191 ( voir corollaire [ cor.compositionregulues ] ) .
soit @xmath104 une varit relle algbrique affine lisse .
soit @xmath192 une sous - varit relle algbrique lisse de @xmath104 .
soient @xmath193 des fonctions rgulires sur @xmath104 engendrant lidal de @xmath192 dans @xmath104 .
notons @xmath4 lapplication rgulire de @xmath194 dans @xmath105 de coordonnes homognes @xmath193 .
_ @xmath195 de @xmath104 le long de @xmath192 est la clture de zariski du graphe de @xmath4 dans le produit @xmath196 .
lclat @xmath195 est encore une varit relle algbrique affine lisse .
lapplication de projection @xmath197 est une application rgulire , et birgulire au - dessus de @xmath194 .
limage rciproque de @xmath192 par @xmath198 peut sidentifier avec le fibr normal projectif de @xmath192 dans @xmath104 . on aura
besoin du rsultat suivant : [ thindeterminations ] soit @xmath104 une varit relle algbrique affine lisse et @xmath4 une fonction rationnelle sur @xmath104 .
alors il existe une composition finie dclatements de centres lisses @xmath199 telle que lapplication rationnelle @xmath200 est rgulire sur @xmath201 toute entire .
pour une dmonstration , on renvoit le lecteur vers le livre de kollr @xcite , notamment le corollaire 3.18 amlior en utilisant thorme 3.21 au lieu de thorme 3.17 , quon applique une complexification lisse @xmath202 de @xmath104 .
soit @xmath104 une varit relle algbrique affine lisse .
une fonction relle @xmath4 sur @xmath104 est _
rgulire aprs clatements _ ( _ blow - regular _ , en anglais ) sil existe une composition dclatements centres lisses @xmath203 telle que la fonction relle @xmath204 est rgulire .
cette notion est lanalogue algbrique des fonctions qui sont nash aprs clatements ou encore des fonctions analytiques relles aprs clatements @xcite .
[ threguliereaeclt ] soit @xmath104 une varit relle algbrique affine lisse .
soit @xmath4 une fonction relle sur @xmath104 .
la fonction @xmath4 est rgulue sur @xmath104 si et seulement si @xmath4 est rgulire aprs clatements . plus prcisment , @xmath4 est rgulue si et seulement sil existe une composition finie dclatements
centres lisses @xmath205 telle que la composition @xmath206 est rgulire .
supposons que @xmath4 est rgulire aprs clatements , et montrons que @xmath4 est rgulue .
il existe une suite dclatements de centres lisses @xmath199 telle que @xmath204 est rgulire .
comme @xmath207 est birationnelle , il existe un ouvert de zariski , qui est dense dans @xmath104 , sur lequel @xmath4 est rgulire .
il nous reste montrer que @xmath4 est continue . or
, la topologie euclidienne sur @xmath104 est la topologie quotient induite par la topologie euclidienne sur @xmath208 .
le fait que @xmath204 soit continue implique donc que @xmath4 est continue .
cela montre bien que @xmath4 est rgulue lorsque @xmath4 est rgulire aprs clatements .
montrons la rciproque .
soit @xmath4 une fonction rgulue sur @xmath104 .
il existe un ouvert de zariski @xmath5 , dense dans @xmath104 , tel que @xmath123 est rgulire . en appliquant thorme [ thindeterminations ] @xmath4 , vue comme fonction rationnelle sur @xmath104 , il existe une suite dclatements de centres lisses @xmath199 telle que lapplication rationnelle @xmath209 est rgulire sur @xmath208 toute entire .
par dfinition de composition dapplications rationnelles , cela veut dire que lapplication rgulire @xmath210 de @xmath211 dans @xmath212 stend @xmath213 toute entire comme application rgulire .
cette extension est ncessairement continue . or
, la composition dapplications continues @xmath204 est galement une extension continue de @xmath210 .
comme @xmath211 est dense dans @xmath213 pour la topologie euclidienne , ces deux extensions sont la mme ce qui veut dire que @xmath204 est rgulire , et @xmath4 est rgulire aprs clatements .
une fonction semi - algbrique @xmath214 est appele _ analytique par arcs _
@xcite si @xmath215 est analytique pour tout arc analytique @xmath216 , o @xmath55 est un intervalle ouvert de @xmath12
. dans ( * ? ? ?
1.1 ) , bierstone et milman montrent que @xmath214 est analytique par arcs si et seulement si @xmath4 devient analytique ( et mme nash ) aprs une suite finie dclatements centres algbriques lisses bien choisis .
comme une fonction rgulire est analytique , on obtient la consquence suivante du thorme [ threguliereaeclt ] , grce la proposition [ regulsa ] : [ reguluestarcanalytique ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
une fonction @xmath2-rgulue sur @xmath0 est analytique par arcs .
il existe bien videmment des fonctions analytiques par arcs qui ne sont pas rgulues .
par exemple @xmath217 est analytique par arcs @xcite et nest clairement pas rgulue sur @xmath23 . on rappelle quun ensemble @xmath42 semi - algbrique dans @xmath0 est dit _ symtrique par arcs _
si et seulement si pour tout arc analytique @xmath218 - 1,1[\rightarrow { \mathbf{r}}^n$ ] , si @xmath219 - 1,0[)\subseteq e$ ] alors @xmath219 - 1,1[)\subseteq e$ ] , voir @xcite .
le lien avec les fonctions analytiques par arcs est le suivant .
si @xmath4 est une fonction analytique par arcs sur @xmath0 , alors son lieu de zros @xmath39 est symtrique par arcs .
suivant les notations de @xcite , on appelle @xmath220 la topologie do nt les ensembles ferms sont les ensembles symtriques par arcs .
kurdyka a montr que la topologie @xmath220 est noethrienne ( * ? ? ?
* th . 1.4 . ) .
on verra ci - dessous que la topologie @xmath220 est plus fine que la topologie rgulue . on en dduira que la topologie @xmath2-rgulue est noethrienne , pour tout @xmath2 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath60 un ferm @xmath2-rgulu de @xmath0 .
alors @xmath60 est un ensemble symtrique par arcs .
soit @xmath221 un ferm @xmath2-rgulu de @xmath0 .
daprs le corollaire [ reguluestarcanalytique ] , @xmath39 est symtrique par arcs , pour tout @xmath222 .
comme les sous - ensembles symtriques par arcs sont les ferms dune topologie , @xmath223 est encore symtrique par arcs .
[ topregnoeth ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
la topologie @xmath2-rgulue sur @xmath0 est noethrienne .
[ cofkresthyp ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
tout ferm @xmath2-rgulu de @xmath0 est de la forme @xmath39 , pour une certaine fonction @xmath2-rgulue @xmath4 sur @xmath0 .
[ cor - eclt ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
tout ferm @xmath2-rgulu @xmath60 de
@xmath0 est un ferm de zariski aprs clatements , i.e. , il existe une composition dclatements centres lisses @xmath224 telle que limage rciproque @xmath225 est zariski ferme dans @xmath226 .
daprs corollaire [ cofkresthyp ] , il existe une fonction @xmath2-rgulue @xmath4 sur @xmath0 telle que @xmath227 .
daprs thorme [ threguliereaeclt ] , il existe une composition dclatements de @xmath0 centres lisses @xmath207 telle que @xmath204 est rgulire .
en particulier , @xmath228 est un ferm de zariski .
on verra dans la proposition [ recip - eclt ] que la rciproque du corollaire ci - dessus est valable .
mme si la topologie @xmath2-rgulue sur @xmath0 est plus fine , @xmath0 reste irrductible pour cette topologie : soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
alors @xmath0 muni de la topologie @xmath2-rgulue est irrductible .
supposons que @xmath229 , o @xmath60 et @xmath230 sont des ferms @xmath2-rgulus de @xmath0 .
il existe des fonctions @xmath2-rgulues @xmath4 et @xmath89 sur @xmath0 telles que @xmath227 et @xmath231 . on a @xmath232 par consquent , @xmath233 .
lanneau @xmath28 tant un sous - anneau du corps @xmath17 , il est intgre . on a donc @xmath234 ou @xmath235 ,
i.e. , @xmath236 ou @xmath237 .
rappelons quun sous - ensemble localement ferm de @xmath0 , au sens de zariski , est un sous - ensemble de la forme @xmath238 , o @xmath239 est un ferm de zariski et @xmath240 un ouvert de zariski . [ thm.stratif ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel
. soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 .
alors , il existe une stratification finie de @xmath0 @xmath241 en sous - ensembles localement ferms de @xmath0 au sens de zariski telle que la restriction @xmath242 est rgulire , pour tout @xmath87 . il suffit de traiter le cas @xmath126 .
on peut , bien - sr , supposer que @xmath4 est non identiquement nulle sur @xmath0 .
notons @xmath243 lespace affine @xmath244 $ ] sur @xmath12 , de sorte que @xmath245 et @xmath246 .
la conjugaison complexe @xmath247 agit sur @xmath248 . on identifie
@xmath249 avec lensemble des points fixes de @xmath247 sur @xmath248 .
ecrivons @xmath250 avec @xmath251 et @xmath252 des polynmes rels en @xmath253
. soit @xmath192 le sous - schma rduit de @xmath243 dfini par le polynme @xmath254 .
comme @xmath4 est non nulle , @xmath192 est un sous - schma de @xmath243 de codimension @xmath191 . appliquons le thorme de hironaka au sous - schma @xmath192 de @xmath243 en clatant @xmath243 , de manire successive , mais seulement en des centres do nt lensemble des points rels est dense . plus
prcisment , il existe une suite finie de morphismes @xmath255 o chaque morphisme @xmath256 est un clatement de centre lisse @xmath257 de codimension @xmath185 ayant les proprits suivantes . pour tout @xmath87 , lensemble des points rels @xmath258 est dense dans @xmath259 , et le transform strict @xmath260 de @xmath192 dans @xmath261 na que des points rels lisses .
notons @xmath262 le diviseur exceptionnel de @xmath263 .
on peut supposer , de plus , que @xmath264 intersecte @xmath265 transversalement en des points rels lisses de @xmath266 . on note
encore @xmath207 la composition @xmath267 .
par construction , @xmath204 est une fonction rationnelle sur @xmath261 do nt le domaine de dfinition contient tous les points rels de @xmath261 .
soit @xmath67 un point du schma @xmath243 do nt le corps rsiduel @xmath268 est rel .
la fibre @xmath269 est une runion finie connexe despaces projectifs sur @xmath268 .
soit @xmath270 lun de ces espaces projectifs .
comme le domaine de dfinition de @xmath204 contient tous les points rels de @xmath261 , la fonction rationnelle @xmath204 se restreint une fonction rationnelle sur @xmath270 .
il existe un point @xmath268-rationnel @xmath9 de @xmath270 en lequel @xmath204 est dfinie .
soit @xmath60 le sous - schma de @xmath271 do nt le point gnrique est @xmath67 , et @xmath230 le sous - schma de @xmath261 do nt le point gnrique est @xmath9 .
la restriction de @xmath207 @xmath230 est un morphisme birationnel dans @xmath60 .
il existe donc des ouverts non vides @xmath5 de @xmath60 et @xmath66 de @xmath230 tels que @xmath272 soit un isomorphisme de @xmath66 sur @xmath5 . quitte remplacer @xmath5 et @xmath66 par des ouverts plus petits , on peut supposer que @xmath204 est dfinie sur @xmath66 . or @xmath207 induit , sur les points rels , un isomorphisme birgulier de @xmath273 sur @xmath274 .
il sensuit que la restriction de @xmath4 @xmath274 est rgulire .
la construction prcdente montre que pour tout idal premier rel @xmath67 de @xmath62 $ ] , il existe un sous - ensemble localement ferm irrductible @xmath275 dense de @xmath276 tel que @xmath277 est rgulire . il existe alors un nombre fini didaux premiers rels distincts @xmath278 tels que @xmath279 pour simplifier la notation , on notera @xmath280 au lieu de @xmath281 .
on construit ensuite une stratification selon le procd habituel : en agrandissant la famille @xmath282 si ncessaire , on peut supposer que toutes les composantes irrductibles des intersections finies des sous - ensembles de la forme @xmath280 appartiennent encore la famille @xmath282 .
introduisons un ordre partiel sur la collection @xmath282 . on pose @xmath283 si @xmath280 est contenu dans
la clture de zariski @xmath284 de @xmath285 .
soit @xmath286 le sous - ensemble @xmath287 de @xmath0 est localement ferm , et contenu dans @xmath280 .
la restriction de la fonction @xmath4 @xmath287 est donc bien rgulire .
montrons que @xmath288 lorsque @xmath289 .
cest clair lorsque @xmath290 ou linverse .
supposons donc que @xmath280 et @xmath285 ne sont pas comparables .
ecrivons lintersection @xmath291 comme runion des @xmath292 o @xmath2 parcourt un ensemble dindice @xmath55 .
comme on a @xmath293 et @xmath294 , pour tout @xmath295 , on a bien @xmath288 , lorsque @xmath296 .
il nous reste montrer que @xmath297 .
soit @xmath139 et soit @xmath280 le plus petit lment de la collection @xmath282 tel que @xmath298 . on a bien @xmath299 .
rappelons quun sous - ensemble @xmath42 de @xmath0 est _ zariski constructible _ ou _ algbriquement constructible _
sil est runion finie de sous - ensembles localement ferms au sens de zariski .
[ coregfermeestconstructible ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
un ensemble ferm @xmath2-rgulu de @xmath0 possde une stratification finie en sous - ensembles localement ferms de @xmath0 .
en particulier , tout ensemble @xmath2-rgulument ferm de @xmath0 est zariski constructible .
notons quun sous - ensemble @xmath2-rgulument ferm de @xmath0 nest pas forcment localement ferm , comme le montre lexemple suivant .
[ exkregfermeconstructible ] soit @xmath300 la courbe
cubique ayant la singularit isole en @xmath158 de lexemple [ cubique ] .
soit @xmath301 .
le sous - ensemble @xmath302 est un sous - ensemble @xmath2-rgulument ferm de @xmath303 qui nest pas localement ferm au sens de zariski .
en effet , la clture de zariski de @xmath60 est gale @xmath304 .
le sous - ensemble @xmath60 de @xmath304 nest pas zariski ouvert pour la topologie induite car son intersection avec @xmath305 est rduite lorigine de @xmath303
et nest pas ouvert dans @xmath305 .
dans le cas @xmath182 , lnonc du corollaire [ coregfermeestconstructible ] peut tre considrablement renforc car les ferms sont alors les ferms de zariski .
le thorme [ thm.stratif ] a encore lnonc suivant comme consquence : [ cor.compositionregulues ] soient @xmath187 des entiers naturels et @xmath2 un entier surnaturel .
soient @xmath188 deux applications @xmath2-rgulues .
alors la composition @xmath306 est @xmath2-rgulue .
lapplication @xmath307 est bien - sr de classe @xmath1 .
soit @xmath5 lintersection des domaines des fonctions coordonnes de @xmath4 .
la restriction de @xmath4 @xmath5 est donc une application rgulire dans @xmath186 .
daprs le thorme [ thm.stratif ] , il existe une stratification @xmath308 en sous - ensembles localement ferms de @xmath186 au sens de zariski telle que la restriction @xmath309 est rgulire , pour tout @xmath87 .
soit @xmath87 tel que @xmath310 est zariski dense dans @xmath5 .
soit @xmath311 un ouvert zariski dense dans @xmath5 avec @xmath312 .
@xmath311 de @xmath307 est alors rgulire .
[ cokregcont ] soient @xmath313 des entiers naturels et @xmath2 un entier surnaturel .
une application @xmath2-rgulue de @xmath0 dans @xmath186 est continue pour la topologie @xmath2-rgulue .
soit @xmath314 une application @xmath2-rgulue et @xmath60 un ferm @xmath2-rgulu de @xmath186 .
il existe une fonction @xmath2-rgulue @xmath89 sur @xmath186 do nt lensemble des zros est gal @xmath60 .
daprs le corollaire [ cor.compositionregulues ] , @xmath307 est @xmath2-rgulue .
@xmath315 est donc un ferm @xmath2-rgulu de @xmath0 .
grce au corollaire [ cor.compositionregulues ] , on est en mesure de montrer que lanneau @xmath28 nest pas noetherien , lorsque @xmath2 est fini , mme si la topologie @xmath2-rgulue est noethrienne : soient @xmath37 et @xmath2 des entiers naturels .
lanneau @xmath28 nest pas noethrien lorsque @xmath34 .
pour @xmath316 , soit @xmath317 la fonction @xmath2-rgulue sur @xmath0 dfinie par @xmath318 soit @xmath319 lidal de @xmath28 engendr par les fonctions @xmath320 . on montre
comme dans ( * ? ? ?
6.11 ) que la suite didaux croissante
@xmath321 nest pas stationnaire .
en fait , on montre par labsurde que @xmath322 . en effet , supposons que @xmath323 appartient @xmath319 .
on peut donc crire @xmath324 o @xmath325 sont des fonctions @xmath2-rgulues sur @xmath0 . en restreignant la courbe relle algbrique @xmath326 , quon identifie avec @xmath12 , on obtient @xmath327 sur @xmath151 .
daprs corollaire [ cor.compositionregulues ] , les fonctions @xmath328 et @xmath329 sont @xmath2-rgulues et donc rgulires . de plus ,
la fonction @xmath330 sannule au point @xmath331 avec multiplicit @xmath332 , pour @xmath84 .
il sensuit que @xmath333 sannule en @xmath331 avec multiplicit au moins @xmath332 .
contradiction .
soient @xmath334 et @xmath37 des entiers naturels , avec @xmath335 .
soit @xmath4 une fonction de classe @xmath1 sur un ouvert @xmath5 de @xmath0 .
habituellement , on crit le
@xmath336-ime polynme de taylor de @xmath4 en un point @xmath337 sous la forme @xmath338 lorsquon sintresse celui - ci comme fonction de @xmath337 , on est amen le noter @xmath339 o @xmath340 dsigne un systme de coordonnes affine arbitraire sur @xmath0 .
on appelle ce dernier le _
@xmath336-jet _ de @xmath4 , et on le note @xmath341 . le fibr vectoriel trivial @xmath342 sur @xmath0 de base @xmath343 est le _ fibr des @xmath336-jets _ sur @xmath0 .
le @xmath336-jet @xmath341 de @xmath4 est une section de @xmath342 de classe @xmath344 au - dessus @xmath5 .
le jonglage notationnel ci - dessus a lavantage que le @xmath124-jet @xmath345 de @xmath4 coincide avec @xmath4 . plus gnralement , le coefficient de @xmath346 du @xmath336-jet de @xmath4 est gal la drive partielle dordre @xmath55 de @xmath4 .
le fibr vectoriel des @xmath336-jets @xmath342 possde une structure
de fibr en algbres si on dfinit @xmath347 si @xmath4 et @xmath89 sont des fonctions de classe @xmath1 sur un ouvert @xmath5 de @xmath0 , on a @xmath348 sur @xmath5 . comme le fibr vectoriel @xmath342 est libre dune base explicite , on peut identifier ses sections au - dessus dun ouvert @xmath5 de @xmath0 avec des applications de @xmath5 dans @xmath349 , o @xmath350 est le rang de @xmath342 .
du coup , cela a un sens de parler de _ section rationnelle _ ou encore de _
section @xmath51-rgulue _ de @xmath342 , o @xmath51 est un entier surnaturel quelconque .
[ lejellrat ] soit @xmath336 un entier naturel .
soit @xmath4 une fonction rationnelle sur @xmath0 , et soit @xmath5 son domaine .
le @xmath336-jet @xmath341 est une section rationnelle de @xmath342 de domaine @xmath5 .
ecrivons @xmath116 sur @xmath5 , o @xmath67 et @xmath9 sont des fonctions polynomiales , @xmath9 ne sannulant pas sur @xmath5 . on sait
que la drive partielle dordre @xmath55 de @xmath4 est de la forme @xmath351 , o @xmath251 est une fonction polynomiale sur @xmath0 .
il sensuit que @xmath341 est une section rationnelle de @xmath342 de domaine @xmath352 . comme le coefficient de @xmath353 est @xmath354 , le domaine @xmath66 est gal @xmath5 .
[ prj1c0 ] soit @xmath4 une fonction relle sur @xmath0 .
la fonction @xmath4 est @xmath191-rgulue si et seulement si @xmath4 est rationnelle et la section rationnelle @xmath355 de @xmath356 au - dessus de @xmath10 stend de manire continue @xmath0 . limplication directe est vidente compte tenu du lemme [ lejellrat ] .
montrons donc limplication rciproque , et soit @xmath4 une fonction rationnelle telle que @xmath355 stend de manire continue @xmath0 .
notons @xmath5 le domaine de @xmath4 .
daprs lhypothse , @xmath4 est de classe @xmath357 sur @xmath5 .
soit @xmath358 le complmentaire de @xmath5 , i.e. , @xmath359 .
comme @xmath358 est un sous - ensemble zariski ferm strictement contenu dans @xmath0 , il existe un systme de coordonnes affine @xmath360 sur @xmath0 avec la proprit suivante .
aucune des fibres des projections associes @xmath361 omettant la @xmath87-ime coordonne @xmath362 nest contenue entirement dans @xmath358 .
quitte remplacer @xmath65 par @xmath360 , on peut supposer que la proprit est satisfaite par le systme des coordonnes @xmath65 , i.e. , pour tout @xmath363 et pour tout @xmath364 , lintersection @xmath365 est un ensemble fini .
comme @xmath4 est de classe @xmath357 sur @xmath5 , les drives partielles @xmath366 , @xmath363 , de @xmath4 existent et sont continues sur @xmath5 .
de plus , daprs lhypothse , elles stendent par continuit sur @xmath0 tout entier . en particulier , @xmath367 stend par continuit sur @xmath368 pour tout @xmath364 et pour tout @xmath87 .
daprs un rsultat classique en analyse ( * ? ? ?
* theorem 11.7 ) , cela implique que la restriction de @xmath367 @xmath368 est de classe @xmath357 sur @xmath368 , pour tout @xmath337 et @xmath87 .
( nous utilisons ici la finitude de @xmath365 . )
autrement dit , @xmath4 est partiellement diffrentiable sur @xmath0 tout entier , et ses drives partielles dordre @xmath191 sont continues sur @xmath0 .
daprs un autre rsultat classique danalyse ( * ? ?
* thm 9.21 ) , @xmath4 est alors de classe @xmath357 sur @xmath0 . [ thjkc0 ] soit @xmath2 un entier naturel .
soit @xmath4 une fonction relle sur @xmath0 .
la fonction @xmath4 est @xmath2-rgulue si et seulement si @xmath4 est rationnelle et la section rationnelle @xmath369 de @xmath370 au - dessus de @xmath10 stend de manire continue @xmath0 .
limplication directe tant encore vidente , on dmontre limplication rciproque .
le cas @xmath126 est trivial .
supposons que @xmath371 et soit @xmath4 une fonction rationnelle de domaine @xmath5 telle que la section @xmath369 au - dessus de @xmath5 stend de manire continue @xmath0 .
les drives partielles @xmath372 dordre @xmath373 de @xmath4 ont des @xmath191-jets qui stendent de manire continue @xmath0 . daprs la proposition [ prj1c0 ] , ces drives partielles sont toutes @xmath191-rgulues .
il sensuit que @xmath4 est @xmath2-rgulue .
[ thjlck - l ] soient @xmath2 un entier surnaturel et @xmath336 un entier naturel @xmath374 .
soit @xmath4 une fonction relle sur @xmath0 .
la fonction @xmath4 est @xmath2-rgulue si et seulement si @xmath4 est rationnelle et la section rationnelle @xmath341 de @xmath342 au - dessus de @xmath10 stend de manire @xmath344 @xmath0 .
on peut supposer que @xmath2 est fini .
comme prcdemment , il suffit de dmontrer limplication indirecte .
soit @xmath4 une fonction rationnelle do nt toutes les drives dordre @xmath375 sont @xmath376-rgulues .
daprs limplication directe du thorme [ thjkc0 ] , le @xmath2-jet de @xmath4 stend de manire continue @xmath0 .
daprs limplication indirecte du mme thorme , @xmath4 est @xmath2-rgulue .
un rle cl dans la suite est jou par une version @xmath2-rgulue de lingalit de lojasiewicz sous sa forme de ( * ? ? ?
2.6.4 ) : [ lojaregulrn ] soient @xmath37 et
@xmath2 des entiers naturels .
soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 .
si @xmath377 est @xmath2-rgulue , alors il existe un entier naturel @xmath378 tel que lextension @xmath0 par @xmath124 de @xmath379 est @xmath2-rgulue .
daprs lingalit de lojasiewicz classique , il existe un entier naturel @xmath380 tel que pour tout @xmath55 avec @xmath381 , lextension de @xmath382 par @xmath124 est continue sur @xmath0 tout entier .
soit @xmath383 .
comme @xmath4 est @xmath2-rgulue sur @xmath0 , tout coefficient du @xmath2-jet @xmath384 de @xmath385 est de la forme @xmath386 , o @xmath387 est une fonction rgulue sur @xmath0 .
du coup , @xmath384 lui - mme est de la forme @xmath388 , o @xmath252 est une section rgulue de @xmath370 au dessus de @xmath0 .
il sensuit que lextension par @xmath124 de @xmath389 est une section rgulue de @xmath370 au - dessus @xmath0 tout entier .
daprs le thorme [ thjkc0 ] , la fonction @xmath379 est @xmath2-rgulue sur @xmath0 .
le lemme ci - dessus nous sera utile sous la forme suivante : [ lojaregulu ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath5 un ouvert @xmath2-rgulu de @xmath0 , et soit @xmath4 une fonction @xmath2-rgulue sur @xmath5 .
si @xmath377 est une fonction @xmath2-rgulue , il existe un entier naturel @xmath378 tel que lextension @xmath5 par @xmath124 de @xmath379 est @xmath2-rgulue .
soit @xmath387 une fonction @xmath2-rgulue sur @xmath0 telle que @xmath390 .
daprs le lemme [ lojaregulrn ] , il existe un entier naturel @xmath380 tel que lextension par @xmath124 de @xmath391 est @xmath2-rgulue sur @xmath0 .
comme @xmath89 est @xmath2-rgulue sur @xmath392 , il existe encore daprs le lemme [ lojaregulrn ] , un entier naturel @xmath378 tel que lextension par @xmath124 de @xmath393 est @xmath2-rgulue sur @xmath0 .
il sensuit que lextension par @xmath124 de @xmath379 @xmath394 est @xmath2-rgulue .
remarquons que les deux noncs ci - dessus sont faux lorsque @xmath182
. un contre - exemple est le suivant .
soit @xmath4 la fonction sur @xmath23 dfinie par @xmath395 , et soit @xmath89 la fonction sur @xmath22 dfinie par @xmath396 .
les fonctions @xmath4 et @xmath89 sont bien de classe @xmath131 , mais lextension @xmath23 par @xmath124 de @xmath379 nest de classe @xmath131 pour aucun entier naturel @xmath378 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath5 un ouvert @xmath2-rgulu de @xmath0 .
la topologie de zariski sur @xmath5 est la topologie induite par la topologie de zariski sur @xmath0 .
une fonction relle @xmath4 dfinie sur @xmath5 est @xmath2-rgulue sur @xmath5 si @xmath4 est de classe @xmath1 sur @xmath5 , et il existe des fonctions polynomiales @xmath67 et @xmath9 sur @xmath0 telles que @xmath116 sur un ouvert de zariski dense de @xmath5 . on note
@xmath397 lensemble des fonctions @xmath2-rgulues sur @xmath5 .
cest un anneau sous les oprations habituelles .
si @xmath66 est un
ouvert @xmath2-rgulu de @xmath0 contenu dans @xmath5 , lapplication de restriction de @xmath397 dans @xmath398 est un morphisme danneaux .
il est clair que @xmath399 est donc un prfaisceau .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
le prfaisceau @xmath399 est un faisceau sur @xmath0 pour la topologie @xmath2-rgulue .
il suffit de montrer lnonc suivant .
soit @xmath400 un recouvrement ouvert dun ouvert @xmath2-rgulu @xmath5 de @xmath0 .
supposons que @xmath367 est une fonction @xmath2-rgulue sur @xmath401 , pour tout @xmath87 , telles que @xmath402 pour tout @xmath403 .
alors il existe une fonction @xmath2-rgulue @xmath4 sur @xmath5 do nt la restriction @xmath401 est gale @xmath367 .
en effet , il existe une fonction @xmath4 sur @xmath5 de classe @xmath1 do nt la restriction @xmath401 est gale @xmath367 pour tout @xmath87 . pour montrer que @xmath4 est @xmath2-rgulue , on peut supposer que @xmath5 est non vide .
du coup , il existe @xmath87 tel que @xmath401 est non vide .
comme @xmath0 est irrductible pour la topologie @xmath2-rgulue , tout ouvert @xmath2-rgulu non vide de @xmath0 est dense .
en particulier , un ouvert de zariski dense dans @xmath401 est aussi dense dans @xmath5 .
comme @xmath367 est rgulire sur un ouvert de zariski dense dans @xmath401 , la fonction @xmath4 est rgulire sur un ouvert de zariski dense dans @xmath5 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
le faisceau @xmath399 sur @xmath0 est un faisceau en @xmath12-algbres locales .
la paire @xmath404 est un espace localement annel en @xmath12-algbres .
soient @xmath313 des entiers naturels et @xmath2 un entier surnaturel .
une application @xmath2-rgulue de @xmath0 dans @xmath186 induit naturellement un morphisme despaces localement annels en @xmath12-algbres de @xmath404 dans @xmath405 .
cela tablit une bijection entre lensemble des applications @xmath2-rgulues de @xmath0 dans @xmath186 et lensemble des morphismes despaces localement annels en @xmath12-algbres de @xmath404 dans @xmath405 .
soit @xmath4 une application @xmath2-rgulue de @xmath0 dans @xmath186 .
daprs le corollaire [ cokregcont ] , @xmath4 est une application continue de @xmath0 dans @xmath186 , par rapport aux topologies @xmath2-rgulues .
de plus , si @xmath89 est une fonction @xmath2-rgulue sur un ouvert @xmath2-rgulu @xmath5 de @xmath186 , la composition @xmath406 est une fonction @xmath2-rgulue sur louvert @xmath2-rgulu @xmath407 de @xmath0 , daprs le corollaire [ cor.compositionregulues ] . on en dduit donc
un morphisme despaces localement annels en @xmath12-algbres @xmath408 rciproquement , tant donn un tel morphisme @xmath409 , lapplication @xmath4 est une application @xmath2-rgulue de @xmath0 dans @xmath186 car la @xmath87-ime fonction coordonne @xmath367 de @xmath4 est @xmath410 et est une fonction @xmath2-rgulue sur @xmath0 , pour @xmath84 .
grce la version @xmath2-rgulue de lingalit de lojasiewicz , on a une description de lanneau des sections du faisceau @xmath399 au - dessus dun ouvert comme une localisation de @xmath28 , lorsque @xmath2 est fini : [ restriction ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath5 un ouvert @xmath2-rgulu de @xmath0 , et soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 telle que @xmath411 .
alors , le morphisme de restriction de @xmath28 dans @xmath397 induit un isomorphisme @xmath412 soit @xmath413 le morphisme de restriction .
comme la restriction de @xmath4 @xmath5 ne sannule pas , @xmath414 induit un morphisme @xmath415 on montre que @xmath416 est un isomorphisme .
soit @xmath417 .
daprs le lemme [ lojaregulu ] , il existe un entier naturel @xmath378 tel que @xmath379 se prolonge en une fonction @xmath2-rgulue sur @xmath0 .
par consquent @xmath416 est surjectif .
pour montrer que @xmath416 est injectif , il suffit de montrer lnonc suivant .
pour toute fonction @xmath2-rgulue @xmath387 sur @xmath0 do nt la restriction @xmath5 est identiquement nulle , la fonction @xmath418 est identiquement @xmath124 sur @xmath0 .
cet nonc est clair lorsque @xmath419 , car @xmath234 dans ce cas .
si @xmath420 , alors @xmath5 est dense dans @xmath0 pour la topologie euclidienne .
il sensuit que @xmath421 sur @xmath0 , et en particulier , @xmath422 . on rappelle
quun idal @xmath55 dun anneau commutatif @xmath53 est _ rel _ sil satisfait la proprit suivante : si @xmath423 , avec @xmath424 , alors @xmath425 pour chaque @xmath84 .
[ radreel ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath426 un idal radical .
alors @xmath55 est un idal rel . on
suppose @xmath423 avec @xmath427 .
pour @xmath428 on a @xmath429 .
en effet cest la compose des applications @xmath2-rgulues @xmath430 et de @xmath431 , qui est bien @xmath2-rgulue daprs le corollaire [ cor.compositionregulues ] .
donc @xmath432 .
lidal @xmath55 tant radical on obtient @xmath425 .
la proposition prcdente montre que lidal engendr par @xmath433 dans @xmath434 nest pas radical . en effet ,
sil est radical alors il est rel et par consquent @xmath435 et donc @xmath436 avec @xmath437 .
mais @xmath438 nest clairement pas continue en lorigine .
soit @xmath55 un idal radical dans @xmath28 .
on sait que @xmath55 est un idal rel daprs la proposition [ radreel ] .
on montre que cette proprit est conserve en intersectant avec les polynmes .
[ reelintersection ] soit @xmath55 un idal radical dans @xmath28 .
alors @xmath439 $ ] est un idal rel et on a @xmath440 .
lidal @xmath441 est rel car @xmath55 est rel ( proposition [ radreel ] ) . on obtient @xmath440 par le nullstellensatz rel ( * ? ? ?
4.1.4 ) soit @xmath441 un idal de @xmath62 $ ] .
on regarde le comportement de @xmath441 quand on ltend dans les fonctions rgulues . on fixe dabord quelques notations .
soit @xmath139 .
dans la suite , on note @xmath442|\,\,p(x)=0\}\ ] ] @xmath443 les idaux maximaux de @xmath62 $ ] et @xmath28 respectivement , des polynmes et des fonctions rgulues qui sannulent en @xmath65 . plus
gnralement , si @xmath444 , on note dans la suite @xmath445|\,\,p(x)=0\,\,\forall x\in a\}\ ] ] et @xmath446 examinons lexemple suivant .
soit @xmath447|\,\,p(o)=0\}$ ] lidal maximal de @xmath153 $ ] o @xmath158 est lorigine de @xmath23 .
on montre que @xmath448 nest pas maximal dans @xmath449 .
+ on a clairement @xmath450 . on a aussi @xmath451 . on va montrer par contre que @xmath452 .
supposons que @xmath4 scrive @xmath453 avec @xmath454 .
alors on peut crire @xmath455 mais alors @xmath4 serait diffrentiable lorigine et on obtient une contradiction .
cet exemple montre que lon peut sattendre quelques surprises .
[ radideal ] soit @xmath441 un idal rel de @xmath62 $ ] .
alors @xmath456 on note @xmath457 .
soit @xmath458 , il existe @xmath459 tel que @xmath460 . comme @xmath461 ( * ? ? ? * thm .
4.1.4 ) dans @xmath62 $ ] car @xmath441 est rel , on en dduit que @xmath462 sannule identiquement sur @xmath463 . par consquent @xmath4 sannule aussi identiquement sur @xmath463 i.e. @xmath464 . on a montr une inclusion
. pour linclusion rciproque , on suppose que @xmath441 est lidal @xmath465 avec @xmath466 $ ] .
alors on a @xmath467 avec @xmath468 .
soit @xmath164 tel que @xmath469 .
soit @xmath470 , cest une fonction @xmath2-rgulue sur @xmath471 donc sur @xmath472 .
par le lemme [ lojaregulrn ] , il existe un entier positif @xmath378 tel que @xmath473 tendue par @xmath124 sur @xmath39 est @xmath2-rgulue sur @xmath0 . on a clairement @xmath474 et de plus @xmath475 : en effet
, sur @xmath472 cest vident et si @xmath476 on a aussi @xmath477 ( on utilisera plusieurs fois dans la suite ce mme argument ) .
par consquent @xmath478 car @xmath479 .
dans le cas o @xmath441 est maximal , on obtient [ radmax ] on a @xmath480 [ diff ] on suppose @xmath126 et soit @xmath481 .
il existe un entier strictement positif @xmath378 tel que @xmath385 soit diffrentiable en @xmath65 avec @xmath482 . on
suppose @xmath483 . en regardant la preuve du thorme prcdent
, il existe un entier strictement positif @xmath378 tel que @xmath484 avec @xmath485 . on a donc @xmath486 avec @xmath487 vrifiant @xmath488 .
on peut donc crire @xmath489 ce qui termine la preuve .
soit @xmath53 un anneau .
on dira quun idal @xmath55 de @xmath53 est _ radicalement principal _
sil existe @xmath491 tel que @xmath492 dans ce cas , on dira aussi que @xmath4 est un _ gnrateur radical _ de @xmath55 . on dit que @xmath53 est _ radicalement principal _
si tous ses idaux le sont .
[ prgenrad ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath55 un idal de @xmath28 , et supposons que @xmath491 soit telle que @xmath493 .
alors , @xmath4 est un gnrateur radical de @xmath55 , i.e. , @xmath494 comme @xmath491 , on a bien - sr linclusion @xmath495 .
pour montrer linclusion rciproque , il suffit de montrer que @xmath496 .
soit @xmath497 .
la fonction @xmath498 est bien dfinie et @xmath2-rgulue sur @xmath499 .
daprs le lemme [ lojaregulrn ] , il existe un entier naturel @xmath378 tel que lextension @xmath387 par @xmath124 de @xmath500 sur @xmath501 est encore @xmath2-rgulue . on a @xmath502 sur @xmath0 .
cela implique que @xmath503 .
[ radprinc ] soient @xmath37 et @xmath2 des entiers naturels .
lanneau @xmath28 est radicalement principal . plus prcisment , soit @xmath55 un idal de @xmath28 .
alors il existe une fonction @xmath2-rgulue @xmath491 telle que @xmath494 daprs le corollaire [ topregnoeth ] , la topologie @xmath2-rgulue sur @xmath0 est noethrienne .
il existe donc un nombre fini de fonctions @xmath2-rgulues @xmath56 dans @xmath55 telles que @xmath504 soit @xmath505 comme @xmath491 et @xmath493 , la fonction @xmath4 est un gnrateur radical de @xmath55 daprs la proposition [ prgenrad ] .
[ prfinradi ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 , et soit @xmath55 un idal de @xmath28 .
alors , @xmath506 si et seulement si @xmath507 . daprs la proposition [ radprinc ] , il existe @xmath508 telle que @xmath509 .
en particulier , on a @xmath510 lnonc est donc une consquence de la proposition suivante .
[ prfinradg ] soient @xmath37 et @xmath2 des entiers naturels .
soient @xmath4 et @xmath89
des fonctions @xmath2-rgulues sur @xmath0 .
alors , @xmath511 si et seulement si @xmath512 . supposons que @xmath511 , et soit @xmath513 .
il existe un entier naturel @xmath378 tel que @xmath514 , i.e. , @xmath515 pour une certaine fonction @xmath2-rgulue @xmath387 sur @xmath0 .
comme @xmath516 , on a @xmath517 et donc @xmath518 , i.e. , @xmath519 .
rciproquement , supposons que @xmath512 .
cela veut dire que @xmath89 ne sannule pas sur @xmath472 .
en particulier , la fonction @xmath520 existe sur @xmath472 et y est @xmath2-rgulue .
daprs la version @xmath2-rgulue de lingalit de lojasiewicz , il existe un entier naturel @xmath378 tel que lextension par @xmath124 de @xmath521 sur @xmath39 est @xmath2-rgulue sur @xmath0 .
notons cette extension par @xmath387 . on a donc
@xmath515 , i.e. , @xmath511 .
soient @xmath37 et @xmath2 des entiers naturels .
soient @xmath4 et @xmath89
des fonctions @xmath2-rgulues sur @xmath0 .
alors , @xmath522 si et seulement si @xmath523 . en utilisant proposition [ radprinc ] encore , on obtient la caractrisation gomtrique suivante des gnrateurs radicaux dun idal de fonctions @xmath2-rgulues : soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 , et soit @xmath55 un idal de @xmath28 .
alors , @xmath524 si et seulement si @xmath493 .
ou encore plus gnralement : soient @xmath37 et @xmath2 des entiers naturels .
soient @xmath55 et @xmath441 des idaux de @xmath28 .
alors , @xmath525 si et seulement si @xmath526 .
en particulier , deux idaux radicaux de @xmath28 sont gaux si et seulement sils ont le mme ensemble des zros .
en rsum : [ thmradprinc ] soient @xmath37 et @xmath2 des entiers naturels .
lanneau @xmath28 est radicalement principal . plus
prcisment , soit @xmath55 un idal de @xmath28 .
alors il existe une fonction @xmath2-rgulue @xmath491 telle que @xmath494 i.e. telle que @xmath527 soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
si @xmath66 est un sous - ensemble de @xmath0 , on note @xmath528 , ou maintenant simplement @xmath529 sil ny a pas de confusion possible , lensemble des fonctions @xmath2-rgulues sur @xmath0 sannulant sur @xmath66 , i.e. , @xmath530 il est clair que @xmath529 est un idal radical dans lanneau @xmath28 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath55 un idal de type fini de lanneau @xmath28 des fonctions @xmath2-rgulues sur @xmath0 .
si @xmath531 dans @xmath0 , alors @xmath532 .
supposons que @xmath531 .
soient @xmath56 des gnrateurs de @xmath55 .
en posant @xmath533 on a @xmath534 cela veut dire que la fonction @xmath2-rgulue @xmath4 na pas de zros dans @xmath0 .
elle est donc inversible dans @xmath28 .
par consquent @xmath535 i.e. @xmath532 . [ null.fini ]
soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath55 un idal de type fini de lanneau @xmath28 des fonctions @xmath2-rgulues sur @xmath0 .
alors , @xmath536 on adapte largument de rabinowitsch @xcite : il est clair que @xmath537 .
comme le premier est radical , on a bien @xmath538 .
pour montrer linclusion inverse , soit @xmath4 une fonction @xmath2-rgulue sur @xmath0 sannulant sur @xmath539 .
on peut supposer que @xmath4 nest pas identiquement nulle .
on identifie le corps des fractions @xmath16 avec son image dans @xmath540 .
le sous - anneau @xmath28 est alors bien un sous - anneau de @xmath541 .
soit @xmath542 lidal dans @xmath541 engendr par @xmath55 et @xmath543 .
il est de type fini car @xmath55 est de type fini .
observons que @xmath544 dans @xmath137 . en effet
, si @xmath545 est un zro commun de @xmath542 , alors @xmath546 est un zro de @xmath55 dans @xmath0 , et un non zro de @xmath4 , ce qui contredit lhypothse que @xmath4 sannule sur @xmath539 . on a donc bien @xmath544 .
daprs le nullstellensatz faible pour idaux de type fini , @xmath547 .
il existe donc @xmath548 dans @xmath541 et @xmath56 dans @xmath55 telles que @xmath549 dans @xmath541 .
soit @xmath550 , et soit @xmath551 lapplication @xmath2-rgulue dfinie par @xmath552 .
comme limage de @xmath387 est de codimension @xmath191 dans @xmath137 , on a un morphisme danneaux @xmath553 induit par la composition avec @xmath387 . on a alors
@xmath554 sur @xmath5 .
daprs le lemme [ lojaregulrn ] , il existe un entier naturel @xmath378 tel que lextension par @xmath124 de @xmath555 soit @xmath2-rgulue sur @xmath0 tout entier , pour tout @xmath87 .
du coup , @xmath556 et donc @xmath506 .
remarquons que le nullstellensatz [ null.fini ] nest pas valable pour les fonctions @xmath26-rgulues .
un contre - exemple est le suivant .
soit @xmath55 lidal de @xmath557 engendr par @xmath433 .
lensemble des zros @xmath539 de @xmath55 est gal lorigine de @xmath23 .
la fonction @xmath65 sannule bien sur @xmath539 , mais nappartient pas lidal radical @xmath558 de @xmath55 dans @xmath557 .
en effet , lextension @xmath23 de la fonction @xmath559 nest de classe @xmath131 pour aucun entier naturel @xmath378 .
[ nullstellensatz ] soient @xmath37 et @xmath2 des entiers naturels .
1 . soit @xmath60 un ferm @xmath2-rgulu de @xmath0 .
alors @xmath560 .
soit @xmath55 un idal ( non - ncessairement de type fini ) de @xmath28 .
alors , @xmath561 .
en particulier , on a une correspondance entre les ferms rgulus de @xmath0 et les idaux radicaux de @xmath28 .
le ( 1 ) est une consquence du ( 2 ) . en effet , en crivant @xmath562 pour un idal @xmath55 de @xmath28 , on a @xmath563 il suffit donc de montrer le ( 2 ) .
soit @xmath55 un idal quelconque de @xmath28 .
daprs la proposition [ radprinc ] , il existe @xmath564 telle que @xmath524 .
daprs le nullstellensatz pour idaux de type fini , on a @xmath565 .
comme @xmath493 et @xmath524 , on a bien @xmath561 .
on donne une preuve diffrente du ( 2 ) sans utiliser le nullstellensatz pour idaux de type fini .
linclusion @xmath566 est vidente .
pour linclusion inverse on considre @xmath567 .
daprs le thorme [ thmradprinc ] , il existe @xmath564 telle que @xmath524 et @xmath493 .
par consquent @xmath568 i.e. @xmath569 .
daprs le lemme [ lojaregulrn ] , il existe un entier naturel @xmath378 tel que lextension par @xmath124 de @xmath570 sur @xmath501 soit @xmath2-rgulue sur @xmath0 .
notons @xmath387 cette extension .
on obtient finalement @xmath571 i.e. @xmath572 . remarquons que lingalit de lojasiewicz [ lojaregulrn ] pour les fonctions @xmath2-rgulues est loutil principal pour la dmonstration du nullstellensatz [ nullstellensatz ] .
ce mme type dide est utilis dans @xcite pour montrer que lanneau des fonctions semi - algbriques continues est un anneau de gelfand .
soit @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath164 .
lensemble des zros de @xmath4 dans le spectre @xmath32 est not @xmath573 , i.e. , @xmath574 son complmentaire est not @xmath575 . plus
gnralement , si @xmath42 est un sous - ensemble de @xmath28 , lensemble des zros communs dans
@xmath32 des lments de @xmath42 est not @xmath576 , et son complmentaire est @xmath577 . plus prcisment , @xmath578 et @xmath579 il est bien connu que les sous - ensembles de la forme @xmath577 constituent une topologie sur @xmath32 ( * ? ? ?
* chapitre 1 ) .
lespace topologique @xmath32 est quasi - compact . plus
gnralement , les ouverts de la forme @xmath580 sont quasi - compacts .
soient @xmath37 un entier naturel et
@xmath2 un entier surnaturel .
soit @xmath139 . on note
encore @xmath581 lensemble des fonctions @xmath2-rgulues sur @xmath0 sannulant en @xmath65 , i.e. , @xmath582 il est clair que @xmath581 est un idal maximal de @xmath28 .
une consquence du nullstellensatz ( [ nullstellensatz ] ) est que tout idal maximal de @xmath28 est de cette forme .
[ maximal ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath583 un idal maximal de @xmath28 .
il existe un et un seul @xmath139 tel que @xmath584
. regardons dabord le cas @xmath182 .
lanneau des fonctions @xmath26-rgulues est gal lanneau des fonctions rgulires sur @xmath0 .
ce dernier est la localisation de lanneau des polynmes @xmath62 $ ] par rapport la partie multiplicative des polynmes ne sannulant pas sur @xmath0 .
@xmath583 est un idal maximal de @xmath20 , cest par consquent un ideal rel et le nullstellensatz classique fonctionnne i.e. @xmath585 ( * ? ? ?
4.1.4 ) .
soit @xmath2 un entier surnaturel .
daprs le corollaire [ nullstellensatz ] si @xmath2 est fini et ce qui prcde si @xmath2 est infini , @xmath585 .
en particulier , @xmath586 est non - vide .
soit @xmath587 . on a donc @xmath588 .
par maximalit , on obtient @xmath589 .
cela montre lexistence de @xmath65 .
lunicit est claire car
@xmath590 lorsque @xmath65 et @xmath340 sont des lments distincts de @xmath0 .
soit @xmath591 lapplication dfinie par @xmath592 , o @xmath581 est lidal maximal de @xmath28 des fonctions @xmath2-rgulues sannulant en @xmath65 .
lapplication @xmath593 est bien continue lorsquon considre @xmath0 avec la topologie @xmath2-rgulue .
daprs la version faible du nullstellensatz rgulu , @xmath593 est une bijection sur lensemble des idaux maximaux de @xmath28 .
la version forte du nullstellensatz ( corollaire [ nullstellensatz ] ) peut se reformuler ainsi : [ thbijferirrprem ] soient @xmath37 et @xmath2 des entiers naturels .
lapplication @xmath593 induit une bijection entre @xmath32 et lensemble des sous - ensembles @xmath2-rgulument ferms et irrductibles de @xmath0 .
plus prcisment , pour tout sous - ensemble ferm @xmath2-rgulument irrductible @xmath104 de @xmath0 il existe un et un seul idal premier @xmath67 de @xmath28 tel que @xmath594 .
comme prcdemment , cet nonc est faux si @xmath182 .
par exemple , lidal de @xmath153 $ ] engendr par le polynme irrductible @xmath433 est un idal premier @xmath67 . or , @xmath595 soit @xmath239 un ferm @xmath2-rgulu .
notons par @xmath596 le plus petit sous - ensemble ferm de @xmath32 tel que @xmath597 , i.e. , @xmath596 est ladhrence de @xmath598 dans @xmath32 i.e. @xmath599 . [ letf = vif ] soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
1 . soit @xmath600 , @xmath48 une collection de sous - ensembles @xmath2-rgulument ferms de @xmath0 .
alors @xmath601 2 .
soient @xmath602 des sous - ensembles @xmath2-rgulument ferms de @xmath0 .
alors @xmath603 la deuxime proprit est vidente .
la premire est valable car @xmath593 est injectif .
soient @xmath37 et @xmath2 des entiers naturels .
lapplication @xmath593 induit une bijection entre lensemble des sous - ensembles ferms de @xmath32 et lensemble des sous - ensembles @xmath2-rgulument ferms de @xmath0 . plus
prcisment , pour tout sous - ensemble @xmath2-rgulument ferm @xmath104 de @xmath0 il existe un et un seul ensemble ferm @xmath192 de @xmath32 tel que @xmath604 , savoir @xmath605 .
soit @xmath104 un ferm @xmath2-rgulu dans @xmath0 .
posons @xmath605 et montrons que @xmath606 .
comme @xmath0 est noethrien , @xmath104 possde un nombre fini de composantes irrductibles @xmath2-rgulues @xmath602 .
daprs le lemme prcdent , @xmath607 .
daprs le thorme [ thbijferirrprem ] , on a alors @xmath608 montrons lunicit de @xmath192 .
soit @xmath358 un sous - ensemble ferm de @xmath32 tel que @xmath609 .
comme @xmath192 est le plus petit ferm de @xmath32 contenant @xmath610 , on a @xmath611 .
montrons linclusion inverse .
soit @xmath612 , et soit @xmath613 .
montrons dabord que @xmath614 .
soit @xmath615 . on a @xmath616 .
comme @xmath617 est contenu dans @xmath358 , et comme @xmath609 , on a @xmath618 . on a galement
@xmath606 , donc @xmath619 .
cela montre bien que @xmath614 . or , daprs le thorme [ thbijferirrprem ] , @xmath60 est un ferm @xmath2-rgulument irrductible dans @xmath0 , et @xmath620 .
il sensuit que @xmath621 .
cela montre linclusion @xmath622 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
lespace topologique @xmath32 est noethrien .
si @xmath2 est fini , cest une consquence du thorme prcdent .
pour le cas @xmath182 , il suffit de rappeler que lanneau @xmath20 est noethrien . pour @xmath5 un ouvert rgulu de @xmath0 , soit @xmath623 le complmentaire de @xmath624 dans @xmath32 , o @xmath60 est le complmentaire de @xmath5 dans @xmath0 .
dans lanneau @xmath625 des fonctions rgulires sur @xmath23 , on peut remarquer que @xmath626 o @xmath627 et @xmath628 sont deux idaux premiers distincts de @xmath625 .
voici un corollaire qui montre bien la diffrence entre lanneau des fonctions rgulues et lanneau des fonctions rgulires sur @xmath0 : [ cobijouv ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath240 un ouvert @xmath2-rgulu .
il existe un et un seul ouvert @xmath66 de @xmath32 tel que @xmath629 , savoir @xmath630 .
l encore , cet nonc est faux lorsque @xmath182 , et ce pour presque tous les ouverts lorsque @xmath34 !
comme consquence du corollaire prcdent , on donne lnonc suivant : soient @xmath37 et @xmath2 des entiers naturels .
tout ouvert de @xmath32 est de la forme @xmath575 .
dans la suite de ce paragraphe , on identifiera @xmath0 avec son image dans @xmath32 via lapplication @xmath593 , pour simplifier .
notons @xmath631 le faisceau structural sur le schma affine @xmath32 .
il induit par restriction un faisceau en anneaux locaux sur @xmath0 que lon notera encore @xmath631 .
soit @xmath5 un ouvert @xmath2-rgulu de @xmath0 , et @xmath4 une fonction @xmath2-rgulue sur @xmath0 telle que @xmath411 .
daprs le corollaire [ cobijouv ] , on a @xmath632 daprs la proposition [ restriction ] , @xmath633 il sensuit lnonc suivant : soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
la restriction du faisceau structural sur @xmath32 @xmath0 est le faisceau @xmath399 des fonctions @xmath2-rgulues sur @xmath0 .
le cas @xmath182 vient du fait suivant .
soit @xmath5 un ouvert de zariski de @xmath0 .
lanneau des fonctions rgulires @xmath634 sur @xmath5 peut sidentifier avec la limite inductive des localisations @xmath635 , o @xmath4 parcourt lensemble des fonctions rgulires sur @xmath0 qui ne sannule pas sur @xmath5 . soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
un @xmath28-module .
le faisceau @xmath636 sur le schma @xmath32 induit encore par restriction un faisceau @xmath636 sur @xmath0 .
l aussi , pour un ouvert @xmath2-rgulu @xmath5 de @xmath0 , on a @xmath637 lorsque @xmath2 est fini , o @xmath4 est une fonction @xmath2-rgulue sur @xmath0 avec @xmath411 .
dans le cas @xmath182 , le module @xmath638 peut sidentifier avec la limite inductive des localisations @xmath639 , o @xmath4 parcourt lensemble des fonctions rgulires sur @xmath0 qui ne sannulent pas sur @xmath5 . plus
gnralement , soit @xmath66 un ouvert @xmath2-rgulu de @xmath0 , et @xmath89 une fonction @xmath2-rgulue sur @xmath0 telle que @xmath640 .
un @xmath398-module .
la restriction @xmath66 du faisceau @xmath636 sur @xmath641 est un faisceau en @xmath642-modules , encore not @xmath636 , dtermin par @xmath637 pour tout ouvert @xmath2-rgulu @xmath5 de @xmath66 , et pour toute fonction @xmath2-rgulue @xmath4 sur @xmath66 avec @xmath411 .
dans le cas @xmath182 , le module @xmath638 peut sidentifier avec la limite inductive des localisations @xmath639 , o @xmath4 parcourt lensemble des fonctions rgulires sur @xmath66 qui ne sannulent pas sur @xmath5 .
un faisceau _ @xmath2-rgulu _
sur @xmath0 est un faisceau en @xmath399-modules sur @xmath0 .
soit @xmath112 un faisceau @xmath2-rgulu sur @xmath0 .
on dira que le faisceau @xmath112 est _ quasi - cohrent _
sil existe un recouvrement de @xmath0 par des ouverts @xmath2-rgulus @xmath401 , @xmath643 , tel que , pour tout @xmath87 , il existe un @xmath644-module @xmath645 avec @xmath646 isomorphe @xmath647 .
nous obtenons alors une version @xmath2-rgulue du thorme a de cartan , cf .
@xcite et ( * ? ? ?
* chapitre 12 ) .
[ tha ] soient @xmath37 et @xmath2 des entiers naturels .
le foncteur qui associe un @xmath28-module @xmath380 le faisceau @xmath648 est une quivalence de catgories sur la catgorie des faisceaux @xmath2-rgulus quasi - cohrents sur @xmath0 .
le foncteur rciproque est le foncteur `` sections globales '' @xmath649 . en particulier ,
le faisceau @xmath636 est engendr par ses sections globales .
comme signal dans lintroduction , ce rsultat est faux pour @xmath182 . en effet ,
il existe des faisceaux rguliers quasi - cohrents sur @xmath0 qui ne sont pas engendrs par leurs sections globales : soit @xmath650 $ ] le polynme dfini par @xmath651 remarquons que @xmath67 possde exactement deux racines relles savoir
@xmath652 et @xmath653 .
pour @xmath655 . les ouverts rguliers @xmath656 et @xmath657 constituent un recouvrement de @xmath23 .
on dfinit un faisceau rgulier quasi - cohrent @xmath112 par rapport ce recouvrement . en fait , @xmath112 va tre localement
libre de rang @xmath191 .
explicitement , on construit @xmath112 partir des faisceaux @xmath658 et @xmath659 en les recollant au - dessus de @xmath660 laide de la fonction de transition @xmath661 sur @xmath660 . plus prcisment , deux sections @xmath662 et @xmath663 de @xmath658 et @xmath659 au - dessus douverts de zariski @xmath664 et @xmath665 , respectivement , se recollent si @xmath666 sur @xmath667 .
montrons que toute section rgulire globale @xmath252 de @xmath112 sannule en @xmath668 .
la restriction @xmath85 de @xmath252 @xmath401 est une fonction rgulire sur @xmath401 , pour @xmath655 .
la condition de recollement est @xmath666 sur @xmath660 .
ecrivons @xmath669 , o @xmath670 $ ] , avec @xmath671 sur @xmath401 et @xmath672 premiers entre eux , pour @xmath655 .
la condition de recollement implique que @xmath673 sur @xmath23 .
comme @xmath67 est irrductible , et comme @xmath674 , le polynme @xmath67 divise @xmath675 .
en particulier , @xmath676 et donc aussi @xmath677 .
il sensuit que le faisceau rgulier quasi - cohrent @xmath112 sur @xmath23 nest pas engendr par ses sections globales .
pour complter , montrons que le faisceau @xmath2-rgulu quasi - cohrent @xmath678 induit est bien engendr par ses sections globales , pour @xmath2 un entier naturel . on montre mme plus : le faisceau @xmath679 est isomorphe @xmath399 en exhibant une section globale de @xmath679 qui ne sannule jamais .
soit @xmath252 la section rgulue globale de @xmath679 dfinie par @xmath680 o @xmath336 est le plus petit entier @xmath681 .
la fonction @xmath85 est bien @xmath2-rgulue sur @xmath401 et non nulle , pour @xmath655 .
le faisceau @xmath2-rgulu @xmath679 est donc isomorphe @xmath399 .
en particulier , il est engendr par ses sections globales .
il suit du thorme a de cartan version @xmath2-rgulue , que le foncteur `` sections globales '' est exact sur les faisceaux @xmath2-rgulus quasi - cohrents sur @xmath0 .
do le la version @xmath2-rgulue du thorme b de cartan : [ thb ] soient @xmath37 et @xmath2 des entiers naturels .
soit @xmath112 un faisceau @xmath2-rgulu quasi - cohrent sur @xmath0 .
alors , @xmath682 pour tout @xmath683 .
soit @xmath60 un sous - ensemble @xmath2-rgulument ferm de @xmath0 .
la _ topologie @xmath2-rgulue
_ sur @xmath60 est la topologie induite par la topologie @xmath2-rgulue de @xmath0 .
soit @xmath77 le faisceau didaux de @xmath399 des fonctions sannulant sur @xmath60 .
le faisceau quotient @xmath684 possde un support gal @xmath60 , et est donc un faisceau sur @xmath60 , le _ faisceau des fonctions @xmath2-rgulues _
sur @xmath60 , not @xmath685 .
soit @xmath5 un ouvert @xmath2-rgulu de @xmath60 .
une _ fonction @xmath2-rgulue _
sur @xmath60 est une section du faisceau @xmath685 .
en particulier , une fonction @xmath2-rgulue sur @xmath60 est une section globale du faisceau @xmath685 sur @xmath60 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath60 un ferm @xmath2-rgulu de @xmath0 , et @xmath5 un ouvert @xmath2-rgulu de @xmath60 .
soit @xmath66 un ouvert @xmath2-rgulu de @xmath0 tel que @xmath686 .
si @xmath4 est une fonction @xmath2-rgulue sur @xmath5 , alors il existe une fonction @xmath2-rgulue @xmath89 sur @xmath66
do nt la restriction @xmath5 est gal @xmath4 . montrons lnonc dabord dans le cas o @xmath2 est fini . on a une suite exacte
de faisceaux quasi - cohrents @xmath687 sur @xmath0 .
daprs la version @xmath2-rgulue du thorme b de cartan , on en dduit une suite exacte @xmath688 comme @xmath689 , il existe bien une fonction @xmath2-rgulue @xmath89 sur @xmath66 do nt la restriction @xmath5 est gale @xmath4 , pour toute fonction @xmath4 @xmath2-rgulue sur @xmath5 , lorsque @xmath2 est fini .
il nous reste montrer lnonc lorsque @xmath182 .
soit @xmath4 une fonction rgulire sur @xmath5 .
il existe des fonctions polynomiales @xmath67 et @xmath9 sur @xmath0 telles que @xmath75 pour tout @xmath135 .
soit @xmath252 une fonction polynomiale sur @xmath0 do nt lensemble des zros est gal @xmath60 .
lensemble des zros de la fonction polynomiale @xmath690 est contenu dans le ferm de zariski @xmath691 .
la fonction dfinie sur @xmath692 par @xmath693 est bien rgulire , et sa restriction @xmath5 est gale @xmath4 .
soient @xmath37 un entier naturel et @xmath2 un entier surnaturel .
soit @xmath60 un ferm @xmath2-rgulu de @xmath0 .
si @xmath4 est une fonction @xmath2-rgulue sur @xmath60 , alors il existe une fonction @xmath2-rgulue @xmath89 sur @xmath0 do nt la restriction @xmath60 est gale @xmath4 .
plus prcisment , lapplication de restriction de @xmath28 dans @xmath694 induit un isomorphisme @xmath695 [ rem.here ] daprs le corollaire prcdent , les fonctions @xmath2-rgulues telles que nous venons de les dfinir sur des sous - ensembles ferms @xmath2-rgulus concident avec les fonctions hereditarily rational de kollr @xcite , lorsque @xmath126 .
lexemple ( * ? ? ?
2 ) dune fonction rationnelle continue sur une surface singulire do nt la restriction laxe des @xmath696 nest pas rationnelle nest pas une fonction rgulue .
soit @xmath60 un sous - ensemble @xmath2-rgulument ferm de @xmath0 .
la paire @xmath697 est un espace localement annel en @xmath12-algbres .
varit @xmath2-rgulue affine _ est un espace localement annel en @xmath12-algbres isomorphe @xmath697 pour un certain ferm @xmath2-rgulu @xmath60 de @xmath0 , o @xmath37 est un entier naturel .
un _ morphisme _
entre varits @xmath2-rgulues affines est un morphisme despaces localement annels en @xmath12-algbres .
soient @xmath51 et @xmath37 des entiers naturels et @xmath2 un entier surnaturel .
soient @xmath698 et @xmath699 des ferms @xmath2-rgulus .
une application @xmath700 est un morphisme de varits affines @xmath2-rgulues si et seulement sil existe des fonctions @xmath2-rgulues @xmath701 sur @xmath186 telles que @xmath702 pour tout @xmath703 .
soit @xmath2 un entier surnaturel .
soit @xmath101 une varit @xmath2-rgulue affine .
soit @xmath4 une fonction @xmath2-rgulue sur @xmath104 , i.e. , @xmath4 est une section globale de @xmath102 .
lensemble des zros de @xmath4 dans le spectre @xmath704 est not @xmath573 , i.e. , @xmath705 son complmentaire est not @xmath575 . plus
gnralement , si @xmath42 est un sous - ensemble de @xmath706 , lensemble des zros communs dans
@xmath704 des lments de @xmath42 est not @xmath576 , et son complmentaire est @xmath577 . plus prcisment , @xmath707 et @xmath708 comme prcdemment , les sous - ensembles de la forme @xmath577 constituent une topologie sur @xmath704 ( * ? ? ?
* chapitre 1 ) .
lespace topologique @xmath704 est quasi - compact . plus
gnralement , les ouverts de la forme @xmath580 sont quasi - compacts .
soit @xmath618 . on note
@xmath581 lidal maximal de @xmath706 des fonctions @xmath2-rgulues sur @xmath104 sannulant en @xmath65 .
la proposition [ maximal ] implique alors lnonc suivant : soit @xmath2 un entier surnaturel .
soit @xmath104 une varit @xmath2-rgulue affine .
soit @xmath583 un idal maximal de @xmath706 .
il existe un et un seul @xmath618 tel que @xmath584 .
comme dans le cas @xmath0 , on a une injection @xmath709 le thorme [ thbijferirrprem ] se gnralise alors ainsi : soit @xmath2 un entier naturel .
soit @xmath104 une varit @xmath2-rgulue affine .
lapplication @xmath593 ci - dessus induit une bijection entre lensemble des sous - ensembles ferms de @xmath704 et lensemble des sous - ensembles ferms de @xmath104 . plus prcisment , pour tout sous - ensemble ferm @xmath60 de @xmath104 il existe un et un seul ensemble ferm @xmath230 de @xmath704 tel que @xmath710 .
pour les ouverts cela donne lnonc suivant : soit @xmath2 un entier naturel .
soit @xmath104 une varit @xmath2-rgulue affine .
soit @xmath711 un ouvert .
il existe un et un seul ouvert @xmath66 de @xmath704 tel que @xmath629 . on a encore : soit @xmath2 un entier naturel .
soit @xmath104 une varit @xmath2-rgulue affine .
tout ouvert de @xmath704 est de la forme @xmath575 , pour une certaine fonction @xmath2-rgulue sur @xmath104 .
comme auparavant , on identifiera @xmath104 avec son image dans @xmath704 via lapplication @xmath593 .
soit @xmath2 un entier surnaturel .
soit @xmath104 une varit @xmath2-rgulue affine .
la restriction du faisceau structural sur @xmath704 @xmath104 sidentifie avec le faisceau @xmath102 .
soit @xmath380 un @xmath706-module .
le faisceau @xmath636 sur le schma @xmath704 induit encore par restriction un faisceau @xmath636 sur @xmath104 .
pour un ouvert @xmath5 de @xmath104 , on a @xmath637 lorsque @xmath2 est fini , o @xmath4 est une fonction @xmath2-rgulue sur @xmath104 avec @xmath411 .
un faisceau _ @xmath2-rgulu _
sur @xmath104 est un faisceau en @xmath102-modules sur @xmath104 .
soit @xmath112 un faisceau @xmath2-rgulu sur @xmath104 .
on dira que le faisceau @xmath112 est _ quasi - cohrent _
sil existe un recouvrement de @xmath104 par des ouverts @xmath2-rgulus @xmath401 , @xmath643 , tels que , pour tout @xmath87 , il existe un @xmath712-module @xmath645 avec @xmath713 isomorphe @xmath647 .
la version @xmath2-rgulue du thorme a de cartan implique alors lnonc suivant : [ thax ] soit @xmath2 un entier naturel .
soit @xmath104 une varit @xmath2-rgulue affine .
le foncteur qui associe un @xmath706-module @xmath380 le faisceau @xmath714 est une quivalence de catgories sur la catgorie des faisceaux @xmath2-rgulus quasi - cohrents sur @xmath104 .
le foncteur rciproque est le foncteur `` sections globales '' @xmath649 . en particulier ,
le faisceau @xmath636 est engendr par ses sections globales .
de mme , on a une version @xmath2-rgulue du thorme b de cartan pour les varits @xmath2-rgulues affines : [ thbx ] soit @xmath2 un entier naturel .
soit @xmath104 une varit @xmath2-rgulue affine .
soit @xmath112 un faisceau @xmath2-rgulu quasi - cohrent sur @xmath104 .
alors , @xmath715 pour tout @xmath683 . en particulier les thormes [ thax ] et [ thbx ] sappliquent pour tout @xmath716 lorsque @xmath104 est une varit relle algbrique affine singulire .
lnonc est alors en tout point identique au cas algbriquement clos !
* chapitre 12 ) .
le but de cette section est de dterminer les sous - ensembles @xmath2-rgulument ferms de @xmath0 pour @xmath2 et @xmath37 des entiers naturels .
ceci fait , si @xmath60 est un sous - ensemble @xmath2-rgulument ferm de @xmath0 ( par exemple si @xmath60 est une sous - varit relle algbrique de @xmath0 ) , on saura caractriser les sous - ensembles @xmath2-rgulument ferms de @xmath60 via la topologie induite . on rappelle
quun sous - ensemble semi - algbrique de @xmath0 est algbriquement constructible sil appartient lalgbre boolenne engendr par les sous - ensembles zariski ferms de @xmath0 .
ces ensembles forment une catgorie constructible , de la mme manire que les ensembles symtriques par arcs . en particulier la topologie pour laquelle les ferms sont les ensembles algbriquement constructibles ferms ( au sens de la topologie euclidienne ) est noethrienne . on renvoie
@xcite pour une introduction aux catgories constructibles .
le but de cette section est de montrer que les sous - ensembles @xmath2-rgulument ferms de @xmath0 coincident avec les sous - ensembles algbriquement constructibles ferms de @xmath0 . on
commence par deux rsultats prcisant la place des points rguliers ainsi que la dimension dun ensemble @xmath2-rgulument ferm vis - - vis de son adhrence de zariski .
[ hypersurface ] soit @xmath55 un idal premier de @xmath29 . on note
@xmath718 $ ] et on suppose que @xmath441 est principal , i.e. @xmath719 avec @xmath720 $ ] irrductible . on note
@xmath66 lensemble @xmath463 et @xmath721 lensemble des points lisses de @xmath66 .
alors latexmath:[\[\overline{v_{\mathrm{reg}}}^{\mathrm{eucl}}\subseteq { \mathcal{z}}(i ) = \ { x\in{\mathbf{r}}^n|\,\ , i\subseteq { \mathfrak{m}}_x\}=\{x\in{\mathbf{r}}^n les fonctions de @xmath29 tant continues , il est suffisant de montrer que @xmath723 pour @xmath724 , on doit montrer que @xmath725 .
supposons lexistence de @xmath491 telle que @xmath726 , cest - - dire @xmath727 . on crit
@xmath728 avec @xmath729 $ ] premiers entre eux .
on obtient alors @xmath730=j=(s)\ ] ] et donc @xmath731 avec @xmath732 $ ] .
ceci implique que @xmath67 sannule identiquement sur @xmath66 , i.e. @xmath733 .
le lemme [ reelintersection ] nous montre
que @xmath719 est un idal rel et par consquent @xmath734 ( * ? ? ?
4.5.1 ) . on en dduit que @xmath735 .
comme @xmath167 puisque @xmath736 , on a aussi @xmath737 car @xmath738 .
on sait que @xmath739 par la proposition [ codim ] .
comme @xmath724 , il existe un voisinage ouvert semi - algbrique de @xmath65 dans @xmath740 de dimension @xmath741 ( * ? ? ?
3.3.10 ) . on note @xmath53 le semi - algbrique @xmath742 .
daprs ce quon a dit prcdemment , le point @xmath65 est un point adhrent @xmath53 .
le lemme de slection des courbes ( * ? ? ?
2.5.5 ) pour le semi - algbrique @xmath53 nous fournit une fonction semi - algbrique continue @xmath743\rightarrow { \mathbf{r}}^n$ ] telle que @xmath744 et @xmath7450,1])\subseteq a$ ] .
par consquent @xmath746\rightarrow { \mathbf{r}}$ ] est une fonction semi - algbrique continue telle que @xmath7470,1])=0 $ ] et @xmath748 , ce qui contredit la continuit de @xmath4
. [ dimension ] soit @xmath55 un idal radical de @xmath29 et @xmath718 $ ] sa trace sur les polynmes . on note
@xmath66 lensemble @xmath463 .
alors @xmath749 o @xmath750 est ladhrence pour la topologie de zariski de @xmath539 . en particulier @xmath751 .
supposons maintenant que @xmath753 soit strictement inclus dans @xmath66 .
alors @xmath754 contient strictement @xmath529 , sinon @xmath755 .
par consquent il existe @xmath756 tel que @xmath757 .
mais @xmath67 est une fonction rgulue qui sannule identiquement sur @xmath539 et donc , par le nullstellensatz rgulu , @xmath758 = i\cap { \mathbf{r}}[x_1,\ldots , x_n]=j.\ ] ] par consquent @xmath759 , et donc @xmath67 appartient @xmath529 , en contradiction avec le choix de @xmath67 .
dans la cas du plan , on peut construire _
la main _ une fonction rgulue qui spare les points isols de ladhrence euclidienne des points rguliers dune courbe , dmontrant au passage que ladhrence au sens de la topologie euclidienne des points rguliers de la courbe est un ensemble rgulument ferm .
[ revhypersurface ] soit @xmath55 un idal premier de @xmath760 .
soit @xmath761 $ ] et notons
@xmath66 lensemble @xmath463 . on note
aussi @xmath721 lensemble des points lisses de @xmath66 et @xmath762 .
alors @xmath763 alors @xmath764 et le rsultat est vident .
si @xmath441 est maximal , le corollaire [ maximal ] donne le rsultat voulu . on suppose donc que @xmath441 est de hauteur @xmath191 , cest - - dire @xmath719 avec @xmath765 $ ] irrductible .
ainsi @xmath766 est une courbe algbrique affine irrductible .
lidal @xmath441 tant rel , on a @xmath440 .
linclusion @xmath767 provient du thorme [ hypersurface ] .
comme @xmath768 , on obtient par ailleurs @xmath769 .
en rsum , on a les inclusions suivantes @xmath770 pour terminer la preuve , pour tout @xmath771 on doit trouver @xmath564 telle que @xmath772 et vrifiant @xmath738
. soit @xmath771 .
supposons @xmath65 gal lorigine @xmath158 pour simplifier les notations
. alors @xmath158 est un zro isol de @xmath252 . il existe donc un disque ferm @xmath773 centr en @xmath158 tel que @xmath158 soit lunique zro de @xmath252 dans @xmath773 .
en utilisant lingalit de lojasiewicz pour @xmath774 et @xmath775 dans @xmath773 , il existe un entier strictement positif @xmath378 tel que @xmath776 stende de faon continue par @xmath124 en @xmath158 dans @xmath773 . on pose @xmath777 .
la fonction @xmath4 est clairement continue en dehors de @xmath158 et peut tre tendue de manire continue par @xmath191 en @xmath158 car @xmath778 par consquent @xmath779 et la restriction de @xmath4 @xmath721 est identiquement nulle puisque la restriction de @xmath4 @xmath780 est identiquement nulle .
comme @xmath4 est continue on a aussi @xmath772 .
il reste donc montrer que @xmath4 appartient @xmath55 .
dj @xmath781 et puisque @xmath55 est un idal premier , au moins un des deux termes du produit appartient @xmath55 .
supposons quil sagisse du premier , cest - - dire @xmath782=j.\ ] ] alors @xmath783 appartient @xmath441 car @xmath441 est un idal rel .
lidal @xmath441 tant de plus premier , on en dduit que @xmath784 .
mais alors @xmath785 et @xmath786 en invoquant une nouvelle fois la ralit de @xmath441 . on en dduit que @xmath441 est
maximal , en contradiction avec notre hypothse .
par consquent @xmath4 appartient @xmath55 , ce qui termine la preuve de la proposition .
en particulier , on constate que les courbes @xmath124-rgulument fermes du plan sont des ensembles zariski constructibles ferms . on montre
ci - dessous que
cest en fait le cas pour tout ensemble @xmath2-rgulu . on sait
dj que tout ensemble @xmath2-rgulument ferm de @xmath0 est algbriquement constructible daprs le corollaire [ coregfermeestconstructible ] . on montre la rciproque par rcurrence sur la dimension .
en dimension nulle , le rsultat est clair .
soit @xmath151 un sous - ensemble algbriquement constructible ferm de @xmath0 .
on peut supposer que @xmath151 est irrductible , autrement dit que son adhrence de zariski @xmath66 est une sous - varit relle algbrique irrductible de @xmath0 , quitte raisonner composante par composante . on note
@xmath358 la runion des composantes algbriquement constructibles de @xmath66 de dimension strictement plus petite que @xmath66 .
ainsi @xmath787 .
il existe une fonction @xmath2-rgulue @xmath89 dfinie sur @xmath0 telle que @xmath788 par hypothse de rcurrence .
par ailleurs , soit @xmath4 une fonction rgulire sur @xmath0 telle que @xmath789 .
daprs le lemme [ lojaregulu ] , il existe un entier naturel @xmath378 tel que la fonction @xmath790 , @xmath2-rgulue sur @xmath472 , se prolonge par @xmath124 sur @xmath791 en une fonction @xmath2-rgulue .
posons @xmath792 alors la fonction @xmath387 est @xmath2-rgulue sur @xmath499 , et sannule sur @xmath793 . on la prolonge par @xmath191 sur @xmath358 .
elle est alors @xmath2-rgulue en dehors de @xmath794 par choix de @xmath378 .
lensemble @xmath795 est algbriquement constructible de dimension strictement plus petite que @xmath151 , il existe donc une fonction @xmath2-rgulue @xmath796 telle que @xmath797 par hypothse de rcurrence .
daprs le lemme [ lojaregulrn ] , il existe alors un entier @xmath798 tel que la fonction @xmath799 soit @xmath2-rgulue sur @xmath0 .
elle satisfait @xmath800 , ce qui fait de @xmath151 un ensemble @xmath2-rgulument ferm .
[ recip - eclt ] soit @xmath60 un sous - ensemble semi - algbrique de @xmath0 .
alors @xmath60 est un sous - ensemble algbriquement constructible de @xmath0 si et seulement sil existe une composition dclatements centres lisses @xmath224 telle que limage rciproque @xmath225 est zariski ferme dans @xmath226 . on sait
quun sous - ensemble constructible de @xmath0 peut devenir zariski ferm aprs clatements en combinant le thorme [ const ] et le corollaire [ cor - eclt ] .
rciproquement , remarquons quil suffit de traiter le cas dun seul clatement , et raisonnons par rcurrence sur la dimension de @xmath60 .
soit @xmath60 un sous - ensemble semi - algbrique de @xmath0 tel que @xmath801 soit un sous - ensemble zariski ferm de @xmath226 , avec @xmath802 lclatement de @xmath0 le long du centre @xmath151 , avec diviseur exceptionnel @xmath42 .
on peut supposer que @xmath803 est irrductible , quitte raisonner composante par composante .
dj que @xmath60 est ferm pour la topologie euclidienne de @xmath0 comme image dun sous - ensemble ( zariski-)ferm par une application ferme . de plus @xmath207 est un isomorphisme entre @xmath804 et @xmath805 , donc @xmath806 est algbriquement constructible .
par ailleurs @xmath807 est un ensemble semi - algbrique ferm qui a pour image rciproque par @xmath207 lensemble zariski ferm @xmath808 , donc @xmath807 est constructible par hypothse de rcurrence , condition que @xmath807 soit de dimension strictement plus petite que @xmath60 .
si ce nest pas le cas , alors @xmath66 est inclus dans
@xmath42 par irrductibilit de @xmath66 et donc @xmath60 est inclus dans le centre @xmath151 .
en particulier , @xmath60 est un sous - ensemble zariski ferm de @xmath151 . on sait
quun sous - ensemble @xmath2-rgulument ferm de @xmath0 est un sous - ensemble algbriquement constructible ferm daprs le thorme [ const ] .
en particulier , un ensemble @xmath2-rgulument ferm irrductible est gal la composante algbriquement constructible de dimension maximale de son adhrence de zariski .
le but de cette section est de dcrire cette composante algbriquement constructible de dimension maximale en terme densembles symtriques par arcs .
[ annulregulue ] soit @xmath66 une sous - varit algbrique irrductible de @xmath0 de dimension @xmath809 .
soit @xmath810 un sous - ensemble semi - algbrique de @xmath66 de dimension @xmath809 et soit @xmath811 sannulant identiquement sur @xmath810 .
alors @xmath4 sannule identiquement sur la runion @xmath108 des composantes @xmath220-irrductibles de dimension maximale de @xmath66 .
la fonction @xmath4 reste rgulue en restriction @xmath66 ( cf .
remarque [ rem.here ] ) .
il existe donc une composition dclatements centres lisses @xmath812 , avec @xmath813 une varit relle algbrique irrductible non singulire , telle que @xmath204 soit rgulire sur @xmath813 daprs le thorme [ threguliereaeclt ] .
comme @xmath810 est de dimension @xmath809 , la fonction rgulire @xmath204 sannule sur un sous - ensemble zariski dense de @xmath813 , donc sur @xmath813 qui est irrductible .
alors @xmath4 sannule sur les points rguliers de @xmath66 , donc aussi sur @xmath108 car @xmath814 est un ensemble symtrique par arcs .
soit @xmath42 un ensemble semi - algbrique de @xmath0 .
daprs ( * ? ? ?
* cor . 2.15 ) , lintersection des ensembles symtriques par arcs contenant @xmath42 est un ensemble symtrique par arcs de mme dimension que @xmath42 , et not @xmath815
. de mme , on note @xmath816 ladhrence de @xmath42 pour la topologie algbriquement constructible ( ou la topologie rgulue ) .
[ hypersurfacegen ] soit @xmath55 un idal premier de @xmath29 . on note
@xmath817 $ ] et @xmath66 lensemble @xmath463 . on note
aussi @xmath721 lensemble des points lisses de @xmath66 et @xmath108 la runion des composantes @xmath220-irrductibles de @xmath66 de dimension @xmath809 , o @xmath818 .
alors @xmath819 notons dj que ladhrence symtrique par arcs @xmath820 est gal @xmath108 daprs ( * ? ? ?
* thm . 2.21 , lem . 2.22 ) .
on sait que @xmath539 est un sous - ensemble algbriquement constructible ferm de @xmath66 daprs le thorme [ const ] , de dimension maximale daprs la proposition [ dimension ] .
comme @xmath66 est irrductible , @xmath539 est donc gal lunique composante algbriquement constructible irrductible de dimension maximale de @xmath66 , et en particulier @xmath539 contient @xmath108 .
[ hypersurface2 ] soit @xmath66 une sous - varit algbrique irrductible de @xmath0 . on note
@xmath822 la runion des composantes @xmath220-irrductibles de dimension maximale de @xmath66 .
alors @xmath823 est un idal premier de @xmath29 .
soient @xmath4 et @xmath89 deux fonctions rgulues telles que le produit @xmath827 sannule identiquement sur @xmath721 . les fonctions @xmath4 et @xmath89 restent rgulues en restriction
@xmath66 daprs la remarque [ rem.here ] , il existe donc une composition dclatements centres lisses @xmath812 telle que @xmath204 et @xmath828 soient rgulires sur @xmath813 , avec @xmath813 une varit relle algbrique irrductible non singulire , daprs le thorme [ threguliereaeclt ] .
le produit de fonctions rgulires @xmath829 sannule identiquement sur la varit irrductible @xmath813 . on peut
donc conclure que , par exemple , @xmath830 sannule identiquement sur @xmath813 et donc @xmath4 sannule identiquement sur
@xmath721 i.e. @xmath824 , ce qui termine la preuve .
[ irred ] soit @xmath66 une sous - varit algbrique irrductible de @xmath0 . on note
@xmath822 la runion des composantes @xmath220-irrductibles de dimension maximale de @xmath66 .
si @xmath831 ( en particulier si @xmath66 est lisse ) , alors @xmath66 est un ensemble rgulument ferm et rgulument irrductible .
supposons que @xmath66 soit la runion @xmath833 de deux ferms rgulus non - vides . daprs le thorme [ thmradprinc ] , il existe des fonctions rgulues @xmath834 telles que @xmath835 et @xmath836 .
alors le produit @xmath837 appartient lidal premier @xmath838 et on en dduit que @xmath839 ou @xmath840 .
soit @xmath841 une sous - varit algbrique irrductible . on note @xmath822 la
runion des composantes @xmath220-irrductibles de dimension maximale de @xmath66 .
la mme preuve que celle de la proposition prcdente montre que @xmath108 est un ensemble rgulument irrductible .
mais ce nest pas en gnral un ensemble rgulument ferm .
sur la figure , on a reprsent la fibre de @xmath845 au dessus de chaque point des deux composantes connexes de la courbe hyperelliptique dquation @xmath846 .
au dessus de lun des ovales la fibre est un point , alors quau dessus de lautre ovale , la fibre est forme de deux droites scantes .
la varit @xmath66 possde deux composantes connexes @xmath108 et @xmath358 , avec @xmath108 lunique composante symtrique par arcs irrductible de dimension @xmath843 de @xmath66 .
le lieu singulier de @xmath66 est la courbe hyperelliptique @xmath151 dquations @xmath846 et @xmath847 , possdant deux composantes connexes @xmath358 et @xmath192 avec @xmath848 . on suppose que @xmath108 est un ferm rgulu .
il existe donc @xmath849 telle que @xmath850 et par consquent @xmath4 sannule identiquement sur @xmath192 mais pas sur @xmath358 .
une telle fonction ne peut exister par la proposition [ annulregulue ] en considrant sa restriction @xmath851 .
soit @xmath853 une suite dinclusions de ferms rgulus irrductibles avec @xmath854 .
pour @xmath855 , on note @xmath856 , @xmath857 $ ] et @xmath858 . on a @xmath859 et de plus @xmath860 est irrductible pour la
topologie de zariski car @xmath861 daprs la proposition [ dimension ] ( lidal @xmath862 est bien un idal premier ) .
on obtient donc une suite dinclusions de ferms de zariski irrductibles @xmath863 puisque @xmath0 est de dimension @xmath37 pour la topologie de zariski , il existe @xmath864 tel que @xmath865 .
lensemble @xmath866 est un ouvert rgulu de @xmath867 , il est donc dense dans @xmath867 pour la topologie rgulue . on en dduit que @xmath867 est le plus petit ferm rgulu contenant @xmath810 . un ferm de zariski tant un ferm rgulu , un ferm de zariski contenant @xmath810 contient aussi @xmath867 donc @xmath868 .
par consquent @xmath869 et @xmath870 .
daprs le thorme [ hypersurfacegen ] , @xmath871 et par consquent @xmath810 est contenu dans le lieu singulier de @xmath868 qui est un ferm de zariski de dimension strictement infrieure celle de @xmath868 , ce qui fournit une contradiction .
la dimension rgulue de @xmath0 est donc infrieure @xmath37 .
il est clairement possible de construire une suite dinclusions @xmath872 de ferms de zariski irrductibles lisses .
daprs la proposition [ irred ] , les @xmath860 sont des ferms rgulus irrductibles ce qui prouve que la dimension rgulue de @xmath0 est bien @xmath37 .
soit @xmath852 .
on rappelle que les topologies @xmath2-rgulue et @xmath124-rgulues coincident . en utilisant le nullstellensatz
@xmath2-rgulu , on en dduit que la dimension de krull de @xmath873 est @xmath37 . * le parapluie de whitney : + considrons la sous - varit @xmath66 dquation @xmath874 dans @xmath303 .
le parapluie @xmath66 est un ensemble rgulument ferm et vrifie la relation @xmath875 .
par la proposition [ irred ] , cest un ensemble rgulument ferm et irrductible .
* le parapluie de cartan : + considrons la sous - varit @xmath66 dquation @xmath876 de @xmath303 .
la dcomposition de @xmath66 en ensembles @xmath220-irrductibles est @xmath877 o @xmath822 est la toile du parapluie et @xmath358 est le manche .
lensemble @xmath108 est rgulument ferm car @xmath878 avec @xmath879 .
par la proposition [ hypersurface2 ] , @xmath108 est rgulument irrductible .
lensemble @xmath358 est un ensemble rgulument ferm car @xmath880 .
lensemble @xmath66 est donc un ensemble rgulument ferm et non - rgulument irrductible . *
un parapluie cornu : + considrons la sous - varit @xmath66 dquation @xmath881 de @xmath303 .
la dcomposition de @xmath66 en ensembles @xmath220-irrductibles est @xmath877 o @xmath822 est la corne du parapluie et @xmath358 est le manche .
lensemble @xmath108 est rgulument ferm car @xmath878 avec @xmath882 .
par la proposition [ hypersurface2 ] , @xmath108 est rgulument irrductible .
le manche @xmath358 est un ensemble rgulument ferm car @xmath880 .
comme pour le parapluie de cartan , lensemble @xmath66 est donc un ensemble rgulument ferm et non - rgulument irrductible . [ thmcaracterisation ] les ferms irrductibles rgulus de @xmath0 sont les ensembles du type @xmath883 o @xmath721 dsigne lensemble des points lisses dune varit relle algbrique irrductible @xmath841 .
le rsultat suivant permet de construire de faon algorithmique le plus petit ferm algbriquement constructible contenant les points rguliers dune varit relle algbrique affine et donc de caractriser les ferms irrductibles rgulus de @xmath0 .
[ sym - necessaire ] soit @xmath841 une sous - varit relle algbrique irrductible .
soit @xmath108 la runion des composantes symtriques par arcs de dimension maximale de @xmath66 .
soit @xmath358 une composante symtrique par arc irrductible de dimension strictement plus petite que @xmath66 .
si @xmath884 alors @xmath358 est dans @xmath885 . notons que ladhrence de zariski de @xmath358 forme une varit relle algbrique irrductible de mme dimension que @xmath358 .
soit @xmath4 une fonction rgulue dfinie sur @xmath0 et sannulant sur @xmath108 .
par hypothse , la fonction @xmath4 sannule sur un ensemble semi - algbrique @xmath886 de mme dimension que @xmath887 , donc sur @xmath358 aussi daprs la proposition [ annulregulue ] .
par consquent @xmath358 appartient ladhrence pour la topologie algbriquement constructible de @xmath108 . on dcrit
maintenant de manire algorithmique lunique composante algbriquement constructible irrductible de dimension maximale dune varit relle algbrique affine en termes densembles symtriques par arcs irrductibles .
soit @xmath841 une sous - varit relle algbrique affine irrductible de dimension @xmath809 . on note
@xmath888 la dcomposition de @xmath66 en sous - ensembles symtriques par arcs irrductibles , avec @xmath889 pour @xmath890 . on note
@xmath108 la runion des composantes symtriques par arcs de dimension maximale de @xmath66 . 1 .
si @xmath892 satisfait la condition de la proposition [ sym - necessaire ] , alors @xmath893 .
2 . sinon , @xmath892 nest pas tout entier contenu dans @xmath894 .
nanmoins , un sous - ensemble symtrique par arcs de @xmath892 peut tre contenu dans @xmath894 .
lintersection @xmath895 est un ensemble symtrique par arcs de dimension au plus @xmath896 , et on rajoute les composantes symtriques par arcs irrductibles de son adhrence de zariski @xmath897 celles de dimension au plus @xmath896 de @xmath66 . procdant de mme pour tout @xmath891 , on obtient ainsi une runion densembles symtriques par arcs irrductibles de dimension au plus @xmath896 @xmath898 o @xmath899 est un ensemble dindices fini contenant @xmath900 .
notons @xmath901 la runion de @xmath108 et des composantes @xmath892 , pour @xmath902 , satisfaisant la condition de la proposition [ sym - necessaire ] . on renouvelle
maintenant les oprations @xmath903 et @xmath904 avec @xmath901 la place de @xmath108 et en considrant les indices @xmath905 appartenant @xmath906 .
ainsi , on construit pas pas , en au plus @xmath907 tapes , ladhrence pour la topologie algbriquement constructible de @xmath108 .
[ ex - algo ] on modifie lgrement lexemple [ c - ex ] de la faon suivante .
on considre la sous - varit algbrique irrductible @xmath753 de @xmath908 de dimension @xmath909 donne par les quations @xmath910 notons @xmath151 la courbe donne par les quations de @xmath753 en faisant @xmath911 .
la dcomposition en sous - ensembles symtriques par arcs irrductibles de @xmath753 est de la forme @xmath912 avec @xmath913 lovale de @xmath151 contenant le point de coordonnes @xmath914 , @xmath915 la surface contenant lovale de @xmath151 contenant le point de coordonnes @xmath916 , et @xmath917 la partie de dimension trois contenant le dernier ovale de @xmath151 .
soit @xmath918 $ ] lquation dun tore contenant en son intrieur lovale @xmath919 ( cest - - dire lovale de droite sur le dessin ) . on
suppose de plus que @xmath920 est ngatif lintrieur du tore et positif lextrieur ( cf .
example 3.2.8 dans @xcite ) .
on dfinit alors @xmath66 dans @xmath921 , en ajoutant une variable @xmath252 , par les quations de @xmath753 auxquelles on rajoute lquation @xmath922 .
la varit @xmath66 ainsi obtenue forme un revtement double de @xmath753 lextrieur du tore , et efface la partie lintrieur du tore .
notons @xmath108 la partie symtrique par arcs de dimension maximale de @xmath66 .
soit @xmath923 une composante symtrique par arcs de dimension un de @xmath66 senvoyant sur lovale @xmath913 ( de gauche sur le dessin ) de la courbe @xmath151 par le revtement .
ladhrence de zariski de @xmath923 , forme de quatre ovales vivant au dessus des ovales @xmath913 et @xmath924 de la courbe @xmath151 , ne rencontre pas @xmath108 par construction de @xmath66 .
ainsi @xmath923 ne satisfait pas lhypothse de la proposition [ sym - necessaire ] .
pour autant , @xmath923 est bien dans ladhrence pour la topologie algbriquement constructible de @xmath108 .
en effet , les deux composantes symtriques par arcs de dimension deux de @xmath66 satisfont les hypothses de la proposition [ sym - necessaire ] puisque @xmath925 , et donc font partie de @xmath885 .
notons @xmath926 leur runion .
alors ladhrence de zariski de @xmath923 rencontre @xmath926 le long dune courbe , donc daprs la proposition [ sym - necessaire ] , @xmath923 est dans ladhrence pour la topologie algbriquement constructible de @xmath926 , donc dans celle de @xmath108 .
k. kurdyka , a. parusiski , _ arc - symmetric sets and arc - analytic mappings _ , arc spaces and additive invariants in real algebraic and analytic geometry , 3367 , panor .
synthses * 24 * , soc . math .
france , paris , 2007 . | nous tudions lanneau des fonctions rationnelles qui se prolongent par continuit sur @xmath0 .
nous tablissons plusieurs proprits algbriques de cet anneau do nt un nullstellensatz fort .
nous tudions les proprits schmatiques associes et montrons une version rgulue des thormes a et b de cartan .
nous caractrisons gomtriquement les idaux premiers de cet anneau travers leurs lieux dannulation et montrons que les ferms rgulus concident avec les ferms algbriquement constructibles . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ msc 2000 : _
14p99 , 14e05 , 14f17 , 26c15 _ keywords : _ regular function , regulous function , rational function , real algebraic variety _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
although sometimes sidestepped , it is a general requirement in studying non vacuum spacetimes in general relativity that the energy momentum tensor , that is the matter ( field ) contents , should have a clear , although possibly highly idealized , physical interpretation . among these choices
the case where matter is described as a large ensemble of particles that interact only through the gravitational field that they themselves , at least partially , create , is of particular interest , both because of their usefulness in modeling physical systems such as star o galaxy clusters , and of the possibility of a relatively detailed analysis , at least in some restricted cases . as usual in theoretical treatments ,
one starts imposing as many restrictions as compatible with the central idea , and then tries to generalize from these cases . in this respect , the restriction to static spherically symmetric systems provides an important simplification , although even with this restriction the problem is far from trivial , and further restrictions have been imposed in order to make significant advances .
one of the first concrete examples is that provided by the einstein model @xcite , where the particles are restricted to move on circular orbits .
this model is static , and it is not easy to generalize as such to include a dynamical evolution of the system .
this generalization can , however , be achieved if the particle world lines are restricted to a shell of vanishing thickness ( `` thin shell '' ) , as considered by evans in @xcite .
the analysis in @xcite , although , motivated by the einstein model , considers only shells where all the component particles have the same value of their ( conserved ) angular momentum . in a recent study @xcite of the dynamics of spherically symmetric thin shells of counter rotating particles ,
of which @xcite is an example , it was found that the analysis can be extended to shells where the particles have angular momenta that take values on a discrete ( but possibly also continuous ) set , and is not restricted to a single value .
it was also found that in the non trivial thin shell limit of a thick einstein shell the angular momentum of the particles acquires a unique continuous distribution , and , therefore , the models in @xcite and @xcite are not approximations to an einstein model .
a relevant question then is what , if any , are the ( thick ) shells that are approximated by those in @xcite and @xcite . in this paper
we look for an answer to this question by considering a generalization of the einstein model where instead of circular orbits we impose , at first , the restriction to a single value of the angular momentum .
the particle contents is described by a distribution function @xmath1 in phase space , and , because of the assumption of interaction only through the mean gravitational field , @xmath1 must satisfy the einstein - vlasov equations @xcite . in the next section we set up the problem and show that it leads to a well defined set of equations . in section iii we set up and analyze a particular model , obtaining expansions for the metric functions at the boundary of the support of @xmath1 , appropriate for numerical analysis .
further properties are analyzed in section iv , where we show that all these shells have finite thickness .
section v contains numerical results for a generic example .
the `` thin shell '' limit is considered in section vi , both through analytic arguments and a concrete numerical example , with the results showing total agreement with the thin shell results of @xcite .
a further comparison with @xcite is carried out in section vii , where the stability of a shell approaching the thin shell limit is considered . the generalization to more than one value of @xmath0 is given in section viii , where we find that particles with different values of @xmath0 may be distributed on shells that overlap completely , or do so only partially or not at all .
numerical examples and a comparisons with @xcite are finally developed in section ix . some comment and conclusions
are given in section x.
the metric for a static spherically symmetric spacetime may be written in the form , @xmath2 where @xmath3 is the line element on the unit sphere and @xmath4 . for a static , spherically symmetric system ,
the matter contents , in this case equal mass collisionless particles , is microscopically described by a distribution function @xmath5 , where @xmath6 are the components of the particle momentum , taken per unit mass .
then , as a consequence of the assumption that the particles move along geodesics of the space time metric , the distribution function @xmath7 satisfies the vlasov equation , which , in this case , takes the form , @xmath8 where @xmath9 correspond to @xmath10 .
it is understood in ( [ vlasov1 ] ) that @xmath11 is to be computed using @xmath12 , so as to satisfy the `` mass shell restriction '' @xmath13 , where @xmath14 is the particles mass .
therefore , in what follows we set , @xmath15}\ ] ] we also notice that @xmath16 , where @xmath17 is proper time along the particle s world line .
+ the einstein equations for the system are , @xmath18 with the energy momentum tensor given by , @xmath19 where @xmath20 is the determinant of @xmath21 , and @xmath22 . equations ( [ vlasov1 ] ) , ( [ einst1 ] ) and ( [ tmunu1 ] ) define the einstein - vlasov system restricted to a static spherically symmetric space time , with the metric written in the form ( [ one1 ] ) .
the assumption that the metric is static and spherically symmetric implies conservation of the particle s energy , @xmath23}\end{aligned}\ ] ] and of the square of its angular momentum per unit mass , @xmath24\ ] ] it is easy to check that the ansatz , @xmath25 where @xmath26 , and @xmath27 are the functions of @xmath28 , and @xmath29 given by ( [ energy1],[lsquare ] ) , solves the vlasov equation for an arbitrary function @xmath30 . + to construct and solve explicit models based on ( [ ansatz01 ] ) for the metric ( [ one1 ] ) , it is convenient to change integration variables in ( [ tmunu1 ] ) .
we set , @xmath31 and write ( [ tmunu1 ] ) in the form , @xmath32 where we should set , @xmath33 andreasson and rein @xcite have explored the properties of models where @xmath30 takes the form , @xmath34 in this note we consider a different type of models , based on the ansatz , @xmath35 where @xmath36 is the heaviside step function , and @xmath37 is dirac s @xmath38 ; namely , we assume that @xmath0 takes only the single value @xmath39 , and that there is an upper bound on @xmath26 , given by @xmath40 .
@xmath41 is assumed to be a smooth function of @xmath26 .
we then have , @xmath42 where @xmath43 depends on @xmath28 and is given by , @xmath44 if @xmath45 , and @xmath46 otherwise .
this simply states the fact that @xmath47 only in those regions where a ( test ) particle with energy @xmath40 and angular momentum @xmath39 can actually move .
we may now use the previous results to construct simple models and analyze their interpretation for a range of possible parameters .
this analysis may be carried out in a number of ways . here
we choose the following ; we first use a gauge freedom in the metric ( [ one1 ] ) to set , @xmath48 then , from the einstein equations and the form ( [ tmunu2 ] ) of @xmath49 , we find two independent equations for @xmath50 and @xmath51 , @xmath52 where , @xmath53 is the energy density , and @xmath54 is the radial pressure , given by ( [ tmunu2 ] ) , with @xmath55 . there is also an equation for @xmath56 , but , as can be checked , this is not independent of ( [ two2 ] ) . equations ( [ two2 ] ) are deceivingly simple , because the explicit dependence of @xmath57 and @xmath58 on @xmath59 and @xmath60 is in general quite complicated . here
we consider a simple example and propose a method for constructing the solutions , that is illustrated by the example .
it can be seen that some simplification is attained if we choose , @xmath61 where @xmath62 is a constant . with this choice
we may perform the integrals in ( [ tmunu2 ] ) explicitly and , after some simplifications , we get , @xmath63\sqrt{r^2 e_0 ^ 2 - ( l_0 ^ 2+r^2)b}}{r^3 b } \\ \label{two4b } \frac{d b}{dr } & = & \frac{2 m b}{r(r-2 m ) } + \frac{2q_2\left[r^2 e_0 ^ 2- ( l_0 ^ 2+r^2)b\right]^{3/2}}{r^3(r-2m)}\end{aligned}\ ] ] where @xmath64 .
we also find , @xmath65\sqrt{r^2 e_0 ^ 2 - ( l_0 ^ 2+r^2)b}}{4 \pi r^5 b } \nonumber \\
p(r ) & = & \frac{q_2\left[r^2 e_0 ^ 2- ( l_0 ^ 2+r^2)b\right]^{3/2}}{4 \pi b r^5 } \nonumber \\
p_t(r ) & = & \frac{3 q_2 l_0 ^ 2 \sqrt{r^2 e_0 ^ 2- ( l_0 ^
2+r^2)b}}{8 \pi r^5}\end{aligned}\ ] ] considering ( [ two4a],[two4b ] ) , we find that it is a simple , but rather difficult to handle , system of equations for @xmath50 and @xmath51 .
we have not found closed ( analytical ) solutions for the system , and , therefore , we must resort to numerical methods .
the application of these methods requires , however , considering and solving several subtleties inherent in the system .
as indicated above , in all these equations ( ( [ two4a],[two4b ] ) and ( [ two5 ] ) ) , the terms involving @xmath66 should be set equal to zero if @xmath67 .
we notice that for @xmath68 we have @xmath69 with @xmath70 , and @xmath71 , where @xmath72 is also a constant , corresponding to the standard schwarzschild solution . for a shell type solution , these solutions correspond to the inner and outer regions , to be matched to the region where @xmath47 .
when @xmath73 , since we must have @xmath74 , we must also have @xmath75 , but , even though @xmath76 we might end up with @xmath77 , and still have all equations satisfied . for shell like solutions , either with an empty interior or with a central mass ( black hole ) , a further difficulty can be seen considering that there should exist an `` allowed region '' where @xmath78 , with @xmath28 taking values in the interval @xmath79 , where @xmath80 and @xmath81 are , respectively , the inner and outer radii of the shell .
we must impose continuity in both @xmath51 and @xmath50 to avoid @xmath38 functions in @xmath49 .
this implies that @xmath82 is continuous in @xmath79 and approaches continuously the value zero at the boundaries .
therefore , both @xmath83 and @xmath84 are also continuous inside and at the boundaries of this interval , and actually we have @xmath85 .
we also find that @xmath86 should be continuous , but @xmath87 must be singular , and this makes the construction of numerical solutions where we try to fix from the beginning the values of @xmath80 and @xmath81 rather difficult .
nevertheless , the above analysis indicates that for @xmath88 , but @xmath89 , we should have , @xmath90 where @xmath72 , @xmath91 , and @xmath92 are constants and @xmath93 and @xmath94 are functions of @xmath28 that vanish respectively faster than @xmath95 and @xmath96 for @xmath97 .
it is straightforward to extend this analysis to higher order by imposing consistency between the right and left hand sides of ( [ two4a],[two4b ] ) as @xmath97 .
we find , @xmath98 where @xmath99 , @xmath100 , @xmath101 and @xmath102 are constants , and @xmath103 and @xmath104 stand for higher order terms .
the constants appearing in ( [ two7 ] ) are not independent .
they may be written , e.g. , in terms of @xmath80 , @xmath92 , @xmath39 , @xmath40 and @xmath66 .
we notice that @xmath92 is the schwarzschild mass for the region inside the shell ( @xmath105 ) . moreover , the system ( [ two4a],[two4b ] ) is invariant under the the rescaling @xmath106 , @xmath107 , and @xmath108 .
the condition that @xmath109 corresponds to the inner boundary of the shell implies , @xmath110 similarly , we find , @xmath111 the explicit expressions for @xmath100 and @xmath102 are also easily obtained but are rather long and will not be included here , although they were used in the numerical computations described below .
an interesting question regarding the model of the previous section is related to the possible values that the thickness of the shells can attain .
this may be analyzed by considering the limit of solutions of the system ( [ two4a],[two4b ] ) as @xmath112 , under the restrictions that @xmath113 , and @xmath114 .
the first , according to ( [ two4a],[two4b ] ) and ( [ two5 ] ) , implies @xmath115 , and @xmath116 .
we remark that @xmath117 .
therefore , @xmath51 must approach @xmath72 monotonically from below .
consider first the case @xmath118 .
replacing in the first equation in ( [ two4a],[two4b ] ) , for large @xmath28 we find that @xmath83 approaches a constant value , and , therefore , @xmath50 grows linearly with @xmath28 .
but then , replacing in the second equation in ( [ two4a],[two4b ] ) , we find that @xmath84 decreases as @xmath119 , leading to a logarithmic growth in @xmath51 , incompatible with the assumed conditions
. therefore , any possible solution should have @xmath120 .
then , for large @xmath28 we should have , @xmath121 with @xmath122 , as @xmath123 . replacing now in ( [ two4a],[two4b ] ) , to leading order we find , @xmath124 and this implies that @xmath125 as @xmath112 . then , using again ( [ two4a],[two4b ] ) , we should have , @xmath126 and this implies @xmath127 , which contradicts the assumption @xmath128 .
thus we conclude that the equation , @xmath129 must always be satisfied for finite @xmath28 , and , therefore , _ all shells constructed in accordance with the prescription ( [ two3 ] ) have finite mass and finite thickness_. we nevertheless believe that this results is more general , and applies to all shells satisfying the ansatz ( [ model01 ] ) , although we do not have a complete proof of this statement .
as indicated , we do not have closed solutions of the equations for @xmath51 and @xmath50 , even for the simple model of the previous section .
nevertheless , since ( [ two4a],[two4b ] ) is a first order ode system , we can apply numerical methods to analyze it .
we may use the expansions ( [ two7 ] ) , ( disregarding the terms in @xmath103 and @xmath104 ) , to obtain appropriate initial values for @xmath51 and @xmath50 , for @xmath28 close to @xmath80 , in the non trivial region @xmath88 .
we may illustrate this point with a particular example .
we take @xmath130 , @xmath131 , @xmath132 , @xmath133 and @xmath134 . using these values and ( [ two7 ] )
( truncated as indicated above ) , we find @xmath135 , and @xmath136 ( actually , the computations were carried out to 30 digits , using a runge - kutta integrator ) .
the numerical results are plotted in figure 1 and figure 2 .
one of the motivations for studying the type of shells considered in this paper is the possible existence of a non trivial `` thin shell '' limit , where the thickness of the shell goes to zero , with the restriction to a single or a finite set of values of @xmath0 , and how this limit compares with the thin shells considered in @xcite and @xcite .
we remark that the existence of thin shell limits of einstein - vlasov systems has already been analyzed in the literature @xcite . here
we are interested not only in the existence of this limit for our particular models , but especially in the limiting values of the parameters characterizing our shells .
since this type of analysis is not immediately included in , e.g , @xcite , we consider it relevant to provide an explicit proof of the properties of our models in the thin shell limit .
we first recall that for a static thin shell constructed according to evans prescriptions @xcite , we have the following relation between the radius @xmath137 , inner ( @xmath92 ) and outer ( @xmath101 ) mass , and angular momentum @xmath138 of the particles , @xmath139 we may now prove that the non trivial thin shell limits of the shells constructed according to the prescription ( [ two3 ] ) effectively coincide with the evans shells of reference @xcite as follows .
we first take the @xmath28 derivative of ( [ two4b ] ) , and then use ( [ two4a ] ) and ( [ two4b ] , to obtain , @xmath140}{r(e^2 r^2 + 2(l^2+r^2)b)(r-2m)}\frac{dm}{dr } \nonumber \\ & & -\frac{4 } { r^2(r-2m)}\left[r(r-2m)\frac{db}{dr}-m b\right]\end{aligned}\ ] ] next let @xmath80 and @xmath81 be , respectively , the inner and outer radii of the shell , and @xmath141 and @xmath142 the corresponding masses inside and outside the shell .
then , for @xmath79 we have @xmath143 , and , @xmath144 the idea now is to use the fact that , @xmath145 on this account we rewrite ( [ thin01 ] ) in the form , @xmath146}\frac{d^2b}{dr^2 } \nonumber \\ & & + \frac{4 ( e^2 r^2 + 2b(l^2+r^2))\left[r(r-2m)\frac{db}{dr}-m b\right ] } { r ( r-2m)\left [ r(2 e^2 + ( r^2+l^2)b)\dfrac{db}{dr}+2(4e^2r^2+(2l^2-r^2)b)\right]}\end{aligned}\ ] ] and integrate both sides from @xmath109 to @xmath147 .
but now we notice that while both @xmath50 and @xmath84 are rapidly changing but bounded in @xmath79 , the change of @xmath51 , and @xmath28 in that interval is only of order @xmath148
. we may then choose a point @xmath149 , with @xmath150 , and set @xmath151 , and @xmath149 , except in the arguments of @xmath50 , @xmath83 , @xmath152 and @xmath153 , in ( [ thin04 ] ) , as this introduces errors at most of order @xmath148 , in the factors of @xmath83 , and @xmath153 , and in the last term in the right hand side of ( [ thin04 ] ) .
similarly , we may set @xmath154 in ( [ thin04 ] ) , to obtain , up to terms of order @xmath148 , @xmath155 } { r ( r-2m)\left[r(r^2+l^2)\dfrac{db}{dr}+2(r^2 + 2 l^2)b_0\right]}\end{aligned}\ ] ] and , again , we notice that the last term on the right of ( [ thin05 ] ) gives a contribution of order @xmath148 .
we then conclude that , @xmath156 the integration of the terms in @xmath83 , and @xmath153 , is now straightforward .
we use next ( [ thin02 ] ) and the fact that in this limit @xmath157 and @xmath158 , to obtain , @xmath159 solving this equation for @xmath27 , we finally find , @xmath160 which is , precisely , the relation satisfied by the parameters of the shells of ( [ five1 ] ) . we can also check this result , and , in turn , the accuracy of numerical codes , by directly considering initial data for the numerical integration that effectively lead to shells where the thickness is a small fraction of the radius .
a particular example is given in figure 3 , where the values of the initial data is also indicated .
we can see that the mass increases by about @xmath161 percent , while the thickness of the shell is less than @xmath162 percent of the shell radius .
we can check that these results are in agreement with ( [ five1 ] ) . solving this for @xmath101 , @xmath163 } { ( r-2m_1)(r^2 + 3 \tilde{l}_0 ^ 2)^2 } \ ] ] and replacing @xmath164 , @xmath165 , and @xmath166 ,
we find @xmath167 in very good agreement with the numerical results quoted in figure 3 .
consider a shell approaching the thin shell limit .
we restrict to the case of vanishing inner mass @xmath168 .
if @xmath80 and @xmath81 are , respectively the inner and outer boundaries of the shell , the matching conditions in the absence of singular shells at @xmath80 and @xmath81 imply that both @xmath50 and @xmath51 and their first derivatives should be continuous at both @xmath109 and @xmath147 . the form ( [ one1 ] ) of the metric with the choice ( [ two1 ] ) imply that for @xmath169 we have , @xmath170 where @xmath171 is a constant .
then , at @xmath109 we have , @xmath172 for @xmath173 , the metric takes the form , @xmath174 where @xmath175 is another constant satisfying , @xmath176 consider now a test particle moving along a geodesic of the shell space time , with 4-velocity @xmath177 , with @xmath178 . without loss of generality we may choose @xmath179 .
then we have the constants of the motion : @xmath180 and the normalization of @xmath181 implies , @xmath182 therefore , for @xmath183 we have , @xmath184 and the particle radial acceleration is given by , @xmath185 if we assume now that the shell is close to the thin shell limit , with angular momentum @xmath39 , radius @xmath137 , ( with @xmath150 ) , and mass @xmath101 , then we should have , @xmath186 then , for @xmath187 , and @xmath188 we find , @xmath189 similarly , in the region @xmath190 , for @xmath191 we find , @xmath192 and , therefore , all these shells are stable under `` single particle evaporation '' , in total agreement with the results of @xcite .
a related problem is that of the dynamical stability of the shell as a whole , as was also analyzed in @xcite .
there , the shells considered where `` thin '' , and therefore , the motion was described by ordinary differential equations for the shell radius @xmath137 as a function , e.g. , of proper time on the shell , which allowed for a significant simplification of the treatment of the small departures from the equilibrium configurations .
unfortunately , the corresponding equations of motion for the shells considered here would be considerably more complicated and their analysis completely outside the scope of the present research . we , nevertheless , expect that such treatment , if appropriately carried out , would also agree with the results found in @xcite as the `` thin shell '' limit is approached .
it is rather simple to extend the analysis of the previous sections to the case where the angular momentum of the particles takes on a discrete , finite set of values . instead of ( [ model01 ] )
, we have , @xmath193 where the functions @xmath194 are arbitrary , with @xmath195 , and @xmath196 finite .
we will restrict to the case of two separate values , ( n=2 ) , since , as will be clear from the treatment , the extension to a larger number of components is straightforward .
we are actually interested in the behaviour of these shells as they approach a common thin shell limit .
therefore , we will further simplify our ansatz to the form , @xmath197 where @xmath198 are constants . with this choice we may perform the integrals in ( [ tmunu2 ] ) explicitly and , after some simplifications , we get , @xmath199\sqrt{r^2 e_1 ^ 2 - ( l_1 ^ 2+r^2)b}}{r^3 b } \nonumber \\ & & + \frac{c_2 \left[2(l_2 ^ 2+r^2)b+r^2e_2 ^ 2\right]\sqrt{r^2 e_2 ^ 2 - ( l_2 ^ 2+r^2)b}}{r^3 b } \nonumber \\
\frac{d b}{dr } & = & \frac{2 m b}{r(r-2 m ) } \nonumber \\ & & + \frac{2c_1\left[r^2 e_1 ^ 2- ( l_1 ^ 2+r^2)b\right]^{3/2}}{r^3(r-2 m ) } + \frac{2c_2\left[r^2 e_2 ^ 2- ( l_2 ^ 2+r^2)b\right]^{3/2}}{r^3(r-2m)}\end{aligned}\ ] ] where @xmath200 , @xmath201 .
we also find , @xmath202\sqrt{r^2 e_1 ^ 2 - ( l_1 ^ 2+r^2)b}}{4 \pi r^5 b } \nonumber \\ & & + \frac{c_2 \left[2(l_2 ^ 2+r^2)b+r^2e_2 ^ 2\right]\sqrt{r^2 e_2 ^ 2 - ( l_2 ^ 2+r^2)b}}{4 \pi r^5 b } \nonumber \\
p(r ) & = & \frac{c_1\left[r^2 e_1 ^ 2- ( l_1 ^ 2+r^2)b\right]^{3/2}}{4 \pi b r^5 } + \frac{c_2\left[r^2 e_2 ^ 2- ( l_2 ^ 2+r^2)b\right]^{3/2}}{4 \pi b r^5 } \nonumber \\
p_t(r ) & = & \frac{3 c_1 l_1 ^ 2 \sqrt{r^2 e_1 ^ 2- ( l_1 ^ 2+r^2)b}}{8 \pi r^5 } + \frac{3 c_2 l_2 ^ 2 \sqrt{r^2 e_2 ^ 2- ( l_2 ^ 2+r^2)b}}{8 \pi r^5}\end{aligned}\ ] ] it is clear that we recover the results of the previous sections if we set either @xmath203 or @xmath204 equal to zero .
+ it will be convenient to define separate contributions to the density , @xmath205 and @xmath206 , for the particles with @xmath207 and @xmath208 .
@xmath209\sqrt{r^2 e_1 ^ 2 - ( l_1 ^ 2+r^2)b}}{4 \pi r^5 b } \nonumber \\
\rho_2(r ) & = & \frac{c_2 \left[2(l_2 ^ 2+r^2)b+r^2e_2 ^ 2\right]\sqrt{r^2 e_2 ^ 2 - ( l_2 ^ 2+r^2)b}}{4 \pi r^5 b}\end{aligned}\ ] ] then , provided the integrations cover the supports of both @xmath205 and @xmath206 , we have , @xmath210 where @xmath211 these expressions may be considered as the contributions to the mass from each class of particles .
this will be used in the next section to compare numerical results with the thin shell limit .
equations ( [ twodos4 ] ) may be numerically solved for appropriate values of the constants @xmath212 , @xmath213 , and @xmath214 , and initial values , i.e. , for some @xmath28 , of @xmath50 and @xmath51 .
we remark that , as in the previous sections , it is understood in ( [ twodos4 ] ) , ( and also in ( [ twodos5 ] ) ) , that both @xmath215 and @xmath216^{3/2}$ ] must be set equal to zero for @xmath217 . in this general case , it is clear that the shells ( where by a `` shell '' we mean here the set of particles having the same angular momentum @xmath214 ) may be completely separated or they may overlap only partially .
we are particularly interested in the limit of a common thin shell for the chosen values of @xmath214 . one way of ensuring that at least one of the shells completely overlaps the other is the following .
we choose an inner mass @xmath92 , and an inner radius @xmath80 .
this implies @xmath218 , while @xmath219 is arbitrary .
if we choose now arbitrary values for @xmath207 and @xmath208 , the density @xmath57 will vanish at @xmath109 if we choose , @xmath220 actually we also need to impose , @xmath221 to make sure that @xmath109 is the _ inner _ and not the _ outer _ boundary of the shells .
we shall assume from now on that @xmath222 . since , using the same arguments as in the single shell case , the shells have finite extension , it follows that one of the shells will be completely contained in the other . in the next section we display some numerical results , both for thick shells that overlap partially , and for shells approaching the thin shell limit .
we again find that the limit is associated to large values of the @xmath212 , and that the parameters describing the shells approach the thin shell values found in @xcite .
as a first example we take @xmath223 , @xmath224 , @xmath225 , and @xmath165 .
we also set @xmath226 , for simplicity .
then , from ( [ twodos6 ] ) , we set @xmath227 , @xmath228 . finally , we choose @xmath229 , @xmath230 , and carry out the numerical integration .
the results obtained indicate that the particles with @xmath223 are contained in the region @xmath231 , while for @xmath232 the corresponding range is @xmath233 . the resulting value of the external mass in @xmath234 . in figure 4
we display the total density @xmath57 as a function of @xmath28 ( solid curve ) , as well as the contributions @xmath205 and @xmath206 to the density from the particles with respectively @xmath223 ( dashed curve ) and with @xmath224 ( dotted curve ) . as an illustration of the approach to the thin shell configuration we considered again the previous values
@xmath223 , @xmath224 , @xmath225 , @xmath165 , @xmath226 , but choose @xmath235 , @xmath236 , and carried out the numerical integration .
figure 5 displays the functions @xmath237 ( solid curve ) , @xmath238 ( dashed curve ) , and @xmath239 ( dotted curve ) .
we see that now the shell extends only to the region @xmath240 , i.e. , its thickness is less than one percent of its radius . the mass , on the other hand , increases by roughly a factor of two , since @xmath241
. we may compare these results with those of the thin shell limit of @xcite as follows .
it can be seen from ( 29 ) and ( 31 ) in @xcite that for a thin shell of radius @xmath137 , inner mass @xmath92 and outer mass @xmath101 , with two components with angular momenta @xmath207 , and @xmath208 , the ratio of the contributions of each component to the total mass is given by , @xmath242 where @xmath138 is given by ( [ five1 ] ) .
the numerical integration gives @xmath243 , @xmath244 . if now take @xmath166 , and solve ( [ ratio1 ] ) for @xmath138 we find , @xmath245 , while replacement in ( [ five1 ] ) gives @xmath246 , which we consider as a good agreement , with a discrepancy of the order of the ratio of thickness to radius .
the general conclusion from this work is that one can effectively construct a wide variety of models satisfying the restriction that @xmath0 takes only a finite set of values , and that they do seem to contain the models used in @xcite as appropriate thin shell limits .
we remark also that the starting point for our construction is a variant of the ansatz used in @xcite , where @xmath1 is factored in an @xmath26 ( the particle energy ) and an @xmath0 dependent terms .
the possibility of multi - peaked structure in the case of more than one value of @xmath0 obtained here is also in correspondence with the general results obtained in @xcite .
this work was supported in part by grants from conicet ( argentina ) and universidad nacional de crdoba . rjg and mar are supported by conicet .
we are also grateful to h. andreasson for his helpful comments .
99 r. j. gleiser , m. a. ramirez _ class .
quant . grav _ * 26*,045006(2009 ) a. b. evans _ gen .
grav . _ * 8*,155(1977 ) a. einstein , _ ann .
_ * 40*,922(1939 ) h. andreasson and g. rein _ class .
grav . _ * 24*,1809(2007 ) for a recent review of the einstein - vlasov system and references see , for example @xcite .
h. andreasson , `` the einstein - vlasov system / kinetic theory '' , _ living rev .
relativity _ , * 8*,2 ( 2005 ) .
http://www.livingreviews.org/lrr-2005-2 , and references therein .
h. andreasson , _ commun .
phys . _ * 274 * , 409 - 425 ( 2007 ) . | in this paper we study static spherically symmetric einstein - vlasov shells , made up of equal mass particles , where the angular momentum @xmath0 of particles takes values only on a discrete finite set .
we consider first the case where there is only one value of @xmath0 , and prove their existence by constructing explicit examples .
shells with either hollow or black hole interiors have finite thickness .
of particular interest is the thin shell limit of these systems and we study its properties using both numerical and analytic arguments to compare with known results .
the general case of a set of values of @xmath0 is also considered and the particular case where @xmath0 takes only two values is analyzed , and compared with the corresponding thin shell limit already given in the literature , finding good agreement in all cases . |
in a series of publications @xcite an ambitious qcd program has been initiated to comprehensively investigate hadron structure .
the theoretical formulation entails renormalization and utilizes established many - body techniques to approximately diagonalize an effective confining hamiltonian .
this paper , a detailed continuation of our recent letter @xcite , focuses upon the quark sector and reports numerical results for mesons complementing our previous gluon study @xcite . over the years
there have been many meson investigations , from the early , simple non - relativistic constituent quark model calculations to more involved relativistic , field theoretical approaches implementing current quarks and spontaneous chiral symmetry breaking .
a common shortcoming of these analyses is an inability to consistently reproduce the physical mass spectrum of the scalar and pseudoscalar mesons .
our paper addresses this issue and significantly extends the pioneering work of the orsay group @xcite , adler and davis @xcite , and the lisbon investigators @xcite . in our approach the exact qcd hamiltonian in the coulomb gauge is modeled by an effective , confining hamiltonian , @xmath0 , that is fully relativistic with quark field operators and current quark masses . however , before approximately diagonalizing @xmath0 , a similarity transformation is implemented to a new quasiparticle basis having a dressed , but unknown constituent mass . as described in sec .
ii , this transformation entails a rotation which mixes the bare quark creation and annihilation operators . by then performing a variational calculation to minimize the ground state ( vacuum ) energy , a specific angle and corresponding quasiparticle mass
is selected . in this fashion chiral symmetry
is dynamically broken and a non - trivial vacuum with quark condensates emerges .
this treatment is precisely analogous to the bardeen , cooper , and schrieffer ( bcs ) description of a superconducting metal as a coherent vacuum state of interacting quasiparticles combining to form condensates ( cooper pairs ) .
excited states ( mesons ) can then be represented as quasiparticle excitations using standard many - body techniques which in this work will be the tamm - dancoff ( tda ) and random phase approximation ( rpa ) methods .
the two treatments are truncated at the one quasiparticle , one quasihole level and then numerically compared .
our rpa analysis confirms and extends the early work of ref .
@xcite which utilized an extended nambu - jona - lasinio mean field approach .
two other comments are in order before proceeding .
first , there are several reasons for choosing the coulomb gauge framework .
as discussed by zwanziger @xcite , the hamiltonian is renormalizable in this gauge and , equally as important , the gribov problem ( @xmath6 does not uniquely specify the gauge ) can be resolved ( see refs .
@xcite for further discussion ) .
related , there are no spurious gluon degrees of freedom since only transverse gluons enter .
this ensures all hilbert vectors have positive normalizations which is essential for using variational techniques that have been widely successful in atomic , molecular and condensed matter physics .
second , due to fock space truncations our analysis is not lorentz invariant .
however , we only plan to use one frame and do not compute hadron form factors with this method .
interestingly , violating lorentz non - invariance implies a prefered reference frame , which , as selected by chiral symmetry breaking , is the condensate rest frame .
this paper consists of six sections and two appendices .
the next section introduces our effective , qcd inspired hamiltionian and developes the bcs vacuum treatment leading to the quasiparticle mass gap equation .
we also compare our approach to the classic nambu - jona - lasinio model . in sec .
iii we detail our numerical , supercomputer solution of the gap equation along with the quark condensate and constituent mass values .
sections iv a and iv b describe the tda and rpa , respectively , while sec .
iv c addresses weak decays and sec iv d presents a derivation of the gell - mann - oakes - renner relation .
the tda and rpa meson spectra are compared and discussed in sec .
v. this section also includes results from a simple @xmath7 flavor mixing analysis for the @xmath8-@xmath9 system and our predictions for the charmed mesons .
conclusions and future work are summarized in sec .
vi . finally , appendix a provides further details regarding the bcs transformation and vacuum state while appendix b presents the most general tda equation for arbitrary angular momentum .
by introducing a phenomenological confining potential , @xmath10 , the qcd coulomb gauge hamiltonian @xcite for the quark sector can be replaced by an effective hamiltonian @xmath11 where @xmath12 , @xmath13 and @xmath14 are the current ( bare ) quark field , mass and color density , respectively ( for a more complete discussion , especially for the heavy quark sector , consult refs .
@xcite ) . for notational
ease the flavor subscript is omitted ( same @xmath0 for each flavor ) and the color index runs @xmath15 .
motivated by lattice gauge studies we adopt a linear confining interaction , @xmath16 , with slope @xmath17 @xmath18 also specified by lattice and regge phenomenology . in our analysis
we have also performed calculations with and without the leading qcd canonical or coulomb ( one - gluon exchange ) interaction , @xmath19 , with @xmath20 @xmath21 @xmath22 . for most observables , especially the meson mass spectrum ,
the coulomb interaction is not important and can be omitted .
this can be understood by noting that in momentum space , where we perform all calculations , the two interactions have the same sign , i.e. @xmath23 @xmath24 because the meson wavefunctions have a finite momentum distribution , most static meson properties are predominantly governed by the infrared ( ir ) , or low , momentum region where the confining potential dominates . including the coulomb interaction is then roughly equivalent to using a slightly larger string tension , @xmath25 .
there are certain observables , and in particular the gap equation detailed below , for which the coulomb interaction is ultra - violet ( uv ) divergent . in such cases
we regularize with a cut - off parameter and then could renormalize to remove cut - off sensitivity using one of our renormalization procedures detailed in refs .
@xcite for the gluon sector . in this paper we only present unrenormalized results since this program is still in progress @xcite and
has not yet been completed for the quark sector .
this is an additional reason for omitting the coulomb interaction .
hence , with the exception of the current quark masses ( we use @xmath26 @xmath27 , @xmath28 @xmath27 , @xmath29 @xmath27 ) , our approach entails only one pre - determined parameter which also sets the hadronic scale , @xmath30 @xmath27 .
we further note that even though the confining potential is ir divergent , this singularity is cancelled ( see ref .
@xcite ) in both the mass gap equation and all calculations for associated observables . hence
, the problem is the delicate numerical evaluation of this integrable singularity which we discuss in sec .
finally , we stress that in constituent quark models free quarks can exist which requires imposing color confinement .
however , as demonstrated in refs .
@xcite the lorentz structure of our coulomb gauge density - density confining interaction only permits stable solutions for color singlet states .
therefore , confinement naturally emerges in our approach .
we now wish to solve @xmath31 as accurately as possible . in this subsection
we focus on the ground state and introduce the bogoliubov - valatin , or bcs , transformation .
we begin by recalling the plane wave , spinor expansion for the quark field operator @xmath32 e^{i\vec{k } \cdot \vec{x}}\ ] ] with free particle , anti - particle spinors @xmath33 and bare creation , annihilation operators @xmath34 for current quarks , respectively . here the spin state ( helicity ) is denoted by @xmath35 and color index by @xmath36 ( which is hereafter suppressed ) .
because we can expand @xmath12 in terms of any complete basis we may equally well use a new quasiparticle basis @xmath37 e^{i\vec{k } \cdot \vec{x}}\ ] ] entailing quasiparticle spinors @xmath38 and operators @xmath39 .
the hamiltonian is equivalent in either basis and the two are related by a similarity ( bogoliubov - valatin or bcs ) transformation .
the transformation between operators is given by the rotation @xmath40 @xmath41 involving the bcs angle @xmath42 .
similarly the rotated quasiparticle spinors are @xmath43\ ] ] @xmath44\ ] ] where @xmath45 is the standard two - dimensional pauli spinor .
we have also introduced the gap angle , @xmath46 , which is related to the bcs angle , @xmath47 , by @xmath48 where @xmath49 is the current , or perturbative , mass angle satisfying @xmath50 with @xmath51 . hence @xmath52
@xmath53 similarly , the perturbative , trivial vacuum , defined by @xmath54 , is related to the quasiparticle vacuum , @xmath55 , by the transformation @xmath56 in this paper we will denote the bcs vacuum by @xmath57 ( in sec .
iv we introduce the rpa vacuum labeled @xmath58 ) . expanding the exponential and noting that the form of the operator @xmath59 is designed to create a current quark / antiquark pair with the vacuum quantum numbers , clearly exhibits the bcs vacuum as a coherent state of quark / antiquark excitations ( cooper pairs ) representing @xmath60@xmath61p@xmath62 condensates .
one can regard @xmath63 as the momentum wavefunction of the pair in the center of momentum system .
we now seek an approximate ground state for our effective hamiltonian by minimizing the bcs vacuum expectation , @xmath64 .
we do this variationally using the gap angle , @xmath65 , ( not the bcs angle ) which leads to the gap equation , @xmath66 .
after considerable mathematical reduction , the nonlinear integral gap equation follows @xmath67 \ .\ ] ] the angular integrals can be analytically evaluated ( see appendix b ) to give @xmath68\ ] ] where @xmath69 @xmath70\ ] ] corresponding to the linear potential above .
similar expressions for the coulomb potential are given in appendix b. there are several alternative ways to derive this same gap equation .
one is through the ward identites .
another is by requiring cancellation of the anomalous bogoliubov terms in the 2-body part of the newly normal ordered hamiltonian .
the latter is necessary to stabilize the vacuum and is also equivalent to minimizing the 0-body constant energy splitting the bcs and trivial vacua ( see ref .
@xcite ) .
numerically we actually solve a different form of the gap equation , originally obtained by adler and davis @xcite , that is more familiar to the solid state community .
they use the function @xmath71 related to our gap angle by @xmath72 @xmath73 with corresponding gap equation @xmath74 examination of eqs .
( [ gap1d],[gap1d ] ) reveals that the divergence at @xmath75 is an integrable singularity for the linear potential since for all @xmath76 the integrands vanish at @xmath75 .
this is not the case for the coulomb potential uv singularity .
however , since it naturally emerges from the canonical qcd hamiltonian ( one gluon exchange ) , we retain the option of including this potential for selected calculations and use a cut - off to regulate its ultraviolet ( uv ) divergence .
the solution of the gap equation ( see sec .
iii ) leads to a vacuum quark - antiquark condensate given by @xmath77 which is quadratically divergent for non - zero current quark mass @xmath78 .
we regulate this by subtracting the trivial condensate contribution giving @xmath79 our model is color confining and does not permit free solitary particles since the self - energy or dispersion relation @xmath80 obtained from the 1-body part of @xmath0 , eq .
( [ hamiltonian ] ) , is divergent ( now there is no cancellation at the singular point @xmath81 ) .
further , this divergence is also cancelled in the bound state equation but only for color singlet states ( see below ) . even though the self - energy diverges it
is still useful to introduce the concept of an effective quasiparticle ( constituent ) mass , @xmath82 , which can be extracted from the low momentum behavior of the gap angle .
we introduce a running , dynamical mass , @xmath83 , by an effective dirac spinor in canonical form @xmath84\ ] ] with normalization @xmath85 and @xmath86 . then using this equation and eq .
( [ inversebv ] ) we equate the two relative normalizations between upper and lower spinor components yielding a relation between the running dynamical mass and gap angle @xmath87 or @xmath88 we identify the dressed quark or quasiparticle mass as @xmath89 and extract it from the low momentum behavior of the gap angle ( essentially inverse of the slope near zero momentum ) .
the value of @xmath90 characterizes the degree of chiral symmetry breaking and can be loosely regarded as the constituent quark mass associated with phenomenological quark models .
note that our expression for the running mass is functionally identical to the perturbative expression @xmath91 .
related , since the rotated quasiparticle spinors have a running momentum dependence , they no longer rigorously provide a representation of the lorentz group .
our form of dynamical chiral symmetry breaking violates lorentz invariance which implies a preferred reference frame , namely the condensate rest frame . for most static observables such as masses , condensates and decay constants ,
lorentz symmetry is not important .
however , for some observables , such as electromagnetic form factors , care is necessary and boost corrections may be important .
this issue is under investigation and will be reported in a future communication . the classical effective model of nambu - jona - lasinio ( njl ) ( see ref .
@xcite for review ) entails various lagrangian formulations , a common one being @xmath92\ ] ] where @xmath93 is a constant .
it is customary to introduce the approximations @xmath94 and @xmath95 to linearize the equations of motion and then extract a constituent quark mass from the njl mass gap equation .
in this fashion chiral symmetry breaking is achieved .
our formulation extends beyond the njl model in several important ways . 1 .
our approach is more general and permits explicit gluonic degrees of freedom ( see refs .
the unification of the quark and glue sectors is crucial for a comprehensive treatment of hadron structure , especially for glueball and hybrid meson systems .
our formulation includes confinement and is renormalizable while the njl model has neither .
the njl pointlike interaction would be recovered in the limit @xmath96 which removes all important nonlocalities .
3 . our model has a density - density interaction kernel with a different lorentz structure , @xmath97 , which is the product of four - vector time components . as discussed in ref .
@xcite , a density - density ( vector - vector ) interaction is superior to the scalar - pseudoscalar displayed by the njl model .
the chiral symmetry breaking mode of the njl is extremely restrictive yielding a constant quasiparticle mass , @xmath98 , and simple dispersion @xmath99 .
related , the njl limit of our model also yields a more restricted gap angle since @xmath100 .
our method has a running mass and different quasiparticle dispersion which yields more realistic tda and rpa hadron masses .
the gap equation ( [ gap1d ] ) has been previously solved for the harmonic oscillator potential , where it takes its simplest form as a differential equation ( see ref .
@xcite ) , and also for the linear potential ( see ref .
@xcite ) . here
we summarize our analysis which confirms and extends the latter results . to numerically treat the integrable ir singularity a regularization
must still be implemented even though the final results are independent of this procedure .
we considered several different regularizations .
we first tried an analytical regulator ( equivalent to a deconfining correction to the potential ) .
this was an unstable algorithm and and convergence could not be achieved .
we next examined the method of ref .
@xcite which off - sets the q - discretization by half a step in the kernel with respect to the k - discretization .
this procedure was also rejected as it was less amenable for documenting the regulator sensitivity .
we finally adopted the simplest method of omitting the singular point @xmath101 .
this also facilitated a controlled senstivity study by just increasing the number of mesh points .
related , we adopted a variable mesh size to integrate more efficiently and mapped the integration variable @xmath102 to @xmath103 @xmath104 for n points uniformly distributed in the interval @xmath105 .
following ref .
@xcite we elected to solve the gap equation in form specified by eq .
( [ gap1d ] ) and also utilized the gauss algorithm as described there .
the gauss method assures convergence but is rather inefficient for extensive sensivity studies in parameter space .
we therefore modified our numerical approach by first finding a good approximate solution , @xmath106 , to the non - linear gap equation and then obtained a linear equation for the desired correction , @xmath107 , giving the final solution @xmath108 to arbitrary accuracy . substituting eq .
( [ guess ] ) in the gap equation , eq .
( [ gap1d ] ) , and dropping higher powers of @xmath107 yields the approximate linear equation @xmath109 \right ] \\ & + \frac{2}{3(2\pi)^2 } \int_0^{\infty } \frac{q^2dq}{1+\psi_q^{0 ^ 2 } } \left [ \hat{v}_0 \left ( 1-\frac{2\psi_q^{0 ^ 2 } } { 1+\psi^{0 ^ 2}_q } \right ) ( 1-\psi_k^{0 ^ 2 } ) + 4\hat{v}_1 \frac{\psi_k^0 \psi^0_q}{1+\psi_q^{0 ^ 2 } } \right]\delta_q= \\ & = -k\psi_k^0+\frac{m}{2}(1-\psi^{0 ^ 2}_k ) - \frac{2}{3(2\pi)^2 } \int_0^{\infty } \frac{q^2dq}{1+\psi_q^{0 ^ 2 } } \left [ \hat{v}_0 \psi_q^0 ( 1-\psi_k^{0 ^ 2})-\hat{v}_1\psi_k^0(1-\psi^{0 ^ 2}_q ) \right ]
\ .\end{aligned}\ ] ] this equation is of the form @xmath110 which can be solved for @xmath107 by matrix inversion .
we found the gauss algorithm sufficient for obtaining the initial approximate solution @xmath106 . to achieve full convergence required up to 12,000 mesh points , a factor of 60 more than the early calculations of ref .
@xcite .
we checked our computer codes by calculating two different toy kernels @xmath111 , @xmath112 and @xmath113 , @xmath114 , each designed to yield a known constant value for @xmath115
. we then performed a series of cut - off sensitivity runs and mapped out the convergence rate as a function of mesh point number @xmath85 which ranged from 100 to 12,000 .
we used the quark condensate as a test observable and also performed calculations for zero and non - zero current quark mass , @xmath13 , with and without the coulomb potential using @xmath116 . for @xmath117 , @xmath118
, we determined the sensitivity to @xmath85 ( the effective cut - off parameter ) was slightly higher than previously reported @xcite and given by @xmath119 ^ 3 \ .\ ] ] note that this number is somewhat smaller than the commonly accepted lattice value of about @xmath120 .
including the coulomb potential only increases the condensate to @xmath121 .
we therefore conclude an improved model ground state is needed which can be provided by including additional terms in the hamiltonian , such as the quark - gluon minimal coupling ( hyperfine ) interaction .
this point is also affirmed below in our rpa treatment which does yield a more realistic condensate value .
our other key result , which will be of interest in connection with chiral perturbation theory @xcite , is for the constituent quark mass and the bcs condensate as a function of the @xmath122 , @xmath123 quark mass .
now it is necessary to use eq .
( [ cond1 ] ) and also impose an additional integration cut - off limit ( @xmath124 around 10 @xmath125 ) .
this yields @xmath126 @xmath127 where the units are @xmath27 .
a precision calculation for @xmath90 with @xmath128 yields the slightly higher value @xmath129 @xmath130 ( all values for the linear potential only ) .
finally , we note that our hamiltonian is @xmath7 flavor symmetric , broken only by the small current flavored quark mass term . however , and quite significant , the vacuum properties and gap angle exhibit substantial @xmath7 violations as evidenced by our strange quark calculations using a current mass of 150 @xmath27 .
important violations occur even for a strange quark mass as low as 50 @xmath27 .
while this result may be model dependent it does suggest that certain chiral symmetry arguments in the literature regarding the strange quark sector should be taken with care .
we now formulate mesons as excited states consisting of quasiparticles and seek approximate eigensolutions of our hamiltonian .
we first develop the tda and then treat the rpa in the next subsection . of the two , only the rpa preserves chiral symmetry , as we detail below .
it is therefore more closely related to the bethe - salpeter formalism @xcite incorporating the schwinger - dyson quark propagator using an instantaneous interaction in the rainbow approximation ( equivalent to our gap equation ) .
we will document this connection more formally in a future publication .
the principle advantage of the tda is that it is a controllable approximation which truncates the fock - space expansion for a chosen level of calculational effort and resources . in terms of the quasiparticle operators introduced in sec .
ii , we introduce the tda meson creation operator @xmath131 a meson with quantum numbers @xmath132 ( radial - node number , @xmath133 , total angular momentum , @xmath134 , and parity , @xmath135 ) is then represented by the fock space expansion @xmath136 containing a quasiparticle and quasihole excited from the bcs vacuum . the hamiltonian equation is then projected onto this 1p-1h truncated fock sector giving the tda equation @xmath137 \arrowvert \omega \rangle= ( e_{njp}-e_0)\psi^{njp}_{\alpha \beta } \ .\ ] ] in evaluating the commutator we note @xmath138=0\ ] ] @xmath139\arrowvert \omega \rangle = ( \epsilon_{k}+ \epsilon_{k } ) \psi^{njp}_{\alpha\beta}\ ] ] and for the two body potential @xmath140 \arrowvert \omega \rangle = \frac{4}{3 } \sum_{\gamma \delta } \psi^{{njp}}_{\delta \gamma } \hat{v } ( { \vec k}_\delta - { \vec k}_\gamma ) u^{\dagger}_\alpha u_{\delta } v^{\dagger}_\gamma v_\beta\ ] ] where @xmath141 is the bcs gap energy given by ( [ dispersion ] ) and @xmath142 is the hamiltonian component containing n field operators ( after normal ordering with respect to the bcs vacuum ) .
we can exploit the rotational invariance of our hamiltonian and reduce the linear tda equation to a one dimensional , nonlocal equation by an angular momentum decomposition . introducing the orbital and spin angular momenta @xmath143 and @xmath144 , respectively ,
the meson state vector can be expanded in partial - waves involving a one - dimensional ( radial ) wavefunction @xmath145 @xmath146 where again the color index is omitted .
note the phase factor and negative magnetic substate sign in the clebsch - gordan coefficient due to the transformation properties of antiparticles under the @xmath147 rotation group . a thorough discussion is given in ref .
@xcite . inserting eq .
( [ njp ] ) into eq .
( [ tda ] ) yields the tda partial - wave equation of motion @xmath148 where the function @xmath149 contains the gap angle from contractions involving rotated spinors @xmath150 \end{aligned}\ ] ] with @xmath151 denoting the meson mass for state @xmath132 by @xmath152 and using the multipole expansion formulas for the interaction yields the final tda equation appropriate for numerical calculation @xmath153 note the hamiltonian spin dependence generates a kernel that couples different orbital and spin states .
we now apply these equations to the low lying meson spectrum with quantum states specified by @xmath154 having @xmath155 parity , @xmath156 , and @xmath93 parity , @xmath157 . in our model
we neglect the small electromagnetic ( isospin violating ) effects as well as coupling to the gluon sector so that @xmath158 and @xmath159 states are degenerate for the same @xmath160 . for pseudoscalar states , @xmath161 , @xmath162 giving only one wavefunction component ( no coupling ) .
this is also the case for scalar mesons ( @xmath163 ) having @xmath164 .
however for the vector meson sector @xmath165 both @xmath166 and @xmath167 waves are allowed and , in general , will be coupled .
similarly for low lying pseudovector mesons having @xmath168 or @xmath169 and tensor mesons , with @xmath170 and @xmath171 , there will be coupled equations .
although these equations are not difficult to solve , we only include the lowest orbital partial - wave component and neglect all coupling since it has been computed small for the harmonic oscillator potential @xcite .
there is then only one kernel @xmath172 for each meson state with structure : * pseudoscalar , @xmath173 , @xmath174 * scalar , @xmath175 , @xmath176 , @xmath177 * vector , @xmath178 , @xmath179 ( neglecting the tensor @xmath180 coupling ) , @xmath181 * pseudovector , @xmath182 , @xmath183 ( degenerate with @xmath184 ) , @xmath185 * tensor , @xmath175 , @xmath186 ( neglecting @xmath187 coupling ) , @xmath188 note for the pseudovector mesons the kernels for @xmath189 and @xmath190 are identical which differs from ref .
@xcite .
we have also applied our approach to other flavored ( @xmath191 and @xmath2 ) meson systems and have obtained similar , but more complicated tda equations .
as a representative result , consider the pseudoscalar @xmath192 meson with a @xmath122 ( or @xmath123 ) and @xmath2 quark .
the gap equation for the @xmath2 quark remains the same , except for current mass , now @xmath193 , which gives a different gap energy , @xmath194 , and angle , @xmath195 .
the tda equation , however , has a different form and generalizes to @xmath196 \\end{aligned}\ ] ] with obvious form for other mixed flavors .
all equations are finite for @xmath75 as the ir divergence terms from the confining potential again cancel .
we have also derived and solved the tda equations for other spin parity states which is further detailed in sec . v. the tda can be improved by utilizing a better vacuum with additional quasiparticle correlations beyond the bcs .
consistent with many - body applications in other disciplines we now formulate the rpa @xcite and introduce a new vacuum , @xmath197 , having both fermion ( two quasiparticles or cooper pairs ) and boson ( four quasiparticles or meson pairs ) correlations .
the rpa meson state @xmath198 involves a meson creation operator which is a generalization of eq .
( [ tdaop ] ) @xmath199 \ .\ ] ] the rpa vacuum then satisfies @xmath200 which , because of additional correlations from admixtures of particle - hole excitation states , is not true for the bcs vacuum . to derive the rpa equations of motion we use eq . ( [ tda ] ) and
replace the bcs vacuum with @xmath201 and also substitute @xmath202 for @xmath203 to generate one equation for the @xmath204 component .
we then repeat , changing the commutator to @xmath205 $ ] to obtain the @xmath206 equation .
following standard treatments in other fields of physics , we also invoke the quasiboson approximation and treat the fermion pair operator @xmath207 as a pure boson operator .
this significantly reduces the commutator algebra complexity and generates one of the two coupled equations for the rpa wavefunctions @xmath204 and @xmath206 . for the important pseudoscalar meson channel we obtain for the excited state @xmath133 @xmath208 = m_{n } x^{n}(k ) \\
2\epsilon_k y^{n}(k ) + \frac{1}{3 } \int_0 ^{\infty } \frac{q^2dq } { ( 2\pi)^2 } \left [ y^{n}(q)f(k , q ) + x^{n}(q ) g(k , q ) \right ] = -m_{n } y^{n}(k)\end{aligned}\ ] ] where @xmath209 similar expressions directly follow for the other spin - parity states .
we adopt the standard normalization for the rpa wavefunctions @xmath210 yielding @xmath211 the rpa equations , which reduce to the tda equations in the limit @xmath206 or @xmath93 @xmath212 , are again an eigenvalue problem for @xmath213 which can be easily diagonalized .
related , the matrix size can be reduced by a factor of 2 using the variables @xmath214 and @xmath215 .
finally , the equations are also ir finite for the sigular point @xmath75 .
a crucial test of any approach is the ability to describe hadronic decays . in this paper
we compute weak decays and defer our analysis of hadronic decays to a subsequent publication . for a pseudoscalar meson @xmath135 with momentum @xmath216 , energy @xmath217 , mass @xmath218 , the weak decay constant , @xmath219 , is defined for our normalization by @xmath220 here @xmath221 is the axial current which specifies the chiral charge operator @xmath222 simplifying eq .
( [ fdecay ] ) for a meson at rest yields @xmath223 applying this result for the tda pion wavefunction gives the tda pion decay constant @xmath224 similarly , for the rpa pion wavefunction we obtain @xmath225 our results easily generalize to the @xmath7 flavor nonet .
now there are nine axial charges given by @xmath226 where the eight gell - mann @xmath227 matrices are supplemented by @xmath228 to obtain both the octet and the singlet under @xmath7 transformations .
the appropriate generalizations of eq .
( [ f1 ] ) are then : @xmath229 @xmath230 @xmath231 we will use the above results in the next subsection to derive a generalized gell - mann - oakes - renner relation and also in sec .
v were we report numerical results .
chiral symmetry is a significant element of hadronic qcd and should be present in all realistic models .
even though our vacuum properly exhibits dynamic chiral symmetry breaking , our model hamiltonian does indeed respect this symmetry since the commutator @xmath232 = m \int d\vec{x } \psi^\dagger ( \vec{x } ) \psi ( \vec{x } ) \simeq 0\ ] ] essentially vanishes , consistent with the small @xmath122 , @xmath123 quark mass .
related , our rpa states also preserve chiral symmetry as the rpa meson creation operator commutes with the chiral charge in the chiral limit ( @xmath233 ) @xmath234 = 0 \ .\ ] ] however , the tda operator , eq .
( [ tdaop ] ) above , does not commute with @xmath235 and violates chiral symmetry since it is not fully symmetric in operator structure ( only contains @xmath236 and not @xmath207 ) .
this can also be documented by chiral transforming the tda meson state verifing that @xmath237 rotates to combinations of @xmath238 , @xmath239 and @xmath240 .
hence , the tda ansatz is not closed under a chiral rotation and goldstone bosons will not appear in the tda spectrum .
we therefore expect significant , but unphysical , chiral symmetry violations in the tda calculations and anticipate the tda pion mass to be much larger than in the rpa which is confirmed in sec .
v as only the rpa calculations yield a goldstone pion in the chiral limit . on the other hand , the rpa ansatz , eq .
( [ rpa1 ] ) , is chirally invariant since it is closed under this rotation in the chiral limit ( @xmath241 ) .
hence the mechanism of spontaneous symmetry breaking is present and the pion mass is zero according to its goldstone boson nature as we numerically verify in the next section . with these results
we now derive two different gell - mann - oakes - renner ( gmor ) relations , one based upon exact model eigenstates while the other relates rpa states and the bcs vacuum .
assume we have the complete set of exact eigenstates , @xmath242 , to our qcd model hamiltonian , including the vacuum ground state @xmath243 .
evaluating the double commutator @xmath244 \right ] \arrowvert \omega_{exact } \rangle = 4 m \langle \overline{q}q\rangle_{exact}\ ] ] then generates the exact quark condensate .
evaluating the double commutator again , but now invoking twice the completeness relation , @xmath245 , and identifying the decay constant relation , eq .
( [ decay ] ) , leads to the generalized gmor relation @xmath246 summed over all ( ground and excited ) pseudoscalar meson states with mass @xmath247 and decay constants @xmath248 .
this can be extended to flavor using two of the nine @xmath7 axial charge operators , @xmath249 and @xmath250 , to derive the following gmor relations : * a= b = 1 , 2 , or 3 @xmath251 * a = b = 4 , 5 , 6 , or 7 @xmath252 * a = b = 8 , @xmath253 * a = b = 0 @xmath254 because the exact eigenstates will generally not be available , these equations are of limited value .
however , they do provide testing criterion for approximation solutions . further , since decay constants are suppressed for excited states @xmath133 we can drop higher terms to obtain more useful relations involving ratios such as @xmath255 which we will use in sec .
v in connection with our discussion of the kaon mass calculation .
next we utilize thouless theorem @xcite applied to the chiral charge operator @xmath256 \right ] \arrowvert \omega \rangle = 2 \sum_{n}\arrowvert\langle\psi^{njp}_{rpa } \arrowvert q_5 \arrowvert \omega_{rpa } \rangle \arrowvert^2(e_n - e_0)_{rpa}\ ] ] to immediately derive the rpa gmor relation @xmath257 where we have again dropped the excited state decay constants .
note that the left hand side entails the bcs vacuum while the right hand side involves the rpa states and energies .
this relation clearly predicts the rpa pion is a golstone boson in the chiral limit ( i.e @xmath258 for @xmath233 ) which we numerically confirm in the next section .
finally , calculating the rpa meson mass spectrum does not require obtaining the rpa vacuum ( see eq.([rpaeq ] ) ) , but computing the rpa decay constants does .
determining @xmath259 is actually quite difficult , however , the leading correction can be approximately calculated using another theorem by thouless @xcite @xmath260 where @xmath261 is a tda type operator given by @xmath262 with the fermion pair operator @xmath237 now obeying bosonic commutation relations . here
@xmath263 is an unknown amplitude assumed to be small ( this is not true in the chiral limit as discussed in the next section ) .
the approximate rpa vacuum is thus described as a mixture of the bcs vacuum having cooper pairs ( eq . [ bcs ] ) and two quasibosons ( mesons ) coupled to vacuum quantum numbers ( @xmath264 ) . since , as shown in the next section , the rpa is of importance for mainly the low - lying pseudoscalar states , we assume that only these states contribute to @xmath265 and neglect all others .
our result easily generalizes to scalar or other mesons ( appropriately coupled to @xmath266 ) . within this approximation
we obtain an improved quark condensate to be compared with eq .
( [ cond ] ) @xmath267 to obtain @xmath268 , we impose @xmath269 and use eq .
( [ rpavacuum ] ) neglecting higher order terms corresponding to fock states with more than two pions .
we also only retain ground state meson ( pion ) contributions from @xmath204 and @xmath206 yielding @xmath270 with a normalization constant depending on both rpa wavefunction components @xmath271 the improved rpa quark condensate is @xmath272 where the constant c is given by @xmath273 also the improved pion decay constant to this vacuum is @xmath274 in contrast to eq .
( [ pdecay ] ) .
we have found that for meson masses above 800 @xmath27 @xmath206 is very small and there is no essential difference between rpa and tda .
we now comprehensively apply the above formulas and conduct a comparative analysis of the tda and rpa approaches .
in this section we first present and discuss our tda spectra for the pseudoscalar and vector mesons and then in subsection b we compare to our rpa results for both meson masses and the decay constants .
subsection c treats mixing for the @xmath8 and @xmath9 mesons while subsection d details applications to charmed mesons , especially @xmath5 states .
solving the tda eigenvalue problem is straight forward .
the diagonal part of the matrix contains the ir singularity that is rigorously controlled by cancellation , again permitting numerical regulation by simply skipping the point @xmath275 as in the gap equation solution . due to the linear nature of this system , results are convergent for a mesh as sparce as 700 points
. tables i and ii summarize the resulting tda meson spectrum corresponding to the five @xmath276 kernels specified in sec .
iv a. the energy difference between pseudoscalar and vector states is about 200 @xmath27 for the @xmath122/@xmath123 quark mesons , 80 @xmath27 for the open flavored ( @xmath277 , @xmath278 ) and only 50 @xmath27 for the pure strange composites ( @xmath279 ) .
our one parameter model can not accurately describe the entire observed splitting indicating additional dynamics beyond simple spin interactions from a dirac spinor field is needed . in general , the masses are in good agreement with the pdg @xcite accepted values for the vector mesons , but the tda low - lying pseudoscalar sector is deficient .
this is expected since in this channel vacuum ( chiral symmetry ) effects are most prominent . in the scalar channel
the situation is more confusing , since other hadron states , some with explicit gluonic structure , can more easily mix .
since our unified model allows us to treat glueballs , mesons and hybrids comprehensively , future work will further address understanding this channel .
interestingly , our lightest @xmath280 mass , which has a p wave oribital excitation , is below 1 @xmath125 .
we also find that the computed tda masses are not very sensitive to details of the vacuum that enters via the gap function characterizing the bcs ground state .
as described above , we may also use a simplified tda equation to extract a constituent quark mass . in the chiral limit ,
the mass obtained is 51 @xmath27 and this dressing is roughly constant , consistent with @xmath7 symmetry , up to current masses of 150 @xmath27 , where the generated constituent mass is 203 @xmath27 .
we are therefore dealing with light dressed quarks , even for strange quarks which may prove a deficiency when calculating electromagnetic form factors .
we shall also address this issue in the near future .
another important feature of our relativistic effective hamiltonian is that it naturally includes the kinematical and spin dependent interactions ( e.g. spin - spin , spin - orbit , tensor ) .
these effects are very important in the light quark sector even after chiral symmetry breaking , because the light quark constituent mass generated in our scheme , around 80 @xmath27 , is still small as compared to our interaction scale ( 424 @xmath27 ) .
in particular , notice in table ii the spin splitting between the @xmath264 , @xmath281 and @xmath282 mesons , all having the same @xmath283 and @xmath284 quantum numbers ( also observe the large radial excitation in each channel ) .
the level spacing is consistent with ref .
@xcite but very different from the naive expectations from the constituent quark model .
the is due to the confining ( nonperturbative ) @xmath285 coupling , which is only of order @xmath286 in the constituent quark model .
the spin spacing is governed by the matrix element @xmath287\ ] ] which , for @xmath288 , reduces to @xmath289 .
the @xmath290 in this expression describes the light scalar meson , whereas the @xmath291 contribution explains the splittings in table ii .
the calculated @xmath282 mass is much heavier than the lightest observed @xmath292 at 1270 @xmath27 .
clearly , including coupling to the @xmath293 , @xmath294 channel as well as multi - quark fock states will alter , but improve , our prediction .
further , and as mentioned previously , additional quark - gluon interaction terms should also be included in our hamiltonian which have a different lorentz spin structure . in our model
this would generate a weaker hyperfine interaction of order @xmath286 .
this would also provide additional splitting between @xmath295 and @xmath190 levels that would improve our tda , and especially rpa , @xmath3-@xmath4 mass underprediction .
such an analysis would fully clarify the relative importance of chiral symmetry versus spin interactions as the latter is generally attributed the dominant effect in conventional constituent quark models having color magnetic , effective one gluon exchange potentials ( see ref .
we are currently examining this issue and will report results in a future paper .
finally and also related is the spin - orbit splitting for other flavors which is summarized in tables i and iv . in table
iv we illustrate the familiar @xmath296 dependence by fitting the tda spectrum for different flavors .
we now present our rpa results and compare with both tda and observation .
table iii and figs . 2 through 7 highlight our key results .
in general the rpa and tda masses agree except for the light pseudoscalar mesons where the rpa provides a better description .
this is because , as discussed above , only the rpa correctly implements chiral symmetry as illustrated in fig .
3 where the lightest scalar and pseudoscalar masses are plotted as a continuous function of the current quark mass .
note that only the pseudoscalar mass approaches zero in the chiral limit consistent with goldstone s theorem ( the observed pion mass is reproduced for a @xmath122 quark mass of about 2 @xmath27 ) .
the kaon system reveals the largest model deficiency ( see table i ) .
even the rpa in the chiral limit produces a too massive kaon , about 850 @xmath27 . as indicated by fig . 4 , to reproduce the observed kaon requires a strange quark mass of about 50 @xmath27 .
a detailed analysis reveals that the explicit contribution of the current quark mass to the rpa equation -through its appearance in @xmath297- is only additive .
it is the gap angle , introducing an implicit flavor dependence , which inhibits a lighter kaon mass .
we could fit both pion and kaon masses by adjusting the current quark masses , but we prefer awaiting improvements from renormalizing the quark gap equation ( work in progress and will be subsequently publish ) .
the gap angle for a non - zero current quark mass is very sensitive at high momentum to the dominanting coulomb potential and sizeable corrections are expected .
also for higher lying excited meson states there is also the issue of two particle , two hole fock state contributions . in fig .
5 we compare our scalar , pseudovector and tensor meson tda and rpa predictions to data . in general
there is qualitative agreement .
note that for the higher excited states above 1 @xmath125 the tda and rpa results are identical and it is clear that these systems are not governed by chiral symmetry .
we also illustrate the behavior of various wavefunctions . in fig .
6 we detail the difference between tda wavefunctions for the @xmath280 and @xmath3 mesons .
figure 7 compares the tda and rpa wavefunctions for the pion .
finally we discuss decay constants .
both the tda and rpa pion decay constants are too small , about 17 @xmath27 , in contrast to the observed value of 93 @xmath27 .
this is consistent with the orsay @xcite results .
we again attribute this to the model hamiltonian and vacuum as the bcs angle does not have sufficiently high momentum components . however ,
if we use the approximate rpa vacuum in the quasiboson approximation ( see eq .
( 4.25 ) ) the improved rpa decay constant increases to 57 @xmath27 . appropriately
, the condensate also significantly increases to @xmath298 , in much better agreement with lattice results @xmath299 .
the latter result is consistent with applications in nuclear physics where the rpa tends to overcorrelate the ground state .
since we have only approximately evaluated the rpa vacuum by truncating at the two pion level it is not surprising that @xmath300 does not agree with measurement ( or chiral perturbation theory results ) and therefore needs further refining .
we expect the truncation to the two pion level , eq . ( [ rpavacuum ] ) , to be reasonable provided we are not in the chiral limit , since then @xmath301 .
however , we would like to be able to calculate in that regime .
further , the calculation in ref .
@xcite points to a necessary decrease of the bcs condensate when including coupled channels .
they argue that for a pure chiral pion , coupling additional channels would decrease its mass which might even become negative , destabilizing the vacuum .
this argument seems sound and since we are above the chiral quark mass limit , at a model value @xmath302 @xmath27 where our pion is too massive ( 277 @xmath27 ) , this decrease is a welcome improvement .
we defer further discussion until publication by our collaborative effort with the lisbon group which will also clarify pionic correlations in the ground state of our approach . recalling that our computed condensates require renormalization except in the chiral limit
, we can only test the gmor relation for @xmath117 which is trivially satisfied in the rpa .
however , we note that the decay constants for excited pion states are much smaller and rapidly approach zero in the chiral limit .
hence the gmor relation is satisfied with predominatly the first state .
this is not true for heavier quarks and to numerically satisfy the gmor requires including several eigenstates .
an extremely challenging but still not understood problem is the @xmath8 , @xmath9 system and attending flavor mixing of the light quarks .
although our effective hamiltonian has an explicit flavor dependence through the current quark masses , it still conserves flavor .
however , if the gluon sector is included , such as through the hyperfine , minimal coupling interaction , an effective flavor dependence naturally emerges though higher order quark - glue - quark effects and dynamic mixing of flavor states is possible .
we are currently deriving such a term which is similar , but more rigorous than the thooft interaction based upon instantons ( classical glue ) .
this will be reported in a future communication , however , it is still of interest to perform a simple @xmath8 , @xmath9 mixing analysis by introducing a flavor off - diagonal interaction as we now detail . with no dynamic flavor mixing ,
the @xmath8 and @xmath9 are ( poorly ) modeled as @xmath7 octet , @xmath304 , and singlet , @xmath305 , states respectively given by ( we adopt the convention used in refs .
@xcite ) @xmath306 involving the isoscalar @xmath307 and @xmath308 states and pure @xmath7 mixing angle @xmath309 .
we have computed the tda masses of the pure @xmath310 and @xmath308 meson states to be @xmath311 612 @xmath27 and @xmath312 1002 @xmath27 , respectively . hence the predicted , pure @xmath7 , @xmath8 ,
@xmath9 masses are @xmath313 for the rpa , @xmath311 290 @xmath27 and @xmath312 978 @xmath27 , yielding @xmath314 @xmath27 and @xmath315 @xmath27 .
similarly the @xmath7 @xmath8 , @xmath9 decay constants are given in terms of the @xmath316 isoscalar , @xmath248 , and strange quark , @xmath317 , decay constants @xmath318 note the quadratic dependence on angles due to expressing both the axial current and meson states in their @xmath7 representations . using the tda computed values of @xmath319 @xmath27 and @xmath320 @xmath27 , yields @xmath321 @xmath27 and @xmath322 40 @xmath27 .
the rpa values are @xmath323 @xmath27 and @xmath320 @xmath27 , giving @xmath324 @xmath27 and @xmath322 63 @xmath27 .
we can now improve these results by generalizing our hamiltonian , which is diagonal in flavor space , to have off - diagonal matrix elements .
the simplest prescription is to just add a constant , @xmath325 , to both diagonal , @xmath326 , @xmath327 , and off - diagonal @xmath328 , terms giving @xmath329 diagonalizing @xmath0 leads to the new , mixed mass eigenvalues @xmath330 and eigenstates @xmath331 involving rotation by an additional angle @xmath332 that is a function of @xmath35 .
the mixed , presumably more physical , decay constants are then @xmath333 performing a least squares fit to the observed masses ( @xmath334 547 @xmath27 , @xmath335 958 @xmath27 ) , yields @xmath336 ( @xmath337 @xmath338 ) for the tda , which in turn produces @xmath334 541 @xmath27 , @xmath335 974 @xmath27 , @xmath339 62 @xmath27 and @xmath340 -30 @xmath27 . for the rpa ,
@xmath341 82 @xmath27 ( @xmath342 @xmath343 ) generating @xmath334 433 @xmath27 , @xmath335 1081 @xmath27 , @xmath339 93 @xmath27 and @xmath340 -3 @xmath27 .
it is interesting that while the simple mixing provides improvement , the tda masses are in better agreement than the rpa .
we could also improve the decay constants utilizing a two angle mixing formalism @xcite but refrain since clearly a more sophisticated treatment is necessary which will be provided by our quark - gluon coupling formulation in the near future .
the constituent quark models and non - relativistic expansions of qcd offer more reliable results for heavy quark systems where physical intuition from quantum mechanical bound states is more appropriate .
hence the charmed mesons afford a good limiting testing for our relativistic approach .
we again calculate the spectrum of the charmed mesons ( see figures [ [ ucspectrum],[scspectrum ] ] ) and of charmonium ( fig . [ [ ccspectrum ] ] ) using our many - body model .
the tda is now sufficient since chiral symmetry , crucial for the light mesons , is not a constraint and the rpa will produce the same results . using a charmed quark mass of 1200 @xmath27 ,
the general features of the spectra are well reproduced and the radial excitations , @xmath344 and @xmath345 , are adequately described .
we therefore expect our predictions for the remaining unconfirmed states to be reasonable .
the angular momentum splittings of these systems are known to be dominated by the one gluon exchange potential ( oge ) which we have not included in this calculation .
hence there will be improvement from future calculations based upon our renormalized project . a general feature reflected by our charmed spectra
is the near vanishing of the spin - spin interaction , leading to degenerate @xmath346 and @xmath347 states .
since the 100 - 150 @xmath27 hyperfine splittings in these systems will presumably be recovered when we include the perturbative oge , we do not comment further . as for the spin - orbit splittings , our results are too large for the @xmath192 mesons and too small for the @xmath348 mesons .
these splittings are adequately explained in non - relativistic quark models ( see ref .
@xcite ) where the absence of a large spin - orbit effect for light quark masses is attributed to the cancellation between the thomas precession in the confining potential and the one gluon exchange effective potential , although it is a bit concerning that the actual splitting between the @xmath349 and the @xmath350 is twice the size of the @xmath350-@xmath351 splitting , when according to the spin - orbit @xmath291 rule , it should be a half .
finally , we note that by including d waves in the charmonium spectrum we are able to resolve the long standing `` overpopulation '' problem of @xmath5 states relative to observation .
this is clearly illustrated in fig .
this ( previous ) deficiency in number of @xmath352 states has been characterized as evidence for glueballs and/or hybrid mesons because the @xmath5 system is believed to be gluon rich .
our result suggests , however , that simple level counting may not be effective in identifying hadrons with explicit gluonic degrees of freedom .
before summarizing our results we comment on the strengths and weaknesses of our many - body approach as well as some attending , open hadronic physics issues . beginning with our hamiltonian , @xmath0 , the current - current (
density - density ) color interaction forbids free , isolated colored objects in the theory . for the vacuum
this is realized in the bcs by an infinite shift in the free quark self - energy due to the integral of @xmath353 .
similarly for hadrons in both tda and rpa , colored composite objects ( e.g. diquarks ) are precluded by the appearance of @xmath354 which is divergent , whereas in the singlet channel this divergence is removed by vanishing color factors .
next we note that @xmath0 conserves chiral symmetry yet our bcs vacuum properly exhibits dynamic chiral symmetry breaking .
further , our rpa pion emerges as a goldstone boson in the chiral limit .
this is not true for the tda pion since only the rpa excitation operator commutes with the chiral charge .
another , significant model feature is that fock state truncation is a controlable approximation amendable to systematic improvement .
thus our hamiltonian many - body approach is an attractive , promising method for comprehensively investigating hadronic structure as it embodies confinement , chiral symmetry breaking and orderly construction of multiparticle fock states .
it also provides an excellent vehicle for testing more fundamental effective hamiltonians as well as affording a powerful phenomenological framework for hadron structure .
considering the form of @xmath0 , with only a single predetermined dynamical parameter , it is encouraging that the chiral limit is adequately reached in the rpa and that the meson spectrum is in qualitative agreement with experiment . to achieve detailed
, quantitative descriptions will require further improvements in both the hamiltonian and effects from including higher fock space components .
in particular , both the high energy behavior ( coulomb potential ) and quark - gluon coupling effects ( efectively instantons ) will be incorporated and reported in a future publication . finally , our current study is similar , but more extensive than the orsay analysis @xcite due to our application to multiflavor systems .
our results are also more realistic ( and numerically more difficult ) then that work since we have utilized a linear confining interaction , determined by lattice and regge phenomenology , which generates complicated nonlocal integral equations , rather than solving a simpler differential equation for a harmonic oscillator potential . summarizing , we have performed approximate , but large - scale diagonalizations of an effective coulomb gauge hamiltonian utilizing standard many - body techniques . using the bcs ,
a non - linear gap equation has been derived and accurately solved to provide vacuum properties ( quasiparticles and condensates ) .
incorporating only predetermined parameters ( string tension , @xmath25 , and reasonable current quark masses ) , we have qualitatively reproduced the low energy @xmath355 and @xmath2 meson spectra .
most importantly , we have obtained a chiral pion , detailed that chiral symmetry is responsible for the large @xmath356 mass splitting and resolved the problem of overpopulation of theoretical @xmath5 states .
future work will address the full hamiltonian in the combined quark and gluon sectors .
in particular , we will obtain improved spin ( hyperfine ) and flavor ( thooft ) interactions from quark - glue coupling .
this should provide a better description of vacuum properties and the scalar / pseudoscalar masses , especially the @xmath357 and @xmath9 .
we will also include more complex 2 quasiparticle-2 quasihole fock states for heavier @xmath316 mesons as well as 3 quasiparticles for baryons and hybrids .
much of this work is in progress and will soon be reported .
the authors are grateful for comments and discussions with p. bicudo , j. e. ribeiro , a. szczepaniak and members ( present and former ) of the ncsu theory group .
this work is supported in part by grants doe de - fg02 - 97er41048 and nsf int-9807009 . f. j. llanes - estrada was a sura - jefferson laboratory graduate fellowship recipient .
supercomputer time from nersc is also acknowledged .
we further discuss the relation between the bcs rotated , @xmath358 , and the trivial or perturbative , @xmath359 , vacua .
we first note that the bcs vacuum state given by eq .
( [ bcs ] ) is not a unitary transformation and does not have a finite normalization .
this is because the operator in the exponential is not antihermitian .
it is therefore necessary to normalize matrix elements by dividing with @xmath360 and this is implicit in our presentation . alternatively , and equivalently ,
@xmath361 can be represented by a norm preserving unitary transformation of the form @xmath362 where @xmath363 and all flavor and color indices are suppressed . here
@xmath364 are matrix elements of the pauli matrices @xmath365
@xmath366 it is interesting to note that the bcs vacuum state is orthogonal to the trivial vacuum , @xmath367 , in the infinite volume limit .
further , the hilbert space vectors constructed from the two different vacua are also orthogonal provided the bcs angle is nonzero ( vacuum condensates are present ) . because of this property the bcs rotation has been called a pseudounitary transformation @xcite . consult this reference for further details ( note they have a different phase convention and use @xmath368 ) .
using the phase convention of ref .
@xcite , the general tda meson equation for arbitrary angular momentum is p. bicudo and j. e. ribeiro , phys .
d * 42 * , 5 ( 1990 ) ; p. bicudo , j. e. ribeiro , and j. rodrigues , phys .
c * 52 * , 4 ( 1995 ) ; p. bicudo , dissertation presented to `` instituto superior tecnico de lisboa '' , 1987 ; p. bicudo and j. e. ribeiro , private communication .
r. brockmann , w. weise , and e. werner , phys . lett . *
122b * , 201 ( 1983 ) ; v. bernard , r. brockmann , m. schaden , w. weise , and e. werner , nucl . phys . *
a412 * , 349 ( 1984 ) ; w. weise , internal review of nuclear physics ( world scientific ) , vol . 1 , 57 ( 1984 ) . | we represent qcd at the hadronic scale by means of an effective hamiltonian , @xmath0 , formulated in the coulomb gauge . as in the nambu - jona - lasinio model , chiral symmetry
is explicity broken , however our approach is renormalizable and also includes confinement through a linear potential with slope specified by lattice gauge theory .
this interaction generates an infrared integrable singularity and we detail the computationally intensive procedure necessary for numerical solution .
we focus upon applications for the @xmath1 and @xmath2 quark flavors and compute the mass spectrum for the pseudoscalar , scalar and vector mesons .
we also perform a comparative study of alternative many - body techniques for approximately diagonalizing @xmath0 : bcs for the vacuum ground state ; tda and rpa for the excited hadron states .
the dirac structure of the field theoretical hamiltonian naturally generates spin - dependent interactions , including tensor , spin - orbit and hyperfine , and we clarify the degree of level splitting due to both spin and chiral symmetry effects .
significantly , we find that roughly two - thirds of the @xmath3-@xmath4 mass difference is due to chiral symmetry and that only the rpa preserves chiral symmetry .
we also document how hadronic mass scales are generated by chiral symmetry breaking in the model vacuum .
in addition to the vacuum condensates , we compute meson decay constants and detail the nambu - goldstone realization of chiral symmetry by numerically verifying the gell - mann - oaks - renner relation . finally , by including d waves in our charmonium calculation we have resolved the anomalous overpopulation of @xmath5 states relative to observation . + pac number(s ) : 12.39.pn , 12.40.yx |
most previous papers on glass transitions so far are concerned with near - equilibrium properties such as relaxations of the density time correlation functions or dielectric response . however , these quantities are too restricted or indirect , and there remains a rich group of unexplored problems in far - from - equilibrium states .
in particular , shear is a relevant perturbation drastically changing the glassy dynamics when the shear rate @xmath0 exceeds the inverse of the structural relaxation time @xmath1 @xcite . in supercooled liquids ,
the microscopic rearrangement processes occur on the time scale of @xmath1 in quiescent states , whereas they are much accelerated even by extremely small @xmath0 ( if larger than @xmath2 ) .
jamming rheology _ has been studied in systems composed of large elements such as colloidal suspensions , dense microemulsions , and granular materials @xcite .
experimentally , simmons _ et al_. @xcite observed shear - thinning behavior , where the steady - state viscosity @xmath3 was represented by ( ) = _ xy/(0)/(1 + _ ) , [ eq:1 ] in the range @xmath4 ( poise ) in soda - lime - silica glass .
the characteristic time @xmath5 is expected to be of order @xmath6 . remarkably , the shear stress @xmath7 tends to a limiting shear stress , @xmath8 , of order @xmath9 , @xmath10 being the shear modulus for infinitesimal strains .
after application of shear , they also observed an overshoot of the shear stress before approach to a steady state .
as a closely related problem , understanding of mechanical properties of amorphous metals has been of great technological importance @xcite .
they are usually ductile in spite of their high strength . at low temperatures @xmath11 , localized shear bands , where zonal slip occurs ,
have been observed above a yield stress . at relatively high temperatures @xmath12 , on the other hand ,
shear deformations are induced _ homogeneously _
( on macroscopic scales ) throughout samples , giving rise to viscous flow with strong shear thinning behavior .
in a model amorphous metal in 3d , takeuchi _ et al . _
@xcite numerically followed atomic motions after application of a small shear strain to observe heterogeneities among poorly and closely packed regions .
such dynamic heterogeneities have been reported in recent simulations in sheared states @xcite and also in quiescent states @xcite . in the study of elasto - plastic dynamics of solids , microscopic simulations
are informative @xcite , while mean - field theories are instructive @xcite . in the latter theories ,
the problem was reduced to to that of one element obeying a stochastic process under the influence of the average stress .
the aim of this paper is then to present a space - time dependent , elasto - plastic dynamical model on the basis of well - defined nonlinear elastic theory .
although our theory is still preliminary , we shall see that introduction of an order parameter , representing the vacancy or the local free volume , can give rise to dramatic effects similar to those reported in ref.@xcite .
to introduce the fundamental concepts in our problem in the simplest manner , we first present a one - dimensional model which mimics a solid with shear deformations varying only in one direction ( 1d slip model ) .
we here write down the dynamic equations in the continuum representation ( for simplicity ) .
the velocity @xmath13 of the shear displacement @xmath14 ( along the @xmath15 axis ) is governed by v = _ x [ _ 0 _ 0 ( _ 0 ^ -1_xu ) ] + _ 0
_ x^2v + _ x _ r , [ eq:2 ] where @xmath16 , @xmath17 is the mass density , @xmath10 is the shear modulus , @xmath18 is a constant determined by the underlying crystal structure , and @xmath19 is a vicosity .
the period of the strain is given by @xmath20 .
to ensure equilibrium in the absence of applied force , we introduce the random stress @xmath21 related to @xmath19 via the fluctuation - disipation relation @xmath22 , where @xmath23 is the temperature .
hereafter space and time will be measured in units of @xmath24 and @xmath25 , respectively .
the dimensionless noise strength is given by @xmath26 .
we numerically solve ( 2 ) on a 1d chain ( @xmath27 with @xmath28 , setting @xmath29 and @xmath30 .
we apply a constant shear rate @xmath0 at @xmath31 with @xmath32 and @xmath33 . in fig.1 the average dimensionless stress @xmath34
is plotted as a function of the scaled strain @xmath35 .
the system undergoes a pure elastic deformation @xmath36 in the initial stage , but slips ( jumps of @xmath37 by multiples of @xmath38 ) appear randomly throughout the system with increasing the average strain @xmath39 .
the inset shows slips and continuously strained regions .
the average slope of the latter regions gives the average stress .
for the range of the shear rates in fig.1 , we can see shear - thinning @xmath40 in steady states ( if averaged over time ) .
it is also straighforward to integrate ( 2 ) under a fixed shear stress @xmath41 applied at one end with the other end being pinned . for small @xmath41 below a yield stress
@xmath42 , we observe no slip formation , while for @xmath43 plastic flow is produced . for the noise strength of @xmath28 , we find @xmath44 .
as a direct generalization of ( 2 ) , we set up a plastic flow model in two dimensions ( 2d slip odel ) . in the continuum limit
the strain components are defined by e_1=_x u_x+ _ yu_y , e_2= _ x u_x- _ yu_y , e_3= _ x u_y + _ yu_x , [ eq:3 ] where @xmath16 and @xmath45 . if we suppose a triangular lattice , the elasticity is isotropic in the harmonic approximation @xcite , being characterized by the bulk and shear moduli , @xmath46 and @xmath10 , but it depends on the orientational angle @xmath47 of one of the crystal axes for large shear deformations .
note that , under rotation of the reference frame by @xmath47 , the shear strains @xmath48 and @xmath49 are changed to @xmath50 and @xmath51 , where @xcite e_2= e_2 2 + e_3 2 , e_3= -e_2 2 + e_3 2 . [ eq:4 ]
we write the elastic energy density in the form , @xmath52 .
the simplest form of the scaling function @xmath53 is of the form ( e_3 , e_2)= [ eq:5 ] and is shown in fig.2 , the @xmath54 is invariant with respect to the rotation @xmath55 , is a periodic function of @xmath51 with period @xmath56 for @xmath57 ( simple shear deformation ) , and becomes @xmath58 for small strains .
if we assume @xmath59 for the angle rotation rate , its time integration yields = ( _ x u_y- _ y u_x ) + _ 0 , [ eq:6 ] where @xmath60 is independent of @xmath61 but may depend on @xmath62 . for @xmath63 one of the crystal axes
is along the @xmath64 axis .
then the elastic energy @xmath65 is a functional of @xmath66 .
we assume that the lattice velocity @xmath67 obeys = - f_el + _ 0 ^2v + _ r , [ eq:7 ] where the first term on the right hand side is also written as @xmath68 in term of the elastic stress tensor @xmath69 .
the symmetric random stress tensor @xmath70 satisfies @xmath71 , because the bulk viscosity is neglected in ( 7 ) , and @xcite = 2k_bt _ 0 ( r - r)(t - t ) .
[ eq:8 ] we measure space and time in units of @xmath72 and @xmath73 and the strains in units of @xmath74 . if these scaling units are used , the noise strength @xmath75 in ( 8) is replaced by = k_bt /_p^2 _ 0 ^ 2 .
[ eq:9 ] we integrate ( 7 ) on a @xmath76 square lattice by applying a constant shear rate @xmath0 at @xmath31 with @xmath77 .
the periodic boundary condition is imposed in the @xmath64 direction , while @xmath78 at the bottom @xmath79 and @xmath80 and @xmath81 at the top @xmath82 . at @xmath31 ,
the values of @xmath83 at the lattice sites are gaussian random numbers with variance @xmath84 but @xmath85 and @xmath86 . in fig.3
we show the average scaled stress @xmath87 as a function of the scaled strain @xmath35 for @xmath88 and @xmath89 . in the inset
we display snapshots of @xmath49 and @xmath48 . because we start with a perfect crystal with the initial fluctuations only in the velocity , the stress curve drops sharply after the peak with catastrophic formation of slips .
then structurally disordered states are produced where defects are proliferated ( _ strain - induced disordering _ ) .
note that slips are accumulations of dislocations @xmath90 $ ] . in the initial stage of plastic flow
we have simple slips consisting of two dislocations with opposite burgers vectors @xmath91 with @xmath92 being the lattice constant . the elastic energy to create such a slip
is minimum in the @xmath64 and @xmath15 directions under shear deformation .
this is the reason why the slips in fig.3 are parallel to the @xmath64 or @xmath15 direction . in fig.4
the shear @xmath93 is swiched off at ( a ) @xmath94 before the peak time of the stress , ( b ) @xmath95 just after the peak time , and ( c ) @xmath96 .
the top and bottom boundaries are kept at rest afterwards .
pure elastic deformation is maintained in ( a ) , while no appreciable time evolution is detected after transients in ( b ) and ( c ) .
this means that the structurely disordered states are metastable .
usually in the literature , the density deviation @xmath97 is equated with @xmath98 in solids @xcite , where @xmath99 is the average density and @xmath100 is assumeed . in the presence of vacancies or interstitials , however , there can be a small difference between these two quantities .
cohen , flemming , and gibbs @xcite constructed a hydrodynamic description of solids including the new variable @xmath101 . according to their theory , the _ vacancy concentration _ @xmath102 may be defined by c= c_0- m= c_0- ( /|+e_1 ) [ eq:10 ] where @xmath103 is the average @xmath104 dependent on @xmath23 and @xmath105 . on the other hand , granato claimed relevance of interstitials in amorphous solids because they can greatly decrease the shear modulus @xcite .
however , because vacancies and interstitials are point defects , it may be more appropriate to treat @xmath102 as the local free - volume fraction @xcite which can take continuous values at each lattice site . then @xmath106 is the average free - volume fraction and we may assume @xmath107 .
the role of @xmath108 or @xmath102 is expected to be crucial in amorphous solids .
we try to include the vacancy variable in our nonlinear slip model ( 2d vacancy model ) , the dynamic equations obeyed by the mass density and the momentum density are of the usual forms , = -| , [ eq:11 ] = + _ 0 ^2v + _ r .
[ eq:12 ] in ( 12 ) the force density , the first term on the right hand side , is written in terms of the free energy functional @xmath109 as = -|(f/)_u - ( f/)_= - ( f/)_m , [ eq:13 ] where use has been made of the identities @xmath110 and @xmath111 .
the random stress tensor @xmath112 satisfies ( 8) . it is important that , in the presence of the vacancy field , the lattice velocity @xmath113 is different from the mass velocity @xmath114 and is assumed to be of the form , = v - _ 0 ( f ) _ + _ r , [ eq:14 ] where @xmath115 is the kinetic coefficient and the components @xmath116 of the random force vector @xmath117 are characterized by = 2k_bt _ 0 _ ( r - r)(t - t ) . [ eq:15 ] from ( 12 ) and ( 14 ) the equation for @xmath108 is expressed as m= _ 0 + _
r. [ eq:16 ] it is worth noting that ( 16 ) is similar to the dynamic equation for the concentration in visccoelastic fluid mixtures @xcite . in passing , owing to the special form of ( 13 )
, the time derivative of the total free energy @xmath118 becomes nonnegative - definite in the absence of applied stress if the random noises are neglected .
this is a self - consistent condition of langevin equations ensuring attainment of equilibrium @xcite .
the free energy @xmath119 is a functional of @xmath66 and @xmath97 ( or @xmath108 ) .
we assume the free energy density @xmath120 in the form , f= m^2 + m^4 + m e_1 + + ||^2 + e_1 ^ 2 + ( m ) f(e_3 , e_2 ) , [ eq:17 ] where @xmath121 , @xmath122 , and @xmath123 are positive constants and @xmath124 is a coupling constant . the gradient term ( @xmath125 )
is introduced to suppress the density fluctuations with short length scales .
( we are considering the density fluctuations longer than the peak distance of the pair correlation function . ) for simplicity , we neglect the other gradient terms involving the gradients of @xmath108 and the strains . in our theory
it is most important that the shear modulus @xmath126 sensitively depends on @xmath108 . near glass transitions , with decreasing @xmath108 or increasing @xmath102 , @xmath126
is expected to decrease abruptly from a finite value @xmath10 to zero ( @xmath127 fluid value ) around a threshold value of @xmath108 or @xmath102 .
analytic calculations of the above model are difficult except for idealized situations . as a simple exercise ,
let us consider small deviations with wave vector @xmath128 around a homogeneously strained state with @xmath129 , where the sytem is at rest and the harmonic approximation can be made for the elastic energy .
then the mechanical equilibrium condition allows elimination of @xmath66 from ( 16 ) @xcite .
we consider the steady - state variance of m in the long wavelength limit , @xmath130 , in 2d . in the linear approximation
it becomes dependent on the angle of @xmath128 : = a- -_12- ( _ 1)^2[1 + ( 2)^2k_0/_0 ] , [ eq:18 ] where @xmath131 , @xmath132 and @xmath133 . the vacancy diffusion constant is written as @xmath134 .
the steady - state density variance @xmath135 in 2d is written as _
= + ^2 _ v. [ eq:19 ] the first term is the usual term , while the second term arises from the vacancy fluctuations .
the dependence of these quantities on @xmath136 and the angle @xmath137 can become appreciable with increasing @xmath138 and@xmath139or decreasing @xmath121 . for @xmath140 , for example , the homogeneous state is unstable for @xmath141 against the flluctuations with @xmath142 along the @xmath64 or @xmath15 axis .
note that @xmath143 can hold if @xmath144 .
it is easy to confirm that essentially the same result follows in 3d in both shear and elongational deformations .
thus , in elastically deformed amorphous solids ( before onset of plastic flow ) , we predict enhancement of the density fluctuations in particular directions of the wave vector .
we then show our numerical results of our model in ( 11)-(17 ) obtained on a @xmath145 square lattice for the case of applying a constant @xmath0 . again
the units of space , time , and strains are @xmath72 , @xmath73 , and @xmath74 , respectively .
the @xmath108 is also divided by @xmath38 .
the dimensionless mass density is given by @xmath146 . though it is not clear what functional form of @xmath126 is appropriate , we tentatively use ( m ) = _ 0 [ eq:20 ] for @xmath147 and @xmath148 for @xmath149 . here
@xmath150 is a constant and is set equal to 0.1 in our simulation , while @xmath151 is a control parameter representing the closeness to the glass transition .
in fact , if @xmath126 goes to zero at @xmath152 , we have @xmath153 from ( 10 ) .
however , essentially the same numerical results follow as long as @xmath126 strongly depends on @xmath108 ( for example , from the simpler form @xmath154 instead of ( 20 ) ) .
the other parameter values used are @xmath155 , @xmath156 , @xmath140 , @xmath157 and @xmath77
. while the initial conditions for @xmath83 , @xmath66 and @xmath47 are the same as in the previous case of figs.3 and 4 , we assign random gaussian randum numbers with variance @xmath158 to @xmath108 .
in fig.5 we show time evolution of the scaled shear stress as a function of the scaled average strain @xmath35 at @xmath88 for @xmath159 , and @xmath160 .
fig.6a and fig.6b display snapshots of @xmath49 , @xmath48 , @xmath161 , and @xmath162 at @xmath163 and @xmath164 on the curve of @xmath165 in fig.5 .
fig.7 demonstrates that these strain - induced disordered states are metastable as in fig.4 .
salient features are as follows : ( i ) plastic flow is induced earlier than in the absence of vacancies , because slip formation is easier in regions with smaller @xmath108 .
onset of plastic flow is sensitive to the initial randomness of @xmath108 .
( ii ) the effective viscosity @xmath166 becomes smaller with decreasing the parameter @xmath151 in ( 17 ) or increasing the elastic inhomgeneity .
( iii ) the @xmath167 also decreases as a function of time on long time scales because of slow accumulation of vacancies around slips . in our model , though this tendency is considerably suppressed by the quartic term ( @xmath168 in ( 16 ) , vacancy accumulation can lead to fluidization at long times after percolation of the slips .
( iii ) the density fluctuations are induced and are frozen after cessation of shear . as demonstared in fig.7 ,
the fluctuation variances of the vacancy and the density are much enhanced than in the initial state .
owing to the gradient term in ( 16 ) , the density smoothly varies in space ( compared with @xmath108 ) .
we have presented a first time - dependent ginzburg - landau theory accounting for inhomogeneous nonlinear elastic deformations .
it is a coarse - grained theory and the sharp peak of the structure factor , which is essential in the mode coupling theory @xcite , does not come into play . instead ,
periodicity of the elastic energy with respect to the strains @xmath51 and @xmath50 and vacancy - dependence of the shear modulus are two major ingredients of our theory , which bring about metastable structural disordered states with heterogeneities in the vacancy and mass densities . as a closely related effect ,
fischer @xcite observed excess scattering from nearly static density fluctuations with sizes in a range of 20 - 200 nm in glass - forming fluids .
we believe that such large - scale frozen fluctuations can arise only as a result of nonlinear elastic deformations induced by structural disorder .
we also mention previous simulations .
et al_. @xcite applied a tensile ( elongational ) strain to a 2d model to produce slips and shear bands making angles of @xmath169 with respect to the stretched direction in agreement with experiments .
the same results also follow from our model and will be reported shortly .
we will show that the elastic energy to create a slip under tension is minimum for these directions .
@xcite applied a tensile strain to a 3d model to induce a change from a perfect crystal to an amorphous solid .
however , we admit that our results are still preliminary and it remains unclear how our theory corresponds to real physical effects . in particular , the initial conditions ( @xmath85 and @xmath170const . ) should be inappropriate for amorphous solids ( where structural disorder should preexist even before applying strain ) . in future work
, we should start with initial states with various amounts of disorder and should also examine what parameter values are appropriate for amorphous solids .
we mention analogous and instructive examples . in solids undergoing phase separation or structural transitions , the dependence of the shear modulus on the order parameter , written as @xmath108 , in the form @xmath171 is of crucial importance in phase ordering @xcite . in all these cases , local minimization of the nonlinear elastic energy ( @xmath172 )
gives rise to heterogeneous metastable states where time - evolution is pinned .
in viscoelastic polymer systems , the composition - dependence of the viscoelastic properties can give rise to shear - induced phase separation @xcite , where the scattered light intensity takes a form similar to @xmath173 in ( 18 ) .
99 + fig.1 , stress vs strain in units of @xmath178 and @xmath38 obtained from the discretized version of the 1d model ( 1 ) after application of constant shear rate . here
@xmath179 and @xmath180 from above .
the inset displays a snapshot of the deviations @xmath181 ( @xmath182 at @xmath183 for @xmath88 .
+ fig.2 , function @xmath184 defined by ( 5 ) , which represents the 2d shear deformation energy density divided by @xmath10 for @xmath63 .
+ fig.3 , stress vs strain for @xmath88 and @xmath89 obtained from the 2d slip model ( 7 ) . in the inset @xmath49 and @xmath48 at @xmath185 for @xmath88 are shown .
+ fig.4 , scaled stress and elastic energy vs time , which demonstates freezing of strain - induced diordered states of ( 7 ) at long times .
the shear is switched off at ( a ) @xmath94 , ( b ) @xmath95 , and ( c ) @xmath96 , as indicated by arrows , using the same parameters as for the curve of @xmath88 in fig.3 .
+ fig.5 , stress vs strain for @xmath88 obtained from integration of ( 11 ) , ( 12 ) , and ( 14 ) using ( 16 ) and ( 17 ) at @xmath88 .
here @xmath186 , and @xmath160 from above .
+ fig.6 , snapshots of @xmath49 , @xmath48 , @xmath161 , and @xmath162 at ( a ) @xmath163 and ( b ) @xmath164 for @xmath88 on the curve of @xmath165 in fig.5 .
+ fig.7 , freezing of strain - induced diordered states of the vacancy model . shown
are the density variance @xmath187^{1/2}/\rho \gamma_p$ ] , the vacancy variance @xmath188^{1/2}/\gamma_p$ ] , and the average elastic energy @xmath189 . as in fig.4
, the shear rate is switched off at @xmath95 .
the frozen density profile at @xmath190 is shown in the inset . | we present a phenomenological time - dependent ginzburg - landau theory of nonlinear plastic deformations in solids . because the problem is very complex
, we first give models in one and two dimensions without vacancies and interstitials , where large strains produce densely distributed slips but the mass density deviations remain small except near the tips of slips .
next we set up a two - dimensional model including a vacancy field ( or local free - volume fraction ) , where relevant is the sensitive dependence of the elastic shear modulus on the vacancy density . in our simulation , if strains are applied to nearly defectless solids but in the presence of such elastic inhomogeneity , the vacancy density and the mass density can become considerably heterogeneous for large strains on spatial scales much longer than the atomic size .
these strain - induced disordered states are metastable or long - lived once they are created . |
discovered by @xcite , u geminorum is the prototype of a subclass of dwarf novae , a descriptive term suggested by @xcite due to the small scale similarity of the outbursts in these objects to those of novae . after the work by @xcite , who found u gem to be a single - lined spectroscopic binary with an orbital period around 4.25 hr , and from the studies by @xcite , who establish the eclipsing nature of this binary , @xcite and @xcite , established the classical model for cataclysmic variable stars .
the model includes a white dwarf primary surrounded by a disc accreted from a roche - lobe filling late - type secondary star .
the stream of material , coming through the l@xmath12 point intersects the edge of the disc producing a bright spot , which can contribute a large fraction of the visual flux .
the bright spot is observed as a strong hump in the light curves of u gem and precedes a partial eclipse of the accretion disk and bright spot themselves ( the white dwarf is not eclipsed in this object ) . a mean recurrence time for u gem outbursts of @xmath13 118 days , with @xmath14m@xmath15=5 and outburst width of 12 d ,
was first found by @xcite .
however , recent analysis shows that the object has a complex outburst behavior @xcite .
@xcite , using the aavso data on the 1985 outburst , has discovered the presence of super - humps , a fact that challenges the current theories of super - outbursts and super - humps for long period system with mass ratios above 1/3 .
the latter author also points out the fact that calculations of the radius of the disc
obtained from the separation of the emission peaks @xcite in quiescence are in disagreement with the calculations of the disc radii obtained from the photometric eclipse data @xcite .
several radial velocity studies have been conducted since the first results published by @xcite . in the visible spectral range , where the secondary star has not been detected , their results are mainly based on spectroscopic radial velocity analysis of the emission lines arising from the accretion disc @xcite .
in other wavelengths , works are based on absorption lines : in the near - infrared , on the na i doublet from the secondary star @xcite and in the ultraviolet , on lines coming from the white dwarf itself @xcite .
although the research work on u gem has been of paramount importance in our understanding of cataclysmic variables , the fact that it is a partially - eclipsed and in the visual range a single - lined spectroscopic binary , make the determination of its physical parameters difficult to achieve through precise measurements of the semi - amplitudes @xmath16 and of the inclination angle @xmath17 of the orbit .
spectroscopic results of @xmath16 differ in the ultraviolet , visual and infrared ranges .
therefore , auxiliary assumptions have been used to derive its more fundamental parameters @xcite . in this paper
we present a value of @xmath18 , obtained from our high - dispersion echelle spectra , which is in agreement with the ultraviolet results , and of @xmath19 from a new method applicable to optical spectroscopy . by chance ,
the system was observed at a peculiar low state , when the classical hot spot was absent .
u geminorum was observed in 1999 , january 15 with the echelle spectrograph at the f/7.5 cassegrain focus of the 2.1 m telescope of the observatorio astrnomico nacional at san pedro mrtir , b.c .
a thomson 2048@xmath202048 ccd was used to cover the spectral range between @xmath215200 and @xmath219100 , with spectral resolution of r=18,000 .
an echellette grating of 150 l / mm , with blaze around 7000 , was used .
the observations were obtained at quiescence ( @xmath22 ) , about 20 d after a broad outburst ( data provided by the aavso : www.aavso.org ) . the spectra show a strong h@xmath0 emission line .
no absorption features were detected from the secondary star .
a first complete orbital cycle was covered through twenty - one spectra , each with 10 min exposure time .
thirteen further spectra were subsequently acquired with an exposure of 5 min each .
the latter cover an additional half orbital period .
the heliocentric mid - time of each observation is shown in column one in table [ tab : radvel ] .
the flux standard hr17520 and the late spectral m star hr3950 were also observed on the same night .
data reduction was carried out with the iraf package .
the spectra were wavelength calibrated using a th - ar lamp and the standard star was also used to properly subtract the telluric absorption lines using the iraf routine _
in this section we derive radial velocities from the prominent h@xmath0 emission line observed in u gem , first by measuring the peaks , secondly by using a method based on a cross - correlating technique , and thirdly by using the standard double - gaussian technique designed to measure only the wings of the line . in the case of the secondary star
, we were unable to detect any single absorption line in the individual spectra ; therefore it was not possible to use any standard method .
however , here we propose and use a new method , based on a co - adding technique , to derive the semi - amplitude of the orbital radial velocity of the companion star . in this section
, we compare our results with published values for both components in the binary .
we first discuss the basic mathematical method used here to derive the orbital parameters and its limitation in the context of cataclysmic variables ; then we present our results for the orbital parameters calculated from the different methods and finally discuss an improved ephemeris for u gem . to find the orbital parameters of the components in a cataclysmic variable in which no eccentricity is expected @xcite
we use an equation of the form @xmath23,\ ] ] where @xmath24 are the observed radial velocities as measured from the emission lines in the accretion disc or from the absorption lines of the red star ; @xmath25 is the systemic velocity ; @xmath26 are the corresponding semi - amplitudes derived from the radial velocity curve ; @xmath27 is the heliocentric julian time of the inferior conjunction of the companion ; and @xmath28 is the orbital period of the binary . a minimum least - squares sinusoidal fit is run , which uses initial values for the four ( @xmath28 , @xmath25 , @xmath29 , and @xmath27 ) orbital parameters .
the program allows for one or more of these variables to be fixed , i.e. they can be set to constant values in the initial parameters file .
if the orbital period is not previously known , a frequency search using a variety of methods for evenly- or unevenly - sampled time series data @xcite may be applied to the measured radial velocities in order to obtain an initial value for @xmath28 , which is then used in the minimum least - squares sinusoidal fit .
if the time coverage of the observations is not sufficient or is uneven , period aliases may appear and their values have to be considered in the least - squares fits .
a tentative orbital period is selected by comparing the quality of each result . in these cases ,
additional radial velocity observations should be sought , until the true orbital period is found unequivocally .
time series photometric observations are usually helpful to find orbital modulations and are definitely important in establishing the orbital period of eclipsing binaries . in the case of u gem ,
the presence of eclipses and the ample photometric coverage since the early work of @xcite , has permitted to establish its orbital period with a high degree of accuracy @xcite . although in eclipsing binaries a zero phase is also usually determined , in the case of u gem the variable positions of the hot spot and stream , causes the zero point to oscillate , as mentioned by the latter authors .
accurate spectroscopic observations are necessary to correctly establish the time when the secondary star is closest to earth , i.e in inferior conjunction .
further discussion on this subject is given in section [ ephem ] . to obtain the real semi - amplitudes of the binary , i.e k@xmath30=k@xmath31 ,
some reasonable auxiliary assumptions are made .
first , that the measurements of the emission lines , produced in the accretion disc , are free from distortions and accurately follow the orbital motion of the unseen white dwarf .
second , that the profiles of the measured absorption lines are symmetric , which implies that the brightness at the surface of the secondary star is the same for all its longitudes and latitudes .
certainly , a hot spot in the disc or irradiation in the secondary from the energy sources related to the primary will invalidate either the first , the second , or both assumptions .
corrections may be introduced if these effects are present . in the case of u gem ,
a three - body correction was introduced by @xcite in order to account for the radial velocity distortion produced by the hot spot , and a correction to heating effects on the face of the secondary star facing the primary was applied by @xcite before equating @xmath32 . as initial values in our least - squared sinusoidal
fits , we use @xmath33 and @xmath34 from @xcite , a systemic velocity of 42 km s@xmath3 from @xcite , and k@xmath35 km s@xmath3 and k@xmath36 km s@xmath3 from @xcite and @xcite , respectively . in our calculations ,
the orbital period was set fixed at the above mentioned value , since our observations have a very limited time coverage .
this allow us to increase the precision for the other three parameters
. .measured h@xmath0 radial velocities .
[ cols= " < , < , > , > , > " , ] in this section we compare three methods for determining the radial velocity of the primary star , based on measurements of the h@xmath0 emission line .
although , as we will see in the next subsections , the last method results in far better accuracy and agrees with the ultraviolet results , we have included all of them here because the first method essentially provides an accurate way to determine the separations of the blue and red peaks , which is an indicator of the outer radius of the disc @xcite , and the second yields a @xmath37 value much closer to that obtained from uv results than any other published method .
this cross - correlation method might be worthwhile to consider for its use in other objects .
furthermore , as we will see in the discussion , all three methods yield a consistent value of the systemic velocity , which is essential to the understanding of other parameters in the binary system . to match the signal to noise ratio of the first twenty - one spectra , we have co - added , in pairs , the thirteen 5-minute exposures .
the last three spectra were added to form two different spectra , in order to avoid losing the last single spectrum . a handicap to this approach is that , due to the large read - out time of the thomson ccd , we are effectively smearing the phase coverage of the co - added spectra to nearly 900 s. however , the mean heliocentric time was accordingly corrected for each sum .
this adds to a total sample of twenty - eight 600 s spectra .
we have measured the position of the peaks using a double - gaussian fit , with their separation , width and position as free parameters .
the results yield a mean half - peak separation @xmath38 of about 460 km s@xmath3 .
the average value of the velocities of the red and blue peaks , for each spectrum , is shown in column 3 of table [ tab : radvel ] .
we then applied our nonlinear least - squares fit to these radial velocities .
the obtained orbital parameters are shown in column 2 of table [ orbparam ] .
the numbers in parentheses after the zero point results are the evaluated errors of the last digit . we will use this notation for large numbers throughout the paper .
the radial velocities are also shown in figure [ fig : dob - peak ] , folded with the orbital period and the time of inferior conjunction adopted in the section [ ephem ] .
the solid lines in this figure correspond to sinusoidal fits using the derived parameters in our program .
although we have not independently tabulated the measured velocities of the blue and red peaks , they are shown in figure [ fig : dob - peak ] together with their average .
the semi - amplitudes of the plotted curves are 154 km s@xmath3 and 167 km s@xmath3 for the blue and red peaks , respectively . .
the curve at the middle is the mean from blue ( bottom curve ) and red ( top curve ) . ]
template near phase 0.02 .
the half - separation of the peaks has a value of 470 km s@xmath3 . ]
we have also cross - correlated the h@xmath0 line in our spectra with a template constructed as follows : first , we selected a spectrum from the first observed orbital cycle close to phase 0.02 when , in the case of our observations , we should expect a minimum distortion in the double - peaked line due to asymmetric components ( see section [ tom ] ) .
the blue peak in this spectrum is slightly stronger than the red one .
this is probably caused by the hot spot near the @xmath11 point ( see section [ ugemb ] ) , which might be visible at this phase due to the fact that the binary has an inclination angle smaller than 70 degrees .
the half - separation of the peaks is 470 km s@xmath3 , a value similar to that measured in a spectrum taken during the same orbital phase in next cycle .
the chosen spectrum was then highly smoothed to minimize high - frequency correlations .
the resulting template is shown in figure [ fig : halfxc ] .
a radial velocity for the template was derived from the wavelength measured at the dip between the two peaks and corrected to give an heliocentric velocity .
the iraf _ fxc _ task was then used to derive the radial velocities , which are shown in column 4 of table [ tab : radvel ] . as in the previous section ,
we have fitted the radial velocities with our nonlinear least - squares fit algorithm .
the resulting orbital parameters are given in column 3 of table [ orbparam ] . in figure
[ fig : rvhalfxc ] the obtained velocities and the corresponding sinusoidal fit ( solid line ) are plotted . ) and open circles to a = 280 pixels ; ] the h@xmath0 emission line was additionally measured using the standard double gaussian technique and its diagnostic diagrams , as described in @xcite .
we refer to this paper for the details on the interpretation of our results .
we have used the _ convolve _
routine from the iraf _ rvsao _ package , kindly made available to us by thorstensen ( private communication ) .
the double peaked h@xmath0 emission line with a separation of about @xmath39 shows broad wings reaching up to @xmath40 from the line center . unlike the case of low resolution spectra where for over - sampled data the fitting is made with individual gaussians having a fwhm of about one resolution element in our spectra , with resolution @xmath13 0.34 @xmath41
, such gaussians would be inadequately narrow , as they will cover only a very small region in the wings . to measure the wings appropriately and , at the same time , avoid possible low velocity asymmetric features
, we must select a @xmath42 value which fits the line regions corresponding to disc velocities from about 700 to 1000 @xmath43 . as a first step
, we evaluated the width of the gaussians by setting this as a free parameter from 10 to 40 pixels and for a wide range of gaussian separations ( between 180 and 280 pixels ) . for each run
, we applied a nonlinear least - squares fit of the computed radial velocities to sinusoids of the form described in section [ orbparcal ] .
the results are shown in figure [ fig : diag1 ] , in particular for three different gaussian separations : @xmath44 , 230 and 280 pixels .
these correspond to the low and upper limits as well as to the value for a preferred solution , all of which are self - consistent with the second step ( see below ) . in the bottom panel of the figure
we have plotted the overall rms value for each least - squares fit , as this parameter is very sensitive to the selected gaussian separations . as expected at this high spectral resolution , the parameters in the diagram change rapidly for low values of @xmath42 , and there are even cases when no solution was found . at low values of @xmath45 ( e.g. crosses )
there are no solutions for widths narrower than 20 pixels .
the rms values increase rapidly with width , while the @xmath46 , @xmath25 and phase shift values differ strongly from the other cases . for higher values of @xmath45 ( open circles )
we obtain lower values for @xmath46 , but the rms results are still large , in particular for intermediate values of the width of the gaussians . for the middle solution ( dots ) the results are comparable with those for large @xmath45 values , but the rms is much lower .
similar results were found for other intermediate values of @xmath45 , and they all converge to a minimum rms for a width of 26 pixels at @xmath47 pixels . for the second step
we have fixed the width to a value of 26 @xmath41 and ran the double - gaussian program for a range of @xmath45 separations , from about 60 to 120 @xmath41 .
the results obtained are shown in figure [ fig : diag2 ] . , corresponding to a @xmath48 . ]
if only an asymmetric low velocity component is present , the semi - amplitude should decrease asymptotically as @xmath45 increases , until @xmath49 reaches the correct value .
here we observe such behavior , although for larger values of @xmath45 , there is a @xmath49 increase for values of @xmath45 up to @xmath50 , before it decreases strongly with high values of @xmath45 .
this behavior might be due to the fact that we are observing a narrow hot - spot near the @xmath11 point ( see section [ tom ] ) .
on the other hand , as expected , the @xmath46 vs @xmath45 curve has a change in slope , at a value of @xmath45 for which the individual gaussians have reached the velocity width of the line at the continuum . for larger values of @xmath45 the velocity measurements
become dominated by noise . for low values of @xmath45 ,
the phase shift usually gives spurious results , although in our case it approaches a stable value around 0.015 .
we believe this value reflects the difference between the eclipse ephemeris , which is based mainly on the eclipse of the hot spot , and the true inferior conjunction of the secondary star .
this problem is further discussed in section [ tom ] . finally , we must point out that the systemic velocity smoothly increases up to a maximum of about 40 km s@xmath3 at gaussian separation of nearly 42 , while the best results , as seen from the figure , are obtained for @xmath45 = 31 .
this discrepancy may be also be related to the narrow hot - spot near the @xmath11 point and might be due to the phase - shift between the hot - spot eclipse and the true inferior conjunction .
this problem will also be address in section [ ephem ] .
the radial velocities , corresponding to the adopted solution , are shown in column 5 of table [ tab : radvel ] and plotted in figure [ fig : velrad ] , while the corresponding orbital parameters obtained from the nonlinear least - squares fit are given in column 4 of table [ orbparam ] . ) .
the solid line , close to the points , correspond to the solution with @xmath51 ( see text ) , while the large amplitude line correspond to the solution found for @xmath1 ( see section [ secon ] ) . ] we were unable to detect single features from the secondary star in any individual spectra , after careful correction for telluric lines .
in particular we found no radial velocity results using a standard cross - correlation technique near the nai @xmath52 doublet .
as we will see below , this doublet was very weak compared with previous observations @xcite .
we have been able , however , to detect the nai doublet and the tio head band around @xmath217050 with a new technique , which enables us to derive the semi - amplitude @xmath53 of the secondary star velocity curve .
we first present here the general method for deriving the semi - amplitude and then apply it to u gem , using not only the absorption features but the @xmath54 emission as well . in many cataclysmic variables
the secondary star is poorly visible , or even absent , in the optical spectral range .
consequently , no @xmath56 measurements are feasible for this component . among these systems
are dwarf novae with orbital periods under 0.25 days , for which it is thought that the disc luminosity dominates over the luminosity of the roche - lobe filling secondary , whose brightness depends on the orbital period of the binary @xcite . for such binaries ,
the orbital parameters have been derived only for the white - dwarf- accretion disc system , in a way similar to that described in section [ orbparcal ] . in order to determine a value of @xmath53 from a set of spectra of a cataclysmic variable , for which the orbital period and time of inferior conjunction have been already determined from the emission lines , we propose to reverse the process :
derive @xmath57 using @xmath58 as the initial value for the semi - amplitude , and set the values of @xmath59 and @xmath27 , derived from the emission lines , as constants .
the initial value for the systemic velocity is set to zero , and its final value may be calculated later ( see below ) .
the individual spectra are then _ co - added _ in the frame of reference of the secondary star , i.e. by doppler - shifting the spectra using the calculated @xmath60 from the equation given in section [ orbparcal ] , and then add them together .
hereinafter we will refer to this procedure as the _ co - phasing process_. ideally , as the proposed @xmath58 is changed through a range of possible values , there will be a one for which the _ co - phased _
spectral features associated with the absorption spectrum will have an optimal signal - to - noise ratio .
in fact , this will also be the case for any emission line features associated with the red star , if present . in a way
, this process works in a similar fashion as the double gaussian fitting used in the previous section , provided that adequate criteria are set in order to select the best value for @xmath53 .
we propose three criteria or tests that , for late type stars , may be used with this method : the first one consists in analyzing the behavior of the measured depths or widths of a well identified absorption line in the _ co - phased _
spectra , as a function of the proposed @xmath58 ; one would expect that the width of the line will show a minimum and its depth a maximum value at the optimal solution .
this method could be particularly useful for k - type stars which have strong single metallic lines like ca i and fe i. the second criterion is based upon measurements of the slope of head - bands , like that of tio at @xmath217050 .
it should be relevant to short period systems , with low mass m - type secondaries with spectra featuring strong molecular bands . in this case
one could expect that the slope of the head - band will be a function of @xmath58 , and will have a maximum negative value at the best solution .
a third test is to measure the strength of a narrow emission arising from the secondary .
this emission , if present , would be particularly visible in the _ co - phased _ spectrum and will have minimum width and maximum height at the best selected semi - amplitude @xmath58 .
we have tested these three methods by means of an artificial spectrum with simulated narrow absorption lines , a tio - like head band and a narrow emission line .
the spectrum with these artificial features was then doppler shifted using pre - established inferior conjunction phase and orbital period , to produce a series of test spectra .
an amount of random gaussian noise was added to each doppler shifted spectrum , sufficient to mask the artificial features .
we then proceeded to apply the _ co - phasing process _ to recover our pre - determined orbital values .
all three criteria reproduced back the original set of values , as long as the random noise amplitude was of the same order of magnitude as the strength of the _ clean _ artificial features .
we have applied the above - mentioned criteria to u gem .
the time of the inferior conjunction of the secondary and the orbital period were taken from section [ ephem ] . to attain the best signal to noise ratio we have used all the 28 observed spectra .
although they span over slightly more than 1.5 orbital periods , any departure from a real @xmath1 value will not depend on selecting data in exact multiples of the orbital period , as any possible deviation from the real semi - amplitude will already be present in one complete orbital period and will depend mainly on the intrinsic intensity distribution of the selected feature around the secondary itself ( also see below the results for @xmath25 ) .
figure [ fig : sodcor ] shows the application of the first test to the nai doublet @xmath61 8183,8195 .
the spectra were _ co - phased _ varying @xmath58 between 250 to 450 km s@xmath3 .
the line depth of the blue and red components of the doublet ( stars and open circles , respectively ) , as well as their mean value ( dots ) are shown in the diagram .
we find a best solution for @xmath1 = 310 @xmath62 5 km s@xmath3 .
the error has been estimated from the intrinsic modulation of the solution curve .
as it approaches its maximum value , the line depth value oscillates slightly , but in the same way for both lines .
a similar behavior was present when low signal to noise features were used on the artificial spectra process described above .
figure [ fig : sodspec ] shows the _ co - phased _ spectrum of the nai doublet of our best solution for @xmath1 .
these lines appear very weak as compared with those reported by @xcite and @xcite .
we have also measured the gamma velocity from the co - phased spectrum by fitting a double - gaussian to the na i doublet ( dotted line in figure [ fig : sodcor ] ) and find a mean value @xmath63 10 km s@xmath3 ( corrected to the heliocentric standard of motion ) .
we did a similar calculation for @xmath25 by _ co - phasing _ the selected spectra used in section [ tom ] , covering a full cycle only .
the results were very similar to those obtained by using all spectra .
( top ) , @xmath64 ( bottom ) and mean ( middle ) as a function of @xmath58 . ] the second test , to measure the slope of the tio band has at @xmath217050 was not successful .
the solution curve oscillates strongly near values between 250 and 350 @xmath65 .
we believe that the signal to noise ratio in our spectra is too poor for this test and that more observations , accumulated during several orbital cycles , have to be obtained in order to attain a reliable result using this method .
however , we have co - phased our spectra for @xmath1 = 310 km s@xmath3 , with the results shown in figure [ fig : tioband ] .
the tio band is clearly seen while the noise is prominent , particulary along the slope of the head - band .
we have used this co - added spectrum to compare it with several late - type m stars extracted from the published data by @xcite fitted to our _ co - phased _ spectrum .
a gray continuum has been added to the comparison spectra in order to compensate for the fill - in effect arising from the other light sources in the system , so as to obtain the best fit . in particular , we show in the same figure the fits when two close candidates gj406 ( m6 v , upper panel ) and gj402 ( m4 - 5 v , lower panel ) are used .
the best fit is obtained for the m6 v star , to which we have added a 95 percent continuum . for the m4 - 5 v star
the fit is poor , as we observe a flux excess around 7000 and a stronger tio head - band .
increasing the grey flux contribution will fit the tio head band , but will result in a larger excess at the 7000 region . on the other hand ,
the fit with the m6 v star is much better all along the spectral interval .
there are a number of publications which assign to u gem spectral types m4 @xcite , m5 @xcite and possibly as far as m5.5 @xcite . even in the case
that the spectral type of the secondary star were variable , its spectral classification is still incompatible with its mass determination @xcite .
for the third test , we have selected the region around h@xmath0 , as in the individual spectra we see evidence of a narrow spot , which is very well defined in our spectrum near orbital phase 0.5 . in this test we have _ co - phased _ the spectra as before , and have adopted as the test parameter the peak intensity around the emission line .
the results are shown in figure [ fig : hotflux ] . a clear and smooth maximum is obtained for @xmath58 = 310 @xmath62 3 km s@xmath3 . the co - phased spectrum obtained from this solution
is shown in figure [ fig : maxiflux ] .
the double - peak structure has been completely smeared as expected when co - adding in the reference frame of the secondary star , as opposed to that of the primary star- and instead we observe a narrow and strong peak at the center of the line .
we have also fitted the peak to find the radial velocity of the spot .
we find @xmath66 10 km s@xmath3 , compatible with the gamma velocity derived from the radial velocity analysis of the emission line , @xmath67 2 km s@xmath3 ( see section [ hdgram ] ) .
this is a key result for the determination of the true systemic velocity and can be compared with the values derived from the secondary star ( see section [ discus ] ) .
spectra as a function of @xmath58 ] spectrum for @xmath68 . ]
as mentioned in section [ orbparcal ] , the presence of eclipses in u gem and an ample photometric coverage during 30 years has permitted to establish , with a high degree of accuracy , the value of orbital period . this has been discussed in detail by @xcite . however , as pointed by these authors , this object shows erratic variations in the timing of the photometric mid - eclipse that may be caused either by orbital period changes , variations in the position of the hot spot , or they may even be the consequence of the different methods of measuring of the eclipse phases .
a variation in position and intensity of the gas stream will also contribute to such changes .
a date for the zero phase determined independently from spectroscopic measurements would evidently be desirable .
@xcite discuss two spectroscopic measurements by @xcite and @xcite , and conclude that the spectroscopic inferior conjunction of the secondary star occurs about 0.016 in phase prior to the mean photometric zero phase .
there are two published spectroscopic studies @xcite , as well as one in this paper , that could be used to confirm this result .
unfortunately there is no radial velocity analysis in the former paper , nor in the excellent doppler imaging paper by @xcite based on their original observations .
however , the results by @xcite are of particular interest since he finds the spectroscopic conjunction in agreement with the time of the eclipse when using the photometric ephemerides by @xcite , taken from @xcite .
the latter authors introduce a small quadratic term which is consistent with the o - c oscillations shown in @xcite .
it is difficult to compare results derived from emission lines to those obtained from absorption lines , especially if they are based on different ephemerides .
furthermore , the contamination on the timing of the spectroscopic conjunction either caused by a hot spot , by gas stream or by irradiation on the secondary has not been properly evaluated . however , since our observations were made at a time when the hot spot in absent ( or , at least , is along the line between the two components in the binary ) and the disc was very symmetric ( see section [ tom ] ) , we can safely assume that in our case , the photometric and spectroscopic phases must coincide .
if we then take the orbital period derived by @xcite and use the zero point value derived from our measurements of the h@xmath0 wings , ( section [ hdgram ] ) , we can improve the ephemeris : @xmath69 for the inferior conjunction of the secondary star .
these ephemeris are used throughout this paper for all our phase folded diagrams and doppler tomography .
doppler tomography is a useful and powerful tool to study the material orbiting the white dwarf , including the gas stream coming from the secondary star as well as emission regions arising from the companion itself .
it uses the emission line profiles observed as a function of the orbital phase to reconstruct a two - dimensional velocity map of the emitting material .
a detailed formulation of this technique can be found in @xcite .
a careful interpretation of these velocity maps has to be made , as the main assumption invoked by tomography is that all the observed material is in the orbital plane and is visible at all times . the doppler tomography , derived here from the h@xmath0 emission line in u gem ,
was constructed using the code developed by @xcite .
our observations of the object cover 1.5 orbital cycles . consequently to avoid disparities on the intensity of the trailed and reconstructed spectra , as well as on the tomographic map we have carefully selected spectra covering a full cycle only . for this purpose
we discarded the first 3 spectra ( which have the largest airmass ) and used only 18 spectra out of the first 21 , 600 s exposures , starting with the spectrum at orbital phase 0.88 and ending with the one at phase 0.86 ( see table [ tab : radvel ] ) .
in addition , in generating the tomography map we have excluded the spectra taken during the partial eclipse of the accretion disc ( phases between 0.95 and 0.05 ) .
the original and reconstructed trailed spectra are shown in figure [ fig : spec2 ] .
they show the sinusoidal variation of the blue and read peaks , which are strong at all phases . the typical s - wave is also seen showing the same simple sinusoidal variation , but shifted by 0.5 in orbital phase with respect to the double - peaks .
the doppler tomogram is shown in figure [ fig : doptom ] ; as customary , the oval represents the roche - lobe of the secondary and the solid lines the keplerian ( upper ) and ballistic ( lower ) trajectories .
the tomogram reveals a disc reaching to the distance of to the inner lagrangian point in most phases .
a compact and strong emission is seen close to the center of velocities of the secondary star .
a blow - up of this region is shown in figure [ fig : spottom ] . both maps
have been constructed using the parameters shown at the top of the diagrams and a @xmath25 velocity of 34 km s@xmath3 .
the velocity resolution of the map near the secondary star is about 10 km s@xmath3 .
the @xmath70 position of the hot - spot ( in km s@xmath3 ) is ( -50,305 ) , within the uncertainties .
the tomography shown in figure [ fig : doptom ] is very different from what we expected to find and from what has been observed by other authors .
we find a very symmetric full disc , reaching close to the inner lagrangian point and a compact bright spot also close to the l@xmath12 point , instead of a complex system like that observed by @xcite , who find u gem at a stage when the doppler tomographs show : emission at low velocity close to the center of mass ; a transient narrow absorption in the balmer lines ; as well as two distinct spots , one very narrow and close in velocity to the accretion disc near the impact region and another much broader , located between the ballistic and keplerian trajectories .
they present also tentative evidence of a weak spiral structure , which have been seen as strong spiral shocks during an outburst observed by @xcite .
our results also differ from those of @xcite , who also find that the bulk of the bright spot arising from the balmer , he i and he ii emission come from a region between the ballistic and keplerian trajectories .
we interpret the difference between our results and previous studies simply by the fact that we have observed the system at a peculiar low state not detected before ( see sections [ intro ] and [ discus ] ) .
this should not be at all surprising because , although u gem is a well observed object , it is also a very unusual and variable system .
figure [ fig : spottom ] shows a blow - up of the region around the secondary star .
the bright spot is shown close to the center of mass of the late - type star , slightly located towards the leading hemisphere .
since this is a velocity map and not a geometrical one , there are at two possible interpretations of the position in space of the bright spot ( assuming the observed material is in the orbital plane ) .
the first one is that the emission is been produced at the surface of the secondary , i.e. still attached to its gravitational field .
the second is that the emission is the result of a direct shock front with the accretion disc and that the compact spot is starting to gain velocity towards the keplerian trajectory .
we believe that the second explanation is more plausible , as it is consistent with the well accepted mechanism to produce a bright spot . on the other hand , at this peculiar low state it is difficult to invoke an external source strong enough to produce a back - illuminated secondary and especially a bright and compact spot on its leading hemisphere .
assuming that the radial velocity semi - amplitudes reflect accurately the motion of the binary components , then from our results @xmath71 km s@xmath3 ; @xmath72 km s@xmath3 and adopting @xmath73 we obtain : @xmath74 @xmath75 @xmath76 and @xmath77 using the inclination angle derived by @xcite , @xmath78 , the system parameters become : @xmath6 ; @xmath79 ; and @xmath8 . a first order estimate of the dimensions of the disc the inner and outer radius can be made from the observed balmer emission line .
its peak - to peak velocity separation is related to the outer radius of the accreted material , while the wings of the line , coming from the high velocity regions of the disc , can give an estimate of the inner radius @xcite .
the peak - to - peak velocity separation of the 31 individual spectra were measured ( see section [ double - peaks ] ) , as well as the velocity of the blue and red wings of @xmath54 at ten percent level of the continuum level .
> from these measurements we derive mean values of @xmath80 and @xmath81 .
these velocities can be related to the disc radii from numerical disc simulations , tidal limitations and analytical approximations ( see @xcite and references therein ) .
if we assume the material in the disc at radius @xmath82 is moving with keplerian rotational velocity @xmath83 , then the radius in units of the binary separation is given by @xcite : @xmath84 the observed maximum intensity of the double - peak emission in keplerian discs occurs close to the velocity of its outer radius @xcite . from the observed @xmath38 and @xmath85 values we obtain an outer radius of @xmath86 and an inner radius of @xmath87 . if we take @xmath88 from the last section we obtain an inner radius of the disc @xmath89 equivalent to about 97000 km .
this is about 25 times larger than the expected radius of the white dwarf ( see section [ discus ] ) . on the other hand ,
the distance from the center of the primary to the inner lagrangian point , @xmath90 , is @xmath91 where @xmath92 ( @xcite ) . using @xmath93
we obtain @xmath94 .
the disc , therefore , appears to be large , almost filling the roche - lobe of the primary , with the matter leaving the secondary component through the @xmath11 point colliding with the disc directly and producing the hot spot near this location .
for the first time , a radial velocity semi - amplitude of the primary component of u gem has been obtained in the visual spectral region , which agrees with the value obtained from ultraviolet observations by @xcite and @xcite . in a recent paper , @xcite present high - resolution spectroscopy around h@xmath0 and h@xmath95 and conclude that they can not recover the ultraviolet value for @xmath49 to better than about 20 percent by any method .
although the spectral resolution at h@xmath0 of the instrument they used is only a factor of two smaller than that of the one we used , the diagnostic diagrams they obtain show a completely different behavior as compared to those we present here , with best values for @xmath18 of about 95 km s@xmath3 from h@xmath0 and 150 km s@xmath3 from h@xmath95 ( see their figures 13 and 14 , respectively ) .
we believe that the disagreement with our result lies not in the quality of the data or the measuring method , but in the distortion of the emission lines due to the presence of a complex accretion disc at the time of their observations , as the authors themselves suggest .
their doppler tomograms show emission at low velocity , close to the center of mass , two distinct spots , a narrow component close to the @xmath11 point , and a broader and larger one between the keplerian and the ballistic trajectories .
there is even evidence of a weak spiral structure .
in contrast , we have observed u gem during a favorable stage , one in which the disc was fully symmetric , and the hot - spot was narrow and near the inner lagrangian point .
this allowed us to measure the real motion of the white dwarf by means of the time - resolved behavior of the h@xmath0 emission line .
our highly consistent results for the systemic velocity derived from the h@xmath0 spot ( @xmath66 10 km s@xmath3 and those found from the different methods used for the radial velocity analysis of the emission arising from the accretion disk ( see section [ prim ] and table [ orbparam ] ) , give strong support to our adopting a true systemic velocity value of @xmath67 2 km s@xmath3 .
if we are indeed detecting the true motion of the white dwarf , we can use this adopted value , to make an independent check on the mass of the primary : the observed total redshift of the white dwarf ( gravitational plus systemic) found by @xcite is 172 km s@xmath3 , from which , after subtraction of the adopted systemic velocity , we derive a gravitational shift of the white dwarf of 138 km s@xmath3 . from the mass - radius relationship for white dwarfs
@xcite , we obtain consistent results for @xmath96 and @xmath97 = 3900 km ( see figure 7 in @xcite ) .
this mass is in excellent agreement with that obtained in this paper from the radial velocity analysis .
> from our new method to determine the radial velocity curve of the secondary ( section [ ugemb ] ) , we obtain a value for the semi - amplitude close to 310 km s@xmath3 .
three previous papers have determinations of the radial velocity curves from the observed na i doublet in the near - infrared . in order to evaluate
if our method is valid , we here compare our result with these direct determinations .
the published values are : @xmath98 km s@xmath99 ( @xcite )
; @xmath100 km s@xmath101 , ( before correction for irradiation effects , @xcite ) ; and @xmath102 km s@xmath3 @xcite . @xcite notes that an elliptical orbital ( @xmath103 ) may better fit his data , as the velocity extremum near phase 0.25 appears somewhat sharper than that near phase 0.75 ( see his figure 3 ) . however , he also finds a very large systemic velocity , @xmath104 km s@xmath3 , much larger than the values found by @xcite ( @xmath105 km s@xmath3 ) and @xcite ( @xmath106 6 km s@xmath3 ) , both obtained from the emission lines . since the discrepancy with the results of these two authors was large , @xcite defers this discussion to further confirmation of his results .
instead , and more important , this author discusses two scenarios that may significantly alter the real value of @xmath19 : the non - sphericity and the back - illumination of the secondary . in the latter effect , each particular absorption line may move further away from , or closer to the center of mass of the binary .
he estimates the magnitude of this effect and concludes that the deviation of the photocenter would probably be much less than 0.1 radii .
@xcite further discusses the circumstances that might cause the photocenter to deviate , and concludes that their observed value for the semi - amplitude should be corrected down by 3.5 percent , to yield @xmath107 km s@xmath108 .
although they discuss the results by @xcite which indicate that the relatively small heating effects in quiescent dwarf novae always lead to a decrease in the measured @xmath109 for the na i lines
they argue that line quenching , produced by ionization of the same lines , may also be important , and result in an increased @xmath109 .
another disturbing effect , considered by the same authors , is line contamination by the presence of weak disc features , like the paschen lines . in this respect
we point out here that a poor correction for telluric lines will function as an anchor , reducing also the amplitude of the radial velocity measurements .
@xcite also find an observed systemic velocity of @xmath110 6 km s@xmath3 and a small eccentricity of @xmath111 .
@xcite also discuss the distortion effects on the na i lines and , based on their fit residuals , argue in favor of a depletion of the doublet in the leading hemisphere of the secondary , around phases 0.4 and 0.6 , as removing flux from the blueward wing of the lines results in an apparent redshift , which would explain the observed residuals .
however , they additionally find that fitting the data to an eccentric orbit , with @xmath112 , results in a significant decrease in the residuals caused by this depletion , and conclude that it may be unnecessary to further correct the radial velocity curve .
we must point out that a depletion of the blueward wing of the na i lines will results in a contraction of the observed radial velocity curves , as the measured velocities especially around phases 0.25 and 0.75 will be pulled towards the systemic velocity .
@xcite present their results derived from the na i doublet and the ki / tio region ( around 7550 - 7750 ) , compared with several spectral standards , all giving values between 289 and 305 km s@xmath3 ( no errors are quoted ) . based on the radial velocity measurements for na i , obtained by these authors in 2001 january ( 115 spectra ) , and using gj213 as template ( see their table 1 ) ,
we have recalculated the circular orbital parameters through our nonlinear least - squares fit .
we find @xmath113 km s@xmath114 , in close agreement with their published value .
it would be advisable to establish a link between the observed gamma velocity of the secondary and the semi - amplitude @xmath1 , under the assumption that its value may be distorted by heating effects .
we take as a reference our results from the radial velocity analysis of the broad @xmath54 line and the hot - spot from the secondary , which support a true systemic velocity of 34 km s@xmath3 . however , we find no positive correlation in the available results derived from the na i lines , either between different authors or even among one data set . in the case of @xcite ,
the gamma values show a range between 11 and 43 km s@xmath3 , depending on the standard star used as a template , for @xmath1 velocities in the range 289 to 305 km s@xmath3 .
@xcite finds @xmath115 10 km s@xmath3 for a low @xmath116 value of 283 km s@xmath3 , while @xcite finds @xmath117 6 km s@xmath3 for @xmath1 about 309 km s@xmath3 , and we obtain a large gamma velocity of about 69 km s@xmath3 for a @xmath1 value of 310 km s@xmath3 .
we believe that further and more specific spectroscopic observations of the secondary star should be conducted in order to understand the possible distortion effects on lines like the na i doublet , and their implications on the derived semi - amplitude and systemic velocity values .
e. de la f wishes to thank andrs rodriguez j. for his useful computer help .
the thomson detector , used in our observations , was obtained through pacime - conacyt project f325-e9211 .
anderson , n. , 1988 , apj .
, 326 , 266 arnold , s. , berg , r.a . & duthie , j.g . , 1976 ,
apj . , 206 , 790 berriman , g. , beattie , i.a . , lee , t.j . ,
mochnacki , s.w . & szkody , p. , 1983 ,
mnras , 204 , 1105 cannizzo , j. , k. , gehrels , n. , & mattei , j.a . , 2002 , apj , 579 , 760 cook , l.m . , 1987 , javso , 16 , 83 echevarra , j , 1983 , rmaa , 8 , 109 echevarra , j & jones , d.h.p . , 1984 , mnras , 206 , 919 friend , m. t. , martin , j. s. , connon smith , r. , & jones , d. h. p. , 1990 ,
mnras , 246 , 637 groot , p.j . , 1991 , apj , , 2649 harrison , t.e . , mcnamara , b.j . , szkody , p. & gilliland , r.l .
, 2000 , apj , 120 , 2649 hind , j. r. , 1856 , mnras , 16 , 56 honeycutt , r. k. , kaitchuck , r. h. , & schlegel , e. m. , 1987 , apjs , 65 , 451 horne , k. , wade , r.a .
& szkody , p. , 1986 ,
mnras , 219 , 791 kopal , z. , _ close binary systems _ , champan & hall , london .
kraft , r. p. , 1962 ,
apj , 135 , 408 kraft , r. p. , 1975
, private communication in smak ( 1976).64 , 637 kreminski , w. , 1965 , aj , 142 , 1051 long , k. s. , & gilliland , r. l. , 1999 , apj , 511 , 916mass . marsh , t.r . , & horne , k. , 1988 , mnras , 235 , 26997 , a&asuppl.ser . , 123 , 473 marsh , t. r. , horne , k. , schlegel , e. m. , honeycutt , r. k. & kaitchuck , r. h. , 1990 , apj , 364 , 637 martin , j.s . , 1988 , _
d phil thesis _ , university of sussex no .
5 , cambridge , mass . mattei , j.a . ,
saladyga , m. , wagen , e.o . & jones , c.m . , 1987 ,
aavso , monograph , cambridge , mass .
montes , d. , martn , e.l .
, fernndez - figueroa , m.j .
, cornide , m. & de castro , e. , 1997 , a&asuppl.ser . , 123 , 473 naylor , t. , allan , a. & long , k. s. , 2005 , mnras , 361 , 1091apj , 496 , 449 payne - gaposchkin , c. & gaposchkin , 1938 , _ variable stars _ , harv . obs .
5 , cambridge , mass .
schwarzenberg - czerny , a. , 1999 , astrophys .
j. , 516 , 315 shafter , a.w . , szkody , p. & thorstensen , j. r. 1986 , apj , 308 , 765 sion , e.l . ,
cheng , f.h .
, szkody , p. , sparks , w. , gnsicke , b.huang , m & mattei , , j. 1998 , apj , 496 , 449 smak , j. , 1971 , acta astr . , 21 , 15 smak , j. , 1976 , acta astr . , 26 , 277 smak , j. , 1981 , acta astr . , 31 , 395 smak , j. , 2001 , acta astr . , 51 , 279 smak , j. , 2004 , acta astr . , 54 , 433 spruit , h. c. , 1998 , astro - ph/9806141 stover , r. j. , 1981 , apj , 248 , 684 szkody , p. , & mattei , j. a. , 1984 , pasp , 96 , 988 unda - sanzana , e. , marsh , t. r. & morales - rueda , l. , 2006 , mnras , 369 , 805 wade , r. a. , 1981 , apj , 246 , 215 warner , b. , 1995 , _ `` cataclysmic variable stars '' _ , cambridge university press warner , b. & nather , r.e . , 1971 , mnras , 152 , 219 zahn , j .-
ann . dastrophys . , 29 , 489 .
zhang , e. h. , & robinson , e. l. , 1987 , apj , 321 , 813 | high - resolution spectroscopy of was obtained during quiescence .
we did not find a hot spot or gas stream around the outer boundaries of the accretion disk .
instead , we detected a strong narrow emission near the location of the secondary star .
we measured the radial velocity curve from the wings of the double - peaked h@xmath0 emission line , and obtained a semi - amplitude value that is in excellent agreement with the obtained from observations in the ultraviolet spectral region by @xcite .
we present also a new method to obtain @xmath1 , which enhances the detection of absorption or emission features arising in the late - type companion .
our results are compared with published values derived from the near - infrared nai line doublet . from a comparison of the tio band with those of late type m stars , we find that a best fit is obtained for a m6v star , contributing 5 percent of the total light at that spectral region . assuming that the radial velocity semi - amplitudes reflect accurately the motion of the binary components , then from our results : @xmath2 km s@xmath3 ; @xmath4 km s@xmath3 , and using the inclination angle given by @xcite ; @xmath5 , the system parameters become : @xmath6 ; @xmath7 ; and @xmath8 . based on the separation of the double emission peaks ,
we calculate an outer disk radius of @xmath9 , close to the distance of the inner lagrangian point @xmath10 .
therefore we suggest that , at the time of observations , the accretion disk was filling the roche - lobe of the primary , and that the matter leaving the @xmath11 point was colliding with the disc directly , producing the hot spot at this location . |
sipms are excellent photosensors for low - intensity light detection . with increasing light intensities , however , their response becomes non linear requiring monitoring . in the calice ahcal prototype we installed an led / pin - diode - based monitoring system to measure the sipm gain ,
monitor the sipm response for fixed light intensities and record the full sipm response function when necessary .
we also record the reverse bias voltage of each sipm and the temperature measured by five sensors in each layer with a slow - control system , since the sipm response is very sensitive to changes in these parameters .
we studied the sipm response function in test beam calibration runs .
if we find an analytic function that parameterizes the sipm response , the monitoring system might be simplified .
the exact shape would be measured once on the test bench before installation .
the raw energy in a cell measured in units of adc bins , @xmath0 $ ] , is converted into units of mips via a calibration factor @xmath1 $ ] @xmath2 = \frac{q^{meas}_{cell}[adc]}{c^{mip}_{cell}[adc ] } \cdot f^{-1}_{sat}(q^{meas}_{cell}[pixel]),\ ] ] where @xmath3)$ ] is the non - linear response parameterized as a function of pixels .
it includes an intercalibration factor matching the pixel and mip calibration scales @xcite .
we measure the sipm gain with low - intensity led light , the position of the mip peak with muons and the non - linearity by varying the led light intensity . gain and light yield vary with temperature ( reverse bias voltage ) as @xmath4 ) and @xmath5 ) , respectively .
the standard procedure consists of adjusting @xmath6 and @xmath7 for temperature changes . here , corrections are instantaneous but non - local , since @xmath8 is measured frequently at five positions per layer during a run . since the gain of each cell
is measured several times a day , we could also correct @xmath7 for gain changes by @xmath9 @xcite .
this procedure is local but not instantaneous .
figure [ fig : energy ] shows energy and energy resolution measurements of 10 50 gev positrons in comparison to simulations @xcite .
the data is corrected for saturation and temperature effects using cell - wise gain and mip calibration constants and layer - wise average temperature correction factors .
the reconstructed energy is linear up to 30 gev . at 50 gev deviations from linearity increase to @xmath10 suggesting the need for a refined analysis procedure .
the energy resolution fits the standard form containing a stochastic term , a constant and a noise term .
the bare simulations demonstrate that detector effects are important .
simulations that include detector effects , however , are still too optimistic and require refinements .
data ( dots ) , bare simulations ( open circles ) and simulations including detector effects ( triangles ) . lines show fits and shaded regions systematic uncertainties.,title="fig:",scaledwidth=48.0% ] data ( dots ) , bare simulations ( open circles ) and simulations including detector effects ( triangles ) . lines show fits and shaded regions systematic uncertainties.,title="fig:",scaledwidth=48.0% ]
we extract sipm and pin diode raw data from lcio files , perform pedestal subtraction with vcalib runs , apply gain corrections , and use intercalibration constants . for each vcalib value we perform gaussian fits to the sipm and pin diode response to determine mean values and their errors .
we plot the pin - diode response versus the sipm response after rescaling the pin - diode values to start at a common origin with a slope of one .
so far we analyzed four calibration runs from the cern test beam in 2006 and 2007 @xcite . at itep , the response of all 7608 sipms was measured prior to installation into the ahcal prototype using calibrated led light shone directly onto the sipm .
figure [ fig : curves ] ( top ) shows the itep measurements after scaling raw data to start at a common origin with slope one .
the curves fit to @xmath11 , where the saturation @xmath12 and @xmath13 are free parameters . in the test beam the sipm response
is measured with high - gain and low - gain preamplifier settings .
since the curves do not fit to @xmath14 , we constructed the function @xmath15,\ ] ] where @xmath16 ( with @xmath17 and @xmath18 are free parameters in the fit .
the latter is a scale factor that accounts for mismatch between high - gain and low - gain regions . with simpler forms we obtained fewer successful fits .
r0.5 figures [ fig : curves](middle , bottom ) show response curves of a typical sipm for itep and test beam data , respectively .
fitting all available sipms in the four calibration runs , we get 5060@xmath19 successful fits . in the 31 - 07 - 07 run @xmath20 of all fits were successful , @xmath21 failed due to a small confidence level and @xmath22 failed due to malfunctioning of the sipm or pin diode . for successful fits , parameter @xmath23 ( @xmath24 ) peaks at @xmath25 @xmath26 with fwhm of @xmath27 .
apart from a few outliers at higher values , a spike near zero is visible .
parameter @xmath13 peaks near @xmath28 with fwhm of @xmath29 , while parameter @xmath30 lies around @xmath31 with fwhm of @xmath32 .
figure [ fig : saturation](left ) shows saturation values for 12 - 07 - 07 and 31 - 07 - 07 calibration runs . for the later run , the saturation peaks near @xmath33 pixels and has a width of fwhm@xmath34 pixels , while for the earlier run the peak is shifted upward by @xmath35 pixels .
this is consistent with expected temperature variations that are not corrected for .
in addition , we see a long tail up to @xmath36 pixels in both runs that is also visible in itep data .
this results from a particular batch of sipms with a lower internal resistor causing multiple pixel excitations during illumination .
a comparison of saturation values of sipms in modules 315 for four runs taken in 2006 and 2006 shows no degradation thus confirming stable operation in the test beam .
figure [ fig : saturation](right ) shows a comparison of saturation values measured in the 31 - 07 - 07 run and at itep .
the itep data peak around @xmath37 pixels with a fwhm of @xmath38 pixels indicating that nearly all 1156 pixels in the sipms are triggered . here , the sipms were illuminated directly by leds , while the light transport in the test beam is rather complex .
the sipm records the light from a 1 mm thick wavelength - shifting ( wls ) fiber coupled via an air gap .
though a 0.2 mm air gap is sufficient for a full illumination of all pixels , losses may occur due to imperfect alignment of the fiber and sipm .
because of the large discrepancy between itep and test beam data , the non - linearity corrections are presently based on the itep saturation measurements with an additional scaling by the ratio of saturation values measured for each sipm in the test beam and at itep .
we found an analytical function that successfully parameterizes about @xmath20 of the sipm response curves in the calice ahcal prototype . in order to improve this close to @xmath39 we need to investigate causes for poor fits and test other analytical functions .
the present studies show that the sipm operation in the cern test beam is stable .
we observe a discrepancy of saturation values measured at itep and in the test beam of about @xmath40 , which may be caused by a misalignment of the wls fiber and the sipm .
we need to verify that we can model temperature and voltage changes in the sipm response curves and we need to extend our studies to include 2008 test beam data at fermilab . at the present level of understanding a full monitoring system
is necessary that allows us to measure the full sipm response any time .
the system , however , may be simplified .
two options are under discussion .
the first option is based on the present monitoring system but foresees long clear fibers , each illuminating one row of tiles rather than individual tiles .
this would reduce the number of leds but it may not achieve sufficient light intensities in all tiles ( see j. zalesak s talk @xcite ) .
the second option consists of embedding one led per tile eliminating fibers but requiring a huge number of led s .
system tests started to optimize led positions , check the homogeneity of the response and test different led types .
the light calibration is compared to the response of a radioactive source .
the system will be temperature controlled .
first tests show no cross - talk , but optimization for dynamic range and led uniformity is needed .
99 v. andreev et al . , nucl.instrum.meth.a540 , 368 ( 2005 ) .
n. feege , diplom thesis , 78 pp , universitt hamburg ( 2008 ) . n. meyer et al ( calice collaboration ) , can-014 , 5pp ( 2008 ) .
t. buanes , phd thesis , 129 pp , university of bergen ( 2008 ) .
j. zalesak , proceedings of lcws08 , 4pp ( 2009 ) . | we present herein our experience with the calibration system in the calice ahcal prototype in the test beam and discuss characterizations of the sipm response curves . |
[ sec : introduction ] compton scattering @xmath15 at energies below @xmath16 explores the two - photon response of the internal low - energy degrees of freedom in the nucleon and in the lightest nuclei . since the electric and magnetic fields of a real photon induce radiation multipoles by displacing the charged constituents and currents in the target , energy - dependence and multipolarity of the emitted radiation test the symmetries and strengths of the interactions between and with them ;
see a recent review for details @xcite . in deuteron compton scattering , one not only has access to the proton and neutron response , but also to how photons couple to the charged pion - exchange currents , thus testing nuclear binding in the simplest stable few - nucleon system .
in addition , the constructive interference with the @xmath10 wave component of the deuteron can be expected to lead to increased sensitivity of the hadronic response to the quadrupole components of the photon fields .
a new generation of high - luminosity facilities like hi@xmath17s , mami and max - lab with near-@xmath18 linear or circular beam polarisation have started to explore these opportunities .
dense deuteron targets with vector polarisations approaching @xmath19 are standard .
since tensor and vector polarisations are related when in thermal equilibrium with a solid lattice , most vector - polarised deuteron targets automatically also provide tensor polarisation degrees of @xmath20 and the potential for greater values in dedicated set - ups @xcite . now is thus an opportune moment for a comprehensive classification of independent deuteron amplitudes and observables .
the spin-@xmath21 case has been discussed by babusci et al .
@xcite . for the deuteron , chen , ji and li
@xcite constructed a basis for those @xmath3 amplitudes which remain linearly independent after parity and time - reversal invariance have been invoked on the @xmath22(photon helicities)@xmath23(deuteron helicities)@xmath24 ^ 2$](both in- and out - state)@xmath25 helicity amplitudes . however , a corresponding list of 23 independent observables ( @xmath3 complex amplitudes minus an overall phase ) is missing . while several single and double polarisation observables have been constructed and their sensitivity to the nucleon polarisabilities explored @xcite , no systematic study of vector and tensor polarisation observables exists .
only one tensor observable has been considered explicitly , namely for an unpolarised beam @xcite .
what is more , some deuteron `` vector '' observables which were defined analogous to the spin-@xmath21 case will be shown to actually receive contributions from both vector and tensor polarisations .
for the case that the polarisations of the final state are not detected , this work aims to classify all @xmath1 independent observables and their relation to the helicity amplitudes . at present , this seems to be the experimentally most feasible situation . instead of simply extending the work by babusci et al .
to the spin-@xmath0 case , the starting point is the most general cross section of an arbitrarily polarised photon beam on an arbitrarily polarised spin-@xmath0 target , in a form which is well - known e.g. from deuteron photo - disintegration @xcite .
it is parametrised in terms of the unpolarised cross section , @xmath0 beam and @xmath2 target asymmetries as well as @xmath3 double asymmetries and has the added benefit that experiments in less - than - ideal settings can easily be described as well , like when residual or mixed target and beam polarisations exist .
a future publication will define and study @xmath7 additional independent polarisation transfer observables @xcite .
a complete set of independent compton scattering observables will then be available from which the @xmath26 real parameters which characterise deuteron compton scattering ( i.e. its independent amplitudes ) can be reconstructed in full .
the second part of this article explores the sensitivity of the @xmath1 observables to the two - photon response of the individual nucleon .
remember that the proportionality constants between the electric or magnetic field of the incident photon and the radiation multipoles induced in each nucleon are the energy - dependent ( dynamical ) polarisabilities of the nucleon @xcite .
they parametrise the stiffness of the nucleon @xmath27 ( spin @xmath28 ) against transitions @xmath29 of definite photon multipolarity at frequency @xmath30 ( @xmath31 ; @xmath32 ; @xmath33 ; @xmath34 ) ; see e.g. @xcite and references therein .
re - written as point - like interactions between photons and nucleons , the terms which contain photon dipoles read : @xmath35\;n \end{split}\ ] ] since each interaction with a photon leaves a unique signal in such dispersive effects , compton scattering allows one to study the symmetries and dynamics of the hadronic constituents in detail .
the zero - energy values , @xmath36 etc . , are often quoted as `` the ( static ) polarisabilities '' .
two scalar polarisabilities @xmath37 and @xmath38 parametrise electric and magnetic dipole transitions .
the four dipole spin polarisabilities @xmath39 , @xmath40 , @xmath41 and @xmath42 encode the response of the nucleon spin - structure . these are particularly interesting since , intuitively interpreted , they parametrise the bi - refringence which the electromagnetic field associated with the spin degrees causes in the nucleon , in analogy to the classical faraday - effect @xcite .
the information accessible in compton scattering thus goes well beyond that in tests of the one - photon response e.g. in form - factor experiments .
theoretical input is of course needed to carefully evaluate data - consistency in one model - independent framework for hidden systematic errors ; identify the underlying mechanisms using minimal theoretical bias , like the detailed chiral dynamics of the pion cloud and of the @xmath4 as the lowest nucleon resonance ; and , most importantly , explain how these findings emerge from qcd by relating to emerging lattice simulations ( see most recently @xcite ) .
the polarisabilities also enter as one of the bigger sources of uncertainties in theoretical determinations of the proton - neutron mass shift ( see e.g. most recently @xcite ) and of the two - photon - exchange contribution to the lamb shift in muonic hydrogen @xcite .
while presumably not providing a solution to the proton - charge - radius puzzle , they also contribute in radiative corrections to this process , see e.g. @xcite .
for all these goals , chiral effective field theory ( @xmath43eft ) , the low - energy theory of qcd and extension of chiral perturbation theory to few - nucleon systems , adds objective estimates of the theoretical uncertainties . indeed , @xmath43efthas been particularly successful in describing proton and few - nucleon compton scattering , starting with the first calculation and sensitivity study of the scalar polarisabilities in @xmath43eft @xcite . ref .
@xcite contains details on its history and status in compton scattering , as well as on @xmath43eftvariants not discussed here .
having established a consistent database from all available proton and deuteron data below @xmath44 in ref .
@xcite , the static scalar polarisabilities of the proton were recently extracted in this framework with a @xmath45 per degree of freedom of @xmath46 @xcite : @xmath47 throughout , polarisabilities without superscripts denote isoscalar quantities , and the canonical units of @xmath48 for scalar and @xmath49 for spin dipole polarisabilities are understood . since the deuteron is an isoscalar , elastic scattering on it provides of course only access to the isoscalar ( average ) nucleon polarisabilities . in ref .
@xcite , these were found to have much larger errors since deuteron data is less accurate and more scarce ( with @xmath50d.o.f.@xmath51 ) : @xmath52 these results were derived using the baldin sum rules , whose isoscalar variant reads @xcite : @xmath53 these publications also discuss in detail the fit procedure and residual theoretical uncertainties . comparing eqs . and shows that within the data - dominated error , the two - photon responses of the proton and neutron as parametrised by the scalar polarisabilities are identical . a particularly interesting prediction of @xmath43eftis that small proton - neutron differences stem from chiral - symmetry breaking interactions with and in the pion cloud around the nucleon , probing details of qcd
experiments are therefore underway and planned to improve the compton scattering database ; see e.g. @xcite for details .
their other focus are the spin polarisabilities . only the linear combinations @xmath54 and @xmath55 of scattering under @xmath56 and @xmath57 are somewhat constrained by data or phenomenology .
conflicting results from mami and legs exist for the proton , and large error - bars are found for the neutron @xcite .
the isoscalar values are in the range ( see also @xcite ) : @xmath58 \end{split}\ ] ] a comprehensive classification of independent amplitudes and observables is thus warranted , including a detailed study of dependencies on scalar and spin polarisabilities .
insofar , this publication extends the so - far most thorough work in ref .
@xcite , including its erratum .
after defining the most general cross section without detection of the polarisations of the final state in subsec .
[ sec : parameterising ] , the remainder of sec .
[ sec : construction ] is devoted to the more technical issues of relating its observables to the helicity amplitudes of deuteron compton scattering and to other parameter - combinations found in the literature , including the babusci - classification .
section [ sec : observables ] discusses the sensitivity of the observables to the dipole polarisabilities , with an eye towards potential experiments .
it also proposes a road - map to the isoscalar , spin - independent and spin - dependent nucleon polarisabilities from high - accuracy experiments with deuteron targets . a customary summary in sec . [ sec : conclusions ] rounds off the article .
preliminary results were presented in a recent proceeding @xcite .
[ sec : construction ] this presentation follows the reviews of arenhvel and sanzone @xcite , and paetz @xcite .
inspired by the former , the kinematics is pictorially represented in fig .
[ fig : kinematics ] .
-axis , linearly polarised at angle @xmath59 relative to the scattering plane ( @xmath60-plane ) ; scattering angle @xmath61 ; deuteron polarisation axis @xmath62 with azimuthal angle @xmath63 from the @xmath64-axis to @xmath62 , and polar angle @xmath65 from the @xmath66-axis to the projection of @xmath62 onto the @xmath67-plane ; @xmath68-axis the normal of the scattering plane ; @xmath69 ( @xmath70 ) the momentum of the incident ( outgoing ) photon.,scaledwidth=55.0% ] the photon beam polarisation is described by a density matrix with entries @xmath71\;\;.\ ] ] here , @xmath72 $ ] is the degree of right - circular polarisation , i.e. the difference between right and left circular polarisation , with @xmath73 describing a fully right / left circularly polarised photon ( positive / negative helicities @xmath74 by @xmath75 ) .
the degree of linear polarisation is parametrised by @xmath76 $ ] , with @xmath77 the angle from the @xmath66-axis to the polarisation plane is counted _ from _ the polarisation plane _ to _ the normal of the scattering plane , i.e. @xmath78 . ] , i.e. a photon polarisation @xmath79 .
today s deuteron targets are both vector- and tensor - polarised along the same axis @xcite .
let the axis @xmath62 in which @xmath80 is diagonal be oriented as in fig .
[ fig : kinematics ] , i.e. , but still @xmath81 . ] @xmath82 with azimuthal angle @xmath83 $ ] and polar angle @xmath84 $ ] .
the entries of the polarisation density matrix are then in the basis @xmath85 of magnetic quantum numbers along @xmath62 : @xmath86\;\;.\ ] ] the subscript denotes of course that the system is quantised along the @xmath62-axis , not the @xmath64-axis , and is kept for comparison with the literature @xcite . here , @xmath87 parametrises the part of the deuteron density matrix which behaves like a scalar under rotations , while @xmath88 and @xmath89 characterise the parts which transform like an ( irreducible ) spherical vector and tensor operator , respectively .
these can be related to the degrees of vector and tensor polarisation in cartesian coordinates : along this quantisation axis , a fraction @xmath90 populates the state with magnetic quantum number @xmath91 .
the overall norm is @xmath92 .
the degree of vector polarisation is in cartesian coordinates @xmath93 $ ] , and that of tensor polarisation is @xmath94 $ ] . because @xmath95 lies between 0 and 1 , they are subject to the combined constraint @xmath96 . when the deuteron spins are in thermal equilibrium with a solid lattice , tensor and vector polarisations are related by @xmath97 , i.e. @xmath98 @xcite .
the advantage to decompose @xmath80 into irreducible representations of the rotation group is that it is then particularly simple to change the quantisation axis from @xmath62 to the beam axis @xmath99 ; cf .
* subsec .
@xcite finally provides the angular momentum representation of the spin-@xmath0 polarisation density matrix which is diagonal along @xmath62 : @xmath100 the conventions for @xmath101-symbols and reduced wigner-@xmath102 matrices are those of rose @xcite and edmonds @xcite , also listed in the particle data booklet @xcite . like any reaction @xmath103 , deuteron compton scattering and deuteron photo - disintegration share
the same in - state .
as long as the final - state polarisations are not detected ( i.e. are summed over ) , their differential cross sections are thus characterised by the same dependence on the initial - state deuteron and photon polarisations .
one can therefore adopt the decomposition familiar from deuteron photo - disintegration @xcite to compton scattering : @xmath104\\ & + \sum\limits_{i=1,2\atop 0\le m\le i}\;t_{im}^\text{circ}(\omega,\theta)\ ; p^{({\mathrm{d}})}_i\;d^i_{m0}({\ensuremath{\vartheta_\text{d}}})\ ; p^{(\gamma)}_\text{circ}\;\sin[m{\ensuremath{\varphi_\text{d}}}+\frac{\pi}{2}\delta_{i1}]\\ & + \sum\limits_{i=1,2\atop -i\le m\le i}\;t_{im}^\text{lin}(\omega,\theta)\ ; p^{({\mathrm{d}})}_i\;d^i_{m0}({\ensuremath{\vartheta_\text{d}}})\ ; p^{(\gamma)}_\text{lin}\;\cos[m{\ensuremath{\varphi_\text{d}}}-2{\ensuremath{\varphi_\text{lin}}}-\frac{\pi}{2}\delta_{i1 } ] \bigg ] \end{split}\ ] ] besides the trivial limitations @xmath105 and @xmath106 , the summations in eq . are easily shown to be constrained by trivial zeros and double counting of angular dependencies : * @xmath107 , i.e. the first factor in eq .
could also be written as @xmath108 ; * @xmath109 , and in particular @xmath110 ; * @xmath111 , and in particular @xmath112 ( the circular - beam asymmetry on an unpolarised target , identical zero due to rotation invariance ) and @xmath113 ; * @xmath114 .
the cross section is thus fully parametrised by the following linearly independent functions : * @xmath0 differential cross section @xmath115 of unpolarised photons on an unpolarised target ; * @xmath0 beam asymmetry of a linearly polarised beam on an unpolarised target @xmath116 ; * @xmath0 vector target asymmetry on an unpolarised beam @xmath117 ; * @xmath118 tensor target asymmetries of an unpolarised beam @xmath119 , @xmath120 ; * @xmath121 double asymmetries of circular photons on a vector polarised target @xmath122 , @xmath123 ; * @xmath121 double asymmetries of circular photons on a tensor polarised target @xmath124 , @xmath125 ; * @xmath118 double asymmetries of linear photons on a vector target @xmath126 , @xmath127 ; * @xmath7 double asymmetries of linear photons on a tensor target @xmath128 , @xmath129 .
since these 18 real , independent functions of scattering energy and angle are of course process - dependent , those discussed in compton scattering differ from those in e.g. deuteron photo - disintegration .
the decomposition of eq . holds in any frame , but the functions are frame - dependent .
it also applies when the polarisation of the target and/or scattered photon is detected in the final state , without specifying the initial state .
the 18 recoil polarisations are thus identical to the functions above .
deuteron compton scattering amplitudes @xmath130 are usually described in the helicity basis ( dependencies on @xmath131 and other parameters are dropped for brevity in this section ) : @xmath132 where @xmath133 is the circular polarisation of the initial / final photon , and @xmath134 is the magnetic quantum number of the initial / final deuteron spin . in the following ,
the indices and summations over the final - state polarisations are suppressed as self - understood , e.g. @xmath135 .
in addition , it is convenient to introduce an abbreviation for the sum over all polarisations of the squared amplitude : @xmath136 the cross section of compton scattering of a photon beam with the density matrix @xmath137 from a target with density matrix @xmath80 , without detection of the final state polarisations , is then @xmath138\;\;,\ ] ] where the trace is taken over the polarisation states and @xmath139 is the frame - dependent flux factor , e.g. in the centre - of - mass and lab frames : @xmath140 transformations between lab and cm kinematics are found in a recent review ( * ? ? ?
2.3 ) . by inserting the density matrices of eqs . and into eq .
, one obtains the cross section in terms of the amplitudes , as function of photon polarisations @xmath141 and @xmath142 with polarisation angle @xmath143 and deuteron polarisation @xmath144 with orientation @xmath145 .
the functional dependence of the result on these parameters is easily matched to the parametrisation in eq . .
for the unpolarised part , one finds of course : @xmath146 the asymmetries are then ( these definitions obey the constraints discussed in sec . [
sec : parameterising ] ) : @xmath147 these explicit forms can also be used to determine which observables are nonzero only due to inelasticities . cross sections and , concurrently , the functions @xmath116 , @xmath148 , @xmath149 are of course real .
the compton amplitudes @xmath150 are real below the first inelasticity , so that the occurrence of the imaginary unit in six of the observables in eqs .
to indicates that they are zero there , namely @xmath151 the deuteron compton amplitude contains 2 independent complex amplitudes for a scalar target , 4 more for a vector target , and 6 more for a tensor target ; see e.g. @xcite .
how many and which of them are accessible with polarised beam and/or target , but without measuring outgoing polarisations ( or , by time - reversal invariance , vice versa ) ?
those which can not be determined must be probed in polarisation transfer experiments .
these are significantly harder because of the difficulties to measure recoil and scattered - photon polarisations . as a warm - up
, one could consider first the compton scattering below the first inelasticity , where all amplitudes are real .
this is however of limited use in deuteron compton scattering , where the first appreciable breakup process , @xmath152 , starts at a cm photon energy of @xmath153 , namely so low that the amplitudes have significant imaginary parts in the experimentally interesting region etc . ) , but is suppressed by powers of @xmath154 and hence does not significantly contribute in experiments .
it is not considered in today s theoretical descriptions , whose first inelasticity thus is the deuteron breakup . ] . in contradistinction ,
the first appreciable inelasticity on the proton starts at the one - pion production threshold . above the first inelasticity , @xmath26
independent real amplitudes exist , namely 3 for a scalar target ( 2 complex minus an overall phase ) , 8 more for a vector target , and 12 more for a tensor target . since the 6 observables of eq .
are nonzero there , one finds : * for scalar targets , only 2 of 3 observables are accessible , leaving 1 to be determined from a polarisation transfer observable . * for vector polarised targets , 6 of 8 observables are accessible , leaving 2 to be determined from polarisation transfer observables . * for tensor polarised targets , 10 of 12 observables are accessible , leaving again 2 to be determined from polarisation transfer observables .
the @xmath7 correlations between beam and recoiling target polarisation which are necessary for complete experiments on the deuteron will be discussed in a future publication @xcite .
this concludes the classification itself ; results in @xmath43eftwill be presented in sec . [ sec : results ] . since some observables in compton scattering with vector and tensor polarised targets have been constructed before
, it is appropriate to relate these to the classification in eq . .
often , observables are expressed not in terms of the degrees of deuteron vector and tensor polarisations , but via the occupation numbers @xmath95 of a state quantised along @xmath62 . from eq .
, the density matrix of a pure deuteron state @xmath155 is : @xmath156 the first tensor observable was constructed by chen @xcite , and also used by karakowski and miller @xcite . his definition of a cross section combination for an unpolarised beam on a deuteron which is tensor polarised along the @xmath64 axis translates into the observables of eq . with @xmath157 , @xmath158 and eq . into : @xmath159=-\frac{3}{2\sqrt{2}}\;t_{20 } \left.\frac{{\mathrm{d}}\sigma}{{\mathrm{d}}\omega}\right|_\text{unpol}\;\;,\ ] ] where the subscript in @xmath160 denotes that @xmath62 points along the @xmath64-axis for the initial state . from now on , the bracketed superscript of an observable
indicates the bibliographic reference from which the notation is taken verbatim .
this is the only tensor - observable for which calculations exist , namely at @xmath161 and @xmath162 both by chen and by karakowski and miller .
nonetheless , these will not be compared in detail with those of the @xmath43eftapproach taken in sec .
[ sec : results ] .
chen s are derived in `` pion - less '' eft , i.e. for typical momenta well below the pion mass and typical photon energies @xmath163 @xcite . these predictions are thus more of qualitative interest . shape and size of the angular dependence
differ indeed considerably from those presented later .
karakowski and miller used an approach similar to that which will be outlined in sec .
[ sec : theory ] , but without a dynamical @xmath4 and without some pion - exchange diagrams dictated by chiral symmetry @xcite .
their results at @xmath161 and @xmath162 agree up to about @xmath164 in shape and magnitude with the ones presented below .
the difference does not stem from the @xmath4 , but may be attributed to the fact that their photon - nucleon interaction for rescattering terms is expanded only to first order , while hildebrandt et al . demonstrated that terms up to @xmath165 should be kept for convergence @xcite . tensor
- observables should be more susceptible to this difference .
babusci et al .
@xcite were the first to identify a complete set of independent observables for compton scattering , namely for a spin-@xmath21 target .
their classification applies of course also to a scalar- or vector - polarised deuteron target , _ provided _ one sets the tensor - component to zero , @xmath166 .
following the discussion in sec .
[ sec : kinematics ] , this constrains @xmath167 , so the vector polarisation can not reach the maximal value of @xmath0 allowed for a spin-@xmath168 target . while one should be aware of this difference , we choose in the following to quote results with a pretense value `` @xmath169 '' .
those for a deuteron target which is maximally vector polarised but not tensor polarised are obtained from these by multiplying the right - hand sides of eqs . to by @xmath170 .
experimentally , these observables are measured as asymmetries between cross sections with different target and beam polarisation angles @xmath171 , normalised to their sum . the configurations
are chosen such that their cross sections sum to twice the total unpolarised cross section .
specifically , the beam asymmetry in refs .
@xcite is the difference of the cross sections of a linearly polarised beam ( @xmath172 , @xmath173 ) either in the scattering plane ( @xmath174 ) or perpendicular to it ( @xmath175 ) on an unpolarised target ( @xmath176 ) , normalised to their sum . inserting these choices into eq
. identifies @xmath177 for readability , the differential cross section symbol is dropped in each term in the second line , and an abbreviation `` @xmath178 '' is introduced for a denominator which is the sum , rather than the difference , of the terms in the numerator .
not surprisingly , all definitions of the beam asymmetry coincide .
the vector target asymmetry with unpolarised beam ( @xmath179 , @xmath180 ) translates as @xmath181 the vector target asymmetries with right - circularly polarised beam ( @xmath179 , @xmath182 , @xmath183 ) as @xmath184 and finally those with linearly polarised beam on a vector target ( @xmath179 , @xmath172 , @xmath185 ) as @xmath186- [ ( \frac{\pi}{2},-\frac{\pi}{2};0)- ( \frac{\pi}{2},-\frac{\pi}{2};\frac{\pi}{2 } ) ] } { [ \;.\;+\;.\;]\;+\;[\;.\;+\;.\ ; ] } { \nonumber}\\ & = & -\frac{1}{\sqrt{2}}\ ; \left(t_{11}^\text{lin}+t_{1,-1}^\text{lin}\right)\;\;,\end{aligned}\ ] ] or @xmath187 in @xmath188 , babusci et al .
flip the circular beam polarisation . due to parity symmetry , this is equivalent to the flip of the target polarisation performed above .
these authors define observables in analogy to those introduced by babusci et al .
however , the deuteron is taken to be prepared such that only the magnetic quantum numbers @xmath189 contribute , in the direction @xmath62 in which the density matrix is diagonal . to understand why this difference may lead to confusion ,
consider the single - polarisation observable for scattering an unpolarised ( or circularly polarised ) beam on a deuteron target which is polarised in a pure @xmath190 state perpendicular to the scattering plane ( i.e. parallel or anti - parallel to the @xmath68 axis ) : @xmath191 where the same abbreviation as in eq .
this appears to be the natural application of @xmath192 , eq .
, to the deuteron .
since the deuteron polarisation is flipped in the difference , the numerator should describe a vector - polarised deuteron . according to eq .
, a pure state @xmath193 is described by @xmath194 and @xmath195 . for this observable
, @xmath62 is parallel to the @xmath68-axis , so that @xmath190 corresponds to @xmath196 . with @xmath157 and the same abbreviations as before ,
the numerator becomes : @xmath197 tensor - observables do indeed not contribute . in contradistinction
, the denominator reads : @xmath198\left .
\frac{{\mathrm{d}}\sigma}{{\mathrm{d}}\omega}\right|_\text{unpol}\;\;.\end{aligned}\ ] ] it is no more proportional to the unpolarised cross section since the @xmath199-term is absent , as noted already in refs . @xcite . like in @xmath192 of eq .
, the resulting asymmetry @xmath200 is proportional to @xmath117 , but the prefactor has changed and now depends in addition on the tensor - polarised observables @xmath201 . while the same symbol is used for the vector target polarisation in ref .
@xcite and for that of ref .
@xcite , eq .
, the two are actually different : @xmath202 it is for that reason that the apparent notational degeneracy is lifted throughout this article by including an explicit reference superscript . translating the other observables of refs .
@xcite is now straightforward .
asymmetries with unpolarised targets are of course identical , eq . . since ref .
@xcite considers both differences of polarised cross sections ( denoted by @xmath203 ) and their asymmetries @xmath204 , both are also recorded in the following .
one finds with @xmath194 , @xmath195 , @xmath182 and @xmath205 for the asymmetries built in analogy to @xmath206 : @xmath207^\text{\cite{chen:2004wwa}}= ( m_{iz}=+1;1)-(m_{iz}=-1;1)=({\ensuremath{\vartheta_\text{d}}}=0)-({\ensuremath{\vartheta_\text{d}}}=\pi){\nonumber}\\ & = & \sqrt{6 } \;t_{10}^\text{circ } \left.\frac{{\mathrm{d}}\sigma}{{\mathrm{d}}\omega}\right|_\text{unpol}\\ \label{eq : sigmazcircwrong } \sigma^\text{circ~\cite{griesshammer:2010pz}}_z&\equiv&\sigma_z^\text{\cite{choudhury:2004yz , shuklathesis}}\equiv-\sigma_z^\text{\cite{chen:2004wwa}}= \frac{\delta_z^\text{circ~\cite{griesshammer:2010pz}}}{.\;+\;.}= \frac{\sqrt{3}\;t_{10}^\text{circ}}{\sqrt{2}+t_{20}}\end{aligned}\ ] ] in no case is the denominator just proportional to the unpolarised cross section ; instead , it also depends on @xmath208 .
it should be noted that ref .
@xcite provides formulae for the denominators of @xmath209 which depend only on the scalar and vector parts of the target polarisation .
these results could not be reproduced .
the following additional cross section differences and asymmetries for linearly polarised beam on a polarised deuteron target were described in ref .
@xcite : @xmath210\left.\frac{{\mathrm{d}}\sigma}{{\mathrm{d}}\omega}\right|_\text{unpol}\\ \label{eq : sigmaxlinwrong } \sigma_x^\text{lin~\cite{griesshammer:2010pz}}&= & \frac{\delta_x^\text{lin~\cite{griesshammer:2010pz}}}{.\;+\;. } = \frac{4\;\sigma^\text{lin}+ \sqrt{3}\left(t_{22}^\text{lin}+t_{2,-2}^\text{lin}\right ) -\sqrt{2}\;t_{20}^\text{lin } } { 4+\sqrt{3}\;t_{22}-\sqrt{2}\;t_{20}}\\ \label{eq : deltazlinwrong } \delta^\text{lin~\cite{griesshammer:2010pz}}_z&= & ( m_{iz}=1;{\ensuremath{\varphi_\text{lin}}}=0)-(m_{iz}=1;{\ensuremath{\varphi_\text{lin}}}=\frac{\pi}{2})= ( { \ensuremath{\vartheta_\text{d}}}=0;{\ensuremath{\varphi_\text{lin}}}=0)-(0;\frac{\pi}{2}){\nonumber}\\ & = & \left[2\;\sigma^\text{lin}+\sqrt{2}\;t^\text{lin}_{20}\right ] \left.\frac{{\mathrm{d}}\sigma}{{\mathrm{d}}\omega}\right|_\text{unpol}\\ \label{eq : sigmazlinwrong } \sigma_z^\text{lin~\cite{griesshammer:2010pz}}&= & \frac{\delta_z^\text{lin~\cite{griesshammer:2010pz}}}{.\;+\;. } = \frac{2\;\sigma^\text{lin}+\sqrt{2}\;t^\text{lin}_{20}}{2+\sqrt{2}\;t_{20}}\end{aligned}\ ] ] with @xmath194 , @xmath195 , @xmath173 and @xmath172 . notice that the numerators @xmath211 depend on different and nontrivial combinations of both @xmath116 and @xmath212 .
@xmath213 and @xmath214 would be identical if the tensor - polarised observables were zero .
the additional terms proportional to @xmath201 in each denominator of @xmath215 and @xmath216 will by themselves turn out to be rather large , sensitive to the polarisabilities , and significantly dependent on photon energy and scattering angle ; see figs .
[ fig : comparets ] , [ fig : t22varied ] and [ fig : t20varied ] in sec . [ sec : results ] . without this input ,
no simple conclusions can thus be drawn how the sensitivity of @xmath215 on the polarisabilities translates into the sensitivity of its numerator alone .
on the other hand , @xmath217 is dominated by @xmath116 and @xmath218 since @xmath219 will turn out to be very small .
[ sec : observables ] the following sub - sections explore the sensitivity of the @xmath1 independent observables to the scalar and spin dipole polarisabilities in @xmath43eft .
since this version of the deuteron compton scattering amplitudes is described comprehensively in previous publications @xcite and summarised in a recent review @xcite , its main ingredients are only sketched here . in @xmath43eftwith
explicit @xmath4 degrees of freedom , four typical low - energy scales are found in deuteron compton scattering : the pion mass @xmath220 as the typical chiral scale ; the delta - nucleon mass splitting @xmath221 ; the deuteron binding momentum ( inverse deuteron size ) @xmath222 as the typical scale of the bound @xmath223 system ; and the photon energy @xmath30 . when measured in units of a natural `` high '' scale @xmath224
at which @xmath43eftwith explicit @xmath4 degrees of freedom can be expected to break down because new degrees of freedom become dynamical , each gives rise to a small , dimensionless expansion parameter .
typical values of @xmath225 are the masses of the @xmath30 and @xmath226 as the next - lightest exchange mesons ( about @xmath227 ) . to avoid a fourfold expansion , it is convenient to approximately identify some scales so that only one dimensionless parameter is left . in the @xmath228-expansion of pascalutsa and phillips @xcite , one chooses @xmath229 i.e. numerically @xmath230 .
the identity is exact for @xmath231 .
since present experiments are run at @xmath232 , the nonzero delta - width is not tested , cf .
@xcite .
the two - nucleon dynamics adds the momentum scale @xmath17 of the shallow bound state .
based on refs .
@xcite , chapter 5 of ref .
@xcite provides a `` unified '' deuteron compton amplitude which is complete at order @xmath5 and valid from zero photon energy to just below the pion production threshold , @xmath233 .
this variant is identical to @xmath234 in the `` small scale expansion '' @xcite , used in ref .
@xcite . at this order ,
the compton scattering kernel consists of `` one - nucleon contributions '' in which both photons interact with the same nucleon ( fig .
[ fig : onenucleon ] ) , and `` two - nucleon contributions '' ( fig .
[ fig : twonucleon ] ) .
the latter consists of two classes , each of which contributes at the order @xmath235 of the present formulation : eftup to order @xmath5 ( permuted and crossed diagrams not shown ) .
photons couple to the same pion ( a ) ; rescattering contributions ( b , c ) .
ellipse : two - nucleon @xmath236-matrix ; dot : coupling via minimal substitution or magnetic moment ; from ref .
@xcite.,scaledwidth=82.0% ] 1 . both photons couple to the charged pion - exchange current , fig . [ fig : twonucleon ] ( a ) @xcite .
each photon couples to the nucleon charge , magnetic moment and/or to different pion - exchange currents ; fig .
[ fig : twonucleon ] ( b ) and ( c ) . between the two couplings ,
the nucleons rescatter arbitrarily often via the full nn s - matrix ( including no rescattering at all ) .
these contributions are small for @xmath237 but required for @xmath238 in order to restore the exact low - energy theorem of compton scattering , i.e. the thomson limit @xcite . at zero energy , its emergence in the @xmath43eft power - counting mandates that the contribution of fig .
[ fig : twonucleon ] ( b ) must be exactly minus half that of the one - nucleon thomson term , fig .
[ fig : onenucleon ] ( a ) , and that the pion - exchange contributions of fig .
[ fig : twonucleon ] ( a ) and ( c ) must add to zero @xcite .
such stringent numerical tests are fulfilled to better than @xmath239 . at higher energies
, the significance of this cancellation belies in a considerable reduction of the dependence of the amplitudes on the deuteron wave function and @xmath223 potential @xcite .
the one - nucleon sector is formed by : eftup to @xmath235 ( permuted and crossed diagrams not shown ) .
top : embedding into the deuteron .
bottom : one - nucleon thomson term ( a ) ; pion cloud of the nucleon ( b ) and @xmath4 ( double line ; ( c ) ) ; excitation of an intermediate @xmath240 ( d ) ; short - distance effects to @xmath241 and @xmath242 ( e ) ; from ref .
@xcite.,scaledwidth=82.0% ] 1 . single - nucleon thomson scattering , fig .
[ fig : onenucleon ] ( a ) , is the leading - order term , @xmath243 .
2 . coupling to the chiral dynamics of the single - nucleon pion cloud , fig .
[ fig : onenucleon ] ( b ) , @xmath244 .
excitation of the @xmath4 intermediate state , fig .
[ fig : onenucleon ] ( d ) , and coupling to the pion cloud around it , fig .
[ fig : onenucleon ] ( c ) , each contributing at @xmath235 for @xmath233 .
following ref .
@xcite , the @xmath240 is treated non - relativistically and with zero width , using @xmath245 , @xmath246 and the non - relativistic version of the @xmath247 @xmath248-coupling @xmath249 , obtained from converting the relativistic value of @xmath250 .
this value , in turn , is found by fitting the single - nucleon amplitudes to the data above @xmath251 in the proton compton database established there .
4 . two energy - independent , isoscalar short - distance coefficients , fig .
[ fig : onenucleon ] ( e ) , which encode those contributions to the nucleon polarisabilities @xmath241 and @xmath242 which arise at this order neither from pions nor from the @xmath4
. since they are formally of one order higher , @xmath252 , the order of the resulting total amplitude is called `` modified @xmath235 '' .
while these `` off - sets '' for the static polarisabilities are determined by data , the energy- and isospin - dependence of the spin - independent polarisabilities are at this order predicted in @xmath43eft . here , their values are taken from the determination in ref .
@xcite ; see eq . .
nucleon polarisabilities arise solely from terms ( 2 ) to ( 4 ) . in this power - counting ,
`` switching off '' @xmath4 contributions is equivalent to a calculation at one lower order , @xmath244 , in which the scalar polarisabilities are parameter - free predictions : @xmath253 @xcite .
these kernels are convoluted with deuteron wave functions to obtain the amplitudes @xmath254 of eq . .
results in this article are obtained with the @xmath43eftdeuteron wave function at n@xmath255lo ( cutoff @xmath256 ) in the implementation of epelbaum et al .
@xcite and the av18 potential @xcite for nn rescattering .
this combination provides an adequate @xmath43eftrepresentation of the two - nucleon system ; see discussion in ref .
@xcite and sec .
[ sec : moredependences ] below .
this formulation differs from the previous ones of refs .
@xcite in some numerical improvements , a new parameter set @xmath257 for the @xmath4 from the breit - wigner parameters and the proton compton data , and in slightly changed numbers for the isoscalar , scalar polarisabilities . in a fully consistent eft calculation ,
the kernel , wave functions and potential should of course be derived in the same framework .
this is work in progress . at this
( modified ) order @xmath5 , the static isoscalar dipole polarisabilities are ( with theoretical uncertainties of about @xmath258 from higher - order contributions and in the canonical units of @xmath48 for the scalar polarisabilities and @xmath49 for the spin - dependent ones ) @xcite : @xmath259 i.e. @xmath260 , @xmath261 , which is not incompatible with those of other approaches , see eq . .
the values for the spin polarisabilities differ slightly from these quoted in refs .
@xcite because of the updates to the @xmath262 amplitudes described in sect .
5.3 of ref .
@xcite .
the convergence of the spin polarisabilities from @xmath244 via the @xmath235 values quoted above to the values at @xmath263 is complicated ; see table 4.2 of ref .
therefore , no theoretical uncertainty is assigned for now . since the deuteron is an isoscalar ,
only average nucleon polarisabilities are accessible in elastic deuteron compton scattering . in order to analyse the sensitivity of each observable
, one varies each dipole polarisability about the static central value by adding the parameters @xmath264 , @xmath265 , @xmath266 , @xmath267 , @xmath268 and @xmath269 to the interactions of the single - nucleon sub - system , eq .
their contribution to the amplitudes in the @xmath17n cm system is @xmath270 \,(\vec{\epsilon}^\prime\cdot\vec{\epsilon } ) -\delta\beta_{m1}\ , ( \vec{\epsilon}^\prime\cdot\hat{k})\,(\vec{\epsilon}\cdot\hat{k}^\prime)\nonumber\\ & & -{\mathrm{i}}\,[\delta\gamma_{e1e1}+z\,\delta\gamma_{m1m1}+\delta\gamma_{e1m2 } + z\,\delta\gamma_{m1e2}]\,\omega\,\vec{\sigma}\cdot ( \vec{\epsilon}^\prime\times\vec{\epsilon})\nonumber\\ & & + { \mathrm{i}}\,[\delta\gamma_{m1e2}-\delta\gamma_{m1m1}]\,\omega\,\vec{\sigma}\cdot \left(\hat{k}^\prime\times\hat{k}\right ) ( \vec{\epsilon}^\prime\cdot\vec{\epsilon})\nonumber\\ & & + { \mathrm{i}}\,\delta\gamma_{m1m1}\,\omega\,\vec{\sigma}\cdot \left[\left(\vec{\epsilon}^\prime\times\hat{k}\right ) ( \vec{\epsilon}\cdot\hat{k}^\prime)- \left(\vec{\epsilon}\times\hat{k}^\prime\right ) ( \vec{\epsilon}^\prime\cdot\hat{k})\right]\nonumber\\ & & + { \mathrm{i}}\,\delta\gamma_{e1m2}\,\omega\,\vec{\sigma}\cdot \left[\left(\vec{\epsilon}^\prime\times\hat{k}^\prime\right ) ( \vec{\epsilon}\cdot\hat{k}^\prime)- \left(\vec{\epsilon}\times\hat{k}\right ) ( \vec{\epsilon}^\prime\cdot\hat{k})\right ] \bigg]\;\;. \label{eq : fit}\end{aligned}\ ] ] these variables may be considered as parametrising the difference between predicted and ( so - far un - measured ) experimental static values of the polarisabilities , under the assumption that the energy - dependence from the pion - cloud and @xmath4 is correctly predicted in @xmath43eft .
alternatively , one can view them as parametrising deviations from the order-@xmath271 @xmath43eftamplitudes at fixed nonzero energy , including the theoretical uncertainties of higher - order effects . in that case , the deviations themselves could be seen as energy - dependent .
such an approach forms the basis of a multipole analysis of deuteron compton scattering advocated in refs .
determining the six dipole polarisabilities is then in principle reduced to a multipole - analysis of @xmath272 high - accuracy scattering experiments .
the variation of the isoscalar values by @xmath273 canonical units is chosen since it is roughly at the level of the combined statistical , theoretical and baldin - sum - rule induced error for @xmath241 and @xmath242 . with quadratic contributions of the polarisabilities @xmath274 suppressed in the squared amplitudes , variations by other amounts are easily linearly extrapolated . in practise
, the scalar polarisabilities of the proton are constrained to better than @xmath273 , so that deuteron compton scattering experiments are more likely focused on extracting neutron polarisabilities . in that case , these studies can be interpreted as providing the sensitivities on varying the neutron polarisabilities by @xmath275 units , with fixed proton values .
the spin polarisabilities are however less well known ; besides the constraints of eq . , no experimental information has been published thus far , and theoretical descriptions easily disagree by as much as @xmath121 units @xcite .
for example , a recent determination of the scalar dipole polarisabilities of the proton included varying one of the spin polarisabilities to @xmath276 , which combined with its theoretical accuracy would by itself already suggest a variation by about @xmath121 units .
amplitudes from scalar polarisabilities scale like @xmath277 , while those containing spin polarisabilities scale like @xmath278 ; see eq . .
ideally , one can therefore perform high - accuracy experiments at relatively low energies , @xmath279 , to better determine @xmath241 and @xmath242 and constrain high - energy predictions .
the spin polarisabilities are then extracted at @xmath280 , as already advocated in ref .
the observables considered here follow this pattern .
additionally , one should address : 1 . the baldin sum rule constraint , eq .
however , its independent test by better data at forward angles would be expedient .
2 . weaker constraints for the forward and backward spin polarisabilities , eq . .
these come with considerable theoretical and systematic uncertainties .
logistic constraints like detector placement and available beam energies , as well as detector and polarisation efficiencies .
all these must be taken into account to determine which experiments have the potential for the greatest sensitivity on a given polarisability and of the greatest impact in the network of data already available .
considering asymmetries removes many systematic experimental uncertainties , but the corresponding count rates are necessary for beam - time estimates and follow from multiplying with the unpolarised cross section , cf . .
in general , asymmetries are by @xmath281% less sensitive to variations of the polarisabilities than the corresponding count rates .
sometimes , sensitivity to the nucleon structure is even lost entirely , while an enhancement appears in no case .
it is the purview of our experimental colleagues to determine to what extent such draw - backs outweigh the benefits of measuring asymmetries instead of cross - section differences . to present all 17 asymmetries and their rates , plus the unpolarised cross section , depending on @xmath282 dipole polarisabilities and @xmath121 kinematic variables ( photon energy @xmath30 and scattering angle @xmath61 ) in the cm and lab frame , plus additional theoretical uncertainties and both theoretical and experimental constraints , far exceeds what can adequately be conveyed in an article
, the focus is therefore on some prominent examples . in order to facilitate planning and analysis of experiments ,
the results of all observables are available as an interactive _ mathematica 9.0 _ notebook from ` [email protected] ` .
it contains both tables and plots of energy- and angle - dependencies of the cross - sections , rates and asymmetries from @xmath283 to about @xmath284 mev , in both the cm and lab systems , including sensitivities to varying the scalar and spin polarisabilities independently as well as subject to the baldin sum rule and other constraints .
since it considers all observables with polarised beams and/or targets , it supersedes ref .
@xcite which only dealt with some observables , built in analogy to the babusci - classification ; see sec . [
sec : shuklaobs ] . figure [ fig : screenshot ] shows a sample screen - shot of a cross - section difference with user - defined beam and target polarisations .
it is finally worth re - emphasising that the purpose of this study is to establish _ relative sensitivities _ of compton scattering observables on _ varying _ the polarisabilities @xcite .
credible predictions of their absolute magnitudes are only meaningful when all systematic uncertainties are properly propagated into observables .
such errors include : theoretical uncertainties from discarding contributions in @xmath43eftwhich are higher than order @xmath5 , like including effects of the @xmath4 width and parameter uncertainties ; uncertainties in the data and in the baldin sum rule , eq .
; and to a lesser extend residual dependence on the deuteron wave - function and nn potential used , as well as numerical uncertainties . figures [ fig : crosssectionvaried ] to [ fig : tlin2 - 2varied ] present the @xmath43eftresults of an @xmath235-calculation , with dynamical @xmath4 and nn - rescattering .
let us concentrate on the sensitivity to the polarisabilities at one representative energy in the ( experimentally most relevant ) lab - frame . with an eye on parameters at hi@xmath17s , maxlab , mami and possible future high - luminosity accelerators like mesa @xcite ,
a beam energy of @xmath285 seems appropriate . staying below the pion - production threshold avoids experimental and theoretical complications .
since the asymmetries differ by @xmath118 orders of magnitude , one should keep in mind changes of scale between plots of different observables . comparing them
is simplified by plots of @xmath119 , @xmath122 , @xmath124 , @xmath126 and @xmath128 , each for the different non - trivial values of @xmath286 at @xmath285 . with magnitudes of up to @xmath287 ,
the largest asymmetries are @xmath116 , @xmath288 and @xmath218 , followed by magnitudes on the order of @xmath289 for @xmath290 , @xmath291 , @xmath292 and @xmath293 .
the order of magnitude of @xmath124 , @xmath294 and @xmath295 is @xmath296 , and that of @xmath297 and @xmath298 even @xmath299 , providing considerable experimental challenges .
the observables @xmath300 show a clear hierarchy , with sizes increasing substantially towards the most negative @xmath286-values at given @xmath301 .
the top panel of each single - observable plot , figs .
[ fig : crosssectionvaried ] to [ fig : t11varied ] and [ fig : t22varied ] to [ fig : tlin2 - 2varied ] , shows the energy - dependence of each observable at four scattering angles @xmath302 . in each case , the deuteron breakup point at @xmath303 is clearly visible . only @xmath304 and @xmath298 significantly decrease with increasing photon energy , but @xmath305 and @xmath306 change sign around @xmath307 .
all observables which are zero below the first threshold , eq . , grow rapidly above it in the case of @xmath117 and @xmath308 even to @xmath309 at @xmath6 .
sensitivity on the nucleon polarisabilities grows as expected with increasing photon energy . in the lower panels of figs .
[ fig : crosssectionvaried ] to [ fig : t11varied ] and [ fig : t22varied ] to [ fig : tlin2 - 2varied ] , two plots show the sensitivity to @xmath241 and the combination @xmath310 when the baldin sum rule constraint is used .
this of course also allows one to assess where variations of @xmath242 are ( anti-)correlated to those of @xmath241 . the other 4 panels describe variations of the spin polarisabilities , without imposing additional constraints .
within one observable , all sensitivities are of course plotted on the same scale
. plots of the unpolarised cross section , fig .
[ fig : crosssectionvaried ] , are included for quick rate - estimates .
its overall size is dramatically affected by a variation of @xmath241 , its backward angles by that of @xmath310 , and there is only minor sensitivity on the spin polarisabilities .
the beam asymmetry @xmath116 shows a mildly different angular dependence on @xmath241 and @xmath242 , possibly allowing for extractions .
that sensitivity to the other polarisabilities is small , had already been demonstrated in a @xmath43eftvariant without dynamical @xmath4 in refs .
delta - effects affect this variable only minimally . in a future world of high - accuracy experiments with well - controlled systematic experimental uncertainties , high luminosities and @xmath18 beam and target polarisations , an ideal observable should be very sensitive to one polarisability , while being near - insensitive to all others . for @xmath241 , this singles out @xmath117 ( fig .
[ fig : t11varied ] ) , @xmath308 ( fig . [ fig : tlin1 - 1varied ] ) and @xmath218 ( fig .
[ fig : tlin2 - 2varied ] ) ; for @xmath311 , @xmath291 ( fig .
[ fig : tcirc11varied ] ) .
when one takes @xmath241 and @xmath242 to be know sufficiently well that the influence on varying them can be neglected , then @xmath291 ( fig .
[ fig : tcirc11varied ] ) , @xmath312 ( figs . [
fig : tcirc22varied ] and [ fig : tcirc21varied ] ) and @xmath294 ( fig .
[ fig : tlin10varied ] ) are dominated by sensitivity to @xmath311 only .
curiously , @xmath297 ( fig . [ fig : tlin11varied ] ) is near - exclusively sensitive to the mixed spin polarisability @xmath9 , and both @xmath298 ( fig .
[ fig : tlin22varied ] ) and @xmath295 ( fig .
[ fig : tlin21varied ] ) to its partner @xmath8 albeit all three are very small .
alternatively , different angular dependencies can be used to dis - entangle two polarisabilities from the same observable ; see e.g. @xmath313 for @xmath311 and @xmath9 ( fig . [ fig : t21varied ] ) and to a lesser extend
@xmath306 for @xmath314 and @xmath8 ( fig .
[ fig : tlin20varied ] ) . keeping in mind that none of the tensor observables have an analogue in compton scattering off the nucleon , such an augmentation is absent in the one - nucleon case .
it appears that mixed polarisabilities are much better accessible in scattering from the deuteron .
the photon quadrupole coupling to one nucleon ( @xmath315 in @xmath8 and @xmath316 in @xmath9 ) seems to be enhanced by the @xmath10 wave components of the deuteron wave function and pion - exchange current , fig .
[ fig : twonucleon ] ( a ) and ( c ) .
one may thus speculate that determinations of @xmath8 and @xmath9 will first appear from deuteron data if the necessary accuracy can be reached for these small asymmetries .
references @xcite have argued in detail that sensitivity to a specific polarisability can be maximised or switched off by considering particular target - beam combinations at particular angles . to that end , one either maximises the scalar products between photon polarisations @xmath317 , @xmath318 , photon momenta @xmath69 , @xmath70 and nucleon spin @xmath319 , or one chooses some vectors to be orthogonal or parallel , rendering the associated ( scalar or vector ) products zero .
many of these `` zero sensitivity points '' are preserved when the relative motion of the @xmath320 cm system inside the deuteron is taken into account .
in some cases , the deuteron effect lifts the zero , but only barely , since the nucleons are predominantly in a relative @xmath236 wave , while @xmath10 wave contributions ( also from pion - exchange currents , fig .
[ fig : twonucleon ] ( a / c ) ) are suppressed .
relativistic boost effects are small at the energies considered @xcite .
examples include the following insensitivities ( angles in cm frame ) : @xmath321 to @xmath242 at @xmath322 ; @xmath323 to @xmath311 at @xmath324 and to @xmath8 at @xmath325 ; @xmath313 to @xmath8 and @xmath9 at @xmath322 ; @xmath326 to @xmath311 at @xmath322 ; and @xmath327 to @xmath314 at @xmath322 .
a good example of undesired correlations between variations of different polarisabilities is @xmath304 ( fig .
[ fig : tcirc22varied ] ) , where angular dependencies and magnitudes of changing @xmath241 and @xmath311 are near - identical .
@xmath288 ( fig .
[ fig : tcirc10varied ] ) is near - equally sensitive to all dipole polarisabilities . applying these criteria and assuming that @xmath241 and @xmath242 are known , the following observables could therefore provide an experimentally realistic but challenging complete set from which to cleanly determine the isoscalar spin polarisabilities : @xmath291 for @xmath311 ( variation by @xmath273 translates into @xmath328 of an asymmetry magnitude of about @xmath329 ) , followed by angular dependence of @xmath306 ( @xmath330 of mag .
@xmath331 ) for @xmath314 , followed by @xmath332 ( @xmath328 of mag .
@xmath333 ) for @xmath9 and check on @xmath314 , plus @xmath295 ( @xmath330 of mag .
@xmath334 ) for @xmath8 .
the different angular dependencies of @xmath313 ( up to @xmath335 of mag .
@xmath331 ) can serve as valuable check . as hinted above
, reliable theoretical predictions should include a study of residual theoretical uncertainties .
the aforementioned _ mathematica _ notebook therefore explores the influence of nn rescattering , of the dynamical @xmath4 , and of the particular two - nucleon interaction used .
the results mostly confirm those of refs .
@xcite and thus are only summarised here .
rescattering significantly affects all observables for energies @xmath336 and is important to reduce residual dependence on the nn potential and deuteron wave function up to @xmath11 , as predicted by the power - counting .
details of the nn potential or deuteron wave function are not reflected in observables .
for example , at @xmath6 , the largest wave - function dependencies are @xmath337 of the maximum in @xmath304 and @xmath338 of the maximum in @xmath306 .
these observables are however quite small ( @xmath339 ) ; all other observables suffer from a residual wave - function dependence of @xmath340 at that energy , as tests with av18 @xcite , nijmegen 93 @xcite and other wave functions demonstrate .
not surprising is also that @xmath4-effects become more pronounced with increasing energy .
it is now well - understood that its spin - flip amplitude considerably changes the shape of the unpolarised differential cross section at backward angles @xcite , thereby solving the `` sal puzzle '' of deuteron compton data at @xmath341 @xcite .
while the influence of the delta on some observables like @xmath116 may be very small , it is hard to imagine an eft without it to be reliable at photon energies around @xmath6 . as case in point , @xmath306 is at @xmath6 increased by @xmath342 and changes shape when the delta is included ; @xmath288 increases by @xmath164 , while @xmath313 is reduced by @xmath343 , and @xmath298 even by @xmath342 .
@xmath323 changes shape at forward angles .
delta effects can not be neglected above about @xmath344 , especially in the large momentum transfers at back - angles .
[ sec : conclusions ] based on a well - known decomposition of the deuteron photo - dissociation cross section , this work presented a classification of all @xmath1 independent observables in compton scattering off an unpolarised , vector , tensor or mixed - polarised spin-@xmath0 target with unpolarised , circularly , linearly or mixed - polarised beam when final - state polarisations are not detected .
the unpolarised cross section , beam asymmetry , @xmath2 target asymmetries and @xmath3 double asymmetries were expressed in terms of the helicity amplitudes and related to previously used , incomplete parametrisations .
this decomposition is particularly transparent , with each observable readily translated into specific and well - known beam / target / detector combinations .
the method was then applied to deuteron compton scattering in @xmath43eftwith dynamical @xmath4 degrees of freedom using amplitudes which are complete at order @xmath5 in the energy range from the thomson limit to just below the pion production threshold .
since this process tests the isoscalar two - photon response of the nucleon , embedded in the simplest bound few - nucleon system @xcite , the sensitivity of each observable on the @xmath282 dipole polarisabilities of the nucleon was studied . these , in turn , encode information on the symmetries and strengths of the interactions with and between the hadronic internal low - energy degrees of freedom .
they characterise the radiation multipoles which are generated by displacing the charges and currents inside the nucleon in the electric or magnetic field of a photon with definite energy and multipolarity . to determine in particular the @xmath2 spin polarisabilities is the objective of a large - scale effort including hi@xmath17s , max - lab and mami , since they parametrise the response of the nucleon spin degrees of freedom but are not yet well - constrained .
this study thus aids in planning and analysing experiments to determine the nucleon polarisabilities from deuteron compton scattering .
an interactive _ mathematica 9.0 _ notebook of its results over a wide range of energies is available from ` [email protected] ` . with future high - accuracy determinations of the scalar polarisabilities @xmath241 and @xmath242 at lower energies , the spin polarisabilities seem to be reliably extractable at energies of @xmath280 from the observables @xmath291 ( circularly - polarised beam on vector target ) , @xmath345 ( unpolarised beam on tensor target ) and @xmath346 ( linearly - polarised beam on tensor target ) .
this experimentally challenging but realistic set consists of asymmetries which have maxima from @xmath329 to @xmath331 and are mostly sensitive to only 1 or 2 polarisabilities .
modifying the spin polarisabilities by @xmath347 in them induces variations of @xmath328 to @xmath335 at @xmath6 . since nuclear binding
is mediated by charged pion - exchange currents to which the photons can couple , deuteron compton scattering concurrently tests the detailed symmetries and dynamics of the charged part of the two - nucleon interaction .
the @xmath10 wave contributions of the deuteron wave function and of the pion - exchange currents lead to nonzero tensor observables . by interference with the quadrupole components of the incident and outgoing photon , these ,
in turn , seem to be much more sensitive on the mixed spin polarisabilities @xmath8 and @xmath9 than any single - nucleon observable .
one may thus speculate that their determination will first appear from deuteron data if the necessary accuracy can be reached .
ongoing work includes embedding the @xmath263 single - nucleon amplitudes of ref .
@xcite for an extension to photon energies above the pion - production threshold , also with @xmath240-ful pion - exchange currents ; inclusion of a chirally consistent nn potential ; and a detailed assessment of theoretical uncertainties . in support of ongoing and
planned experiments at hi@xmath17s , max - lab and mami , this effort is pursued in the context of a comprehensive theoretical description of compton scattering on the proton , deuteron and @xmath348hein @xmath43eft , valid from zero photon energy well into the @xmath240 resonance region .
as pendant to the present article , a classification of the independent polarisation transfer observables on a spin-@xmath0 target will determine those @xmath7 which are linearly independent and complement those presented here for the complete set of @xmath26 independent observables @xcite .
from these , the @xmath26 independent real amplitudes can be reconstructed in turn , and hence all information accessible in the two - photon response of the deuteron and its constituents . finally , i offer to embed single - nucleon compton amplitudes , chiral or not , into the available deuteron code , so that other theoretical descriptions can be tested collaboratively .
i am particularly indebted to my compton collaborators j. a. mcgovern and d. r. phillips for discussions and encouragement , and to the organisers and participants of the int workshop 12 - 3 : `` light nuclei from first principles '' as well as the `` workshop to explore physics opportunities with intense , polarized electron beams with energy up to 300 mev '' at the mit , both of which also provided financial support . d. g. crabb and w. meyer shared their expertise on tensor - polarised targets , and r. schiavilla the crucial hint to ref .
discussions with w. briscoe , e. downie , g. feldman and h. weller helped clarify various experimental points .
this work was supported in part by the national science foundation under career award phy-0645498 , by the us - department of energy under contract de - fg02 - 95er-40907 , by the deutsche forschungsgemeinschaft and the national natural science foundation of china through funds provided to the sino - german crc 110 `` symmetries and the emergence of structure in qcd '' , and by the epos network of the european community research infrastructure integrating activity `` study of strongly interacting matter '' ( hadronphysics3 ) .
h. w. griesshammer , j. a. mcgovern , d. r. phillips and g. feldman , prog .
* 67 * ( 2012 )
841 [ arxiv:1203.6834 [ nucl - th ] ] . h. paetz gen .
schieck , `` nuclear physics with polarized particles '' , lecture notes in physics 842 , springer 2012 .
d. babusci , g. giordano , a. i. lvov , g. matone and a. m. nathan , phys .
c * 58 * ( 1998 ) 1013 [ arxiv : hep - ph/9803347 ] .
j. w. chen , x. d. ji and y. c. li , phys .
c * 71 * ( 2005 ) 044321 [ arxiv : nucl - th/0408004 ] .
j. -w .
chen , x. -d .
ji and y. -c .
li , phys .
b * 620 * ( 2005 ) 33 [ arxiv : nucl - th/0408003 ] .
d. choudhury and d. r. phillips , phys .
c * 71 * ( 2005 ) 044002 [ arxiv : nucl - th/0411001 ] .
d. choudhury , ph.d .
thesis ohio university ( 2006 ) .
h. w. griehammer and d. shukla , eur .
j. a * 46 * ( 2010 ) 249 [ arxiv:1006.4849 [ nucl - th ] ] ; erratum ibid .
* 48 * ( 2012 ) , 76 [ arxiv:1006.4849 [ nucl - th ] ] .
j. w. chen , nucl .
phys . a * 653 * ( 1999 ) 375 [ arxiv : nucl - th/9810021 ] .
j. karakowski , arxiv : nucl - th/9901011 . j. j. karakowski and g. a. miller , phys .
c * 60 * ( 1999 ) 014001 [ arxiv : nucl - th/9901018 ] .
h. arenhovel and m. sanzone , few body syst .
suppl . * 3 * ( 1991 ) 1 . h. w. griehammer , in preparation .
h. w. griehammer and t. r. hemmert , phys .
c * 65 * ( 2002 ) 045207 [ arxiv : nucl - th/0110006 ] .
r. p. hildebrandt , h. w. griehammer , t. r. hemmert and b. pasquini , eur .
phys . j. a * 20 * ( 2004 ) 293 [ arxiv : nucl - th/0307070 ] . b. r. holstein , arxiv : hep - ph/0010129 .
w. detmold , b. c. tiburzi and a. walker - loud , phys .
d * 81 * ( 2010 ) 054502 [ arxiv:1001.1131 [ hep - lat ] ] .
m. engelhardt , pos lattice * 2011 * ( 2011 ) 153 .
[ arxiv:1111.3686 [ hep - lat ] ] .
m. lujan , a. alexandru and f. lee , pos lattice * 2011 * ( 2011 ) 165 .
[ arxiv:1111.6288 [ hep - lat ] ] .
a. walker - loud , c. e. carlson and g. a. miller , phys .
* 108 * ( 2012 ) 232301 [ arxiv:1203.0254 [ nucl - th ] ] .
k. pachucki , phys . rev . a * 60 * ( 1999 ) 3593 . c. e. carlson and m. vanderhaeghen , arxiv:1109.3779 [ physics.atom-ph ] . m. c. birse and j. a. mcgovern , eur .
j. a * 48 * ( 2012 ) 120 [ arxiv:1206.3030 [ hep - ph ] ] . g. a. miller , phys .
b * 718 * ( 2013 ) 1078 [ arxiv:1209.4667 [ nucl - th ] ] .
v. bernard , n. kaiser , u. g. meiner , phys .
* 67 * ( 1991 ) 1515 .
v. bernard , n. kaiser and u. g. meiner , int .
j. mod .
e * 4 * ( 1995 ) 193 [ arxiv : hep - ph/9501384 ] .
j. a. mcgovern , d. r. phillips and h. w. griesshammer , eur .
phys . j. a * 49 * ( 2013 ) 12 [ arxiv:1210.4104 [ nucl - th ] ] .
m. schumacher , prog .
* 55 * ( 2005 ) 567 [ arxiv : hep - ph/0501167 ] .
h. w. griesshammer , d. r. phillips and j. a. mcgovern , arxiv:1306.2200 [ nucl - th ] .
m. e. rose , `` elementary theory of angular momentum '' , wiley 1957 .
r. p. hildebrandt , h. w. griehammer and t. r. hemmert , eur .
phys . j. a * 46 * ( 2010 ) 111 [ arxiv : nucl - th/0512063 ] .
r. p. hildebrandt , `` elastic compton scattering from the nucleon and deuteron , '' ph.d .
thesis technische universitt mnchen ( 2005 ) , arxiv : nucl - th/0512064 .
e. e. jenkins , a. v. manohar , phys .
* b255 * ( 1991 ) 558 .
e. e. jenkins , a. v. manohar , in _
dobogokoe 1991 , proceedings , effective field theories of the standard model _ 113 and calif .
san diego - ucsd - pth 91 - 30 ( 91/10,rec.dec . )
26 p. ( 201392 ) ( see conference index ) .
t. r. hemmert , b. r. holstein and j. kambor , phys .
b * 395 * ( 1997 ) 89 [ arxiv : hep - ph/9606456 ] .
t. r. hemmert , b. r. holstein and j. kambor , j. phys .
g g * 24 * ( 1998 ) 1831 [ arxiv : hep - ph/9712496 ] .
v. pascalutsa , d. r. phillips , phys .
* c67 * ( 2003 ) 055202 [ arxiv : nucl - th/0212024 ] .
h. w. griehammer , prog .
phys . * 55 * ( 2005 ) 215 [ arxiv : nucl - th/0411080 ] .
s. r. beane , m. malheiro , d. r. phillips and u. van kolck , nucl .
a * 656 * ( 1999 ) 367 [ arxiv : nucl - th/9905023 ] .
j. l. friar , annals of phys .
( ny ) * 95 * ( 1975 ) 170 .
j. l. friar , phys .
c * 16 * ( 1977 ) 1504 .
h. arenhovel , z. phys .
a * 297 * ( 1980 ) 129 .
m. weyrauch and h. arenhovel , nucl .
a * 408 * ( 1983 ) 425 .
e. epelbaum , w. gloeckle , u .- g . meiner , nucl .
* a671 * ( 2000 ) 295 . [ arxiv : nucl - th/9910064 ] .
l. c. maximon , phys .
c * 39 * ( 1989 ) 347 .
r. p. hildebrandt , h. w. griehammer , t. r. hemmert and d. r. phillips , nucl .
a * 748 * ( 2005 ) 573 [ arxiv : nucl - th/0405077 ] .
s. r. beane , m. malheiro , j. a. mcgovern , d. r. phillips , u. van kolck , phys .
* b567 * ( 2003 ) 200 [ arxiv : nucl - th/0209002 ] .
s. r. beane , m. malheiro , j. a. mcgovern , d. r. phillips and u. van kolck , nucl .
a * 747 * ( 2005 ) 311 [ arxiv : nucl - th/0403088 ] .
d. l. hornidge , b. j. warkentin , r. igarashi , j. c. bergstrom , e. l. hallin , n. r. kolb , r. e. pywell , d. m. skopik et al .
84 * ( 2000 ) 2334 [ arxiv : nucl - ex/9909015 ] . in the lab frame , in @xmath349 .
top : energy - dependence at different angles .
other panels : sensitivity to varying a polarisability abouts its central value ( ) of eq . by @xmath350 ( ) and @xmath351 ( ) units , at @xmath285 . from top left to bottom right : variation of @xmath241 , @xmath310 ( constrained by the baldin sum rule ) , @xmath311 , @xmath314 , @xmath8 , @xmath9 . ] in the lab frame .
top : energy - dependence at different angles .
other panels : sensitivity to varying a polarisability abouts its central value ( ) of eq . by @xmath350 ( ) and @xmath351 ( ) units , at @xmath285 . from top
left to bottom right : variation of @xmath241 , @xmath310 , @xmath311 , @xmath314 , @xmath8 , @xmath9 . ] , @xmath352 , @xmath353 , @xmath354 , @xmath355 , at @xmath285 in the lab frame , with the static polarisabilities given by eq . .
: @xmath356 ; : @xmath357 ; : @xmath358 ; : @xmath359 ; : @xmath360 .
each panel is drawn at a different scale . ] | a complete set of linearly independent observables in compton scattering with arbitrarily polarised real photons off an arbitrarily polarised spin-@xmath0 target is introduced , for the case that the final - state polarisations are not measured .
adopted from the one widely used e.g. in deuteron photo - dissociation , it consists of @xmath1 terms : the unpolarised cross section , the beam asymmetry , @xmath2 target asymmetries and @xmath3 asymmetries in which both beam and target are polarised .
they are expressed by the helicity amplitudes and where available related to observables discussed by other authors . as application to deuteron compton scattering , their dependence on the ( isoscalar ) scalar and spin dipole polarisabilities of the nucleon is explored in chiral effective field theory with dynamical @xmath4 degrees of freedom at order @xmath5 .
some asymmetries are sensitive to only one or two dipole polarisabilities , making them particularly attractive for experimental studies . at a photon energy of @xmath6 , a set of @xmath7 observables
is identified from which one may be able to extract the spin polarisabilities of the nucleon . these are experimentally realistic but challenging and mostly involve tensor - polarised deuterons .
relative to compton scattering from a nucleon , sensitivity to the `` mixed '' spin polarisabilities @xmath8 and @xmath9 is increased because of interference with the @xmath10 wave component of the deuteron and with its pion - exchange current .
an interactive _ mathematica 9.0 _ notebook with results for all observables at photon energies up to @xmath11 is available from ` [email protected] ` .
arxiv:1304.6594 [ nucl - th ] , int - pub-13 - 014 + 25th april 2013 ; revised version 5th july 2013 + accepted by europ .
j. phys . * a * * harald w. griehammer@xmath12 * _ @xmath13 institute for nuclear studies , department of physics , + the george washington university , washington dc 20052 , usa _ + _ and _ + _ @xmath14 institut fr kernphysik ( ikp-3 ) , institute for advanced simulation and + jlich centre for hadron physics , forschungszentrum jlich , d-52428 jlich , germany _ 1.0 cm [ cols= " > , < " , ] 1.0 cm |
during experimental data processing often there is a task to fit arbitrary distribution with the only peak to some function with a set of free parameters .
this can be necessary for more precise determination of peak position , or full width at half maximum ( devise resolution ) , or for the signal form approximation using monte carlo events for subsequent analysis of data distribution and so on .
available set of fitting functions does not always matches the experimental requirements , especially for high statistics ( it is difficult to achieve suitable confidence level ) . in practice most often
the sum of the three gaussian distributions is used to fit such histograms . in most cases
this fit provides suitable confidence level .
at least one can add more and more gaussians until confidence level is admissible .
disadvantage of this method is that every gaussian used is symmetric , so in principle one can not used sum of gaussians with common center to fit asymmetric distributions , and if all gaussians have different centers , then it is difficult to provide the only maximum of fitting function and monotony to the right and left of maximum .
another often used function is a spline of two halves of different gaussians .
this function is good in case that the function `` tails '' can be approximated with gaussian .
however for high statistics the major experimental distributions far of peak have exponential or power - of - x dependence , which do not match gaussian .
logarithmic gaussian distribution is also often used ( at slac it is called `` novosibirsk '' function ) : @xmath0,\;\ ; \int\limits_{\frac{x_m - x}{x_m - x_p}>0}\!\!\!\!\!f_n{\,\mathrm{d}}x=1,\ ] ] where @xmath1 is a peak location ( function maximum ) , @xmath2 is full width at half maximum ( fwhm ) , @xmath3 is asymmetry parameter , @xmath4 is normalization factor . @xmath5
@xmath6 @xmath7 as is seen from ( [ eq : xm - expression ] ) , for @xmath8 the boundary coordinate @xmath9 and all @xmath10 .
if @xmath11 , then @xmath12 and @xmath13 .
using these notations one can rewrite this function , excluding the inconvenient variable @xmath14 : @xmath15 . \end{array}\ ] ] function is equal to 0 for all @xmath16 , for which the logarithm argument is negative .
possible values of parameters : @xmath17 , @xmath3 is arbitrary , @xmath1 is arbitrary . for @xmath18
the formulae have an ambiguity of the type @xmath19 , so some parameters should be expanded to the tailor series .
@xmath20{c}\sim \\[-2mm]\scriptstyle\lambda\sim 0\end{array } 1+\frac{h\left|\lambda\right|}{2}+\frac{h^2\lambda^2}{8},\ ] ] therefore @xmath21{c}\sim \\[-2mm]\scriptstyle\lambda\sim 0\end{array } \lambda\cdot\left[1-\frac{h^2\lambda^2}{8 } + \frac{7h^4\lambda^4}{128}\right],\ ] ] @xmath22{c}\sim \\[-2mm]\scriptstyle\lambda\sim 0\end{array } \frac{h^2\lambda^2}{8\ln 2}\cdot\left [ 1-\frac{h^2\lambda^2}{3 } \right],\;\;\ ; \frac{\sigma}{\left|\lambda\right| } \begin{array}[t]{c}\sim \\[-2mm]\scriptstyle\lambda\sim 0\end{array } \frac{h}{2\sqrt{2\ln 2}}\cdot\left[1-\frac{h^2\lambda^2}{6 } \right],\ ] ] @xmath23{c}\sim \\[-2mm]\scriptstyle\lambda\sim 0\end{array } \frac{4\ln 2 \left(x - x_p\right)^2}{h^2}\cdot\left [ 1-\left(x - x_p\right)\lambda \right].\ ] ] for @xmath24 the function converts to @xmath25,\ ] ] that is gaussian distribution with root mean square @xmath26 this function is convenient for fitting the distributions with abrupt spectrum end .
however many experimental distributions are more smooth , and suggested in this paper function can be more successful .
it is suggested to build the fitting function on the base of convolution of gaussian and exponential distributions , which can be easily derived : @xmath27 \times \\[3mm]\rule{35mm}{0mm}\times \exp\left [ -\frac{x - x_g}{\lambda } + \frac{\sigma_g^2}{2\lambda^2 } \right ] .
\end{array}\ ] ] integral of this function over all @xmath16 equals 1 .
such a function was first used by the author in 2004 at slac ( babar note # 582 ) for fitting the deposited energy distributions in calorimeter with the aim of peak position and resolution determination for the algorithms of absolute photon energy calibration , and despite some technical difficulties , this function proved to be enough convenient for fitting such distributions , especially for high statistics .
technical difficulties appear when the argument of erf function is big @xmath28 if the argument of erf function in formula ( [ eq : fitcurve1explicit ] ) is denoted as @xmath29 @xmath30 then the formula ( [ eq : fitcurve1explicit ] ) looks like : @xmath31 \exp\left[\frac{z\sigma_g\sqrt{2}}{\left|\lambda\right|}- \frac{\sigma_g^2}{2\lambda^2 } \right].\ ] ] if @xmath32 , then exponential index also goes to infinity , and so the ambiguity of the type @xmath33 arises . because of finite accuracy of computer calculations , this difficulty appears rather fast , just for moderate values of @xmath34 . to avoid this problem one can use the asymptotic expansion @xmath35 , \;\;\ ; k_0\leq z^2 .
\end{array}\ ] ] substituting this expansion to ( [ eq : fitcurve1reduced ] ) , we obtain : @xmath36 \cdot \left[1+\sum\limits_{k=1}^{k_0}\frac{(-1)^k(2k-1)!!}{2^kz^{2k } } \right ] .
\end{array}\ ] ] here for big @xmath29 values no ambiguities appear , function goes to 0 .
this very expansion allows to find the limit for @xmath37 . indeed ,
@xmath38 and @xmath39 \ ] ] let us consider the limit @xmath40 .
@xmath41 and substituting this to ( [ eq : fitcurve1reduced ] ) , we get @xmath42 \frac{1}{\left|\lambda\right|}\exp\left[-\frac{(x - x_g)}{\lambda}\right ] , \;\ ; \frac{(x - x_g)}{\lambda}>0 . \end{array}\right.\ ] ] function @xmath43 plots for several sets of parameters are presented in fig.[sampleoffb1 ] .
= 0.49 = 0.49 sometimes the integral distribution can be useful : @xmath44 -\\[1 mm ] \rule{40mm}{0 mm } -\frac{\lambda}{2\left|\lambda\right| } e^{\frac{\sigma_g^2}{2\lambda^2}-\frac{x - x_g}{\lambda}}\left [ 1-\mathrm{erf}\left(\frac{\sigma_g}{\sqrt{2}\left|\lambda\right|}- \frac{\left(x - x_g\right)\left|\lambda\right|}{\sqrt{2}\sigma_g\lambda } \right)\right ] .
\end{array}\ ] ] usage of @xmath43 to fit the distributions would be more convenient , if the free parameter is @xmath14 location of function maximum , instead of @xmath45 .
equation for the search for @xmath14 looks rather complicated : @xmath46 \end{array}\ ] ] or in other notations @xmath47 where @xmath48 . for @xmath49
@xmath50 : @xmath51 } \longrightarrow z_m\approx -\sqrt{\ln\frac{\rho}{2-\frac{2}{\rho\sqrt{\left|\ln\frac{\rho}{2}\right| } \sqrt{\pi}}\cdot\left [ 1-\frac{1}{2\ln\frac{\rho}{2}}\right ] } } \longrightarrow \\[7mm]\rule{5mm}{0mm}\longrightarrow -\frac{\left(x_m - x_g\right)\lambda}{\sigma_g\left|\lambda\right|\sqrt{2 } } + \frac{\sigma_g}{\left|\lambda\right|\sqrt{2}}= -\sqrt{\ln\frac{\left|\lambda\right|}{\sigma_g\sqrt{2\pi } } } \end{array}\ ] ] or @xmath52{c } \approx \\[-2 mm ] \scriptstyle \sigma_g
\ll |\lambda| \end{array } \frac{\sigma_g^2}{\lambda}+\frac{\sigma_g\left|\lambda\right|\sqrt{2}}{\lambda}\cdot \sqrt{\ln\frac{\left|\lambda\right|}{\sigma_g\sqrt{2\pi } } } \to 0.\ ] ] for @xmath53 @xmath54 , and here we also can derive approximate solution : @xmath55 = \rho\exp\left(-z_m^2\right)\ ] ] @xmath56 or @xmath57 from ( [ eq : peakcondition ] ) one can derive more terms of taylor series : @xmath52{c } \approx \\[-2 mm ]
\scriptstyle |\lambda| \ll \sigma_g \end{array } \lambda\cdot\left[1-\left(\frac{\lambda}{\sigma_g}\right)^2 + \left(\frac{\lambda}{\sigma_g}\right)^4+\ldots \right].\ ] ] let us return to the equation ( [ eq : maxfb1 ] ) , which should be solved in order to find @xmath43 maximum location .
let us transform the interval of @xmath58 variable to the interval @xmath59 : @xmath60 at the ends of the interval we know the solution : @xmath61 \mu \sim 1\longrightarrow z_m\approx \frac{\sqrt{\pi}\ln\mu}{2}-\frac{1}{\sqrt{\pi}\ln\mu } \end{array}\ ] ] let us look for approximating function in the form : @xmath62 function @xmath63 plot and the approximating cubic spline with 5 knots are presented in fig.[fpapproximation ] .
= 0.7 root mean square deviation equals @xmath64 , maximum error of interpolation @xmath65 is achieved at @xmath66 .
spline coefficients are cited in table[tab : fpspline ] .
.[tab : fpspline]spline of deficiency 2 coefficients for approximation of @xmath63 function . [ cols="<,^,^,^,^,^",options="header " , ] now we can calculate the shift of peak position vs gaussian center : @xmath67 the function @xmath68 can be easily implemented in any programming language , using the above formulae .
in some cases the function @xmath43 itself can be suitable for fitting .
however more often this function is not enough flexible to provide satisfactory confidence level with experimental distribution .
one could try to use the function which is the convolution of the three distributions : gaussian and two different exponential .
such a function could be useful if the distribution `` tails '' both to the right and left from the peak do not match gaussian distribution .
however in this case and in other difficult ones the distributions are fitted more successfully to the sum of different @xmath43 functions with the common parameter peak position @xmath14 . for the sum of two functions one can use the following expression : @xmath69\rule{51mm}{0mm}+ \sin^2\xi\cdot f_{b1}\left(x;x_m-\delta x_{mg}(\sigma_2,\lambda_2),\sigma_2,\lambda_2\right ) \end{array}\ ] ] sample of using such a function is presented in fig.[fitetram0acfb1 ] (
one more parameter is added common factor @xmath4 ) .
= 0.49 = 0.49 in principle for complicated cases one can use the sum of more @xmath43 functions .
however for so many free parameters the likelihood function minimization can be unstable , and one need to help minuit program .
the simplest and enough effective trick is optimization of parameters in turn , initially fixed at some reasonable values .
for fitting the smooth distributions with one peak a function is suggested , which is the convolution of gaussian and exponential distributions . | in the paper a new fitting function is suggested , which can essentially increase the existing instrumentation for fitting of asymmetric peaks with the only maximum . |
nowadays , mesonic chiral perturbation theory ( chpt ) @xcite is widely accepted as the low - energy theory of the strong interactions based on the underlying symmetries of qcd .
impressive accuracy in the description of data has been achieved in the last decade @xcite . on the other hand , the baryon sector of chpt proved to be more problematic @xcite .
although there has been considerable progress in recent years ( see , e.g. , @xcite ) , the issue of the convergence of perturbative calculations in the nucleon sector of the effective theory is still of great interest . to compare lattice calculations with experimental data , one has to extrapolate the results to physical quark masses . the preferred method to cope with
this problem is to calculate the physical quantities as functions of the quark masses in an effective field - theoretical approach ( see , e.g. , @xcite ) .
of course , these effective theories have a limited range of applicability .
the aim of this work is to probe the issue of convergence of perturbative calculations in bchpt . to that end
we consider the contribution of particular infinite sets of diagrams to the @xmath0 scattering amplitude and to the nucleon self - energy .
these contributions are summed up by solving integral equations analytically . for a related analysis of exact solutions to the bethe - salpeter equations in bchpt in the @xmath1 sector ,
see ref .
@xcite .
throughout this paper we use dimensional regularization in combination with the infrared renormalization scheme @xcite without explicitly showing the counter - terms responsible for the subtractions of loop diagrams .
first , let us specify the lagrangians required for the purposes of this work . from the mesonic sector we need the lowest - order lagrangian of the @xmath2 sector @xcite @xmath3 where @xmath4 is a unimodular unitary matrix containing the goldstone boson fields . in eq .
( [ l2 ] ) , @xmath5 denotes the pion - decay constant in the chiral limit : @xmath6=92.4 $ ] mev . here , we work in the isospin - symmetric limit @xmath7 , and the lowest - order expression for the squared pion mass is @xmath8 , where @xmath9 is related to the quark condensate @xmath10 in the chiral limit @xcite .
next , the nucleon fields are collected in an isospin doublet @xmath11 with two four - component dirac fields @xmath12 and @xmath13 describing the proton and neutron , respectively . from the nucleon sector
we need the leading - order lagrangian omitting external sources @xmath14 where @xmath15,\ ] ] and @xmath16 and @xmath17 refer to the chiral limit of the physical nucleon mass and the axial - vector coupling constant .
let us consider elastic @xmath0 scattering with @xmath18 , @xmath19 the four - momenta of the incoming and @xmath20 , @xmath21 the four - momenta of the outgoing nucleons and pions , respectively ( see fig . 1 ) .
the corresponding vertex function ( amputated green s function ) can be obtained by solving the integral equation @xmath22 where @xmath23 .
@xmath24 stands for the @xmath0 effective potential and @xmath25 is the product of the ( dressed ) nucleon and pion propagators . here , the effective potential is defined as the sum of all diagrams contributing to the vertex function , which can not be reduced to two @xmath0 scattering diagrams by cutting one pion line and one nucleon line .
the standard representation for the vertex function in terms of isospin symmetric and antisymmetric parts reads @xmath26\ , \gamma^- \ , .
\label{tampl}\ ] ] for our purposes it is convenient to decompose the scattering amplitude in isospin - invariant components @xmath27 these vertex functions satisfy the integral equations written symbolically as @xmath28 where @xmath29 or @xmath30 can take one of the two values @xmath31 and @xmath30 . ] and @xmath32\ , v^-\,.\end{aligned}\ ] ] suppose the potential can be written as @xmath33 where the @xmath34 depend only on @xmath35 as is the case , e.g. , for the potential ) corresponds to the weinberg - tomozawa term plus the @xmath36-channel nucleon - pole diagram obtained from the lagrangian of eq .
( [ lolagr ] ) .
note that the @xmath37-channel nucleon - pole diagram can not be written in the form of eq .
( [ vmatrix ] ) . ] @xmath38 in this case the vertex functions @xmath39 can also be written as @xmath40 substituting eqs .
( [ vmatrix ] ) and ( [ tmatrix ] ) in eq .
( [ decoupledequations ] ) results in the following matrix equations @xmath41 where @xmath42 for the undressed propagator @xmath43 we obtain @xmath44\,,\nonumber\\ g_{12 } & = & g_{21 } = \left ( p^2+m\,p \hspace{-.45em}/\hspace{.1em}\right ) \,i_{n \pi } ^{(p)}(-p,0)-m^2\,i_{n \pi}(-p,0)-i_n\,,\nonumber\\ g_{22 } & = & p \hspace{-.45em}/\hspace{.1em}\,\left ( p^2-m^2\right ) \,i_{n \pi } ^{(p)}(-p,0)-m^2\left ( p \hspace{-.45em}/\hspace{.1em}-m\right)\,i_{n \pi}(-p,0)-\left ( p \hspace{-.45em}/\hspace{.1em}-m\right)\,i_n\,.\label{gmatrixelements}\end{aligned}\ ] ] the loop integrals @xmath45 , @xmath46 , and @xmath47 are given in the appendix .
decomposing the matrices as @xmath48 and substituting the result in eq .
( [ matrixcoupledequations ] ) we obtain @xmath49 if we define @xmath50 and substitute in eq .
( [ mdecdeqs ] ) the resulting equations reduce to the decoupled system @xmath51 eqs .
( [ mdecdeqs ] ) and ( [ redeqs ] ) are systems of matrix equations and can be solved exactly . inserting the solutions in @xmath52 and using the most general parity - conserving form for the on - shell @xmath53 matrix ,
@xmath54 one can calculate the four lorentz - invariant amplitudes as @xmath55 on the other hand by expanding eqs .
( [ t1t2 ] ) perturbatively we can compare the results of the term - by - term loop expansion with the non - perturbative expression and estimate the error of the perturbative approximation for various kinematics .
we calculated exactly ( as closed expressions of one - loop integrals ) the non - perturbative ( re - summed ) contribution and the tree plus one - loop order contributions to the @xmath56 , @xmath57 , and @xmath58 scattering processes for the potential due to the weinberg - tomozawa term , @xmath59 the results for differential cross sections are given in figs .
[ pin1forward : fig]-[pin3forward : fig ] .
these figures suggest that the perturbative results ( tree plus one - loop order ) approximate the re - summed contributions very poorly already for @xmath60 .
the full ( dressed ) nucleon propagator has the form @xmath61 where the nucleon self - energy @xmath62 represents the one - particle - irreducible contribution to the two - point function .
the nucleon self - energy contains the contributions of counter - terms so that @xmath16 corresponds to the nucleon pole mass in the chiral limit .
the physical mass @xmath63 of the nucleon is defined as the solution to the equation @xmath64 below we calculate the contribution to the nucleon mass of an infinite number of diagrams shown in fig . [ nse : figdiagrams ] .
the sum of these diagrams can be written in a closed form as @xmath65\left [ q_2 ^ 2-m^2+i\,0^+\right]}\ , , \label{npse}\end{aligned}\ ] ] where the @xmath0 vertex function @xmath66 is obtained by solving eq .
( [ tamplequation ] ) with the potential @xmath67 it is easily verified that @xmath68 depends only on @xmath69 . using the solution for @xmath69 from the previous section in eq .
( [ npse ] ) and integrating over @xmath19 and @xmath21 we obtain an analytic expression for the contribution of the diagrams in fig .
[ nse : figdiagrams ] to the nucleon mass @xmath70 where @xmath71 ^ 2\ , , \nonumber\\ d & = & 2 \,m_n \biggl\ { 8 \,m_n\,f^4 + 4\,\biggl [ \left ( m-\,m_n\right ) i_{\pi } + \left(m^2-m^2 - 2\,m\,m_n + m_n^2\right ) ( m+m_n ) \,i_{n \pi } \biggr ] f^2\nonumber\\ & & + \ , i_{\pi } \biggl[\left ( m_n - m\right ) i_{\pi } - \left(m^2-m^2 - 2\,m\,m_n + m_n^2\right ) \left ( m+m_n\right ) i_{n \pi}\biggr]\biggr\}\ , .
\label{massd}\end{aligned}\ ] ] on the other hand , by expanding eq .
( [ contributioninmass ] ) in powers of @xmath72 we can identify the contributions of each diagram separately . using the ir renormalization scheme and substituting @xmath73 mev @xcite , @xmath74 mev , @xmath75 mev , @xmath76 and @xmath77 mev we obtain @xmath78 as can be seen from eq .
( [ dmbl ] ) the first term in the perturbative expansion reproduces the non - perturbative result well and the higher - order corrections are clearly suppressed .
it is relevant for the chiral extrapolation of lattice data to consider the nucleon mass for larger values of the quark masses .
therefore in fig .
[ nseplot : fig ] we plot @xmath79 of eq .
( [ massd ] ) together with the leading contribution ( first diagram in fig .
[ nse : figdiagrams ] ) and the leading non - analytic correction to the nucleon mass @xmath80 @xcite as functions of @xmath81 . as can be seen from this figure , up to @xmath82 mev the non - perturbative sum of higher - order corrections
is suppressed in comparison with the @xmath83 term .
also , the leading higher order contribution reproduces the non - perturbative result quite well . on the other hand for @xmath84
mev the higher - order contributions are no longer suppressed in comparison with @xmath83 .
in this work we have addressed the issue of convergence of perturbative calculations in bchpt by analyzing pion - nucleon scattering and the nucleon mass . by solving the equation for the @xmath0 vertex function using dimensional regularization
we have obtained an exact expression for a sum of an infinite number of loop diagrams .
the solution is given in a closed form in terms of one - loop integrals .
we have renormalized the obtained non - perturbative expression by applying infrared renormalization @xcite .
we compared the perturbative contributions with the re - summed expression for elastic @xmath0 scattering .
we find that already for @xmath85 the perturbative results approximate the re - summed non - perturbative expression very poorly . as the potential , which has been iterated by solving the equation does not receive contributions from the intermediate @xmath86 state , we conclude that the inclusion of the @xmath86 as an explicit degree of freedom does not solve the problem of convergence of the considered loop contributions . in our opinion , to solve this problem one needs to include the @xmath86 degrees of freedom @xcite and _ simultaneously _ consider the @xmath0 scattering equations .
next , using the non - perturbative result for the @xmath0 scattering amplitude we have obtained an exact expression corresponding to a sum of an infinite number of diagrams contributing to the nucleon self - energy . using the infrared renormalization scheme and comparing the non - perturbative contribution to the nucleon mass with contributions of the first several terms in its perturbative expansion
we conclude that the so obtained perturbative series for the nucleon mass converges very well .
we also considered the correction to the nucleon mass for larger values of quark masses and found that the re - summed higher order contributions become larger than the leading non - analytic contribution for @xmath87 mev .
from this we conclude that for such values of @xmath81 bchpt can not be trusted in extrapolations of lattice data .
even if there are large cancelations these can not be treated systematically in standard bchpt . as we have summed up only a subset of higher - order diagrams ,
our analysis is not complete and therefore our result should be considered as an estimate for an upper limit of the radius of convergence .
we would like to thank d. leinweber for useful comments on the manuscript .
the work of d. d. and j. g. was supported by the deutsche forschungsgemeinschaft ( sfb 443 ) .
one - loop integrals : @xmath88 [ k^2-m^2+i0^+]}\nonumber\\ & = & \frac{p^\mu}{2p^2}\left[(p^2-m^2+m^2)\,i_{n\pi}(-p,0)+i_n -i_\pi\right]\,,\nonumber\\ i_\pi & = & i\,\int \frac{d^n k}{(2\pi)^n}\frac{1}{k^2-m^2+i0^+ } = 2\,m^2\bar{\lambda}+\frac{m^2}{8\pi^2}\ln\left(\frac{m}{m}\right)\ , , \nonumber\\ i_n & = & i\int \frac{d^n k}{(2\pi)^n } \frac{1}{k^2-m^2+i0^+ } = 2 \,m^2\,\bar{\lambda}\,,\nonumber\\ \label{inpi } i_{n\pi}(-p,0)&=&i\int\frac{d^n k}{(2\pi)^n}\frac{1}{[(k - p)^2-m^2+i0^+ ] [ k^2-m^2+i0^+]}\nonumber\\ & = & 2\,\bar{\lambda}+\frac{1}{16\pi^2}\left[-1 + \frac{p^2-m^2+m^2}{p^2}\ln\left(\frac{m}{m}\right ) + \frac{2mm}{p^2}f(\omega)\right],\end{aligned}\ ] ] where @xmath89\right\}\,,\nonumber\\ f(\omega ) & = & \left \ { \begin{array}{ll } \sqrt{1-\omega^2}\arccos(-\omega),&-1\leq\omega\leq 1,\\ \sqrt{\omega^2 - 1}\ln\left(\omega+\sqrt{\omega^2 - 1}\right ) -i\pi\sqrt{\omega^2 - 1},&1\leq \omega , \end{array } \right.\,,\nonumber\\ \omega & = & \frac{p^2-m^2-m^2}{2\,m\,m}\,.\end{aligned}\ ] ] infrared renormalized expression for @xmath90 : @xmath91+f_{ir}(\omega),\end{aligned}\ ] ] where @xmath92 and @xmath93 .
s. weinberg , physica a * 96 * , 327 ( 1979 ) .
j. gasser and h. leutwyler , ann .
( n.y . ) * 158 * , 142 ( 1984 ) .
j. gasser and h. leutwyler , nucl .
* b250 * , 465 ( 1985 ) . g. colangelo , j. gasser , and h. leutwyler , phys . lett .
b * 488 * , 261 ( 2000 ) .
g. colangelo , j. gasser , and h. leutwyler , nucl .
* b603 * , 125 ( 2001 ) . i. caprini , g. colangelo , j. gasser , and h. leutwyler , phys .
d * 68 * , 074006 ( 2003 ) .
i. caprini , g. colangelo , and h. leutwyler , arxiv : hep - ph/0509266 .
i. caprini , g. colangelo , and h. leutwyler , arxiv : hep - ph/0512364 .
g. colangelo , aip conf .
proc . * 756 * , 60 ( 2005 ) .
s. scherer , arxiv : hep - ph/0512291 .
j. bijnens , arxiv : hep - ph/0604043 . j. gasser , m. e. sainio , and a. varc , nucl . phys .
* b307 * , 779 ( 1988 ) . s. scherer , in _ advances in nuclear physics , vol . 27 _ , edited by j. w. negele and e. w. vogt ( kluwer academic / plenum publishers , new york , 2003 ) . s. r. beane , nucl .
* b695 * , 192 ( 2004 ) . m. procura , t. r. hemmert , and w. weise , phys . rev .
d * 69 * , 034505 ( 2004 ) .
d. b. leinweber , a. w. thomas , and r. d. young , pos * lat2005 * , 048 ( 2005 ) . m. procura , b. u. musch , t. wollenweber , t. r. hemmert , and w. weise , arxiv : hep - lat/0603001 .
t. becher and h. leutwyler , eur .
j. c * 9 * , 643 ( 1999 ) .
m. r. schindler , j. gegelia , and s. scherer , phys .
b * 586 * , 258 ( 2004 ) .
t. fuchs , j. gegelia , and s. scherer , eur .
j. a * 19 * , 35 ( 2004 ) . c. hacker , n. wies , j. gegelia , and s. scherer , phys .
c * 72 * , 055203 ( 2005 ) . | using the examples of pion - nucleon scattering and the nucleon mass we analyze the convergence of perturbative series in the framework of baryon chiral perturbation theory . for
both cases we sum up sets of an infinite number of diagrams by solving equations exactly and compare the solutions with the perturbative contributions . |
recently a novel solution for the positron emission tomography scanner was proposed which utilizes plastic scintillators as detectors for the annihilation quanta @xcite .
a single detection unit of this detector is built out of a scintillator strip read out on both sides by photomultipliers .
the position of the interaction of the gamma quanta inside a strip is determined based on the shape and time of photomultiplier signals .
the shape of signals may be distorted by the secondary scattering of gamma quanta inside the scintillator .
this is because a secondary scattering creates an additional light signal and as a result a light pulse reaching the photomultiplier is composed of overlaping signals originating from points of primary and secondary interactions . in this article
we estimate influence of the secondary scattering on the quality of hit position reconstruction in the pet detectors based on the plastic scintillators . to this end
a dedicated simulation programme was elaborated and series of simulations have been performed .
these allowed to determine multiplicity distributions of annihilation quanta interaction inside a scintillator strip , as well as distributions of distance between points of scattering .
the obtained results are then interpreted in view of the distortion of spatial and temporal resolution of the detector due to the secondary interactions of registered gamma quanta .
the main simulation algorithm is elaborated assuming that in plastic scintillators the annihilation gamma quanta with energy of about 511 kev , undergo a compton scattering only .
main aim of the simulation was an estimation of number of gamma quanta interactions in a scintillator strip and determinantion of spatial distribution of scattering centers within the scintillator volume . for this purpose , we assume that gamma quantum originates in a certrain initial position with a user defined four momentum vector . description of scintillator is also provided by user and consists of : attenuation constant at energy for @xmath1 and dimensions of scintillator cuboid . from this input data attenuation constant at any energy is extrapolated using following formula : @xmath2 where @xmath3 denotes inverse of attenutation length , and @xmath4 stands for total cross section of gamma quantum at given energy @xmath5 .
the values of cross sections have been extracted from @xcite .
a given number of events , @xmath6 , and maximum number of interactions in one event , @xmath7 , is also passed to algorithm , which may be in general described by a following scheme : position @xmath8 initial position direction @xmath8 initial direction energy @xmath8 initial energy @xmath9 @xmath10 @xmath11 total cross section length @xmath8 random length of pdf @xmath12 position @xmath8 position + direction @xmath11 length end event polar angle @xmath8 random angle from klein - nishina pdf azimuthal angle @xmath8 random angle from izotropic pdf _ scatter _ direction vector using generated angles print position , deposited energy , time as a result , collection of events containing ordered set of interactions , is obtained .
the generation of varaiables used in the programme according to the given probability density function ( pdf ) is performed based on the distribution of a corresponding cumulative function , which is homogenous by definition .
cumulative function ( @xmath13 ) of the angular probability density function is calculated based on the klein - nishina formula @xcite . in following equations
it is factorized into @xmath14independent ( constant @xmath15 ) and @xmath14dependent functions , and other @xmath14independent function is added ( only relative changes to @xmath16 are important ; additive function is chosen to ensure that @xmath17 , which simplifies calculations ) : @xmath18\\ p(\theta,\gamma)&=\frac { 1}{1+\gamma(1-\cos\theta)}\\ \gamma&=\frac{e}{m_e c^2},\end{aligned}\ ] ] where @xmath5 denotes energy of gamma quantum , and @xmath19 stands for the mass of electron . +
this method is used only for generation of scattering angle @xmath14 . in the case of the distance @xmath20 between subsequent scattering a cumulative distribution function
can be derived analyticaly since pdf posesses an exponential form : @xmath21 where @xmath22 .
hence , cumulative function reads : @xmath23 therefore , for a given random number of uniform distribution @xmath24 $ ] : @xmath25 in order to effectively simulate a series of scatterings , we have elaborated an algorithm in which calculations reduces to the rotations of vectors . for given vector @xmath26 the rotation matrix @xmath27 -matrices are changing reference frame , while @xmath28-ones describe transformation of vector in particular frame . ]
fullfiling the condition @xmath29 is generated .
let @xmath30 .
then , another rotation matrix @xmath31 , such that @xmath32 is computed .
now scattering can be simulated by rotation of vector @xmath33 by means of matrix @xmath34 in such way that @xmath35 forms a desired angle with z - axis . to preserve cylindrical symmetry
, this vector must be rotated by rotation matrix around the z - axis @xmath36 by a random angle . generated vector must be re - transformed to original reference frame , and the final scattered vector is given by : @xmath37 the discussed rotations are visualized in fig .
[ scatter ] .
the gamma quantum passing through the matter may undergo various processes among which compton , rayleigh and photelectric effects are dominant for the energy range below 1 mev . for simulations of the interaction of annihilation
quanta in a plastic scintillator we assumed the dominance of the compton effect and have neglected other processes since they can contribute significantly only for energies below 50 kev ( see fig .
[ fig : tcss ] ) .
the energy of a gamma quantum may be expressed in terms of the _ reduced gamma quantum energy _ @xmath38 .
such notation is especially convenient for calculations involving annihilations quanta for which @xmath39 . for further consideration
it is useful to mention that @xmath40 for gamma quantum with energy of 50 kev . in the scintillator strip annihilation gamma quantum
may in principle undergo many compton scatterings and its minimum energy after @xmath41th interaction may be calculated based on the iterative formula : @xmath42 first terms of this series are @xmath43 .
it shows , that up to 4th interaction energy of gamma quantum is above @xmath44 .
it will be shown in the following sections that in the case of the scintillator strips of the j - pet detector a probability that the gamma quantum scatters more than three times is negligible .
therefore , all efects other than compton scattering can be safely neglected .
+ another assumption applied is the convexity of scintillator s shape : in such case if gamma quantum leaves scintillator , it will never come back and computations can stop .
it is also certain ( from definition ) that line segment bounded by two points of interactions is inside scintillator s volume , which simplifies calculations of next interaction point .
we have performed simulations assuming that the scintillator strip is a cuboid with dimensions of @xmath45 .
the cross section of the strip corresponds to the size of the modules used for the first j - pet prototype , and its length was chosen arbitrarily large .
starting point of gamma quanta was placed on the middle of @xmath46 wall .
its initial energy was set to @xmath47 ( @xmath39 ) .
number @xmath7 was set to 1000 to make sure that all possible interactions were taken into account .
simulations were performed for three directions of the gamma quanta shown in fig .
three cases denoted in fig .
[ cond ] as a , b , c have initial direction vectors @xmath48 , @xmath49 and @xmath50 , respectively .
for each direction @xmath51 events were simulated .
attenuation constant at @xmath39 was set to @xmath52 .
+ , scaledwidth=50.0% ] distribution of multiplicity of interactions determined for three studied directions a , b and c is shown in table [ tab : freq ] .
.frequency of occurence of interactions .
` frequency ' of level @xmath7 is indicating fraction of events with @xmath7 or more number of interactions .
the values are obtained taking into account only these events for which gamma quantum interacted in the scintillator . [ cols="^,^,^,^",options="header " , ]
the table indicates that more than 90@xmath0 of secondary scattering occurs for @xmath53 smaller than 1.4 cm .
however , for the distortion of the signals not only a distance between interaction points is important , but also the energy deposited by the secondary ionisation .
correlations between deposited energy and spatial separation of interaction points is shown in figures [ fig : edxa][fig : edxc ] .
these figures are however not 2d histograms ( i.e. heat maps ) , but sets of dense packed 1d histograms , in which each line ( sub histogram ) is separately normalized in that way , that maximum value in each sub histogram is equal to @xmath54 .
the figures are made this way , because one may approximate total energy stored in scintillator , and could be interested in most probable @xmath53 in this event .
it is important to mention that in this and following paragraph only events including at least two scatterings were taken into account . as a function of total deposited energy .
note that this figure shows a set of one dimensional histograms normalized to unity in maximum .
a more detailed description can be found in the text . ] as a function of total deposited energy .
note that this figure shows a set of one dimensional histograms normalized to unity in maximum .
a more detailed description can be found in the text . ] as a function of total deposited energy .
note that this figure shows a set of one dimensional histograms normalized to unity in maximum .
a more detailed description can be found in the text . ]
results presented in these figures indicate that in order to decrese the blurring of the signals due to the secondary scattering one can select from the full spectrum only these events for which deposited energy is larger than e.g. 0.2 @xmath19 and smaller than 0.7 @xmath19 . for each of the studied cases the broadest @xmath53 distributions
are observed for the energy deposition larger than the maximum energy deposition in the primary scattering ( 0.67 @xmath19 ) . in general
the largest spread of @xmath53 is observed as expected for the case c , for which the primary gamma quantum can travel the longest distance along the scintillator . in this case
a scattering under small angles resulting in a small energy depositions leads to the scatterd gamma quantum which , due to the relatively large energy , can travel on the average large distance before the second scattering , which may again occur at most probable under a small angle with a small energy deposition .
therefore in this case broad @xmath53 distributions are observed also at small values of total deposited energy .
the full energy spectra for all cases are presented in fig .
[ tote ] .
spatial distance of first and subsequent interactions , along with their time difference leads to blurring of signal coming to photomultipliers .
the light signals are read out by photomulitpliers on both edges of the strip , therefore for one of them the signal distortion by the secondary scattering will be much smaller than for the other one .
this is because , between primary and secondary scattering the gamma quantum propagates towards one of the photomulitpliers together with the primary light signal and therefore the signal distortion in this photomultiplier will be smaller than in the other edge where the delay between primary and secondary light pulses is equal to : @xmath55 where @xmath56 denotes difference of x component of interactions positions , @xmath57 denotes their time difference and @xmath58 stands for the effective velocity of light signal propagation trough the scintillator . in this calculations @xmath58 = 14 cm / ns was used .
histogram of @xmath59 ( normalized to pdf ) is shown in fig .
[ fig : dtau ] .
histogram ( normalized to probability density function ) for case a , b and c. ] fig .
[ fig : dtau ] shows that the primary and secondary light signals overlap with a delay of about 40 ps ( fwhm of the distribution ) .
simulations of scattering of annihilation gamma quanta in a strip of polymer scintillator have been conducted .
such strips constitute basic detection modules of the newly proposed pet detector being developed by the j - pet collaboration .
an algorithm simulating chain of compton scatterings was elaborated and series of simulations have been conducted for the scintillator strip with the cross section of 5 mm x 20 mm .
simulations were simplified due to the observation that for the energy range of interest the compton scattering is dominant and photoelectric and rayleigh effects may be neglected . as a result : ( i ) distributions of multiplicity of interactions , ( ii ) spatial distribution of interactions points as a function of the deposited energy , and ( iii ) spectra of time differences between subsequent interactions have been determined . obtained results
indicate that secondary interactions occur only in the case of about 8@xmath0 of events and out of them only 25@xmath0 take place in the distance larger than 0.5 cm .
it was also established that light signals produced at primary and secondary interactions overlap with the delay which is spread by about 40 ps ( fwhm ) . moreover , analysis of histograms of the distance between subsequent interactions points as a function of total deposited energy revealed that the blurring of signals due to the secondary interactions may be decreased by selecting from the full spectrum only these events for which deposited energy is e.g. larger than @xmath60 and smaller than @xmath61 .
we acknowledge technical and administrative support by m. adamczyk , t. gucwa - ry , a. heczko , m. kajetanowicz , g. konopka - cupia , j. majewski , w. migda , a. misiak , and the financial support by the polish national center for development and research through grant innotech - k1/in1/64/159174/ncbr/12 , the foundation for polish science through mpd programme and the eu and mshe grant no .
poig.02.03.00 - 161 00 - 013/09
. 4 moskal p. , salabura p. , silarski m. , smyrski j. , zdebik j. , zieliski m. novel detector systems for the positron emission tomography .
bio - algorithms and med - systems 2012;7:73 - 78 ; e - print arxiv:1305.5187 .
moskal p. , bednarski t , biaas p. , ciszewska m. , czerwiski e. , heczko a. , et al .
tof - pet detector concept based on organic scintillators .
nuclear medicine review 2012;15:c81-c84 ; e - print arxiv:1305.5559.11 .
moskal p. , bednarski t , biaas p. , ciszewska m. , czerwiski e. , heczko a. , et al .
strip - pet : a novel detector concept for the tof - pet scanner .
nuclear medicine review 2012;15:c68-c69 ; e - print arxiv:1305.5562 .
wolfram research .
( dynamic content ) .
online integrator . retrieved april 30 , 2013 from http://integrals.wolfram.com/index.jsp?expr=%28%281%2bg%281-cos+x%29%29+%2b+1%2f+%281%2bg%281-cos+x%29%29+-+sin^2+x+%29%2f%281%2bg%281-cos+x%29%29^2+2+pi+sin+x&random=false[http://integrals.wolfram.com/index.jsp?expr=%28%281%2bg%281-cos+x%29%29+%2b+1%2f+%281%2bg%281-cos+x%29%29+-+sin^2+x+%29%2f%281%2bg%281-cos+x%29%29^2+2+pi+sin+x&random=false ] nist xcom .
( dynamic content ) .
photon cross section database . retrieved april 30 , 2013 from http://physics.nist.gov/physrefdata/xcom/text/intro.html silarski m. polymer scintillators for pet produced commercially .
cracow : jagiellonian university .
( pet uj raport nr 19/2012 ) + | this article describes simulations of scattering of annihilation gamma quanta in a strip of plastic scintillator .
such strips constitute basic detection modules in a newly proposed positron emission tomography which utilizes plastic scintillators instead of inorganic crystals .
an algorithm simulating chain of compton scatterings was elaborated and series of simulations have been conducted for the scintillator strip with the cross section of 5 mm x 19 mm . obtained results indicate that secondary interactions occur only in the case of about 8@xmath0 of events and out of them only 25@xmath0 take place in the distance larger than 0.5 cm from the primary interaction . it was also established that light signals produced at primary and secondary interactions overlap with the delay which distribution is characterized by fwhm of about 40 ps . |
in the framework of modified newtonian dynamics ( mond ) @xcite we get quite the reasonable theoretically and successful phenomenologically explanation for the empirical tully fisher law , that regularly relates the _ visible _ masses of spiral galaxies to asymptotically constant rotation - velocities of stars in the region of dominating `` dark matter halo '' ( see figure [ fig - tf ] ) as well as for profiles of rotational velocities of stars in galaxies , wherein the regime of flattening the rotation curves does not take place . moreover , the mond is able to explain many other phenomena caused by the _ inhomogeneity _ of baryonic matter - distribution in the universe at the scales of gravitationally coupled systems ( see reviews in @xcite ) .
( 50,77 ) ( 5.5,3.5 ) .
the mass is given in units of solar mass @xmath0 .
the figure is taken from @xcite.,title="fig : " ] ( 52,2)@xmath1 , km / s ( 8,3)@xmath2 ( 36,3)@xmath3 ( 3.4,17)@xmath4 ( 3.4,28.5)@xmath5 ( 3.4,40)@xmath6 ( 3.4,51.5)@xmath7 ( 3.4,63.5)@xmath8 ( -2,76)@xmath9 for instance , in disk galaxies , wherein a dominant contribution to the baryonic mass is given by the interstellar gas , the matter density is determined with a high accuracy , since an uncertainty , caused by the extraction of stellar masses from their visible magnitudes , is significantly reduced . the mond gives a precise agreement of its predictions with the experimental data on the rotation speeds of matter around the centers of galaxies without any free parameters , while the uncertainties of results are caused by errors of measurements , only @xcite . by construction , the basic conclusion of mond is the statement that the observational data can be confidently described without any introduction of non - baryonic dark matter at the galactic scales of matter - inhomogeneity as well as at the scales of gravitationally coupled galactic clusters @xcite due to the specially fitted universal modification of the gravity law if the acceleration of free falling @xmath10 is less than the critical value @xmath11 , so that @xmath12 , where @xmath13 is the matter mass , while @xmath14 is a distance to a reference point
. however , such the conclusion conflicts with cosmological modeling the universe evolution and properties of its large scale structure , that is characterized by a high degree of homogeneity for the spatial distribution of matter ( @xmath15 for fluctuations of energy density @xmath16 and temperature of cosmic microwave background radiation ( cmbr ) , @xmath17 ) , because such calculations lead to the necessary introduction of dark matter with the density being approximately 5 times greater than the baryonic density @xcite , if we follow the equations of general relativity ( gr ) .
in cosmology of homogeneous and isotropic universe , the points in the space with coordinates @xmath18 move as given by a dependence of scale factor on the time , @xmath19 , so that velocity @xmath20 and acceleration @xmath21 of material point are determined by @xmath22 where the dot over a symbol denotes the derivative with respect to time .
due to the homogeneity and isotropy , the evolution equations do not contain the co - moving coordinate @xmath18 .
therefore , the evolution law is universal , and it can be considered at small @xmath18 , i.e. in the region of applicability of the non - relativistic mechanics ( @xmath23 ) and gravity law by newton , namely , the acceleration of free falling along the radius vector @xmath24 reads off @xmath25 where we introduce the komar tolman gravitational mass , where the symbol @xmath17 corresponds to the tensor of energy - momentum and its trace , and @xmath26 denotes the metric . therefore ,
the static gravitational potential is determined by not the energy density alone , but by the temporal component of the source .
the integration of temporal component over the volume gives the komar tolman mass , exactly .
] for a spherically symmetric distribution of the matter with the energy density @xmath16 and pressure @xmath27 inside the sphere with radius @xmath28 : @xmath29 so that @xmath30 that naturally does not include the co - moving coordinate @xmath14 .
in addition , the adiabatic law of energy conservation for the substance with the energy density @xmath16 and pressure @xmath27 reads off @xmath31 so under the substitutions of @xmath32 , @xmath33 , we get @xmath34 then , after multiplying ( [ ddot ] ) by @xmath35 we can easily integrate and get the friedmann equation @xmath36 where we have introduced the hubble constant @xmath37 , while in gr the constant of integration @xmath38 is presented by the sum of contributions given by a constant spatial curvature and a matter with equation of state @xmath39 .
these terms do not contribute to the komar tolman mass , i.e. they do not produce the gravitational force .
thus , we get the equations of gr for the evolution of homogeneous and isotropic universe ( [ conserv2 ] ) and ( [ hubble1 ] ) . then , the description of such evolution is reduced to the non - relativistic dynamics at @xmath40 .
the straightforward application of mond paradigm at @xmath40 implies the transition to the regime of modified law of gravity , when @xmath41 hence , @xmath42 notice that a direct observation of the scale factor evolution by the registration of type ia supernovae @xcite reliably shows that the accelerated universe expansion ( @xmath43 ) at present has changed the decelerated expansion ( @xmath44 ) , which took place at the redshift . ]
@xmath45 , where @xmath46 is extracted in the procedure of fitting the deceleration parameter @xmath47 by the linear function @xmath48 .
consequently , in vicinity of transition point @xmath49 , where @xmath50 , the validity of mond in cosmology is quite justified
. however , eq . ( [ mond - old ] ) at @xmath51 is inconsistent with the condition of homogeneity and isotropy of matter , since the actor @xmath52 explicitly depends on the co - moving coordinate @xmath14 , hence , such the modification of gravity law is inherently related to the inhomogeneity of matter distribution . in particular , eq .
( [ mond - old ] ) results in the _ vacuum catastrophe _ : the vacuum with @xmath53 is to be unstable , namely , due to eq .
( [ mond - old ] ) the inhomogeneous distribution of matter is initiated .
thus , the straightforward transfer of mond to the cosmology is theoretically inconsistent .
nevertheless , keeping in mind that mond is related to the spatial inhomogeneity of matter distribution , we can easily suggest its cosmological extrapolation for the homogeneous and isotropic universe , indeed : by setting @xmath54 we arrive to the evolution equation at @xmath55 @xmath56 which is consistent with the initial conditions imposed on the matter as well as with the vacuum stability .
on the other hand , the visible large scale structure ( lss ) is seen at the angle distances of @xmath57 radians , that corresponds to @xmath58 , where the hubble horizon @xmath59 at the current value @xcite @xmath60 empirically in mond @xcite @xmath61 so that if we put @xmath62 then we expect @xmath63 it is convenient to parameterize the quantity @xmath64 in the form @xmath65 whereas by the order of magnitude @xmath66 thus , we can assume that the change of regime in the cosmological extrapolation of mond as given by ( [ g0 ] ) is caused by the presence of spatial inhomogeneity of matter , that allows us theoretically to conform the existence of two regimes for the critical acceleration of gravity . in this paper
we investigate the scheme proposed in ( [ g0 ] ) and ( [ k ] ) for the cosmological extrapolation of mond with the parametrization in the transition region as derived in the holographic description of gravity as the entropic force @xcite at low temperatures ( the mond regime ) @xcite .
this way essentially differs from the relativistic theory of gravity for the mond paradigm offered by j.bekenstein in @xcite , wherein he constructed the action satisfying a set of requirements , which include the appropriate limit to the mond regime as well as , in principle , a correct description of cosmological effects , too .
the modification is very specific due to additional gravitational scalar and vector fields in the action .
additional fields appear in the modified gravity formulated by j.w.moffat in @xcite , too .
then , the gravitational constant becomes the field satisfying the equations of motion .
so , j.w.moffat has obtained the modification of gravity law @xcite similar to the mond , but it is distinguishable from the mond , while some cosmological effects can be treated in agreement with observations @xcite .
thus , approaches in @xcite and @xcite represent the axiomatic attempts to construct the consistent theory of modified gravity from the primary principles of field theory , and they are able to produce the framework for complete calculations including the propagation of cosmological perturbations in the relativistic way , of course . we follow a more pragmatic way : without an introduction of any new additional notions , the modification of gravitational dynamics in the form of ( [ g0 ] ) and ( [ k ] ) can be consistent with the mond at galactic scales , and it is enough for the investigation of modified cosmological evolution essentially different from the evolution in the framework of general relativity if we consider the ordinary components of matter .
therefore , the extrapolated mond could exhibit some general features , which are common for meaningful extensions of general relativity as they tend to exclude the dark matter from the cosmology _
dynamically_. in this way , we understand that the specific additional gravitational degrees of freedom defined explicitly as in @xcite and @xcite should be inevitably considered if we try to take into account the modification of theory for the propagation of matter inhomogeneity in cosmology , hence , in the construction of full range description of large scale structure of universe , that includes the matter power spectrum and anisotropy of cosmic microwave background radiation . in the general relativity , that is the dark matter , which is responsible for the actual description of all features in the large scale structure due to corrections to the matter distributions during the cosmological evolution , but it is commonly known that basic cosmological characteristics such as the accelerated - decelerated regimes of universe expansion , and the visible angular scale of matter inhomogeneity in the sky are mainly given by the appropriate evolution rate of universe , and we show that such the evolution can be achieved in the framework of cosmological extrapolation of mond without any dark matter
. this extrapolation does not provide us with the complete theory of inhomogeneity propagation ( especially , because such the theory is expected to be nonlinear in contrast to the well - working ordinary linear perturbation theory in the general relativity ) , hence , we do not suppose to reproduce the full description of angular dependence of cosmic microwave background radiation subject to the given primary spectrum of matter distribution , of course , or the big - bang nucleosynthesis , for instance .
nevertheless , the baryonic matter inhomogeneities themselves can produce the potential gravitational wells for the further concentration of matter , though such the wells would be affected by the sound propagation in the photon - electron - baryon plasma , and this evolution of inhomogeneities will be mainly determined by the strong field regime of mond , which is consistent with the general relativity , while the nonlinear effects are expected to be subleading .
thus , we can expect the correct reproduction of main features of large scale structure in the framework of cosmological extrapolation of mond .
so , our preliminary studies show that the first acoustic peak in the cosmic microwave background anisotropy is successfully obtained within the cosmological extrapolation of mond under the very standard settings of primary spectrum of matter power without any dark matter , of course , while the second peak can be also correctly described by an appropriate variation of the matter power spectrum , but the subleading structures need the construction of full theory of cosmic perturbations beyond of purposes of our paper .
so , the analysis of anisotropy in the cosmic microwave background radiation would be presented elsewhere , since it requires a deep and comprehensive investigation in details .
the similar note can be addressed to the problem of nucleosynthesis , too . here , we present the general scheme for the cosmological extrapolation of mond and its applications to the basic features of universe evolution . in this paper
, we apply the minimal scheme of ( [ g0 ] ) and ( [ k ] ) to the description of data on type ia - supernovae .
we show that in the redshift region accessible for observations , such the astronomical data can be reliably described in the proposed model if there are the cosmological constant and baryonic matter , only , without any dark matter , whereas the numerical estimate of ( [ estimatek ] ) is correct .
while comparing with the friedmann cosmology at presence of dark matter , we can evaluate the imitated contribution of dark matter @xmath67 to the critical density of @xmath68 .
the ratio of @xmath69 to the baryonic contribution @xmath70 is roughly determined by @xmath71 in order of magnitude . in accordance to the standard friedmann cosmology - model with the cold dark matter and cosmological constant ( @xmath72cdm ) , the parameters correspond to @xmath73 at @xmath74 .
further , we analyze the large scale structure of inhomogeneity in the matter distribution of universe in calculations of baryon acoustic oscillations , measured by observing the luminous matter in the sky versus the redshift @xcite , and in estimation of `` acoustic scale '' in the cmbr anisotropy @xcite .
the estimates obtained in the framework of cosmological extrapolation of mond agree with the measured values of lss without any postulation of hypothetic dark matter : it is enough to adjust the value of baryonic matter in the presence of cosmological constant .
then , the density of baryonic matter is roughly twice its value in @xmath72cdm of gr .
after modeling the acoustic scale in the cmbr anisotropy we calculate the critical acceleration @xmath75 in the framework of the cosmological extrapolation of mond and find preferable values of cosmological parameters of the universe ( the density of baryonic matter ) .
we analyze the evolution into the future , too . in conclusion
we summarize our results and discuss its meaning .
at present , measuring the stellar magnitude of type ia supernovae versus the redshift gives the only direct observation of scale factor versus time in cosmology .
such the observation does not base on any model of universe expansion , so that even the theory beyond the friedmann equation can be confronted to the experiment without any additional assumptions . in this respect , such the data are unique and they are suitable for testing the cosmological extrapolation of mond .
the stellar magnitude @xmath76 depends on the redshift according to the formula @xmath77 where @xmath78 is the absolute stellar magnitude , i.e. the stellar magnitude of light source at the distance of 10 pc .
the photometric distance @xmath79 measured in mpc , is determined by the hubble constant evolution @xmath80 where @xmath81 is the speed of light . in @xcite
the following dependence of the deceleration parameter on the redshift has been used as the working hypothesis : @xmath82 that allows it to describe the supernovae data at values @xmath83 essentially different from zeros , whereas the parameters significantly correlate ( see details in the original paper @xcite ) . according to ( [ q - lin ] ) , the hubble constant @xmath84 that should be reasonably compared with the case of zero deceleration parameter , when @xmath85 from eqs .
( [ mu])([h - nil ] ) we see that normalizing the data at low redshifts , i.e. measuring the value of @xmath86 , allows us to plot a fiducial dependence of supernovae stellar magnitude @xmath87 with the deceleration parameter equal to zero , and further , systematically to study the deviations from this dependence versus the increasing redshift .
data from @xcite are shown in figures [ ev1 ] and [ dif - ev1 ] .
they confidently signalize on the accelerated expansion of universe at present as well as on the existence of transition from the decelerated expansion to the accelerated one at @xmath88 .
the best fit of data takes place at parameter values listen in ( [ q - fit ] ) , as it shown in figure [ dif - ev1 ] by the dashed curve . in the cosmological extrapolation of mond we use the following equation for the acceleration at the presence of both the cosmological term with the energy density @xmath89 and baryonic matter with the density @xmath90 : @xmath91 where the region of transition from the friedmann regime to the mond is described by the 1-dimensional debye function as accepted in @xcite ,
hence , @xmath92 ( 145,108 ) ( 4.2,2.7 ) of type ia supernovae versus the redshift @xmath93 : the bottom dotted curve presents the case of zero deceleration parameter , the solid curve shows the result of cosmological model of mond , the top dotted curve depicts the fit in @xmath72cdm .
the supernovae discovered by the ground based telescopes are marked by rhombuses , the circles show the supernovae discovered by the hubble space telescope ( data from @xcite ) .
, title="fig : " ] ( 140,0)@xmath93 ( 39.75,2)@xmath94 ( 76.4,2)@xmath95 ( 109.5,2)@xmath96 ( 2,5.5)@xmath97 ( 2,32.7)@xmath98 ( 2,60.)@xmath99 ( 2,87.5)@xmath100 ( 2,105)@xmath76 ( 153,100 ) ( 4.5,3.5 ) for the type ia supernovae versus the redshift @xmath93 after subtracting the fiducial magnitude at zero deceleration parameter .
the solid line refers to the cosmological mond , the dashed line shows the best fit at the deceleration parameter @xmath101 @xcite .
the bottom panel serves for the systematical illustration of supernovae data as averaged for the similar redshifts .
the dotted and dashed - dotted lines represent the evolution with appropriate constant positive and negative deceleration parameter , respectively . , title="fig : " ] ( 140,0)@xmath93 ( 40.3,2)@xmath94 ( 78.15,2)@xmath95 ( 112.7,2)@xmath96 ( 0,9.3)@xmath102 ( 3.3,28)@xmath103 ( 3.3,46.75)@xmath95 ( 0,54.3)@xmath102 ( 3.3,73)@xmath103 ( 3.3,91.5)@xmath95 ( 1,99)@xmath104 so that at the acceleration greater than critical one , i.e. in the limit of standard cosmology at @xmath105 , we get @xmath106 while in the deep regime of mond , i.e. at @xmath107 , asymptotically we find @xmath108 and we arrive to eq .
( [ mond - new ] ) .
the critical density defined by @xmath109 allows us to write down the baryonic density as @xmath110 while the requirement of vanishing the acceleration at redshift @xmath49 leads to the following expression for the cosmological contribution : @xmath111 thus , the model for the cosmological extrapolation of mond is described by equation @xmath112 combined with initial data @xmath113 figures [ ev1 ] and [ dif - ev1 ] illustrate the results of cosmological mond at the typical value of @xmath74 as in @xmath72cdm , while @xmath114 so that in the region of available data , the calculated profile is practically indistinguishable from the fit obtained at the linear dependence of deceleration parameter on the redshift as performed in @xcite .
note , that the difference of stellar magnitudes presented in figure [ dif - ev1 ] is independent of the normalization of hubble constant @xmath86 , while data shown in figure [ ev1 ] allows us to fix both the value of @xmath86 at small @xmath93 and the fiducial curve at @xmath115 .
notice that at present , the regime of deep mond takes place in cosmology , so that from eq .
( [ cexmond2 ] ) we find @xmath116 hence , we can see that the value of baryonic matter density , i.e the parameter of @xmath117 , can get quite arbitrary variations in this test on the supernovae data , because the region of mond applicability near @xmath118 is completely described by the factor of @xmath119 , entering in eq .
( [ k0 ] ) , while the hubble constant at low redshifts is mainly determined by initial data , namely , by @xmath86 and @xmath120 , so that the influence of @xmath117 value on the quality of data description is quite weak . for instance , the twice increase of baryonic fraction is almost invisible on the curve derived from the cosmological mond as shown in figure [ dif - ev1 ] , i.e. the region of low redshifts is weakly sensitive to @xmath117 . in the framework of @xmath72cdm the deceleration parameter and redshift of transition from the deceleration to the acceleration
are given by the fractions of densities for the vacuum @xmath121 and matter @xmath122 , @xmath123 so that in the case of spatially flat universe , when @xmath124 , we get @xmath125 the quality of fitting the supernovae data for the flat universe in @xmath72cdm at @xmath126 , @xmath73 is slightly worse than in the case of the linear approximation ( [ q - lin ] ) for the deceleration parameter , but it has got a little bit of better quality at @xmath127 , @xmath128 @xcite .
the parameters brought from @xmath72cdm into mond give @xmath129 at @xmath130 we find @xmath131 .
however , by taking into account the fact that the point of maximal confidence level of data fit in @xmath72cdm is shifted from the standard values @xmath132 , @xmath73 to @xmath133 , @xmath128 , hence , numerically @xmath134 , we deduce @xmath135 , that is in agreement with our estimates obtained in consideration of changing the regime of homogeneous matter distribution to the limit of inhomogeneity in mond . thus , in the framework of cosmological extrapolation of mond we can reliably describe the type ia supernovae data by making use of baryonic matter and cosmological constant , only , without any postulating the dark matter . in this way
, the parameters of extrapolation is naturally conformed with the milgrom s critical acceleration in mond , applicable at scales , when the spatial inhomogeneity of baryonic matter distribution becomes significant .
the cosmological extrapolation of mond allows us , in fact , to calculate the present ratio of density of hypothetical dark matter to the baryonic density @xmath136 in terms of the slope of critical acceleration in its dependence on the distance , i.e. the parameter of @xmath64 , expressed in units of hubble constant at present .
therefore , the supernovae data lead to the necessary introduction of dark matter if we keep the friedmann evolution , i.e. if the evolution is determined by the newton law of gravity , indeed , while the modification of gravity law at low accelerations in the case of inhomogeneous distribution of matter in accordance with mond as well as for the homogeneous distribution of matter in cosmological mond , gives the confident description of data without any dark matter .
the universe evolution essentially transforms the spatial distribution of matter , that is observed as baryonic acoustic oscillations ( bao ) @xcite and cmbr anisotropy @xcite .
so , the results of survey of the matter distribution on the celestial sphere versus the redshift @xcite are particularly reduced to the ratio of sound horizon @xmath137 for the waves in the baryon - electron - photon substance as written in co - moving coordinates @xmath138 to the effective distance defined by @xmath139 where @xmath140 is the sound speed in the medium , while the angular distance is given by @xmath141 the measurements give the value of @xmath142 at @xmath143 and @xmath144 , which are shown in table [ table ] .
here @xmath145 is the redshift of epoch , when the interaction of baryonic and photonic components of substance via the compton scattering of electrons off photons and coulomb attraction of electrons to protons becomes negligible , i.e. when baryons stop to drag photons ( the finish of drag epoch ) .
@xmath146 \mbox{quantity } & \displaystyle\frac{r_s(z_d)}{d_v(0.2 ) } & \displaystyle\frac{r_s(z_d)}{d_v(0.35 ) } & l_a \\[3mm]\hline \mbox{exp.\footnotesize\cite{wmap , bao } } & 0.1905\pm 0.0061 & 0.1097\pm 0.0036 & 302.69\pm 0.76\\[1mm]%%%\hline \mbox{mond } & 0.1924 & 0.1149 & 302.5\\ \br \end{array } $ ] the speed of sound is given by the following expression : @xmath147 where @xmath148 is the baryon - photon ratio depending on the redshift @xmath149 introducing the redshift @xmath150 for the moment , when the density of non - relativistic matter @xmath67 equals to the density of radiation ( photons and neutrinos ) , allows us to write down the formula for the sound horizon as @xmath151 where @xmath152 analytical parameterizations of numerical calculations for @xmath150 , @xmath145 and @xmath148 are given in @xcite , wherein the dependencies on the cmbr temperature @xmath153 k and parameters @xmath154 , @xmath155 with @xmath156 being the coefficient for the hubble constant written down as @xmath157 , are explicitly presented .
thus , the universe evolution at low redshifts determines the angular distance @xmath158 in eq .
( [ da ] ) , which depends on the initial data on @xmath86 and @xmath120 , while the sound horizon @xmath137 and redshift @xmath145 are determined by fractions of baryonic and dark matters in the energy budget of universe .
since the evolution at low redshifts in the cosmological extrapolation of mond conforms with the supernovae data in the same way as it was done in @xmath72cdm , the angular distances are almost coincident in both these models , hence , the description of bao data is determined by fitting the sound horizon @xmath137 . in the case of the cosmological extrapolation of mond all of non - relativistic matter
is identified to baryons , i.e. @xmath159 .
hence , we find @xmath160 , i.e. it is twice and a half greater than the baryonic fraction in @xmath72cdm . at such the value of baryonic density we get the agreement with the bao data ( see table [ table ] ) . in numerical estimations
we have put @xmath161 , @xmath162 k and @xmath163 , @xmath164 as it was in the previous section . then , @xmath165 .
such the doubling can point to necessity of accounting for the baryonic matter outside luminous stars and visible gas , that is also numerically consistent with the lack of visible baryonic matter in some galactic clusters , which dynamics is described in mond without any dark matter ( see @xcite ) . at the same parameters we estimate the `` acoustic scale '' @xmath166 which is measured in the spectra of cmbr temperature anisotropy by locations of peaks of temperature fluctuations versus the multipole number in wmap data @xcite . here
@xmath167 is the redshift of decoupling , when due to the recombination of electrons with protons the medium becomes transparent for photons ( see analytical approximations for @xmath167 in terms of baryonic density , matter density and hubble constant in @xcite . ) . calculating the value of @xmath168
, we take into account for the contribution of relativistic particles into the hubble constant , since @xmath169 as we can see from table [ table ] , the cosmological extrapolation of mond is safely able to describe the large scale structure data given by the measurements of both the baryon acoustic oscillations of visible matter and the anisotropy of cmbr , since , as authors of @xcite have pointed out , this quantity can not be directly extracted from the angular spectrum of cmbr anisotropy because of a poor accuracy of measurements at low values of the multipole number , i.e. at large angles of correlations , wherein the shift parameter makes a significant influence on the amplitude of spectrum , while the high multipoles are not sensitive to @xmath170 .
it means that the choice of model for fitting the spectrum in the region of acoustic oscillations at @xmath171 determines the extrapolation to low values of multipoles , but it is not critical for the data at low multipoles because of poor accuracy of data at low @xmath172 .
consequently , the value of shift parameter given in @xcite is actually the model dependent - extrapolation , indeed . ] .
it is interesting that our argumentation about changing the mond regime at distances of the order of the scale for the spatial inhomogeneity of matter distribution , in fact , is confirmed by the calculation of acoustic scale @xmath173 , which points to the angular size of inhomogeneity about @xmath174 in consistency with our initial assumptions .
since the critical milgrom s acceleration in mond is experimentally measured with the high accuracy in the gas - reach galaxies @xcite @xmath175 it makes sense to use this information in order to refine of parameters in the cosmological extrapolation of mond .
indeed , within our approach , the milgrom s acceleration is determined by the angular scale of inhomogeneity in the matter distribution @xmath176 , so that in hubble units @xmath177 however , the acoustic scale gives @xmath178 then , by expressing @xmath71 in term of deceleration parameter @xmath120 and redshift @xmath49 for the transition from the universe deceleration to its acceleration in accordance with ( [ k0 ] ) , we find @xmath179 that allows us to plot the dependence of milgrom s acceleration on the baryon density at fixed values of @xmath120 and @xmath49 , which are determined by the supernovae data .
further , we can compare the result with the empirical value of @xmath180 known with the 12%-accuracy after taking into account for the uncertainty of the hubble constant .
( 80,56 ) ( 1,1 ) calculated in accordance to ( [ g0t2 ] ) at the deceleration parameter @xmath181 and the redshift of zero acceleration @xmath182 ( the curves from top to bottom , respectively ) as well as at given acoustic scale @xmath183 ( the dashed line ) versus the baryon fraction @xmath117 in comparison with the experimental value ( the shaded band).,title="fig : " ] ( 0,54)@xmath184 ( 77,0)@xmath117 the comparison is shown in figure [ g0tilde ] .
it turns out that the agreement of data on the supernovae ( sn ) , baryonic acoustic oscillations ( bao ) , acoustic scale ( wmap ) and hubble constant ( @xmath86 ) with the milgrom s acceleration is reached at the deceleration parameter @xmath185 and accessible 10%-variation of the transition redshift @xmath186 . in this way
, the account for the small error in the value of acoustic scale leads to @xmath187 \displaystyle \tilde g_0 & = & ( 1.10\pm0.03)\cdot 10^{-10 } { \mbox{\small m}}/{\mbox{\small s}^2},\\[2 mm ] \frac{r_s(z_d)}{d_v(0.2 ) } & = & 0.185\pm 0.006,\\[2 mm ] \frac{r_s(z_d)}{d_v(0.35 ) } & = & 0.112\pm 0.004 .
\end{array}\ ] ] thus , we successfully conform the logical consistency of two mond regimes : the cosmological limit and the locally inhomogeneous case ; and we extract the confident intervals for the universe parameters .
in the future @xmath188 , and the equation of cosmological mond is reduced to @xmath189 hence , according to ( [ k0 ] ) @xmath190 so that the limit is the de sitter unverse with @xmath191 $ ] , where the cosmological hubble constant @xmath192 since @xmath193 and @xmath194 , @xmath195{1-q_0 ^ 2}}-1.\ ] ] therefore , the future epochs at @xmath194 are constrained by condition ( [ constr ] ) as illustrated in figure [ futur ] .
( 79,48 ) ( 1,0 ) and the redshift of zero acceleration @xmath49 .
the preferable fit of wmap+bao+sn+@xmath86+@xmath75 data is depicted by the crossing of the shaded bands.,title="fig : " ] ( 68,-1)@xmath196 ( 2,48)@xmath49 ( 45,13)@xmath197 ( 40,33)@xmath198 substituting of @xmath181 as obtained by fitting the data , into eq . ( [ constr ] )
gives @xmath199 so that according to our estimates for @xmath49 in the previous section , it means that to the end of evolution @xmath200 ( see figure [ futur ] ) .
it is interesting to compare the cosmological term evaluated in mond in terms of @xmath201 with its value in gr : @xmath202 so that the equality of @xmath203 takes place at @xmath204 numerically @xmath205 therefore , the limit of de sitter universe and , hence , the problem of small - scale cosmological constant can not get an adequate description in gr , while in order to consider the vacuum we need essentially to modify the equation for the connection between the gravity and matter .
in this paper we have conformed the mond to its cosmological extrapolation , that allows us to eliminate hypothetical dark matter not only at scales of spatial inhomogeneity of baryon matter distribution in galaxies and galactic clusters , but also in the description of universe evolution up to the redshift of @xmath206 in the observation of type ia supernovae . in this conformation ,
the angular size of inhomogeneity in the large scale structure of universe is essential .
another question , related to the cosmology and modification of gravity , is the necessity to postulate the dark matter in order to describe the cmbr anisotropy in @xmath72cdm . to our point of view
, this problem is also closely connected to the usage of friedmann model of expansion at redshifts from zero to @xmath167 , when the cmbr inhomogeneity was formed .
clearly , parameters such as the contribution of cosmological term into the energy budget of universe and the fraction of dark matter can be mostly significant at low redshifts , exactly when the modification of gravitational law becomes remarkable at low accelerations of expansion .
however , the forming of cmbr essentially depends on the proportion between the baryons and dark matter . on the other hand ,
the evolution of the baryonic density versus the redshift is completely determined by the conservation law of matter , hence , the conformation of the baryon densities in both @xmath72cdm and mond allows us to generate similar conditions for the interaction between photons and electrons , which density correlates with baryons .
the problem of considering the anisotropy evolution versus the redshift at the whole interval in the framework of cosmological mond expects a solution , i.e. the question about a detailed investigation of cmbr anisotropy still remains open within the cosmological extrapolation of mond . in this respect
we have to mention the field theory - approaches in @xcite and @xcite , which can , in principle , provide us with complete calculations for the propagation of perturbations , whereas the theory by j.bekenstein @xcite includes the mond regime for inhomogeneous distribution of matter .
however , we follow another motivation based on the minimal extension of mond that satisfies the reasonable constraint on the transition to the cosmology of homogeneous distribution of matter . in this way
, we note that some integral quantities , which values are deduced by the consideration of cmbr anisotropy , are derived under the assumption of friedmann evolution and they can not be straightforwardly transferred to the modified law of gravity .
this note is also valid as concerns for the observation of large scale structure , i.e. in studying the baryon acoustic oscillationscdm the spatial inhomogeneity of both the matter and dark matter are generated , say , by the primordial spectrum of inhomogeneity as given by quantum fluctuations of a scalar filed having caused the universe inflation @xcite .
therefore , the baryonic and dark matter initially has got common locations of compression - decompression regions , that is consistent with the assumption about the joint concentration of baryonic and dark matter .
however , the baryon oscillations having the electromagnetic nature , suggest that the inhomogeneities of the photon - electron - baryon medium propagate as sound waves decoupled from the dark matter ( except the interaction due to gravitational forces ) , while the inhomogeneities of dark matter do not propagate at all , since there are no sound in the dark matter . during the evolution after the baryon acoustic oscillations stopped , the gravitational interaction attracts the centers of baryonic and dark matter concentration , that results in a partial nearing the locations of matter lumps .
the transfer function of initial perturbations in the energy densities consists of two different parts given by the terms of baryonic and dark matter .
consequently , initially coinciding regions of concentration for the baryons and dark matter become spatially separated , that creates different centers of gravitational contraction of baryons and dark matter in the future evolution .
thus , the existence of baryon acoustic oscillations challenges the dark matter hypothesis by asking for a natural explanation for the coincident locations of baryons and dark matter in galaxies , that rises the additional problem in a dynamical derivation of the tully fisher law in the framework of standard cosmology . ]
( bao ) caused by the spatial inhomogeneity @xcite .
nevertheless , we have analyzed the `` angular scale '' in the cmbr anisotropy and parameters of baryon acoustic oscillations in the framework of cosmological mond and showed that mond can confidently describe these cosmological phenomena .
then , we can conclude that the cosmological mond has exhibited its positive potential for the adequate description of universe evolution at low redshifts as well as up to the redshifts of cmbr forming .
we have to note , of course , that the purpose of this paper has been not a complex fitting the type ia supernovae data , baryon acoustic oscillation and cmbr anisotropy , but the aim has been the demonstration of opportunity to get the successful application of cosmological mond to those problems . finally , we have compared the critical acceleration calculated in the framework of cosmological mond with its empirical value and extract the intervals of cosmological parameters , wherein there is the good agreement of the model with observations of wmap+bao+sn+ @xmath86+@xmath75 .
in addition , we have found that the modification of gravity law is essential in studying the problem of cosmological constant .
thus , the preliminary estimates performed in the framework of cosmological mond conceptually permit for the confidential description of universe evolution and point to the necessity to work out a cumbersome analysis of complete base of the data on the supernovae , large scale structure , baryogenesis and abundance of elements in the universe etc .
, for the accurate extraction of cosmological parameters , in particularly , the baryonic fraction of energy . nevertheless , in this paper we have shown that hypothetical dark matter can be completely excluded from the cosmology by the appropriate modification of gravity law at accelerations less than the critical value in analogy to the case , when it was successfully done at the galactic scale in mond .
notice , there are several papers , wherein the influence of modified law of gravity to the universe evolution at low accelerations was considered @xcite , however , in those papers the acceleration is evaluated by the temperature of apparent horizon @xmath207 giving @xmath208 , while the critical acceleration is the constant value independent of distance . in @xcite
this approach follows the mond at the linear dependence of factor modifying the acceleration @xmath209 on ratio @xmath210 .
hence , it is clear that such the modification of gravity law at the hubble horizon is extrapolated inside the hubble sphere , that can be never conformed with our consideration .
this work was partially supported by the grant of russian foundations for basic research 10 - 02 - 00061 , the grant of special federal program `` scientific and academics personnel '' for the scientific and educational center 2009 - 1.1 - 125 - 055 - 008 , and the work of t.s.a .
was supported by the russian president grant mk-406.2010.2 as well as by the national science support foundation program `` the best graduates of russian academy of sciences '' .
99 milgrom m 1983 _ astrophys .
j. _ * 270 * 365 milgrom
m 2011 md or dm ? modified dynamics at low accelerations vs dark matter _ preprint _
1101.5122 [ astro-ph.co ] milgrom m 2008 the mond paradigm _ preprint _ 0801.3133 [ astro - ph ] milgrom m 2009 mond : time for a change of mind ?
_ preprint _ 0908.3842 [ astro-ph.co ] mcgaugh s s 2011 _ phys .
lett . _ * 106 * 121303 ( _ preprint _ 1102.3913 [ astro-ph.co ] ) komatsu e et al .
[ wmap collaboration ] 2010 _ astrophys .
j. suppl . _
* 192 * 18 ( _ preprint _ 1001.4538 [ astro-ph.co ] ) riess a g et al .
[ supernova search team collaboration ] 2004 _ astrophys .
j. _ * 607 * 665 ( _ preprint _ astro - ph/0402512 ) riess a g et al .
2007 _ astrophys . j. _ * 659 * 98 ( _ preprint _ astro - ph/0611572 )
astier p et al .
[ the snls collaboration ] 2006 _ astron .
astrophys . _ * 447 * 31 ( _ preprint _ astro - ph/0510447 ) wood - vasey w m et al . [ essence collaboration ] 2007 _ astrophys . j. _ * 666 * 694 ( _ preprint _ astro - ph/0701041 ) amanullah r et al .
2010 _ astrophys . j. _ * 716 * 712 ( _ preprint _ 1004.1711 [ astro-ph.co ] ) verlinde e p 2010 on the origin of gravity and the laws of newton _ preprint _ 1001.0785 [ hep - th ] padmanabhan t 2010 _ mod .
a _ * 25 * 1129 ( _ preprint _ 0912.3165 [ gr - qc ] ) kiselev v v and timofeev s a 2011 _ mod .
lett . a _ * 26 * 109 ( _ preprint _ 1009.1301 [ hep - th ] ) bekenstein j d 2004 _ phys . rev . d _ * 70 * 083509 [ erratum 2005 _ ibid _ * 71 * 069901 ] ( _ preprint _ astro - ph/0403694 ) moffat j w 2006 _ jcap _ * 0603 * 004 ( _ preprint _
gr - qc/0506021 ) moffat j w 2011 modified gravity or dark matter ?
_ preprint _ 1101.1935 [ astro-ph.co ] moffat j w and toth v t 2011 cosmological observations in a modified theory of gravity ( mog ) _ preprint _ 1104.2957 [ astro-ph.co ] percival w j , cole s , eisenstein d j , nichol r c , peacock j a , pope a c and szalay a s 2007 _ mon . not .
_ * 381 * 1053 ( _ preprint _ 0705.3323 [ astro - ph ] ) eisenstein d j and hu w 1998 _ astrophys . j. _ * 496 * 605 ( _ preprint _ astro - ph/9709112 ) hu w and sugiyama n 1996 _ astrophys . j. _ * 471 * 542 ( _ preprint _ astro - ph/9510117 ) bond j r , efstathiou g and tegmark m 1997 _ mon . not .
soc . _ * 291 * l33 ( _ preprint _ astro - ph/9702100 ) guth a h 1981 _ phys .
d _ * 23 * 347 linde a d 1982 _ phys
. lett .
b _ * 108 * 389 albrecht a and steinhardt p j 1982 _ phys .
lett . _ * 48 * 1220 linde a d 1983 _ phys .
b _ * 129 * 177 gao c 2010 _ phys . rev .
d _ * 81 * 087306 ( _ preprint _ 1001.4585 [ hep - th ] ) chang z , li m h and li x 2010 unification of dark matter and dark energy in a modified entropic force model _ preprint _ 1009.1506 [ gr - qc ] | regime of mond , which is used in astronomy to describe the gravitating systems of island type without the need to postulate the existence of a hypothetical dark matter , is generalized to the case of homogeneous distribution of usual matter by introducing a linear dependence of the critical acceleration on the size of region under consideration .
we show that such the extrapolation of mond in cosmology is consistent with both the observed dependence of brightness on the redshift for type ia supernovae and the parameters of large - scale structure of universe in the evolution , that is determined by the presence of a cosmological constant , the ordinary matter of baryons and electrons as well as the photon and neutrino radiation without any dark matter . |
we give the salient points of the method of solving the electron nuclear spin dynamics in a gaas quantum dot .
the system consists of an electron with spin vector @xmath122 and @xmath0 nuclear spins , @xmath123 , with zeeman energies
@xmath49 and @xmath124 under a magnetic field @xmath125 , respectively , where @xmath31 denotes both positions and isotope types ( @xmath126as , @xmath127ga and @xmath128ga ) . the interaction can be separated as `` diagonal '' ( or `` longitudinal '' ) terms which involve only the spin vector components along the field ( @xmath32 ) direction and `` off - diagonal '' ( or `` transverse '' ) terms which involve spin flips . because @xmath49 is much larger than @xmath24 and the strength of the direct nuclear - nuclear interactions ( e.g. , nuclear dipolar interaction ) ,
the off - diagonal part of the electron nuclear hyperfine interaction can be eliminated by a standard canonical transformation , with the second - order correction left as the hyperfine - mediated nuclear - nuclear interaction . for the same reason , the off - diagonal part of the nuclear - nuclear interaction includes only terms which conserve the zeeman energies , thus excluding the hetero - nuclear terms .
the total effective hamiltonian can be written as @xmath129 , with @xmath130 , @xmath131 , @xmath132 , and @xmath133 [ hamiltonian ] where @xmath18 are the eigenstates of @xmath134 , the summation with a prime runs over only the homo - nuclear pairs , the subscript @xmath135 denotes the hyperfine mediated nuclear - nuclear interaction , @xmath136 the off - diagonal part of the direct nuclear - nuclear interaction , @xmath137 the diagonal part of the direct nuclear - nuclear interaction , and @xmath138 the diagonal part of the contact electron - nuclear hyperfine interaction .
the hyperfine energy , determined by the electron orbital wavefunction , has a typical energy scale @xmath139 s@xmath140 for a dot with about @xmath76 nuclei @xcite , where @xmath141 is the hyperfine constant depending only on the element type .
the direct nuclear - nuclear interaction , which is short - ranged " ( referred here as decaying no slower than dipolar ) , has the near - neighbor coupling @xmath142 s@xmath140 .
the hyperfine mediated interaction , which couples any two nuclear spins that are in contact with the electron and is associated with opposite signs for opposite electron spin states , has an energy scale dependent on the field strength , @xmath143@xmath144 s@xmath140 for field @xmath145@xmath146 t. this hyperfine mediate interaction is differentiated from the short - range " direct nuclear - nuclear interaction by the qualifier infinite - range " .
we work in the interaction picture defined by @xmath147 and @xmath148 in which the dynamics are determined by @xmath16 .
the basis set of the bath are eigenstates of @xmath148 : @xmath149 . in eqn .
( [ hamiltonian ] ) , @xmath150 and @xmath151 are diagonal in this basis .
the off - diagonal terms @xmath152 , @xmath153 are weak perturbations that will excite the bath initially on an arbitrary configuration @xmath154 .
the elementary excitations in the nuclear spin bath are pair - flip excitations created by operators @xmath155 in the reduced hamiltonian . starting from any initial nuclear configuration , the evolution of the nuclear spin states by these elementary excitations is of the hierarchy as shown in the left side of fig .
[ hierarchy ] .
we can regard the zeroth layer of this hierarchy , the initial state , as the ` vacuum ' of the pair - flip excitations and layer @xmath31 corresponds to @xmath31 pair - flip excitations have been created .
the state at time @xmath8 is a linear superposition of all possibilities : @xmath156 where the summation over the indexes @xmath157 are defined such that @xmath34 , latexmath:[$\hat{j}_m^+\hat{j }
_ n^- latexmath:[$\hat{j}_l^+\hat{j}_p^- \hat{j}_m^+\hat{j}_n^- of @xmath148 orthogonal to each other .
we solve this dynamics in the nuclear spin bath based on a pseudo - spin model as described below .
we have , on sites labelled @xmath159 , the nuclear spin states @xmath160 with @xmath161 for nuclei of spin @xmath3 . as the elementary excitations are pair dynamics driven by @xmath162 , we first sort out the pair states @xmath163 .
these pair states are divided into three categories : \1 . _ down states _ :
a down state @xmath163 has a partner @xmath164 created by @xmath165 a down state must have ( @xmath166 ) .
there are @xmath167 down states for each bond .
_ up states _
: an up state @xmath163 has a partner @xmath168 created by @xmath169 .
the up state must have ( @xmath170 ) .
there are @xmath167 up states for each bond .
\3 . _ bachelor states _ : a single pair state has no partners connected by @xmath171 or its hermitian conjugate , i.e. , @xmath172 or @xmath173 . the bachelor states may be mapped to pseudo - spin 0 states .
since they are scalar states , their hamiltonian terms will commute with every other operators , they can only contribute to the phase factor in the electron spin coherence through the overhauser field , causing inhomogeneous broadening .
we shall find by an explicit construction that the up and down states can be paired to provide states of @xmath167 two - level systems - the pseudo - spins .
these states are divided into : \1 . _ monogamy states _
: each state belongs to only one pseudo - spin although its partner may be a bigamist .
these states are edge states in that at least one of the two spin quantum numbers ( @xmath29 or @xmath174 ) equal to @xmath175 but they can not both be equal to @xmath3 or to @xmath176 .
half of them ( @xmath177 states ) are down states , @xmath178(a ) .
the other @xmath177 states , @xmath179(c ) .
_ bigamy states _
: each belongs to two different pseudo - spins .
they are interior states : @xmath180 and @xmath181 .
there are @xmath182 of them , see fig .
[ map](b ) . for a bond between two sites
@xmath183 , all possible pseudo - spins excitations are sort out and labelled by using two numbers @xmath184 and @xmath185 for @xmath186 , @xmath187 the construction follows from the angular momentum addition of the two sites and is illustrated for @xmath188 , @xmath189 when @xmath190 , the two - spin states are @xmath191 value equal to or between @xmath192 and @xmath193 , the @xmath194 state at the bottom of the ladder is chosen to be the spin down of the doublet and one rung above to be its partner .
thus , the bottoms of the ladders yield the down monogamy states and the top rungs the up monogamy states . in between
, the states are the bigamy states . for any initial configuration @xmath34 , the relevant set of pseudo - spins @xmath195
is determined by examining every possible nuclear spin pair @xmath196 .
each pair will contribute @xmath197 , @xmath146 or @xmath111 pseudo - spins if @xmath198 is in the bachelor state , monogamy state or bigamy state configuration respectively ( see fig .
[ map ] ) .
the many nuclear spin initial state @xmath199 is then replaced by , @xmath200 where @xmath39 labels both the nuclear pair @xmath201 and the pseudo - spin type @xmath202 .
@xmath203 or @xmath204 depending on whether @xmath39 is mapped from up or down monogamy or bigamy state .
different initial nuclear configurations will result in different sets of pseudo - spins . for a randomly chosen initial configuration @xmath34 , the number of pseudo - spins is given by @xmath205 where @xmath0 is the total number of nuclear spins and @xmath206 the number of nuclei coupled to a particular nuclear spin by the nuclear - nuclear interaction . for
short ranged direct interaction , @xmath207 and for the infinite - ranged hyperfine mediated interaction @xmath208 .
the factor @xmath209 arises as the single state , monogamy state and bigamy state are contributing @xmath197 , @xmath146 and @xmath111 pseudo - spins respectively . for convenience ,
when the set @xmath210 is determined from the @xmath34 , we redefine the pseudo - spin up and down states , i.e. @xmath211 so that the initial state @xmath212 in this new definition corresponds to all pseudo - spins pointing ` up ' : @xmath213 .
the redefinition is conditioned on the initial nuclear configuration @xmath34 .
conditioned on the electron state @xmath18 , the pseudo - spins are driven by the effective hamiltonian of the form , @xmath214 the effective magnetic field @xmath215 on the pseudo - spins , conditioned on the electron spin state , are to be determined by reproducing the matrix elements , @xmath216 and @xmath217 , namely the energy cost and transition matrix element for nuclear pair - flips .
the pseudo - spin model for characterizing the nuclear spin bath dynamics is to approximate the exact evolution of eqn .
( [ expansion_exact ] ) by the independent evolution of all pseudo - spins in @xmath218 , @xmath219 this pseudo - spin dynamics can be put into a similar hierarchy as shown in the right part of fig .
[ hierarchy ] which we will refer to as model hierarchy in contrast to the exact hierarchy . with the mapping established for the state ( eqn .
( [ mapping_state ] ) ) and the hamiltonian ( eqn .
( [ mapping_hamiltonian ] ) ) , the first two layers of the exact hierarchy will be reproduced exactly by the model hierarchy , i.e. , there is a one to one correspondence between @xmath220 and @xmath221 with the energy and coupling to the initial state @xmath222 ( @xmath223 ) exactly reproduced .
difference between the model and the exact hierarchies arises when more than one excitations have been created in the system . in fig .
[ ovelap ] , we illustrate with the case when two pair - flip excitations have been created .
if the two pair - flips do not overlap as shown in fig .
[ ovelap](a ) , their dynamics are then independent of each other and well described by the pseudo - spin model .
fig.[ovelap](b - e ) illustrate the various situations that the two pair - flips overlap , by sharing one or two nuclei .
the flip - flop of the first nuclear pair @xmath224 changes the spin configuration of both nuclear @xmath225 and @xmath226 and if a second flip - flop is to take place on pair @xmath224 or @xmath183 or @xmath227 , it is no longer described by the dynamics of the original pseudo - spins assigned to it . instead ,
in the model hierarchy by the independent pseudo - spin model , two successive flip - flops on pair @xmath224 or two successive flip - flops on pair @xmath228 and @xmath201 respectively are shown in fig.[ovelap](f - h ) .
[ ovelap](g ) can be considered as the approximate form of fig .
[ ovelap](c ) and fig .
[ ovelap](h ) as that of fig .
[ ovelap](d ) .
the model hierarchy contains events like fig .
[ ovelap](f ) which is absent in the exact hierarchy and events like fig .
[ ovelap](e ) in the exact hierarchy is not contained in the model hierarchy . therefore , on layer @xmath111 , the model hierarchy coincides with the exact hierarchy in events described by fig .
[ ovelap](a ) and differ by replacing the events of fig.[ovelap](b - e ) with events of fig.[ovelap](f - h ) .
the difference in a general layer can be analyzed in the same way . by the pseudo - spin model
, we are using eqn .
( [ expansion_model ] ) as the bath state at time @xmath8 for calculating physical properties instead of eqn .
( [ expansion_exact ] ) .
the difference of the exact and model hierarchies is estimated below which serves as an upper bound for error estimation ( notice that the physical properties of interest are not necessarily changed by replacing events of fig.[ovelap](b - e ) with events of fig.[ovelap](f - h ) , therefore , this error - estimation is not necessarily a tight bound ) .
if @xmath229 pair - flip excitations have already been generated , to create the next excitation , we have @xmath230 pseudo - spin to choose from and @xmath231 of them overlap with the previous excitations .
therefore , the probability of having a new excitation without overlapping with the previous excitations is given by : @xmath232 . by induction , the probability of creating @xmath31 non - overlapping pair - flip excitations is then given by , @xmath233 \notag \\ & = & \mathrm{exp } \left [ -\frac{n(n-1)}{n}(\frac{2j+1}{2j})^2 \right ] \notag\end{aligned}\ ] ] the second @xmath234 holds if @xmath235 which is always true in the timescale relevant in our study . comparing the model hierarchy and the exact hierarchy , we find from the above analysis that they differ in layer @xmath31 with the relative amount of @xmath236 .
we perform a self - consistent analysis on validity of the pseudo - spin model . here
the relevant timescale plays the crucial role in determine the validity of the model . if only the first @xmath31 layers of the hierarchy ( the exact one and the model one ) are involved , the error in the calculated physical properties is bounded by @xmath237 if @xmath238 . therefore , in the very short time limit , only the first several layers of the hierarchy can be involved and the pseudo - spin model gives an almost exact account of the dynamics . to estimate error upper bound for the longer time limit
, we will calculate the number of layers involved ( denoted as @xmath31 ) based on the model hierarchy of the pseudo - spin model .
if @xmath239 obtained is small , we conclude that @xmath31 also faithfully reflects the number of layers involved in the exact hierarchy .
therefore , the error estimation based on the pseudo - spin model is faithful and any physical properties calculated based on pseudo - spin model is also a good approximation since the difference from the exact dynamics is small . otherwise , the approximation is not good .
it is established below that the condition @xmath238 is the origin of an upper bound and a lower bound on the quantum dot size @xmath0 for which our theory is able to deal with .
the validity of the pair - correlation approach ( pseudo - spin model ) is indeed in the mesoscopic regime . at any given time @xmath8
, an average excitation number @xmath240 can be defined as follows based on the model hierarchy , @xmath241 and our analysis @xcite shows that the layer - distribution of population in the hierarchy is of a normal distribution centered at @xmath242 , i.e. the population is distributed in layers from layer @xmath243 to @xmath244 . therefore , the quantity for characterizing the error upper bound is of a very simple form : @xmath245 . in the pseudo - spin model ,
@xmath240 defined in eqn .
( [ nflip ] ) has an equivalent expression which is more convenient for evaluation : @xmath246 where @xmath247 is the evolution operator for pseudo - spin @xmath39 .
the contribution can be divide into two parts:@xmath248 .
@xmath249 is the number of non - local pair - flip excitations and @xmath250 is the number of local pair - flip excitations that have been created . @xmath251 and @xmath252 have very different behavior and we analyze them separately . in free - induction evolution , the number of non - local pair - flip excitations is given by , @xmath253 where @xmath254 and @xmath255 is the number of non - local nuclear spin pairs . since the evolution of the non - local pair - correlation is completely reversed by the @xmath68 pulses , @xmath256 is also reversed and @xmath257 at each spin echo time .
therefore , @xmath258 does not accumulate in the pulse controlled dynamics and we just need to look at the maximum value of @xmath258 between echoes . for example , in the control with the equally spaced pulse sequence , in order to have the coherence well preserved or restored at spin echo time , the delay time between successive pulses is limited by @xmath259 , where @xmath260 is the hahn echo decay timescale and @xmath261 in gaas @xcite .
therefore , @xmath249 at any time is bounded by @xmath262 in this scenario .
the estimate of the bound on @xmath0 imposed by @xmath263 for the validity of our approach is therefore , @xmath264 for gaas fluctuation dot in a magnetic field of @xmath144 tesla , the above condition is well satisfied for @xmath265 .
estimation for other controlling pulse sequences give a lower bound on @xmath0 which is in the same order .
for the number of local pair - flip excitations , we have a similar expression in the free - induction evolution , @xmath266 where @xmath267 and @xmath268 is the number of local nuclear spin pairs .
@xmath269 ( @xmath270 in gaas ) is determined by number of local neighbors and the nuclear spin quantum number @xmath3 .
in contrast to the non - local pair dynamics , the local pair dynamics is _ not _ reversed under the influence of the electron spin flip and @xmath271 accumulates all through the time .
nonetheless , it turns out that @xmath272 holds for all scenarios of pulse controls being discussed .
therefore , the condition @xmath273 sets an upper bound on @xmath0 : @xmath274 , which depends on the time range @xmath8 we wish to explore .
alternatively speaking , @xmath273 sets an upper bound on the time range @xmath8 we can explore for some fixed @xmath0 using the pseudo - spin model .
we illustrate this bound using the following two examples .
if we wish to calculate the hahn echo signal using the pseudo - spin model , we shall have @xmath275 where the hahn echo decay time @xmath276 @xcite .
therefore , the upper bound on @xmath0 is given by @xmath277 .
for gaas quantum dot , this condition is well satisfied for @xmath278 . \2 .
for bath of an intermediate size in the allowed region of @xmath279 \gg 1 \gg \alpha^2 n^2b^2\mathcal{a}^{-2}$ ] , e.g. , a quantum dot of typical size @xmath280 in our problem , @xmath281 is satisfied for a much longer time range @xmath282 . in summary , for the pair - correlation approximation ( or pseudo - spin model ) to be valid , nuclear spin dynamics of local pair - flips imposes an upper bound on @xmath0 while nuclear spin dynamics of non - local pair - flips imposes a lower bound . within this mesoscopic regime ,
the pair - correlation approximation is well justified .
the error estimation is based on characterizing the difference in the hilbert space structure of the exact dynamics and that of the pseudo - spin model and assuming this difference has a full influence on the electron spin coherence calculation .
therefore , the bound is not necessarily tight and it is possible that the pulse control methodology developed using the pseudo - spin model have actually a much larger validity regime .
investigation is underway . | for a two - state quantum object interacting with a slow mesoscopic interacting spin bath , we show that a many - body solution of the bath dynamics conditioned on the quantum - object state leads to an efficient control scheme to recover the lost quantum - object coherence through disentanglement .
we demonstrate the theory with the realistic problem of one electron spin in a bath of many interacting nuclear spins in a semiconductor quantum dot .
the spin language is easily generalized to a quantum object in contact with a bath of interacting multi - level quantum units with the caveat that it is mesoscopic and its dynamics is slow compared with the quantum object .
coherent superposition of states of a quantum object is the wellspring of quantum properties and key to quantum technology .
decoherence of a quantum object results from the entanglement with an environment by coupled dynamics @xcite .
amelioration of decoherence becomes important in any sustained quantum process .
different types of amelioration include dynamical decoupling @xcite , decoherence - free subspace @xcite , quantum error correction ( for a review , see @xcite ) , and feedback control @xcite . we offer an alternate approach to the restoration of coherence based on the theory that control of the quantum object can direct the quantum evolution of the bath to disentangle the object from the bath .
the operation resembles the spin echo schemes @xcite but it removes the pure decoherence due to bath interaction dynamics as well as the inhomogeneous broadening effect .
the key is that the environment is effectively a mesoscopic system , i.e. , the number of particles @xmath0 is small enough for the timescale of the quantum object decoherence to be much smaller than its energy relaxation time @xmath1 while large enough for ergodicity , specifically for the poincar period to be effectively infinite as compared to @xmath1 .
the theoretical demonstration of coherence restoration uses one electron spin in a semiconductor quantum dot of many ( @xmath2 ) nuclear spins , which serves as a paradigmatic system of a two - level system in a bath of interacting spins for decoherence physics @xcite and for spin - based quantum technology @xcite .
electron spin decoherence due to the hyperfine interaction with the nuclear spins has been much studied ( see @xcite and references therein ) .
theories of the effect of interaction between nuclear spins on the electron decoherence have recently appeared @xcite .
our theory of coherence recovery by disentanglement is based on the previous finding @xcite that the mesoscopic bath of slow dynamics is well described by a simple pseudospin model for the particle pair interaction in the bath .
we start by defining the decoherence process in a simple quantum prescription .
the model for the coupled spin - bath system is a localized electron of spin 1/2 coupled to a bath of finite @xmath0 mutually interacting nuclei with spin @xmath3 in a magnetic field .
the isolation of the electron spin plus the meso - bath , in the timescale of the total processing duration @xmath4 , from the rest of universe arises out of their weak coupling with the outside .
the initial state of the electron spin , @xmath5 , is prepared as a coherent superposition of the spin up and down states @xmath6 in an external magnetic field .
the state of the total system of the spin plus bath at that instant forms an _ unentangled _ state , @xmath7 .
it evolves over time @xmath8 to the _ entangled _
@xmath9 where the bath states @xmath10 are different .
the electron spin state is now given by the reduced density matrix by tracing over the bath states , @xmath11 , @xmath12 .
the environment - driven shifting between @xmath13 and @xmath14 is longitudinal relaxation . either off - diagonal element gives a measure of the coherence of the spin .
the longitudinal relaxation contributes to the decoherence . when this contribution is removed , the remaining decoherence is called pure dephasing . for applications in quantum technology
, the longitudinal relaxation time @xmath1 can be made much longer than the processing duration @xmath4 by a choice of system and of the electron zeeman splitting much larger than the dominant excitation energies in the bath and the spin - bath coupling strength @xcite . when the cause of spin flip is removed , the reduced hamiltonian of the whole system is in the form diagonal in the electron spin basis , @xmath15 .
the bath evolves under the hamiltonians @xmath16 into separate states @xmath17 depending on the electron basis states @xmath18 .
pure dephasing is then measured by @xmath19 .
the electron spin coherence may be restored by exploiting the dependence of the bath dynamics on the electron spin states to make the bifurcated bath pathways intersect at a later time , i.e. , @xmath20 , leading to disentanglement . at temperature ( @xmath21mk@xmath22k ) @xmath23 the nuclear zeeman energy @xmath24 ( @xmath25 mk ) @xmath23 nuclear spin interaction ( @xmath26 nk ) , the nuclear bath initially has no off - diagonal coherence and is described by @xmath27 where @xmath28 , @xmath29 is the quantum number for @xmath30 , the component of the @xmath31th nuclear spin along @xmath32 ( the magnetic field direction ) , and @xmath33 gives thermal distribution .
the essence of electron decoherence is contained in the consideration of each pure bath state @xmath34 and later the ensemble average over @xmath35 is included .
the transverse interaction @xmath36 between two bath spins creates the pair - flip excitation , @xmath37 .
we sort out all such elementary excitations from the vacuum " state @xmath34 and denote each by the flip - process of a _ pseudospin _ @xmath38 indexed by @xmath39 : @xmath40 , characterized by the energy cost @xmath41 and the transition matrix element @xmath42 depending on the electron @xmath18 state .
@xmath43 ( @xmath44 ) is from the longitudinal interaction of the form @xmath45 between the electron spin @xmath46 and each of the two nuclear spins @xcite .
@xmath47 ( @xmath48 , @xmath49 being the electron zeeman energy ) is the extrinsic coupling between two nuclear spins mediated by virtual spin - flips of the single electron @xcite .
its dependence on the number of particles in the bath signifies its mesoscopic nature .
@xmath50 is due to the transverse part of the intrinsic ( e.g. dipolar ) interaction between the nuclear spins with strength @xmath51 .
the extrinsic interaction @xmath52 couples any two spins in the meso - bath , as opposed to the finite - range intrinsic interaction @xmath50 .
@xmath53 ( @xmath51 ) is due to the longitudinal part of the intrinsic nuclear interaction .
in the nuclear bath with the descending order of parameters , @xmath54 , the bath dynamics is slow and the density of pair - flip excitations created from the vacuum " state @xmath34 is much less than unity in timescale of interest @xcite .
the excitations are almost always spatially separated , leading to the pair - correlation approximation @xcite which treats pair - flips as independent of each other .
the bath is then driven by the effective hamiltonian derived from the first - principles interactions @xcite , @xmath55 where @xmath56 is the pauli matrix for pseudospin @xmath39 driven by a pseudo - magnetic field @xmath57 depending on the electron @xmath18 state . from the justification that correlations of more than two spins are negligible @xcite
, we derive the restrictions which the decoherence timescale places on the size of the bath @xmath0 , given by @xmath58 , to be established below .
the upper bound for @xmath0 distinguishes the bath from a macroscopic system .
it comes from the dominance of the pair correlation in the interaction dynamics of the bath spins over the correlations of more than two particles due to the intrinsic interaction .
the lower bound , @xmath59 , is by a similar consideration but due to the extrinsic interaction of the bath spins .
the lower bound @xmath60 simply signifies the necessary statistics for decoherence . in the case of the electron spin in a gaas quantum dot ,
the theory is well justified for @xmath61 which covers quantum dots of all practical sizes .
the theory of the interacting nuclear spin dynamics dominated by the pair excitation in the form of the pseudospin evolution leads to a simple physical picture of coherence decay and restoration .
the initial unpolarized bath state @xmath62 .
each pseudospin , representing a nuclear spin states pair , initially points along the pseudospin @xmath63-axis and then precesses about the pseudo - magnetic field @xmath64 , latexmath:[$|\psi_k^{\pm } \rangle \equiv e^{-\frac{i}{2}{\mathbf h}^{\pm}_k\cdot\hat{\boldsymbol \sigma}_k t } @xmath6 state .
thus , the electron spin coherence is measured by the divergence of the pseudospin paths , @xmath66 .
@xmath67 is the geometric distance between the two conjugate pseudospin paths on bloch sphere .
now we examine the consequences of the pseudospin echo .
a fast @xmath68-pulse applied at @xmath69 to flip the electron spin @xcite would cause the pseudo - spin evolution @xmath70 to find out how to control the decoherence , we neglect for the time being the diagonal nuclear spin interaction @xmath53 , which contributes to the same component of the pseudo - magnetic field as @xmath71 but much smaller . because the pseudo - fields dominated by the extrinsic nuclear spin interaction , @xmath72 , invert exactly into each other , disentanglement of the electron spin from the affected bath spin pairs follows at @xmath73 as in the classic spin echo to remove the inhomogeneous broadening effect .
the pseudo - fields dominated by the intrinsic interaction , @xmath74 , do not exactly invert under the influence of the electron spin flip and the resultant pseudospin trajectories are illustrated in fig .
[ fig1](a ) .
disentanglement from the affected nuclear pairs and , hence , recovery of the electron spin coherence occurs , by the rotation kinematics , at time @xmath75 , distinct from the classic echo .
fig .
[ fig1](b , c ) give the computed results for the electron spin coherence in a dot of @xmath76 nuclear spins in gaas which reveals the coherence recovery after a flip of the electron spin at a range of values for @xmath77 , even after the coherence has apparently vanished .
furthermore , the coherence may be restored by a sequence of electron spin flips .
for example , with a sequence of @xmath68-pulses evenly spaced with interval @xmath77 , the disentanglement from the bath will occur at @xmath78 between the @xmath31th and the @xmath79th pulses , as illustrated in fig .
[ fig1](d ) . consider the correction from the small term @xmath53 in the pseudo - fields , the residue decoherence , at the disentanglement point @xmath80 , is measured by @xmath81 .
compared to the free - induction decay @xcite where @xmath82 , the decoherence is reduced by a factor of @xmath83 ( @xmath84 for @xmath85s ) .
ensemble average over the mixed bath states is necessary in two scenarios , namely , observation of decoherence of an ensemble of quantum objects and observation of a single quantum object repeated in a time sequence @xcite .
the coherence of the electron spin is now @xmath86 , where @xmath87 is the inhomogeneous broadening factor due to the probability distribution @xmath33 of the initial bath state @xmath34 ( different nuclear bath state may result in different overhauser energy - splitting @xmath88 of the electron ) @xcite .
@xmath89 $ ] under the control of a sequence of @xmath68-rotations on electron spin at @xmath90 and @xmath91 .
the coherence factor @xmath92 is insensitive , up to a factor of @xmath93 , to the selection of initial bath state @xmath94 ( verified by numerical evaluations ) , and is taken out of the summation @xcite .
both the inhomogeneous broadening and the pure decoherence due to the extrinsic bath spin interaction are shown to be removed at the classic spin echo time @xmath95 in contrast to the unusual recovery time of @xmath96 in the case of intrinsic bath spin interaction .
we need a pulse sequence to produce a time where the decoherence from all three sources can be removed .
a solution is a two - pulse control .
fig .
[ fig2](a ) shows that , after a second electron spin flip at @xmath97 , the paths of the two pseudospin states corresponding to the electron @xmath18 states driven by the intrinsic nuclear spin pair interaction cross again at @xmath98 , coinciding with the secondary spin echo time for the other two causes .
this two pulse sequence is well - known as carr - purcell sequence in nmr spectroscopies @xcite .
the residual decoherence at @xmath99 is @xmath100 .
the restoration by two - pulse control of coherence in the presence of both pure and ensemble decoherence is demonstrated by the results of numerical evaluation shown in fig .
[ fig2](b , c ) .
the electron spin coherence is restored at @xmath101 by the second pulse even when the first spin - echo at @xmath95 has completely vanished , illustrating the remarkable observation @xcite that the absence of spin echo does not mean irreversible loss of coherence .
the power of concatenation of pulse sequences has been shown in the context of dynamical decoupling in quantum computation @xcite .
similarly , the control of disentanglement of the bath states from the quantum object may be enhanced by concatenation . the pseudo - spin evolution with the two - pulse control of the quantum object
can be constructed recursively from the free - induction evolution @xmath102 , by the concatenation , @xmath103 , @xmath104 .
the process can be extended by iteration , fig .
[ fig2](d ) , to any level , @xmath105 , where @xmath106 is the rotation vector along the axis of rotation through an angle @xmath107 .
disentanglement occurs at @xmath108 coinciding with the classic spin echo . for small @xmath107 ,
the recursion relation is @xmath109 . at each iteration ,
the rotation vectors of the conjugate pseudo - spin states have their mean @xmath110 increased by a factor of @xmath111 and their difference @xmath112 reduced by a factor of @xmath113 ( deduced by induction from @xmath114 ) .
the decoherence is reduced by an order of @xmath115 at @xmath116 for each additional level of concatenation till saturation at the level @xmath117 .
hence , the coherence echo magnitude scales with the echo delay time according to @xmath118 as shown in fig .
[ fig2](e ) .
our result shows the protection of electron spin coherence by pulse sequences with interpulse interval up to @xmath119s in conclusion , we note that our scheme of restoring the coherence depends on the pure decoherence being driven by the interaction in the spin bath and by the domination of the bath pair excitation in the slow bath dynamics .
the pulse sequence design is borrowed from the dynamical decoupling schemes in nmr spectroscopies @xcite and in quantum computation @xcite but the disentanglement method aims directly at the bath dynamics .
our method seeks not to eliminate the object - bath interaction by dynamical averaging , but to disentangle by controlling the quantum object to maneuver the bath evolution .
thus , elimination of coupling between the quantum object and the bath is not a necessary condition for their disentanglement , as illustrated by coherence recovery at @xmath120 where effective object - bath interaction does not vanish even in the first order of hyperfine coupling @xmath121 .
direct observation of coherence echoes at such magic times is possible with the narrowing of inhomogeneous distribution by measurement projection @xcite .
the control of bath spins may well develop into a valuable addition to the collection of armaments of coherence preservation for quantum information processing .
we acknowledge support from nsf dmr-0403465 , aro / lps , and darpa / afosr . 26 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , , , , , , _ _ ( , , ) , ed . , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( )
; , * * , ( ) ; , , , * * , ( ) . , _ _ ( , , ) . , , , * * , ( ) . , _ _ ( , , ) , ed . , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , ( ) . , , , , ,
* * , ( ) . , , , , , , * * , ( ) . , , , , , , , * * , ( ) .
the orbital wavefunction of the electron covers all @xmath0 nuclei , so the contact hyperfine coupling with each nuclear spin is @xmath121 , inverse proportional to the bath size @xmath0 . , , , , * * , ( ) . , , , , , , , , , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , , , , , , ,
* * , ( ) . , , , * * , ( ) .
, , , , , ( ) ; , , , * * , ( ) ; , , , , * * , ( ) .
* supplementary information for `` restoring coherence lost to a slow interacting mesoscopic bath '' * |
in this paper , we study the following hypothesis testing problem introduced by @xcite .
one observes an @xmath0-dimensional vector @xmath1 .
the null hypothesis @xmath2 is that the components of @xmath3 are independent and identically distributed ( i.i.d . )
standard normal random variables .
we denote the probability measure and expectation under @xmath2 by @xmath4 and @xmath5 , respectively . to describe the alternative hypothesis @xmath6 , consider a class @xmath7 of @xmath8 sets of indices such that @xmath9 for all @xmath10 .
under @xmath6 , there exists an @xmath11 such that @xmath12 where @xmath13 is a positive parameter .
the components of @xmath3 are independent under @xmath6 as well .
the probability measure of @xmath3 defined this way by an @xmath11 is denoted by @xmath14 .
similarly , we write @xmath15 for the expectation with respect to @xmath14 . throughout , we will assume that every @xmath11 has the same cardinality @xmath16 .
a test is a binary - valued function @xmath17 .
if @xmath18 then we say that the test accepts the null hypothesis , otherwise @xmath2 is rejected .
one would like to design tests such that @xmath2 is accepted with a large probability when @xmath3 is distributed according to @xmath4 and it is rejected when the distribution of @xmath3 is @xmath14 for some @xmath19 . following @xcite , we consider the risk of a test @xmath20 measured by @xmath21 this measure of risk corresponds to the view that , under the alternative hypothesis , a set @xmath22 is selected uniformly at random and the components of @xmath3 belonging to @xmath23 have mean @xmath24 . in the sequel , we refer to the first and second terms on the right - hand side of ( [ eq : bayes_risk ] ) as the type i and type ii errors , respectively .
we are interested in determining , or at least estimating the value of @xmath24 under which the risk can be made small .
our aim is to understand the order of magnitude , when @xmath0 is large , as a function of @xmath0 , @xmath25 , and the structure of @xmath26 , of the value of the smallest @xmath24 for which risk can be made small .
the value of @xmath24 for which the risk of the best possible test equals @xmath27 is called _ critical_. typically , the @xmath0 components of @xmath3 represent weights over the @xmath0 edges of a given graph @xmath28 and each @xmath11 is a subgraph of @xmath28 .
when @xmath29 then the edge @xmath30 is `` contaminated '' and we wish to test whether there is a subgraph in @xmath26 that is entirely contaminated . in @xcite , two examples were studied in detail . in one case
, @xmath26 contains all paths between two given vertices in a two - dimensional grid and in the other @xmath26 is the set of paths from root to a leaf in a complete binary tree . in both cases ,
the order of magnitude of the critical value of @xmath24 was determined .
@xcite investigate another class of examples in which elements of @xmath26 correspond to clusters in a regular grid .
both @xcite and @xcite describe numerous practical applications of problems of this type .
some other interesting examples are when @xmath26 is : * the set of all subsets @xmath31 of size @xmath25 ; * the set of all cliques of a given size in a complete graph ; * the set of all bicliques ( i.e. , complete bipartite subgraphs ) of a given size in a complete bipartite graph ; * the set of all spanning trees of a complete graph ; * the set of all perfect matchings in a complete bipartite graph ; * the set of all sub - cubes of a given size of a binary hypercube .
the first of these examples , which lacks any combinatorial structure , has been studied in the rich literature on multiple testing ; see , for example , @xcite , @xcite , @xcite and the references therein . as pointed out in @xcite , regardless of what @xmath26 is
, one may determine explicitly the test @xmath32 minimizing the risk .
it follows from basic results of binary classification that for a given vector @xmath33 , @xmath34 , if and only if the ratio of the likelihoods of @xmath35 under @xmath36 and @xmath4 exceeds @xmath37 . writing @xmath38 and @xmath39 for the probability densities of @xmath4 and @xmath14 , respectively , the likelihood ratio at @xmath35 is @xmath40 where @xmath41 .
thus , the optimal test is given by @xmath42 the risk of @xmath32 ( often called the bayes risk ) may then be written as @xmath43 we are interested in the behavior of @xmath44 as a function of @xmath26 and @xmath24 .
clearly , @xmath44 is a monotone decreasing function of @xmath24 .
( this fact is intuitively clear and can be proved easily by differentiating @xmath44 with respect to @xmath24 . ) for @xmath24 sufficiently large , @xmath44 is close to zero while for very small values of @xmath24 , @xmath44 is near its maximum value @xmath37 , indicating that testing is virtually impossible .
our aim is to understand for what values of @xmath24 the transition occurs .
this depends on the combinatorial and geometric structure of the class @xmath26 .
we describe various general conditions in both directions and illustrate them on examples .
@xcite also consider the risk measure @xmath45 clearly , @xmath46 and when there is sufficient symmetry in @xmath20 and @xmath26 , we have equality .
however , there are significant differences between the two measures of risk .
the alternative measure @xmath47 obviously satisfies the following monotonicity property : for a class @xmath26 and parameter @xmath13 , let @xmath48 denote the smallest achievable risk . if @xmath49 are two classes then for any @xmath24 , @xmath50 .
in contrast to this , the `` bayesian '' risk measure @xmath51 does not satisfy such a monotonicity property as is shown in section [ nonmonotone ] . in this paper
, we focus on the risk measure @xmath51 . throughout the paper we assume , for simplicity , that each set @xmath19 has the same cardinality @xmath25 .
we do this partly in order to avoid technicalities that are not difficult but make the arguments less transparent . at the same time , in many natural examples this condition is satisfied . if @xmath26 may contain sets of different size such that all sets have approximately the same number of elements , then all arguments go through without essential changes . however , if @xmath26 contains sets of very different size then the picture may change because large sets become much easier to detect and small sets can basically be ignored .
another approach to handle sets of different size , adopted by @xcite , is to change the model of the alternative hypothesis such that the level @xmath24 of contamination is appropriately scaled depending on the size of the set @xmath23 .
the paper is organized as follows . in section [ simpletests ] , we briefly discuss two suboptimal but simple and general testing rules ( the _ maximum test _ and the _ averaging test _ ) that imply sufficient conditions for testability that turn out to be useful in many examples . in section [ lowerbounds ] ,
a few general sufficient conditions are derived for the impossibility of testing under symmetry assumptions for the class . in section
[ examples ] , we work out several concrete examples , including the class of all @xmath25-sets , the class of all cliques of a certain size in a complete graph , the class of all perfect matchings in the complete bipartite graph and the class of all spanning trees in a complete graph . in section [ nonmonotone ] ,
we show that , perhaps surprisingly , the optimal risk is not monotone in the sense that larger classes may be significantly easier to test than small ones , though monotonicity holds under certain symmetry conditions . in the last two sections of the paper , we use techniques developed in the theory of gaussian processes to establish upper and lower bounds related to geometrical properties of the class @xmath26 . in section
[ hellinger ] , general lower bounds are derived in terms of random subclasses and metric entropies of the class @xmath26 .
finally , in section [ typeone ] we take a closer look at the type i error of the optimal test and prove an upper bound that , in certain situations , is significantly tighter than the natural bound obtained for a general - purpose maximum test .
as mentioned in the , the test @xmath32 minimizing the risk is explicitly determined .
however , the performance of this test is not always easy to analyze . moreover ,
efficient computation of the optimal test is often a nontrivial problem though efficient algorithms are available in many interesting cases .
( we discuss computational issues for the examples of section [ examples ] . ) because of these reasons , it is often useful to consider simpler , though suboptimal , tests . in this section ,
we briefly discuss two simplistic tests , a test based on averaging and a test based on maxima .
these are often easier to analyze and help understand the behavior of the optimal test as well . in many cases ,
one of these tests turn out to have a near - optimal performance .
perhaps the simplest possible test is based on the fact that the sum of the components of @xmath3 is zero - mean normal under @xmath4 and has mean @xmath52 under the alternative hypothesis .
thus , it is natural to consider the _ averaging test _
@xmath53 [ average ]
let @xmath54 .
the risk of the averaging test @xmath20 satisfies @xmath55 whenever @xmath56 observe that under @xmath4 , the statistic @xmath57 has normal @xmath58 distribution while for each @xmath19 , under @xmath14 , it is distributed as @xmath59 .
thus , @xmath60 .
another natural test is based on the fact that under the alternative hypothesis for some @xmath19 , @xmath61 is normal @xmath62 .
consider the _ maximum test _ @xmath63 the test statistic @xmath64 is often referred to as a _ scan statistic _ and has been thoroughly studied for a wide range of applications ; see @xcite . here , we only need the following simple observation .
[ maxtest ] the risk of the maximum test @xmath20 satisfies @xmath55 whenever @xmath65 in the analysis , it is convenient to use the following simple gaussian concentration inequality ; see @xcite .
[ tsirelson ] let @xmath66 be an vector of @xmath0 independent standard normal random variables .
let @xmath67 denote a lipschitz function with lipschitz constant @xmath68 ( with respect to the euclidean distance ) .
then for all @xmath69 , @xmath70 proof of proposition [ maxtest ] simply note that under the null hypothesis , for each @xmath19 , @xmath71 is a zero - mean normally distributed random variable with variance @xmath72 . since @xmath64 is a lipschitz function of @xmath3 with lipschitz constant @xmath73 , by tsirelson s inequality , for all @xmath74 , @xmath75 on the other hand , under @xmath14 for a fixed @xmath19 , @xmath76 and therefore @xmath77 which completes the proof .
the maximum test is often easier to compute than the optimal test @xmath32 , though maximization is not always possible in polynomial time .
if the value of @xmath78 is not exactly known , one may replace it in the definition of @xmath20 by any upper bound and then the same upper bound will appear in the performance bound .
proposition [ maxtest ] shows that the maximum test is guaranteed to work whenever @xmath24 is at least @xmath79 .
thus , in order to better understand the behavior of the maximum test ( and thus obtain sufficient conditions for the optimal test to have a low risk ) , one needs to understand the expected value of @xmath64 ( under @xmath4 ) .
as the maximum of gaussian processes have been studied extensively , there are plenty of directly applicable results available for expected maxima .
the textbook of @xcite is dedicated to this topic . here
, we only recall some of the basic facts .
first , note that one always has @xmath80 but sharper bounds can be derived by chaining arguments ; see @xcite for an elegant and advanced treatment .
the classical chaining bound of @xcite works as follows . introduce a metric on @xmath26 by @xmath81 where @xmath82 denotes the hamming distance . for @xmath74 ,
let @xmath83 denote the @xmath84-covering number of @xmath26 with respect to the metric @xmath85 , that is , the smallest number of open balls of radius @xmath84 that cover @xmath26 .
by dudley s theorem , there exists a numerical constant @xmath86 such that @xmath87 where @xmath88 denotes the diameter of the metric space @xmath26 .
note that since @xmath89 for all @xmath19 , @xmath90 .
dudley s theorem is not optimal but it is relatively easy to use .
dudley s theorem has been refined , based on `` majorizing measures , '' or `` generic chaining '' which gives sharp bounds ; see , for example , @xcite .
[ rmk : vc ] in certain cases , it is convenient to further bound dudley s inequality in terms of the vc dimension ; see @xcite .
recall that the vc dimension @xmath91 of @xmath26 is the largest positive integer @xmath92 such that there exists an @xmath92-element set @xmath93 such that for all @xmath94 subsets @xmath95 there exists an @xmath11 such that @xmath96 .
@xcite proved that the covering numbers of @xmath26 may be bounded as @xmath97 so by dudley s bound , @xmath98 an interesting alternative to the maximum test , proposed and investigated by @xcite and @xcite , is based on the idea that under the null hypothesis the distribution of the vector @xmath3 does not change if the sign of each component is changed randomly , while under the alternative hypothesis the distribution changes . in @xcite and @xcite ,
methods based on symmetrization and bootstrap are suggested and analyzed .
such tests are meaningful and interesting in the setup of the present paper as well and it would be interesting to analyze their behavior .
in this section , we investigate conditions under which the risk of any test is large .
we start with a simple universal bound that implies that regardless of what the class @xmath26 is , small risk can not be achieved unless @xmath24 is substantially large compared to @xmath99 .
an often convenient way of bounding the bayes risk @xmath44 is in terms of the bhattacharyya measure of affinity [ @xcite ] @xmath100 it is well known [ see , e.g. , @xcite , theorem 3.1 ] that @xmath101 thus , @xmath102 essentially behaves as the bayes error in the sense that @xmath44 is near @xmath37 when @xmath102 is near @xmath37 , and is small when @xmath102 is small . observe that , by jensen s inequality , @xmath103 straightforward calculation shows that for any @xmath19 , @xmath104 and therefore we have the following .
[ universal ] for all classes @xmath26 , @xmath105 whenever @xmath106 .
this shows that no matter what the class @xmath26 is , detection is hopeless if @xmath24 is of the order of @xmath99 .
this classical fact goes back to @xcite .
the next lemma is due to @xcite . for completeness
, we recall their proof .
[ pairs ] let @xmath23 and @xmath107 be drawn independently , uniformly , at random from @xmath26 and let @xmath108 . then @xmath109 as noted above , by the cauchy schwarz inequality , @xmath110 since @xmath111 , @xmath112 -1.\ ] ] however , by definition @xmath113 , so we have @xmath114 = \frac{1}{n^2 } \sum_{s , s'\in{\mathcal{c } } } e^{-k \mu^2 } \mathbb{e}_0 e^{\mu(x_s+x_{s'})}.\ ] ] but @xmath115 \\ & = & ( \mathbb{e}_0 e^{\mu x } ) ^{2(k-|s\cap s'| ) } ( \mathbb{e}_0 e^{2\mu x } ) ^{|s\cap s'| } \\ & = & e^{\mu^2 ( k-|s\cap s'|)+2\mu^2|s\cap s'|},\end{aligned}\ ] ] and the statement follows .
the beauty of this proposition is that it reduces the problem to studying a purely combinatorial quantity . by deriving upper bounds for the moment generating function of the overlap @xmath116 between two elements of @xmath26 drawn independently and uniformly at random ,
one obtains lower bounds for the critical value of @xmath24 .
this simple proposition turns out to be surprisingly powerful as it will be illustrated in various applications below .
we begin by deriving some simple consequences of proposition [ pairs ] under some general symmetry conditions on the class @xmath26 .
the following proposition shows that the universal bound of proposition [ universal ] can be improved by a factor of @xmath117 for all sufficiently symmetric classes .
[ symmetric ] let @xmath118 .
assume that @xmath26 satisfies the following conditions of symmetry .
let @xmath119 be drawn independently and uniformly at random from @xmath26 .
assume that : the conditional distribution of @xmath108 given @xmath107 is identical for all values of @xmath107 ; for any fixed @xmath120 and @xmath121 , @xmath122 .
then @xmath123 for all @xmath24 with @xmath124 we apply proposition [ pairs ] . by the first symmetry assumption , it suffices to derive a suitable upper bound for @xmath125=\mathbb{e } [ e^{\mu^2 z}| s']$ ] for an arbitrary @xmath126 .
after a possible relabeling , we may assume that @xmath127 so we can write @xmath128 . by hlder
s inequality , @xmath129 & = & \mathbb{e}\biggl [ \prod_{i=1}^k e^{\mu^2 \mathbh{1}_{\ { i\in s \ } } } \biggr ] \\ & \le & \prod_{i=1}^k ( \mathbb{e } [ e^{k\mu^2 \mathbh{1}_{\ { i\in s \ } } } ] ) ^{1/k } \\ & = & \mathbb{e } [ e^{k\mu^2 \mathbh{1}_{\ { 1\in s \ } } } ] \qquad \mbox{[by assumption ( ii ) ] } \\ & = & ( e^{\mu^2k } -1 ) \frac{k}{n } + 1.\end{aligned}\ ] ] proposition [ pairs ] now implies the statement .
surprisingly , the lower bound of proposition [ symmetric ] is close to optimal in many cases .
this is true , in particular when the class @xmath26 is `` small , '' made precise in the following statement .
[ smallclass ] assume that @xmath26 is symmetric in the sense of proposition [ symmetric ] and that it contains at most @xmath130 elements where @xmath131 .
then @xmath132 for all @xmath24 with @xmath133 and @xmath134 for all @xmath24 with @xmath135 the first statement follows from proposition [ symmetric ] while the second from proposition [ maxtest ] and the fact that @xmath136 .
the proposition above shows that for any small and sufficiently symmetric class , the critical value of @xmath24 is of the order of @xmath137 , at least if @xmath138 for some @xmath139 .
later , we will see examples of `` large '' classes for which proposition [ symmetric ] also gives a bound of the correct order of magnitude .
the bound of proposition [ symmetric ] may be improved significantly under an additional condition of negative association that is satisfied in several interesting examples ( see section [ examples ] below ) .
recall that a collection @xmath140 of random variables is _ negatively associated _
if for any pair of disjoint sets @xmath141 and ( coordinate - wise ) nondecreasing functions @xmath20 and @xmath142 , @xmath143 \le\mathbb{e } [ f(y_i , i\in i ) ] \mathbb{e}[g(y_j , j\in j ) ] .\ ] ] [ negass ] let @xmath118 and assume that the class @xmath26 satisfies the conditions of proposition [ symmetric ] .
suppose that the labels are such that @xmath144 .
let @xmath23 be a randomly chosen element of @xmath26 .
if the random variables @xmath145 are negatively associated , then @xmath123 for all @xmath24 with @xmath146 we proceed similarly to the proof of proposition [ symmetric ] .
we have @xmath129 & = & \mathbb{e}\biggl [ \prod_{i=1}^k e^{\mu^2 \mathbh{1}_{\ { i\in s \ } } } \biggr ] \\ & \le & \prod_{i=1}^k \mathbb{e } [ e^{\mu^2 \mathbh{1}_{\ { i\in s \ } } } ] \qquad\mbox{(by negative association ) } \\ & = & \biggl ( ( e^{\mu^2 } -1 ) \frac{k}{n } + 1 \biggr)^k.\end{aligned}\ ] ] proposition [ pairs ] and the upper bound above imply that @xmath44 at least @xmath147 for all @xmath24 such that @xmath148 the result follows by using @xmath149 with @xmath150 .
in this section , we consider various concrete examples and work out upper and lower bounds for the critical range of @xmath24 .
we start with the simplest possible case , that is , when all @xmath19 are disjoint ( and therefore @xmath151 ) .
fix @xmath118 . then
, under @xmath4 , the @xmath71 are independent normal @xmath152 random variables and the bound @xmath153 is close to being tight . by applying the maximum test @xmath20
, we see that @xmath154 whenever @xmath155 to see that this bound gives the correct order of magnitude , we may simply apply proposition [ pairs ] . here
@xmath156 may take two values : @xmath157 thus , @xmath158 and therefore @xmath159 whenever @xmath160 so in this case the critical transition occurs when @xmath24 is of the order of @xmath161 . in section [ hellinger
] , we use this simple lower bound to establish lower bounds for general classes @xmath26 of sets . note that in this simple case one may directly analyze the risk of the optimal test and obtain sharper bounds .
in particular , the leading constant in the lower bound is suboptimal .
however , in this paper our aim is to understand some general phenomena and we focus on orders of magnitude rather than on nailing down sharp constants .
taking @xmath162 , @xmath163 , and @xmath164 in the above example , we obtain a connection with multiple hypothesis testing . in the latter ,
one tests `` @xmath165 '' against `` @xmath166 '' for every @xmath167 , and traditionally uses as test statistics @xmath168 , to build a multiple testing procedure ( often rejecting all the hypotheses corresponding to large @xmath169 ) , see , for instance , @xcite , @xcite .
such a procedure will reject the global null hypothesis `` @xmath170 '' if at least one of the alternatives `` @xmath166 '' is preferred .
the main difference with the approach taken in this paper concerns the error rate .
a multiple testing procedure is generally calibrated to control measures of the type i error like the family wise error rate or the false discovery rate , while the tests defined in this paper are designed to control the entire risk . finally ,
in this example , the subsets are disjoint which is the traditional framework in multiple testing .
consider the example when @xmath26 contains all sets @xmath171 of size @xmath25 .
thus , @xmath172 .
as mentioned in the , this problem is very well understood as sharp bounds and sophisticated tests are available ; see , for example , @xcite , @xcite , @xcite .
we include it for illustration purposes only and we warn the reader that the obtained bounds are not sharpest possible .
let @xmath118 .
it is easy to see that the assumptions of proposition [ negass ] are satisfied [ this follows , e.g. , from proposition 11 of @xcite ] and therefore @xmath159 for all @xmath146 this simple bound turns out to have the correct order of magnitude both when @xmath173 [ in which case it is of the order of @xmath174 and when @xmath175 ( when it is of the order of @xmath176 ) .
this may be seen by considering the two simple tests described in section [ simpletests ] in the two different regimes . since @xmath177
we see from proposition [ maxtest ] that when @xmath178 for some fixed @xmath179 , then the threshold value is of the order of @xmath180 . on the other hand ,
when @xmath181 is bounded away from zero , then the lower bound implied by proposition [ negass ] above is of the order @xmath176 and the averaging test provides a matching upper bound by proposition [ average ] .
note that in this example the maximum test is easy to compute since it suffices to find the @xmath25 largest values among @xmath182 .
let @xmath26 be the set of all perfect matchings of the complete bipartite graph @xmath183 .
thus , we have @xmath184 edges and @xmath185 , and @xmath186 . by proposition [ average ] ( i.e. , the averaging test ) , for @xmath118 , one has @xmath55 whenever @xmath187 . to show that this bound has the right order of magnitude , we may apply proposition [ negass ] .
the symmetry assumptions hold obviously and the negative association property follows from the fact that @xmath108 has the same distribution as the number of fixed points in a random permutation .
the proposition implies that for all @xmath92 , @xmath159 whenever @xmath188 note that in this case the optimal test @xmath32 can be approximated in a computationally efficient way . to this end , observe that computing @xmath189 ( where the summation is over all permutations of @xmath190 ) is equivalent to computing the permanent of an @xmath191 matrix with nonnegative elements . by a deep result of @xcite ,
this may be done by a polynomial - time randomized approximation . consider a network of @xmath92 nodes in which each pair of nodes interacts .
one wishes to test if there is a corrupted node in the network whose interactions slightly differ from the rest .
this situation may be modeled by considering the class of _ stars_.
a star is a subgraph of the complete graph @xmath192 which contains all @xmath193 edges containing a fixed vertex ( see figure [ asshole ] ) . consider the set @xmath26 of all stars .
in this setting , @xmath194 and @xmath195 . .
] in this case , we are in the situation of corollary [ smallclass ] and propositions [ symmetric ] and [ maxtest ] imply that if @xmath26 is the class of all stars in @xmath192 then for any @xmath179 , @xmath196 consider again a network of @xmath92 nodes in which each pair of nodes interact .
one may wish to test if there exists a corrupted connected subgraph containing each node .
this leads us to considering the class of all spanning trees as follows .
let @xmath197 represent the edges of the complete graph @xmath192 and let @xmath26 be the set of all spanning trees of @xmath192 .
thus , we have @xmath198 spanning trees and @xmath193 .
[ see , e.g. , @xcite .
] by proposition [ average ] , the averaging test has risk @xmath55 whenever @xmath199 .
this bound is indeed of the right order . to see this
, we may start with proposition [ pairs ] .
there are ( at least ) two ways of proceeding .
one is based on negative association . even though proposition [ negass ] is not applicable because of the lack of symmetry in @xmath26 , negative association still holds .
in particular , by a result of @xcite [ see also @xcite and @xcite ] , if @xmath23 is a random uniform spanning tree of @xmath192 , then the indicators @xmath200 are negatively associated .
this means that , if @xmath23 and @xmath107 are independent uniform spanning trees and @xmath108 , @xmath201 & = & \mathbb{e}\mathbb{e}\bigl [ e^{\mu^2 |s\cap s'| } \vert s ' \bigr ] \\ & = & \mathbb{e}\mathbb{e } [ e^{\mu^2 \sum_{i\in s'}\mathbh{1}_{\ { i\in s \ } } } \vert s ' ] \\ & \le & \mathbb{e}\prod_{i\in s'}\mathbb{e } [ e^{\mu^2 \mathbh{1}_{\ { i\in s \ } } } \vert s ' ] \qquad \mbox{(by negative association ) } \\ & \le & \mathbb{e}\prod_{i\in s ' } \biggl ( \frac{2}{m } e^{\mu^2 } + 1 \biggr ) \\ & = & \biggl ( \frac{2}{m } e^{\mu^2 } + 1 \biggr)^{m-1 } \\ &
\le & \exp(2e^{\mu^2 } ) .\end{aligned}\ ] ] this , together with proposition [ pairs ] shows that for any @xmath118 , @xmath159 whenever @xmath202 we note here that the same bound can be proved by a completely different way that does not use negative association . the key is to note that we may generate the two random spanning trees based on @xmath203 independent random variables @xmath204 taking values in @xmath205 as in @xcite [ see also @xcite ] .
the key property we need is that if @xmath206 denotes the number of common edges in the two spanning trees when @xmath169 is replaced by an independent copy @xmath207 while keeping all other @xmath208 s fixed , then @xmath209 ( the details are omitted ) . for random variables satisfying this last property ,
an inequality of @xcite implies the sub - poissonian bound @xmath210 clearly , @xmath211 , so essentially the same bound as above is obtained .
as the bounds above show , the computationally trivial average test has a close - to - optimal performance . in spite of this
, one may wish to use the optimal test @xmath32 .
the `` partition function '' @xmath212 may be computed by an algorithm of @xcite , who introduced a random sampling algorithm that , given a graph with nonnegative weights @xmath213 over the edges , samples a random spanning tree from a distribution such that the probability of any spanning tree @xmath23 is proportional to @xmath214 .
the expected running time of the algorithm is bounded by the cover time of an associated markov chain that is defined as a random walk over the graph in which the transition probabilities are proportional to the edge weights .
if @xmath24 is of the order of a constant ( as in the critical range ) then the cover time is easily shown to be polynomial ( with high probability ) as all edge weights @xmath215 are roughly of the same order both under the null and under the alternative hypotheses .
another natural application is the class of all cliques of a certain size in a complete graph .
more precisely , the random variables @xmath182 are associated with the edges of the complete graph @xmath192 such that @xmath216 and let @xmath26 contain all cliques of size @xmath217 .
thus , @xmath218 and @xmath219 .
this case is more difficult than the class of @xmath25-sets discussed above because negative association does not hold anymore .
( this may be easily seen by considering the indicator variables of two adjacent edges both being in the randomly chosen clique . ) also , computationally the class of cliques is much more complex .
a related , well - studied model starts with the subgraph @xmath192 containing each edge independently with probability @xmath27 , as null hypothesis .
the alternative hypothesis is the same as the null hypothesis , except that there is a clique of size @xmath217 on which each edge is independently present with probability @xmath220 .
this is called the `` hidden clique '' problem ( usually only the special case @xmath221 is considered ) . despite substantial interest in the hidden clique problem ,
polynomial time detection algorithms are only known when @xmath222 [ @xcite , @xcite ] .
we may obtain the hidden clique model from our model by thresholding at weight zero ( retaining only edges whose normal random variable is positive ) , and so our model is easier for testing than the hidden clique model .
however , it seems likely that designing an efficient test in the normal setting will be as difficult as it has proved for hidden cliques .
it would be of interest to construct near - optimal tests that are computable in polynomial time for larger values of @xmath217 .
we have the following bounds for the performance of the optimal test .
it shows that when @xmath217 is a most of the order of @xmath223 , the critical value of @xmath24 is of the order of @xmath224 . the proof below
may be adjusted to handle larger values of @xmath217 as well but we prefer to keep the calculations more transparent . [ cliques ] let @xmath26 represent the class of all @xmath219 cliques of a complete graph @xmath192 and assume that @xmath225 . then : for all @xmath118 , @xmath226 whenever @xmath227 @xmath105 whenever @xmath228 \(i ) follows simply by a straightforward application of proposition [ maxtest ] and the bound @xmath229 . to prove the lower bound ( ii ) , by proposition [ pairs ]
, it suffices to show that if @xmath119 are @xmath217-cliques drawn randomly and independently from @xmath26 and @xmath156 denotes the number of _ edges _ in the intersection of @xmath23 and @xmath107 , then @xmath230 \le2 $ ] for the indicated values of @xmath24 .
because of symmetry , @xmath230 = \mathbb{e } [ \exp(\mu^2z ) \vert s ' ] $ ] for all @xmath107 and therefore we might as well fix an arbitrary clique @xmath107 . if @xmath231 denotes the number of vertices in the clique @xmath232 , then @xmath233 .
moreover , the distribution of @xmath231 is hypergeometrical with parameters @xmath92 and @xmath217 . if @xmath234 is a binomial random variable with parameters @xmath217 and @xmath235 ,
then since @xmath236 is a convex function of @xmath237 , an inequality of @xcite implies that @xmath238 = \mathbb{e } [ e^{\mu^2y^2/2 } ] \le\mathbb{e } [ e^{\mu^2b^2/2 } ] .\ ] ] thus , it remains to derive an appropriate upper bound for the moment generating function of the squared binomial . to this end , let @xmath239 be a parameter whose value will be specified later . using @xmath240 and the cauchy
schwarz inequality , it suffices to show that @xmath241 \cdot\mathbb{e}\bigl[\exp\bigl(\mu^2kb\mathbh{1}_{\ { b > c{k^2}/{m } \ } } \bigr ) \bigr ] \le4.\ ] ] we show that , if @xmath24 satisfies the condition of ( ii ) , for an appropriate choice of @xmath242 , both terms on the left - hand side are at most @xmath243 .
the first term on the left - hand side of ( [ product ] ) is @xmath244 = \biggl ( 1 + \frac{k}{m } \biggl(\exp\biggl(\mu^2c\frac{k^2}{m } \biggr ) -1 \biggr ) \biggr)^k,\ ] ] which is at most @xmath243 if and only if @xmath245 since @xmath246 , this is implied by @xmath247 to bound the second term on the left - hand side of ( [ product ] ) , note that @xmath248 & \le & 1 + \mathbb{e}\bigl[\mathbh{1}_{\ { b > c{k^2/m } \}}\exp(\mu^2 kb ) \bigr ] \\ & \le & 1 + \biggl({\mathbb{p}}\biggl\{b > c\frac{k^2}{m } \biggr\ } \biggr)^{1/2 } ( \mathbb{e}[\exp(\mu^2 kb ) ] ) ^{1/2},\end{aligned}\ ] ] by the cauchy
schwarz inequality , so it suffices to show that @xmath249\le1.\ ] ] denoting @xmath250 , chernoff s bound implies @xmath251 on the other hand , @xmath252 = \biggl ( 1 + \frac{k}{m}\exp(\mu^2k ) \biggr)^k,\ ] ] and therefore the second term on the left - hand side of ( [ product ] ) is at most @xmath243 whenever @xmath253 using @xmath254 , we obtain the sufficient condition @xmath255 summarizing , we have shown that @xmath105 for all @xmath24 satisfying @xmath256 choosing @xmath257 [ which is greater than @xmath37 for @xmath258 , the second term on the right - hand side is at most @xmath224 .
now observe that since @xmath259 is convex , for any @xmath260 , @xmath261 .
choosing @xmath262 , the first term is at least @xmath263 where we used the condition that @xmath264 and that @xmath265 for all @xmath266 . a closely related problem arising in the exploratory analysis of microarray data [ see @xcite ]
is when each member of @xmath26 represents the @xmath25 edges of a @xmath267 biclique of the complete bipartite graph @xmath183 where @xmath268 .
( a biclique is a complete bipartite subgraph of @xmath183 . )
the analysis and the bounds are completely analogous to the one worked out above , the details are omitted .
intuitively , one would expect that the testing problem becomes harder as the class @xmath26 gets larger .
more precisely , one may expect that if @xmath49 are two classes of subsets of @xmath269 , then @xmath270 holds for all @xmath24 .
the purpose of this section is to show that this intuition is wrong in quite a strong sense as not only such general monotonicity property does not hold for the risk , but there are classes @xmath49 for which @xmath271 is arbitrary close to @xmath37 and @xmath272 is arbitrary close to @xmath273 for the same value of @xmath24 .
however , monotonicity does hold if the class @xmath26 is sufficiently symmetric .
call a class @xmath26 _ symmetric _ if for the optimal test @xmath274 the value of @xmath275 is the same for all @xmath276 .
note that several of the examples discussed in section [ examples ] satisfy the symmetry assumption , such as the classes of @xmath25-sets , stars , perfect matchings , and cliques .
however , the class of spanning trees is not symmetric in the required sense .
[ symmetricclass ] let @xmath26 be a symmetric class of subsets of @xmath269 .
if @xmath277 is an arbitrary subclass of @xmath26 , then for all @xmath13 , @xmath278 . in this proof , we fix the value of @xmath13 and suppress it in the notation .
recall the definition of the alternative risk measure @xmath279 which is to be contrasted with our main risk measure @xmath280 the risk @xmath47 is obviously monotone in the sense that if @xmath281 then for every @xmath20 , @xmath282 .
let @xmath283 and @xmath284 denote the optimal tests with respect to both measures of risk .
first , observe that if @xmath26 is symmetric , then @xmath285 .
but since @xmath286 for every @xmath20 , we have @xmath287 this means that all inequalities are equalities and , in particular , @xmath288 .
now if @xmath277 is an arbitrary subclass of @xmath26 , then @xmath289 which completes the proof
. for every @xmath290 there exist @xmath0 , @xmath24 , and classes @xmath291 such that @xmath292 and @xmath293 .
we work with @xmath294 distances . for any class @xmath295 , denote @xmath296 . recall that @xmath297 given @xmath298 , we fix an integer @xmath299 large enough that @xmath300 and that @xmath301 and let @xmath302 .
we let @xmath303 consist of @xmath304 disjoint subsets of @xmath305 , each of size @xmath304 .
we let @xmath306 consist of all sets of the form @xmath307 , where @xmath30 ranges from @xmath304 to @xmath0 , and assume @xmath303 has been chosen so that @xmath308 .
we then let @xmath309 .
we take @xmath310 so that , as seen in section [ canonical ] , we have @xmath311
. we will require an upper bound on @xmath312 , which we obtain by considering the averaging test on variables @xmath313 , @xmath314 just as in proposition [ average ] , we have @xmath315 whenever @xmath316 , which is indeed the case by our choices of @xmath24 and @xmath25 .
it follows that @xmath317 .
we remark that @xmath318 we let @xmath319 ; then @xmath320 , and note @xmath321 thus , @xmath322 . observe that nonmonotonicity of the bhattacharyya affinity also follows from the same argument . to this end
, we may express @xmath323 in function of the hellinger distance @xmath324 as @xmath325 .
recalling [ see , e.g. , @xcite , page 225 ] that @xmath326 we see that the same example as in the proof above , for @xmath0 large enough , shows the nonmonotonicity of the bhattacharyya affinity as well .
in this section , we derive lower bounds for the bayes risk @xmath327 .
the bounds are in terms of some geometric features of the class @xmath26 .
again , we treat @xmath26 as a metric space equipped with the canonical distance @xmath328 [ i.e. , the square root of the hamming distance @xmath329 . for an integer @xmath330
, we define a real - valued parameter @xmath331 of the class @xmath26 as follows .
let @xmath49 be obtained by choosing @xmath332 elements of @xmath26 at random , without replacement .
let the random variable @xmath333 denote the smallest distance between elements of @xmath277 and let @xmath334 be a median of @xmath333 .
[ randomclassbound ] let @xmath335 be an integer .
then for any class @xmath26 , @xmath336 whenever @xmath337 to interpret the statement of the theorem , note that latexmath:[\[k-\tau^2/2 = { \mathop{\max_{s , t\in{\mathcal{a}}}}_{s\neq t } } @xmath277 .
thus , just like in proposition [ pairs ] , the distribution of the overlap between random elements of @xmath26 plays a key role in establishing lower bounds for the optimal risk .
however , while in proposition [ pairs ] the moment generating function @xmath339 of the overlap between two random elements determines an upper bound for the critical value of @xmath24 , here it is the median of the largest overlap between many random elements that counts .
the latter seems to carry more information about the fine geometry of the class .
in fact , invoking a simple union bound , upper bounds for @xmath339 may be used together with theorem [ randomclassbound ] . in applications ,
often it suffices to consider the following special case .
[ medianzero ] let @xmath335 be the largest integer for which zero is a median of @xmath340 where @xmath277 is a random subset of @xmath26 of size @xmath332 [ i.e. , @xmath341 .
then @xmath342 for all @xmath343 .
to illustrate the corollary , consider the following example which is the simplest in a family of problems investigated by @xcite : assume that @xmath0 and @xmath25 are both perfect squares and that the indices @xmath269 are arranged in a @xmath344 grid .
the class @xmath26 contains all @xmath345 sub - squares .
now if @xmath23 and @xmath346 are randomly chosen elements of @xmath26 ( with or without replacement ) then , if @xmath347 , @xmath348 and therefore @xmath349 which is at least @xmath27 if @xmath350 in which case @xmath351 .
thus , by corollary [ medianzero ] , @xmath342 for all @xmath352 .
this bound is of the optimal order of magnitude as it is easily seen by an application of proposition [ maxtest ] . in some other applications ,
a better bound is obtained if some overlap is allowed . a case in point
is the example of stars from section [ stars ] . in that case , any two elements of @xmath26 overlap but by taking @xmath353 , we have @xmath354 , so theorem [ randomclassbound ] still implies @xmath342 whenever @xmath355 . the main tool of the proof of theorem [ randomclassbound ] is slepian s lemma which we recall here [ @xcite ] .
[ for this version , see @xcite , theorem 3.11 . ] [ slepian ] let @xmath356 be zero - mean gaussian vectors such that for each @xmath357 , @xmath358 let @xmath359 be such that for all @xmath360 and @xmath361 , @xmath362 then @xmath363 .
proof of theorem [ randomclassbound ] let @xmath335 be fixed and choose @xmath332 sets from @xmath26 uniformly at random ( without replacement ) .
let @xmath277 denote the random subclass of @xmath26 obtained this way .
denote the likelihood ratio associated to this class by @xmath364 where @xmath365 .
then the optimal risk of the class @xmath26 may be lower bounded by @xmath366 denoting by @xmath367 expectation with respect to the random choice of @xmath277 , we have @xmath368 } \nonumber\\ \eqntext{\mbox{(since the variance of a sample without replacement } } \\ \eqntext{\mbox{is less than that with replacement ) } } \\ & = & \widehat{\mathbb{e } } r^*_{\mathcal{a}}(\mu ) - \frac{1}{2\sqrt{m } } \sqrt { \frac{1}{n } \sum_{t\in{\mathcal{c } } } \mathbb{e}_0 \biggl ( v_t - \frac{1}{n } \sum_{s\in{\mathcal{c } } } v_s \biggr)^2}.\nonumber\end{aligned}\ ] ] an easy way to bound the right - hand side is by writing @xmath369 summarizing , we have @xmath370 where we used the assumption that @xmath371 .
thus , it suffices to prove that @xmath372 .
we bound the optimal risk associated with @xmath277 in terms of the bhattacharyya affinity @xmath373 recalling from section [ lowerbounds ] that @xmath374 and using that @xmath375 is concave , we have @xmath376 therefore , it suffices to show that the expected bhattacharyya affinity @xmath377 corresponding to the random class @xmath277 satisfies @xmath378 in the argument below , we fix the random class @xmath277 , relabel the elements so that @xmath379 , and bound @xmath380 from below .
denote the minimum distance between any two elements of @xmath277 by @xmath333 . to bound @xmath380
, we apply slepian s lemma with the function @xmath381 where @xmath382 .
simple calculation shows that the mixed second partial derivatives of @xmath383 are negative , so slepian s lemma is indeed applicable .
next , we introduce the random vectors @xmath384 and @xmath385 .
let the components of @xmath384 be indexed by elements @xmath386 and define @xmath387 .
thus , under @xmath4 , each @xmath388 is normal @xmath152 and @xmath389 is just the bhattacharyya affinity @xmath380 . to define the random vector @xmath385 ,
introduce @xmath390 independent standard normal random variables : one variable @xmath391 for each @xmath392 and an extra variable @xmath393 . recall that the definition of @xmath333 guarantees that the minimal distance between any two elements of @xmath277 as at least @xmath333 .
now let @xmath394 then clearly for each @xmath395 , @xmath396 and @xmath397 ( @xmath398 ) . on the other hand , @xmath399 and @xmath400 therefore , by slepian s lemma , @xmath401 .
however , @xmath402 to finish the proof , it suffices to observe that the last expression is the bhattacharyya affinity corresponding to a class of disjoint sets , all of size @xmath403 , of cardinality @xmath404 .
this case has been handled in the first example of section [ examples ] where we showed that @xmath405 where again we used the condition @xmath343 and the fact that @xmath406 . therefore , under this condition on @xmath24 , we have that for any fixed @xmath277 , @xmath407 and therefore @xmath408 where @xmath334 is the median of @xmath333 .
this concludes the proof . at the risk of losing a constant factor in the statement of theorem [ randomclassbound ]
, one may replace the parameter @xmath334 by a larger quantity .
the idea is that by thinning the random subclass @xmath277 one may consider a subset of @xmath277 that has better separation properties .
more precisely , for an even integer @xmath335 we may define a real - valued parameter @xmath409 of the class @xmath26 as follows .
let @xmath281 be obtained by choosing @xmath332 elements of @xmath26 at random , without replacement .
order the elements @xmath410 of @xmath277 such that @xmath411 and define the subset @xmath412 by @xmath413 . let the random variable @xmath414 denote the smallest distance between elements of @xmath415 and let @xmath416 be the median of @xmath333 .
it is easy to see that the proof of theorem [ randomclassbound ] goes through , and one may replace @xmath334 by @xmath416 ( by adjusting the constants appropriately ) .
one simply needs to observe that since each @xmath417 is nonnegative , @xmath418 if @xmath416 is significantly larger than @xmath334 , the gain may be substantial .
if the class @xmath26 is symmetric then thanks to theorem [ symmetricclass ] , the theorem above can be improved and simplified . if the class is symmetric , instead of having to work with randomly chosen subclasses , one may optimally choose a separated subset .
then the bounds can be expressed in terms of the metric entropy of @xmath26 , more precisely , by its _ packing numbers _ with respect to the canonical distance @xmath328 .
we say that @xmath49 is a @xmath84-separated set ( or @xmath84-packing ) if for any @xmath419 , @xmath420 . for @xmath421 , define the _ packing number _ @xmath422 as the size of a maximal @xmath84-separated subset @xmath277 of @xmath26 .
it is a simple well - known fact that packing numbers are closely related to the covering numbers introduced in section [ simpletests ] by the inequalities @xmath423 .
let @xmath26 be symmetric in the sense of theorem [ symmetricclass ] and let @xmath424 .
then @xmath425 whenever @xmath426 let @xmath49 be a maximal @xmath84-separated subclass . since @xmath26 is symmetric , by theorem [ symmetricclass ] , @xmath427 so it suffices to show that @xmath428 for the indicated values of @xmath24 .
the rest of the proof is identical to that of theorem [ randomclassbound ] . to interpret this result ,
take @xmath429 for some @xmath430 .
then , by the theorem , @xmath132
if @xmath431 as an example , suppose that the class @xmath26 is such that there exists a constant @xmath432 such that @xmath433 .
( recall that all classes with vc dimension @xmath434 have an upper bound of this form for the packing numbers , see remark on page . ) in this case , one may choose @xmath435 and obtain that @xmath132 whenever @xmath436 ( for some constant @xmath242 ) .
this closely matches the bound obtained for the maximum test by dudley s chaining bound .
in all examples considered above , upper bounds for the optimal risk @xmath44 are derived by analyzing either the maximum test or the averaging test .
as the examples show , very often these simple tests have a near - optimal performance .
the optimal test @xmath32 is generally more difficult to study . in this section ,
we analyze directly the performance of the optimal test .
more precisely , we derive general upper bounds for the type i error ( i.e. , the probability that the null hypothesis is rejected under @xmath4 ) of @xmath32 .
the upper bound involves the expected value of the maximum of a gaussian process indexed by a sparse subset of @xmath26 and can be significantly smaller than the maximum over the whole class that appears in the performance bound of the maximum test in proposition [ maxtest ] .
unfortunately , we do not have an analogous bound for the type ii error .
we consider the type i error of the optimal test @xmath32 @xmath437 an easy bound is @xmath438 so @xmath439 thus , @xmath440 whenever @xmath441 . of course
, we already know this from proposition [ maxtest ] where this bound was derived for the ( suboptimal ) test based on maxima . in order to understand the difference between the performance of the optimal test @xmath32 and the maximum test
, one needs to compare the random variables @xmath442 and @xmath64 .
[ typeonebound ] for any @xmath118 , the type error of the optimal test @xmath32 satisfies @xmath443 whenever @xmath444 where @xmath277 is any @xmath445-cover of @xmath26 . if @xmath277 is a minimal @xmath445-cover of @xmath26 , then @xmath446 by `` sudakov s minoration '' [ see @xcite , theorem 3.18 ] this upper bound is sharp up to a constant factor .
it is instructive to compare this bound with that of proposition [ maxtest ] for the performance of the maximum test . in proposition
[ typeonebound ] , we were able to replace the expected maximum @xmath78 by @xmath447 where now the maximum is taken over a potentially much smaller subset @xmath49 .
it is not difficult to construct examples when there is a substantial difference , even in the order of magnitude , between the two expected maxima so we have a genuine gain over the simple upper bound of proposition [ maxtest ] .
unfortunately , we do not know if an analog upper bound holds for the type ii error @xmath448 of the optimal test @xmath32 . in cases when @xmath449 , we suspect that the maximum test is far from optimal .
however , to verify this conjecture , one would need a similar analysis for the type ii error as well .
proof of proposition [ typeonebound ] introduce the notation @xmath450 then @xmath451 we use tsirelson s inequality ( lemma [ tsirelson ] ) to bound this probability . to this end
, we need to show that the function @xmath452 defined by @xmath453 is lipschitz [ where @xmath454 . observing that @xmath455 we have @xmath456 and
therefore @xmath457 is indeed lipschitz @xmath73 . by tsirelson s inequality
, we have @xmath458 thus , the type i error is bounded by @xmath147 if @xmath459 it remains to bound @xmath460 .
let @xmath461 be a positive integer and consider a minimal @xmath84-cover of the set @xmath26 , that is , a set @xmath49 with cardinality @xmath462 such that , if @xmath463 denotes an element in @xmath277 whose distance to @xmath19 is minimal then @xmath464 for all @xmath19 .
then clearly , @xmath465 to bound the first term on the right - hand side , note that , by jensen s inequality , @xmath466 since for each @xmath23 , @xmath467 and therefore @xmath468 is a centered normal random variable with variance @xmath469 . for the second term
, we have @xmath470 choosing @xmath471 , we obtain the proposition .
we thank ery arias - castro and emmanuel cands for discussions on the topic of the paper .
we also thank the referees for their valuable remarks .
tsirelson , b. s. , ibragimov , i. a. and sudakov , v. n. ( 1976 ) .
norm of gaussian sample function . in _ proceedings of the 3rd japan
symposium on probability theory_. _ lecture notes in math . _ * 550 * 2041 .
springer , berlin . | we study a class of hypothesis testing problems in which , upon observing the realization of an @xmath0-dimensional gaussian vector , one has to decide whether the vector was drawn from a standard normal distribution or , alternatively , whether there is a subset of the components belonging to a certain given class of sets whose elements have been `` contaminated , '' that is , have a mean different from zero .
we establish some general conditions under which testing is possible and others under which testing is hopeless with a small risk .
the combinatorial and geometric structure of the class of sets is shown to play a crucial role .
the bounds are illustrated on various examples .
, , + and . . |
the stationary axisymmetric einstein equations are of great physical importance since their solutions can describe the gravitational field of stars and galaxies in thermodynamical equilibrium . in this case
the matter can be approximated as an ideal fluid , but the field equations in the matter region do not seem to be integrable .
the vacuum equations are however equivalent to the ernst equation @xcite which is completely integrable @xcite . if one considers two - dimensional matter distributions as disks which are discussed as models for the matter in galaxies or in accretion disks around black - holes , the matter leads to a boundary value problem for the vacuum equations .
a solution to this boundary value problem leads to global spacetimes with matter distributions which can be physically interpreted .
the first analytic solution for a stationary disk was identified by neugebauer and meinel @xcite as belonging to korotkin s @xcite algebro - geometric solutions to the ernst equation .
a systematic study of these solutions in @xcite made it possible in @xcite to identify a class of disk solutions which can be interpreted as being made up of counter - rotating dust .
in the first paper of this this series @xcite ( henceforth referred to as i ) , the implications of the underlying riemann surface on the boundary data taken at a disk were discussed .
it was possible to construct the solution @xcite in this way and to identify the range of the physical parameters where the solution is globally regular except at the disk where the boundary data are prescribed .
we will briefly summarize results of i where details of the notation can be found .
we use the weyl lewis
papapetrou metric ( see e.g.@xcite ) @xmath0 where @xmath1 and @xmath2 are weyl s canonical coordinates and @xmath3 and @xmath4 are the two commuting asymptotically timelike respectively spacelike killing vectors . with @xmath5 and the potential @xmath6 defined by @xmath7 and @xmath8 for @xmath9 , we define the complex ernst potential @xmath10 which is subject to the ernst equation @xcite @xmath11 where a bar denotes complex conjugation in @xmath12 .
the metric function @xmath13 follows from @xmath14 in i we have considered disks which can be interpreted as two counter - rotating components of pressureless matter , so - called dust .
the surface energy - momentum tensor of these models is defined on the hypersurface @xmath15 .
it could be written in the form @xmath16 where greek indices stand for the @xmath17 , @xmath1 and @xmath18 component and where @xmath19 .
we gave an explicit solution for disks with constant angular velocity @xmath20 and constant relative density @xmath21 .
this class of solutions is characterized by two real parameters @xmath22 and @xmath23 which are related to @xmath20 and @xmath24 and the metric potential @xmath25 at the center of the disk via , @xmath26 we put the radius @xmath27 of the disk equal to 1 unless otherwise noted .
since the radius appears only in the combinations @xmath28 , @xmath29 and @xmath30 in the physical quantities , it does not have an independent role .
it is always possible to use it as a natural lengthscale unless it tends to 0 as in the case of the ultrarelativistic limit of the one component disk .
the solution of the ernst equation we will consider in this paper is given on a hyperelliptic riemann surface @xmath31 of genus 2 which is defined by the algebraic relation @xmath32 .
we choose @xmath33 , @xmath34 and @xmath35 with @xmath36 .
we use the cut - system of fig .
[ schnitt1 ] .
the base point of the abel map @xmath37 is @xmath38 .
the main result of i was the proof of the following let @xmath39 and @xmath40 where @xmath41 is the smallest positive value of @xmath22 for which @xmath42 .
let @xmath43 with @xmath44 , @xmath45 real and @xmath46 then the solution to the ernst equation for an energy - momentum tensor of the form ( [ 2.0 ] ) with constant @xmath20 and constant @xmath24 can be written in the form @xmath47(\omega(\infty^{+})+u ) } { \theta[m](\omega(\infty^{+})-u)}e^{i } \label{rel1a},\ ] ] where @xmath48 , where @xmath49 , where @xmath50 is the covering of the imaginary axis in the + -sheet of @xmath31 between @xmath51 and @xmath52 , where the characteristic reads @xmath53=\left [ \begin{array}{cc } 0 & 0 \\ 1 & 1 \end{array } \right]$ ] , and where @xmath54 we note that with @xmath44 and @xmath45 given , the riemann surface is completely determined at a given point in the spacetime , i.e. for a given value of @xmath55 .
in contrast to algebro - geometric solutions to non - linear evolution equations , it depends on the physical coordinates exclusively via the branch points @xmath55 and @xmath56 .
since only the modular properties of the theta functions are important , the solutions are neither periodic or quasiperiodic .
the complete metric ( [ vac1 ] ) can be expressed via theta functions .
let the characteristics @xmath57 $ ] be given by @xmath58=\left [ \begin{array}{cc } 0 & 0 \\ 1 & 1 \end{array } \right]$ ] and @xmath59=\left [ \begin{array}{cc } 0 & 0 \\ 0 & 0 \end{array } \right]$ ] .
then the function @xmath60 can be written in the form @xmath61(u)\theta[n_{2}](u ) } { \theta[n_{1}](0)\theta[n_{2}](0 ) } \frac{\theta[n_{1}](\omega(\infty^{-}))\theta[n_{2}](\omega(\infty^{- } ) ) } { \theta[n_{1}](\omega(\infty^{-})+u)\theta[n_{2}](\omega(\infty^{-})+u ) } e^{i } \label{2.5}.\ ] ] the metric function @xmath62 can be expressed via @xmath63(0 ) \theta[n_{2}](0 ) } { \theta[n_{1}](\omega(\infty^{-}))\theta[n_{2 } ] ( \omega(\infty^{- } ) ) } \frac{\theta[n_{1}](u)\theta[n_{2}](u+2\omega(\infty^{- } ) ) } { \theta[n_{1}](u+\omega(\infty^{-}))\theta[n_{2}](u+\omega(\infty^{- } ) ) } -1\right ) \label{2.6},\ ] ] where the constant @xmath64 is determined by the condition that @xmath62 vanishes on the regular part of the axis and at infinity .
the metric function @xmath65 can be put in the form @xmath66(u)\theta[n_{2}](u ) } { \theta[n_{1}](0)\theta[n_{2}](0)}\exp\left ( \frac{2}{(4\pi \mathrm{i})^{2}}\int_{\gamma}^{}\int_{\gamma}^ { } dk_{1}dk_{2}h(k_{1})h(k_{2 } ) \ln \frac{\theta_{o } ( \omega(k_{1})-\omega(k_{2}))}{k_{1}-k_{2}}\right ) \label{2.7},\ ] ] where @xmath67 is a theta function with an odd characteristic , where @xmath68 , and where @xmath69 is a constant which is determined by the condition that @xmath13 vanishes on the regular part of the axis and at infinity .
* proof : * + the metric potential @xmath60 is just the real part of the ernst potential . with the help of fay s trisecant identity @xcite ,
@xmath60 was written in @xcite in the form ( [ 2.5 ] ) .
korotkin @xcite gave an expression for the metric function @xmath62 as a derivative of theta functions with respect to the argument .
in @xcite this formula could be written in the form ( [ 2.6 ] ) free off derivatives by using the trisecant identity .
the metric function @xmath65 is related to the so - called @xmath70-function of the linear system associated to the ernst equation ( see @xcite ) .
this connection made it possible to give the explicit expression for @xmath13 in terms of theta functions of ( [ 2.7 ] ) in @xcite .
the axis of symmetry is of physical importance since the multipole moments such as the arnowitt - deser - misner ( adm ) mass and the angular momentum can be read off from the ernst potential on the axis . on the axis
the branch points @xmath55 and @xmath56 coincide which implies that the riemann surface becomes singular and that some of the periods diverge .
the ernst potential is however regular at the axis except at the disk .
the proof for this result is based on results by fay @xcite and yamada @xcite and was given in @xcite .
we denote here and in the following the elliptic theta functions with @xmath71 where @xmath72 for the characteristics @xmath73 $ ] , @xmath74 $ ] , @xmath75 $ ] and @xmath76 $ ] respectively .
if one observes that @xmath77 and @xmath69 are determined by the condition that the metric functions @xmath62 and @xmath13 vanish on the regular part of the axis , one can summarize the results in the following theorem .
we indicate quantities defined on the elliptic riemann surface @xmath78 given by @xmath79 with a prime .
then the ernst potential on the axis for @xmath80 has the form @xmath81 where @xmath82 , @xmath83 , @xmath49 , and where @xmath84 .
the real part of the ernst potential can be written in the form @xmath85 the constant @xmath77 can be expressed via theta functions , @xmath86 the constant @xmath69 is given by @xmath87 the ernst potential on the axis can be used to determine the ernst potential at the origin , @xmath88 , which is related to the redshift @xmath89 of photons emitted from the center of the disk and detected at infinity , @xmath90 .
the surface @xmath78 admits the involution @xmath91 which implies @xmath92 where @xmath93 denotes equal up to periods and where @xmath94 is the @xmath6-period on @xmath78 .
similarly one has @xmath95 and @xmath96 where the integral @xmath97 is to be understood as the principal value ( this leads to a contribution of @xmath98 times the residue ) .
using @xmath99 and @xmath100 one ends up with the ernst potential @xmath101 at the center of the disk is given by @xmath102 where @xmath103 is the purely imaginary quantity @xmath104 the adm mass @xmath105 and the angular momentum @xmath106 of the spacetime can be obtained by expanding the axis potential ( [ 2.17 ] ) in the vicinity of infinity .
the real part of the ernst potential for @xmath107 reads @xmath108 and the imaginary part @xmath109 ( see e.g. @xcite ) .
we get the adm mass is given by the formula @xmath110 and the angular momentum is given by @xmath111 here @xmath112 denotes the coefficient of the linear term in the expansion of a function @xmath113 in the local parameter in the vicinity of @xmath114 .
since we concentrate on positive values of @xmath2 , the riemann surface can only become singular if @xmath55 coincides with @xmath56 , i.e. on the axis , or if it coincides with @xmath115 . then , as on the axis , several periods of the riemann surface diverge .
with the same techniques as on the axis , it was proven in @xcite that the ernst potential and the metric functions are regular at this point .
we denote with a double prime the quantities defined on the riemann surface @xmath116 of genus 0 given by @xmath117 .
then the differentials on @xmath31 reduce for @xmath118 to differentials on @xmath116 , @xmath119 , @xmath120 and @xmath121 .
the ernst potential reads @xmath122 the function @xmath62 follows from @xmath123 and the function @xmath65 is given by @xmath124
in the equatorial plane , the riemann surface @xmath31 has the additional involution @xmath91 which makes it possible to express the ernst potential in terms of elliptic functions ( see @xcite ) . in i
it was shown that there exist algebraic relations between the real and imaginary parts of the ernst potential , the function @xmath125 and their one - sided derivatives at the disk ( @xmath126 and @xmath127 ) .
the function @xmath60 itself can be expressed via elliptic theta functions on the surface @xmath128 defined by @xmath129 .
we cut the surface in a way that the @xmath62-cut is a closed contour in the upper sheet around the cut @xmath130 $ ] and that the @xmath6-cut starts at the cut @xmath131 $ ] .
the abel map @xmath132 is defined for @xmath133 as @xmath134 .
let @xmath135 .
then the real part of the ernst potential at the disk is given by @xmath136 where @xmath137 at the disk , the relations @xmath138 for @xmath60 and @xmath6 , @xmath139 for @xmath140 and @xmath60 , and @xmath141 for the derivatives are valid .
* proof : * + one can define on @xmath128 the divisor @xmath142 as the solution of the jacobi inversion problem @xmath143 thus @xmath142 can be expressed in standard manner via theta functions on @xmath128 , @xmath144 in i it was shown that @xmath142 is related to @xmath60 ,
@xmath145 entering with this relation in ( [ y4 ] ) and solving for @xmath60 , one ends up after some algebraic manipulation with ( [ y1 ] ) .
the relations ( [ 2.9 ] ) to ( [ 2.10b ] ) were given in i. this completes the proof . to discuss physical quantities in the disk ,
the behaviour of the metric in the vicinity of the center and of the rim of the disk is important .
we have at the disk ( @xmath146 , @xmath80 ) , the metric functions and their derivatives have for @xmath147 and @xmath148 , @xmath149 an expansion of the form @xmath150 where @xmath113 stands for @xmath60 , @xmath140 , @xmath6 , @xmath151 , and @xmath65 , and where the @xmath152 are constants with @xmath153 and @xmath154 . + for @xmath155 and @xmath156 the metric functions and their derivatives have an expansion of the form @xmath157 , @xmath158 , @xmath159 , @xmath160,and @xmath161 .
+ for @xmath162 , the critical value @xmath163 . in this case
the expansion of the metric functions is as for @xmath155 , @xmath156 or of the form @xmath164 , @xmath165 , @xmath166 , @xmath167 , @xmath168 .
+ at the rim of the disk ( @xmath169 ) , the imaginary part of the ernst potential vanishes for @xmath170 and @xmath127 as @xmath171 .
* proof : * + since the riemann surface @xmath128 is regular for @xmath172 , all periods are functions of @xmath173 .
the integral @xmath174 is also a smooth function of @xmath173 which can be seen from @xmath175 where @xmath176 thus the quantity @xmath177 in ( [ y2 ] ) is a smooth function in @xmath173 .
for @xmath148 it has the form @xmath178 this implies via the relations of theorem 4.1 that @xmath60 , @xmath140 , @xmath6 and @xmath151 are also smooth functions in @xmath173 .
the behaviour of @xmath13 is determined by using the relation ( [ 2.10a10 ] ) together with the condition that @xmath13 vanishes on the axis .
+ in the case @xmath179 the ultra - relativistic limit @xmath180 is reached for @xmath181 for @xmath182 i.e. @xmath183 at the center of the disk . since the theta functions in ( [ y2 ] ) are functions in @xmath173 and since @xmath177 is quadratic in the theta functions , @xmath177 is of the form @xmath184 .
expanding @xmath60 in ( [ y1 ] ) for small @xmath177 one gets @xmath185 . for @xmath186 and @xmath187 , equation ( [ y1 ] ) leads to @xmath188 for @xmath189 or @xmath190 for @xmath191 . for @xmath189 relation ( [ 2.9 ] ) can either be satisfied by a function @xmath6 of the form @xmath192 or @xmath193 .
to decide one has to consider @xmath194 which has the form @xmath195 , where @xmath196 for @xmath182 .
thus @xmath194 can only vanish for @xmath162 in this case if @xmath197 , the static limit , where it vanishes identically .
+ the behaviour of @xmath6 at the rim of the disk follows from ( [ 2.9 ] ) which implies with ( [ y1 ] ) @xmath198 since the riemann surface @xmath128 and therefore its periods are regular at @xmath169 , the integral @xmath174 is dominated by the integral in ( [ y7 ] ) which is proportional to @xmath171 for @xmath199 in the disk .
the theta function @xmath200 has zeros of first order at @xmath201 which implies that @xmath177 diverges for @xmath202 .
consequently for @xmath203 we have @xmath204which implies that @xmath205 i.e.@xmath206 in the non - static case .
the one - component disk which was studied by bardeen and wagoner @xcite is obtained by simply putting @xmath155 in the ernst potential ( [ rel1a ] ) .
this gives the solution of neugebauer and meinel @xcite in the notation of @xcite .
the newtonian limit is reached for @xmath207 for arbitrary @xmath23 which follows from ( [ 2.1 ] ) : if @xmath208 and @xmath209 , this means that both the central redshift and the maximal velocity in the disk ( compared to the velocity of light which is 1 in the used units ) are small , which just defines the newtonian regime .
we get equation ( [ 2.11 ] ) just describes the maclaurin disk in the notation of i. the metric has the behaviour one expects from a general post - newtonian expansion . +
* proof : * + in the limit @xmath207 , the branch points @xmath215 and @xmath216 , @xmath217 , tend to infinity . to treat this limiting case
, we use the conformal transformation @xmath218 . on the transformed surface , the cut @xmath219 $ ] collapses as on the axis , whereas the remaining cuts remain finite . in the limit
@xmath220 the abelian integrals can be expressed in terms of quantities on the elliptic riemann surface @xmath221 given by @xmath222 .
we get @xmath223 the @xmath224 and @xmath225 are however not abelian integrals .
we have @xmath226 , whereas the corresponding @xmath227-periods are proportional to @xmath22 . with @xmath228
we thus find that @xmath229 and @xmath225 are proportional to @xmath230 and @xmath231 in lowest order . the leading contributions in ( [ 2.20a ] )
are consequently given by @xmath97 in @xmath232 . since @xmath233 in this case ,
we get ( [ 2.11 ] ) . in a similar way the lowest order contributions to the imaginary part can be obtained from ( [ 2.20a ] ) .
they arise from the term @xmath234 which is an odd and purely imaginary function .
it has zeros of first order which implies that @xmath235 which is of order @xmath230 i.e. of order @xmath213 . thus @xmath62 is also of order @xmath213 . for the metric function
@xmath13 we get with ( [ 2.7 ] ) as above that the leading contributions arise from the integral in the exponent which are quadratic in @xmath236 and thus of order @xmath214 .
this completes the proof .
the function ( [ rel1a ] ) becomes real in the limit @xmath239 with the potential @xmath211 given by @xmath240 the metric function @xmath62 vanishes identically whereas the function @xmath13 is given by @xmath241 * proof : * + in the limit @xmath242 , the branch points @xmath215 and @xmath216 coincide for @xmath217 on the real axis since @xmath243 and thus @xmath244 .
mathematically this limit corresponds to the standard solitonic limit of algebro - geometric solutions of evolution equations , see e.g.@xcite . in the cut - system fig .
[ schnitt1 ] the @xmath6-periods @xmath245 and @xmath246 diverge whereas the remaining periods are finite .
thus in this limit the theta functions in ( [ rel1a ] ) become identical to 1 whereas the differential of the third kind @xmath247 . the limit of ( [ 2.4 ] ) for @xmath248
is ( [ 2.15 ] ) which is identical to @xmath249 , i.e. the form which was given for the morgan and morgan disk with constant @xmath20 in @xcite .
since the theta functions in ( [ 2.6 ] ) all tend to one , we find that @xmath62 is identical to zero in this case with @xmath64 being equal to zero for @xmath250 . in the expression for the metric function @xmath13 ( [ 2.7 ] ) the theta functions again tend to 1 . to evaluate the integral in the exponent , we integrate by parts with respect to both @xmath251 and @xmath252 using the fact that @xmath253 vanishes at the limits of integration .
the differential of the second kind then takes the form given in ( [ 2.15a ] ) . since @xmath13 obviously vanishes on the axis
, the constant @xmath69 is identical to 1 in this limit .
this completes the proof .
the ultra - relativistic limit is reached if the redshift of photons emitted at the center of the disk and detected at infinity diverges which is equivalent to the fact that photons from the center of the disk can not escape to infinity as from a horizon of a black - hole . in the case
@xmath179 the limit is reached for @xmath254 .
this implies with ( [ 2.17b ] ) and ( [ k3 ] ) that both constants @xmath77 and @xmath69 diverge as @xmath255 .
the axis is in fact singular in the sense that the metric function @xmath60 vanishes there identically which can be seen from ( [ 2.17a ] ) .
the ernst potential is identical to @xmath51 on the axis for @xmath80 .
a consequence of the diverging constant @xmath77 is that the angular velocity @xmath256 , which is the coordinate angular velocity in the disk as measured from infinity , vanishes .
since this implies that either @xmath27 or @xmath20 vanish , bardeen and wagoner @xcite argued that the spacetime can be interpreted in the limit @xmath257 and @xmath258 as the extreme kerr metric in the exterior of the disk .
in @xcite it was shown that such a limit ( diverging multipoles , singular axis , ) can occur in general hyperelliptic solutions and can always be interpreted as an extreme kerr spacetime . for an algebraic treatment of the ultra - relativistic limit of the bardeen - wagoner disk
see @xcite .
we have let @xmath259 be an ernst potential of the form ( [ rel1a ] ) , which is regular except at the disk , with @xmath254 and @xmath260 where @xmath261 is a finite non - zero positive real constant .
then in the limit @xmath258 , the ernst potential ( [ rel1a ] ) for @xmath262 takes the form @xmath263 in the ultra - relativistic limit of the above disks for @xmath179 , this implies that the spacetime becomes an extreme kerr spacetime with @xmath264 .
+ * proof : * + the most direct way to prove this statement seems to determine the potential in the limit on the axis and then to use a theorem of hauser and ernst @xcite that an ernst potential is uniquely determined for given sources if it is given on some finite regular part of the axis .
the potential on the axis is given by ( [ 2.17 ] ) .
the limit @xmath258 is equivalent to the limit @xmath265 as used in the calculation of the adm - mass and the angular momentum in section 3 with @xmath254 .
the constant @xmath77 can be written with the help of fay s trisecant identity @xcite as @xmath266 this implies with identities between elliptic theta functions ( see e.g. @xcite ) and the fact that @xmath267 , the @xmath6-period on @xmath78 , that @xmath268 .
expanding the axis potential ( [ 2.17 ] ) in first order of @xmath27 , one thus gets @xmath269 which is the potential ( [ ultra6 ] ) on the axis .
this completes the proof .
for @xmath270 the axis remains regular except at the origin , the constants @xmath77 and @xmath69 in ( [ 2.17b ] ) and ( [ k3 ] ) remain finite here since they can only diverge if @xmath254 which can happen only for @xmath179 .
the integrals in the respective exponents of ( [ 2.17b ] ) and ( [ k3 ] ) are always finite though @xmath271 has a term @xmath272 in the limit @xmath186 as can be easily seen .
the ultrarelativistic limit of the disks with counter - rotation is thus a disk of finite radius with diverging central redshift . | this is the second in a series of papers on the construction of explicit solutions to the stationary axisymmetric einstein equations which can be interpreted as counter - rotating disks of dust .
we discuss the class of solutions to the einstein equations for disks with constant angular velocity and constant relative density which was constructed in the first part .
the metric for these spacetimes is given in terms of theta functions on a riemann surface of genus 2 .
we discuss the metric functions at the axis of symmetry and the disk .
interesting limiting cases are the newtonian limit , the static limit , and the ultra - relativistic limit of the solution in which the central redshift diverges .
pacs numbers : o4.20.jb , 02.10.rn , 02.30.jr |
a central topic of past and modern hadron physics is the determination of the wave function of resonances in terms of quark and gluon degrees of freedom , both in the baryon and meson sectors and both for light and heavy quarks ( for reviews see refs .
@xcite ) . in the mesonic sector , beyond conventional quark - antiquark ( @xmath1 ) mesons , one has glueball states , multiquark states such as tetraquarks and ` dynamically generated resonances ' , most notably molecular states .
indeed , different authors used the term ` dynamical generation ' in rather different contexts . in the works of refs .
@xcite a dynamical generated state is regarded as a resonance obtained via unitarization methods from a low - energy lagrangian ; in refs .
@xcite it is considered as a state which does not follow the quark - antiquark pattern in large-@xmath0 expansion . in refs .
@xcite the concept of ` additional , companion poles ' as dynamically generated states is introduced , while in refs .
@xcite a dynamically generated resonance is regarded as loosely bound molecular state .
these various definitions are not mutually exclusive and describe different points of view of the problem . in this article ( sec . ii and iii ) a definition for a dynamically generated state is proposed and its implications , also in connection with the aforementioned works , are presented .
this definition , in turn , leads to a method allowing to distinguish in principle tetraquark and molecular states , although they are both four - quark states . the concept of ` dynamical reconstruction ' is then introduced and discussed : it applies to resonances which are obtained from low - energy effective lagrangians via unitarization methods , but still correspond to ` fundamental ' ( not dynamically generated ) @xmath2 glueball or multiquark states . in this context , the study of the large @xmath0 behavior of these resonance constitutes a useful tool to discuss their nature . in the end of sec .
iii some general thoughts about the form of an effective theory of hadrons valid up to 2 gev are also presented . in sec .
iv the attention is focused on a simple toy model , in which only two scalar fields are considered .
the parameters of the toy model are chosen in a way that both fields behave as quarkonium - like state in the large @xmath0 limit .
the heavier state is first integrated out in order to obtain an effective low - energy lagrangian in the toy world and then is reobtained via a bethe - salpeter ( bs ) study applied to the low - energy lagrangian .
it is shown that this is ( at least in some cases ) possible and that the corresponding large @xmath0 behavior can be brought in agreement with the expected large @xmath0 limit ; this is a subtle and interesting issue on its own . however , when only the lowest term of the effective , low - energy lagrangian is retained , the large @xmath0 limit of the reconstructed state is not reproduced : instead of the correct large @xmath0 quarkonium limit ( which must hold by construction ) , it fades out as a molecular state would do .
implications of these results are presented : it is proposed that axial - vector , tensor and ( some ) scalar mesons just above 1 gev obtained via the bs approach from the corresponding low - energy , effective lagrangian in which only the lowest term is kept , are ( reconstructed ) quark - antiquark fields , in agreement with the constituent quark model , although they might fade away as molecular states in the large @xmath0 limit . finally , in sec .
v the conclusions are presented .
consider a physical system which is correctly and completely described in the energy range @xmath3 by a quantum field theory in which n fields @xmath4 @xmath5 their masses and interactions are encoded in a lagrangian @xmath6 .
each lagrangian has such an @xmath7 beyond which it can not be trusted .
in particular , we shall refer to @xmath7 in the following sense : all the masses @xmath8 of the states @xmath9 and the energy transfer in a two - body scattering @xmath9 @xmath10 should be smaller than @xmath7 @xmath11 and , in the s - channel , @xmath12 ) .
moreover , we also assume that : ( i ) the theory is not confining , thus , to each field @xmath9 there is a corresponding , measurable resonance ( and at least one with zero width ) .
( ii ) if not renormalizable , an appropriate regularization shall be specified . a resonance @xmath13 emerging in the system described by @xmath14 ,
is said to be dynamically generated if it does _ not _ correspond to any of the original fields @xmath4 @xmath15 @xmath16 @xmath14 in the lagrangian and if its mass @xmath17 lies below @xmath7 ( @xmath18 ) .
the last requirement @xmath18 is natural because the state @xmath19 can be regarded as an additional , dynamically generated resonance only if it belongs to the energy range in which the theory is valid .
this simple consideration plays an important role in the following discussion . clearly , the dynamical generated state @xmath19 emerges via interactions of the original resonances @xmath20 when switching them off , @xmath19 must disappear .
for this reason a dynamically generated mesonic resonance in qcd fades out in the large-@xmath0 limit , which corresponds to a decreasing interaction strength of mesons , see later for more details .
some examples and comments are in order : \(a ) @xmath21 , in which the electron and the photon fields are the basic fields .
this theory is valid up to a very large @xmath7 ( gut scale ) .
positronium states are molecular , electron - positron bound states .
they appear as poles close to the real axis just below the threshold @xmath22 , but slightly shifted due to their nonzero decay widths into photons .
clearly , positronium states are ` dynamically generated ' according to the given definition and should not be included in the original qed lagrangian , otherwise they would be double - counted .
note , the number of positronium states is infinite .
\(b ) in lagrangians describing nucleon - nucleon interaction via meson exchange ( @xmath23 @xmath24 @xmath25 and @xmath26 ) a bound state close to threshold , called the deuteron , emerges via yukawa - interactions , see for instance ref .
@xcite and refs .
the deuteron is a dynamically generated molecular state . in this case , when lowering the interaction strength below a critical value ( by reducing the coupling and/or increasing the mass of the exchanged particle(s ) ) , the bound state disappears .
in fact , the number of molecular states which can be obtained via a yukawa potential is finite @xcite , and eventually zero if the attraction is too weak . in such models
the deuteron should not be included in the original lagrangian in order to avoid double - counting @xcite .
\(c ) @xmath27 the fermi theory of the weak interaction , in which the neutrino and electron fields interact via a local , quartic interaction .
this theory is valid up to @xmath28 where @xmath29 is the boson mediator of the weak force . as already mentioned in ref .
@xcite , the linear rise of the @xmath30 cross section -as calculated from @xmath31- shows a loss of unitarity at high energy .
unitarization applied to @xmath31 implies that a resonance well above @xmath7 exists , and this resonance is exactly the @xmath29 meson .
however , being @xmath32 one can not state in the framework of the fermi theory if the @xmath29 meson is dynamically generated or not . a straightforward way to answer this question
is the knowledge of the corresponding theory valid up to an energy @xmath33 of course , this theory is known : it is the electroweak theory described by the lagrangian @xmath34 , which is part of the standard model @xcite and is valid up to a very high energy ( gut scale ) . in the framework of @xmath34 the neutrino ,
the electron and the @xmath29 meson are all elementary fields .
one can then conclude that the @xmath29 meson is not a dynamically generated state .
indeed , @xmath31 can be seen as the result of integrating out the @xmath29 field from the electroweak lagrangian @xmath34 .
unitarization arguments applied to the fermi lagrangian @xmath35 allow in a sense to ` dynamically reconstruct ' the @xmath36 which is already present as a fundamental field in @xmath34 .
\(d ) it is important to discuss in more depth and to formalize the issue raised in the previous example . to this end
let us consider the lagrangian @xmath37 as the low - energy limit of a lagrangian @xmath38 valid up to an energy @xmath39 beyond the fields @xmath40 @xmath41 depends also on the fields @xmath42 , which are heavier than @xmath7 .
formally , when integrating out the fields @xmath42 from @xmath41 one obtains @xmath14 . in general , a unitarization scheme uses the information encoded in a low - energy effective lagrangian and the principle of unitarity in quantum field theories , in order to deduce the existence and some properties of resonances beyond the limit of validity of the theory itself .
when applying a unitarization scheme to the lagrangian @xmath14 , an energy window between @xmath43and a new energy scale @xmath44 - which depends on the detail of the unitarization- becomes ( partially ) accessible . for our purposes , we assume that @xmath45 . let @xmath19 be a resonance with mass @xmath46 obtained from @xmath14 via a unitarization approach .
is this resonance @xmath19 dynamically generated or not ?
a straightforward way to answer this question would be the knowledge of @xmath41 .
if @xmath19 corresponds to one of the fields @xmath42 is not dynamically generated and vice - versa .
however , if @xmath41 is not known it is not possible to answer this question at the level of the unitarized version of @xmath14 only . in conclusion , although the unitarization approach opens a window between @xmath43and @xmath47 and the existence of resonances in this range can be inferred from the unitarized lagrangian @xmath14 only , still the knowledge of the latter is not complete @xcite . if @xmath48 is unknown , some other kind of additional information is required to deduce the nature of @xmath19 . in the framework of low - energy qcd
this additional information can be provided by large-@xmath0 arguments , see sec .
\(e ) let us consider a scalar field @xmath49 in a 1 + 1 dimensional world @xmath50 subject to the potential @xmath51 we assume that this theory is valid up to high energies .
when expanding around one of the two minima @xmath52 the mass of @xmath53 is found to be @xmath54 in addition , this theory admits also a soliton with mass @xmath55 which is large if @xmath56 is small @xcite . in this example
the solitonic state with mass @xmath57 can be regarded as a ` dynamically generated state ' .
\(f ) mixing can take place among two ` fundamental fields ' @xmath9 and @xmath58 : two physical resonances arise as an admixture of these two fields .
one is predominantly @xmath9 and the other predominantly @xmath59 also , mixing can take place among a dynamically generated resonance @xmath19 and one ( or more ) of the @xmath20 it decreases when the interaction is lowered ( large-@xmath0 limit in the mesonic world ) : one state reduces to the original , preexisting resonance and the other disappears . in conclusion ,
mixing surely represents a source of technical complication which renders the identification of states ( extremely ) more difficult , but it does _ not _ change the number of states and the meaning of the previous discussion .
let us turn to the hadronic world below 2 gev .
the basic ingredients of each low - energy hadronic lagrangian are quark - antiquark mesons and three - quark baryons . in the framework of our formalism
, we shall consider each quark - antiquark ( 3-quark ) state as a fundamental state , which is described by a corresponding field in the hadronic lagrangian ( as long as its mass is below an upper energy @xmath7 ) .
let us formalize this point in the mesonic sector as it follows .
consider the correct , effective theory describing mesons up to @xmath60 gev given by @xmath61{c}% , \end{array } \label{1}%\ ] ] where @xmath0 is the number of colors .
its precise form is , unfortunately , unknown .
in fact , because confinement has not yet been analytically solved , it is not possible to derive @xmath62 from the qcd lagrangian . in the limit @xmath63 the effective lagrangian @xmath64 is expected to be more simple .
although even in this limit a mathematical derivation is not possible , it is known that it must primarily consists of non - interacting quark - antiquark states .
in fact , their masses scale as @xmath65 and their decay widths as @xmath66 respectively @xcite . considering that the @xmath1 mass scales as @xmath67 these states shall be clearly present also when going back from the large @xmath0 limit to the physical world @xmath68 the next - expected states which are present in the large @xmath0 limit are glueballs , i.e. bound states of pure gluonic nature .
their masses also scale as @xmath65 and the decay widths as @xmath69 they thus are also expected to be present in the real world for @xmath68 in particular , the lightest glueball is a scalar field which is strongly related to the trace anomaly , i.e. the breaking of the classical dilatation invariance of the qcd lagrangian ( see also sec .
in addition to quark - antiquark and glueball states , also hybrid states survive in the large-@xmath0 limit @xcite .
they constitute an interesting subject of meson spectroscopy ( see ref .
@xcite and refs .
therein ) , but will not be considered in the following discussion .
all these states are therefore ` preexisting ' and not - dynamically generated states of the mesonic lagrangian under consideration .
an intermediate comment is devoted to baryon states : they have an linearly increasing mass with @xmath0 ( @xmath70 ) which exceeds @xmath7 for a large enough @xmath0 and are therefore not present in the large-@xmath0 limit of @xmath64 .
thus , although they appear in the @xmath71 world , they are not present in the effective lagrangian because of the way in which the limit is constructed . if , instead , we construct the large @xmath0 limit as @xmath72 ( in such a way that at @xmath71 it coincides with eq .
( [ 1 ] ) ) , baryons are well defined states as proved originally in ref . @xcite . for simplicity baryons will not be discussed in this paper , but together with hybrid states , should be included in a more complete treatment .
the reason why the value @xmath60 gev is chosen is that all the resonances under study in this work are lighter than 2 gev .
thus , they either correspond to a field in @xmath73 or arise as additional resonances ( i.e. dynamically generated ) via interaction of preexisting states of @xmath74 .
the full knowledge of the lagrangian @xmath75 gev@xmath76 would allow to answer if a resonance lighter than 2 gev is dynamically generated or not by a simple look at it .
clearly , if one would be interested in a resonance whose mass is heavier than 2 gev , then @xmath7 should be increased .
moreover , one expects to find below 2 gev all the relevant ground - state mesons in the channel @xmath77 @xmath78 @xmath79 , @xmath80 @xmath81 .
thus , @xmath82 gev@xmath83 may be described by an effective lagrangian which exhibits linear realization of chiral symmetry and its spontaneous breakdown , see section iii.d for a closer discussion .
most of the mesonic resonances listed in the particle data group @xcite can be immediately associated to a corresponding , underlying quark - antiquark state . yet , the question if some resonances of ref .
@xcite are not @xmath1 is interesting and at the basis of many studies . in the mesonic sector ,
two alternative possibilities are well - known : @xmath84 molecular states : they are bound state of two distinct quark - antiquark mesons .
they correspond to the example of the positronium ( example ( a ) sec .
just as the positronium states are not included in the qed lagrangian , hadronic molecular states should _ not _ be included directly in @xmath85 they arise upon meson - meson interactions in the @xmath71 physical world , see the general discussion above .
however , they inevitably fade out in the large-@xmath0 limit because the interaction of a n - leg meson vertex decreases as @xmath86 this is therefore a clear example of dynamically generated states within a mesonic system ( see also the next subsection for a closer description of physical candidates below 1 gev ) .
@xmath84 tetraquark states : they consist of two distinct , colored ` bumps ' , in contrast to a molecular state , which is made of two colorless , quark - antiquark bumps @xcite
. loops of @xmath1 mesons , corresponding to the interaction of two colorless states , can not generate the color distribution of a tetraquark .
if present at @xmath71 , they shall be included directly in the effective lagrangian @xmath87 and , in view of the given definition , should not be regarded as dynamically generated states . a second , slightly different way to see
it is the following : let us imagine to construct the lagrangian @xmath88 we first put in quark - antiquark and glueball states , that is those configurations which surely survive in the large-@xmath0 limit and corresponds to non - dynamically generates states : @xmath89 then the question is : ` does this lagrangian describe the physical world for @xmath71 ' ? ( note , loops shall be taken into account and dynamically generated states can eventually emerge out of this lagrangian . )
if the answer is positive , no multiquark states are needed .
if the answer is negative , the basic lagrangian shall be extended to include from the very beginning multiquark states , most notably tetraquark states : @xmath90 a third approach to the problem goes via large-@xmath0 arguments . in refs .
@xcite it has been shown that a tetraquark state also vanishes in the large-@xmath0 limit . however , for @xmath71 the most prominent and potentially relevant for spectroscopy is the ` good ' diquark , which is antisymmetric in color space : @xmath91 ( with @xmath92 @xcite .
the tetraquark is the composition of a good diquark and a good antidiquark : @xmath93 .
the extension to @xmath0 of a good diquark is the antisymmetric configuration @xmath94 with @xmath95 which constitutes of @xmath96 quarks .
thus , the generalization of the tetraquark to the @xmath0 world is not a diquark - antidiquark object , but the state @xmath97 which is made of @xmath96 quarks and @xmath96 antiquarks , see also the discussion in ref .
it is the dibaryonium already described in ref .
@xcite which has a well defined large-@xmath0 limit : its mass scale as @xmath98 and decays into a baryon and an antibaryon .
the state @xmath99 while not present in @xmath100 because its mass overshoots @xmath7 , appears in @xmath101 in which also the baryons survive : this is contrary to a dynamically generated state , which disappears also in this case . as a result of our discussion , tetraquark states and molecular states , although both formally four - quark states , are crucially different : the former are ` elementary ' and should be directly included in the effective lagrangian @xmath87 , the latter can emerge as dynamically generated resonances .
we now turn to the particular case of the light scalar mesons , where all these concepts play an important role .
one of the fundamental questions of low energy qcd concerns the nature of the lightest scalar states @xmath102 @xmath103 @xmath104 and @xmath105 shall these states be included from the very beginning in @xmath106 ?
if yes , they correspond to quark - antiquark or tetraquark nonets ( one of them can be also related to a light scalar glueball ) .
if not , they shall be regarded as dynamically generated states .
the main point of the following subsection is to discuss previous works about light scalar mesons in connection with the proposed definition of dynamical generation .
in fact , it is easy to classify previous works into two classes ( not dynamically generated and dynamically generated ) , thus allowing to order different works of the last three decades in a clear way .
we first review works in which scalar states are not dynamically generated and then works in which they are dynamically generated .
_ the light scalar states are not dynamically generated and should be directly included in _
@xmath107__ : _ _ \(i ) in the quark - antiquark picture these light scalar states form the nonet of chiral partners of pseudoscalar mesons .
their flavor wave functions reads @xmath108 @xmath109 @xmath110 @xmath111 at the microscopic level this is the prediction of the njl model @xcite , where a @xmath26 mass of about @xmath112 is obtained and where @xmath113 mev is the constituent quark mass .
this is usually the picture adopted in linear sigma models at zero @xcite and at nonzero density and temperature @xcite .
however , this assignment encounters a series of problems : it can hardly explain the mass degeneracy of @xmath114 and @xmath115 the strong coupling of @xmath114 to kaon - kaon , the large mass difference with the other p - wave nonets of tensor and axial - vector mesons @xcite , it is at odd with large-@xmath0 studies ( see later on ) and with recent lattice works @xcite .
\(ii ) tetraquark picture , first proposed by jaffe @xcite and revisited in refs .
@xcite : @xmath116[u , d],$ ] @xmath117[u , s]+[\overline{d},\overline{s}][d , s]),$ ] @xmath118[u , d],$ ] @xmath119[u , d]$ ] where @xmath120 $ ] stands for antisymmetric configuration in flavor space ( which , together with the already mentioned antisymmetric configuration in color space , implies also a s - wave and spinless structure of the diquarks and of the tetraquarks ) .
degeneracy of @xmath114 and @xmath121 is a natural consequence .
a good phenomenology of decays can be obtained if also the next - to - leading order contribution in the large-@xmath0 expansion is taken into account @xcite and/or if instanton induced terms are included @xcite .
linear sigma models with an additional nonet of scalar states can be constructed @xcite .
the quark - antiquark states lie above 1 gev @xcite and mix with the scalar glueball whose mass is placed at @xmath122 gev by lattice qcd calculations @xcite .
this reversed scenario directly affects the physics of chiral restoration at non - zero temperature @xcite .
\(iii ) different assignments , in which also the glueball state shows up below 1 gev have been proposed , see refs . @xcite and refs . therein .
\(iv ) in all these assignments the very existence of the scalar mesons is due to some preformed compact bare fields entering in @xmath123 . by removing the corresponding bare resonances from @xmath124 they disappear .
dressing via meson - meson loops , such as @xmath125 for @xmath126 @xmath127 for @xmath128 and @xmath129 for @xmath114 and @xmath115 surely takes place .
in particular , due to the intensity in these channels and the closeness to thresholds , they can cause a strong distortion and affect the properties of the scalar states @xcite .
however , the important point is that in all these scenarios mesonic loops represent a further complication of light scalars , but are not the reason of their existence .
\(v ) as discussed in section iii.a , non - dynamically generated scalar states survive in the large-@xmath0 limit , although in a different way according to quarkonium , glueball or tetraquark interpretations .
_ the light scalars states are dynamically generated and _
_ should not be included in _
@xmath64 : \(i ) in ref .
@xcite the @xmath26 pole arises as a broad enhancement due to the inclusion of @xmath130 mesons in the t - channel isoscalar @xmath125 scattering . in this case
the @xmath26 is ` dynamically generated ' and arises because of a yukawa - like interaction due to @xmath130 meson exchange ( pretty much as the deuteron described above , but above threshold ) . when reducing the @xmath131 coupling @xmath132 ( which , in the large-@xmath0 limit scales as @xmath133 ) , the @xmath26 fades out .
alternatively , the limit @xmath134 also implies a disappearance of the @xmath26-enhancement .
\(ii ) similar conclusions for the @xmath135 and @xmath136 meson , described as molecular @xmath137 bound states just below threshold , have been obtained in refs .
@xcite . in particular , in ref .
@xcite the origin of these states is directly related to a one - meson - exchange potential . within all these approaches
the @xmath136 and the @xmath135 are ` dynamically generated ' .
\(iii ) in the model of ref . @xcite the @xmath114 state also arises as an additional , dynamically generated state , but in a different way .
scalar and pseudoscalar quark - antiquark mesons are the original states .
a bare scalar state with a mass of @xmath138 gev is the original , quark - antiquark ` seed ' .
when loops of pseudoscalar mesons are switched on , the mass is slightly lowered and the state is identified with @xmath139 in addition , a second state , arising in this model as a further zero of the real part of the denominator of the propagator , is identified with the @xmath136 meson : it is dynamically generated and disappears in the large-@xmath0 limit , where only the original quark - antiquark seed survives .
more in general , we refer to @xcite for the emergence of additional , companion poles not originally present as preexisting states in the starting lagrangian . in particular , in ref .
@xcite the conventional scalar quark - antiquark states , calculated within an harmonic oscillator confining potential , lie above 1 gev . when meson loops are switched on , a
complete _ second nonet of dynamically generated states below 1 gev emerge .
note , in all these studies the validity of the employed theories lies well above 1 gev , so that the definition of dynamical generation given in section ii holds for the light scalars .
\(iv ) note , in ( i ) and ( ii ) the emergence of states is due to t - channel forces .
this is not the case in ( iii ) .
however , a common point of them is that the light scalar states disappear in the large-@xmath0 limit .
it is clear that the situation concerning light scalars is by far not understood .
we wish , however , to stress once more that there is a crucial difference among the two outlined options in relation to the lagrangian @xmath85 note also that in this subsection we only discussed works for which it is possible to immediately conclude if the scalar mesons are dynamically generated or not according to the definition given in sec .
unitarization methods were not discussed here ; in fact , when the latter are applied , care is needed .
this is the subject of the next subsection .
the lagrangian * * * * @xmath62 with @xmath60 gev induces breakdown of chiral symmetry @xmath140 where @xmath141 is the number of light flavors . there are therefore @xmath142 goldstone bosons : the pion triplet for @xmath143 in addition four kaonic states and the @xmath144 meson for @xmath145 if we integrate out all the fields in @xmath106 besides the three light pions , we obtain the lagrangian of chiral perturbation theory ( see ref .
@xcite and refs .
therein ) for @xmath146@xmath147{c}% , \end{array}\ ] ] where @xmath148 should be lower than the first resonance heavier than the pions ( @xmath149 mev ) .
@xmath150 is recasted in an expansion of the pion momentum @xmath151 and for each @xmath152 there is a certain number of low - energy coupling constants , which in principle could be calculated from @xmath153 if it were known .
being this not the case , they are directly determined by experimental data .
( similar properties hold when the kaons and the @xmath144 are retained in @xmath150 ) .
for instance , the vector isotriplet @xmath130 meson is predicted by a large variety of approaches ( such as quark models ) to be a preexisting @xmath154 quark - antiquark field . in this sense
it is a fundamental field appearing in @xmath82 gev@xmath155 which is integrated out ( together with other fields ) to obtain @xmath156 however , the @xmath130 meson spectral function can not be obtained from chiral perturbation theory unless a unitarization scheme is employed @xcite . as an example , via the iam ( inverse amplitude method ) unitarization scheme applied to @xmath157 @xcite , a window between the original energy @xmath148 and @xmath158 gev is opened : resonances with masses in this window , such as the @xmath130 meson , can be described within unitarized @xmath159 as discussed in the point ( d ) of sec .
ii , the very last question if the @xmath130 meson is dynamically generated or not _ can not _ be answered at the level of unitarized @xmath160 .
one still does not know if @xmath130 corresponds to a basic , preexisting field entering in @xmath161 gev , @xmath162 or not . some additional information is needed . in the interesting and important case of large-@xmath0 studies of unitarized @xmath160
, the required additional knowledge is provided by the large-@xmath0 scaling of the low - energy constants : it has been shown in ref .
@xcite that the @xmath130 mass scales as @xmath65 and the width as @xmath66 and thus the @xmath130 meson should be considered as a fundamental ( not dynamically generated ) quark - antiquark field , which shall be directly included in @xmath163 .
we also refer to the analytic results of @xcite where the large-@xmath0 limit is evident .
let us turn to the lightest scalar - isoscalar resonance @xmath164 as obtained from ( unitarized ) @xmath160 . in refs .
@xcite precise determinations of the @xmath26 pole are obtained , but -as stated in ref . @xcite- it is difficult to understand its properties in terms of quarks and gluons . in ref .
@xcite a study of the @xmath26 pole within the iam - scheme in the large-@xmath0 limit has been performed : a result which is at odd with a predominantly quarkonium , or a glueball , interpretation of the @xmath26 meson has been obtained .
the mass is not constant and the width does not decrease .
however , even at this stage one still can not say if the @xmath26 is ` dynamically generated ' or ` reconstructed ' in relation to @xmath87 because it is hard to distinguish the molecular and the tetraquark assignments in the large-@xmath0 limit ( see discussion above ) .
recently , the bethe - salpeter unitarization approach has been used to generate various axial - vector @xcite , tensor and scalar mesons above 1 gev @xcite .
the starting point are low - energy lagrangians for the vector - pseudoscalar ( such as @xmath165 ) and vector - vector ( such as @xmath166 interactions .
these lagrangians are also in principle derivable by integrating out heavier fields from the complete @xmath74 . for instance , in the @xmath165 axial - vector channel the @xmath167 meson is obtained and in the @xmath168 tensor and scalar channels the states @xmath169 and @xmath170 are found . are these dynamically generated states of molecular type ?
the answer is : not necessarily .
in fact , the masses of the obtained states lie _ above _ the energy limit of the low - energy effective theories out of which they are derived . even for these states
the possibility of ` dynamical reconstruction ' -just as for the @xmath130 meson described above- is not excluded : in this scenario , these resonances above 1 gev are intrinsic , preexisting quark - antiquark or glueball ( multiquark states are improbable here ) fields of @xmath171 while first integrated out to obtain the low - energy lagrangians , unitarization methods applied to the latter allow to reconstruct them . in the next section a toy model
is presented , in which this mechanism is explicitly shown : although a state obtained via bs - equation looks like a molecule , it still can represent a fundamental , preexisting quark - antiquark ( or glueball ) state . as discussed in the summary of the pdg compilation @xcite and in refs .
@xcite ( and refs .
therein ) the tensor resonances @xmath172 @xmath173 @xmath174 and @xmath175 represent a nonet of quark - antiquark states .
the ideal mixing , the very well measured strong and electromagnetic decay rates @xcite , the masses and the mass splitting are all in excellent agreement with the quark - antiquark assignment . in this case
they are fundamental ( intrinsic ) fields of @xmath124 which can be dynamically reconstructed ( rather than generated ) via unitarization scheme(s ) applied to a low - energy lagrangians .
although experimentally and theoretically more involved , the same can hold in the axial - vector channel : the resonances @xmath176 @xmath177 @xmath167 and @xmath178 are in good agreement with the low - lying @xmath179 quark - antiquark assignment .
even more complicated is the situation in the scalar channel : the low - lying quark - antiquark states mix with the scalar glueball @xcite .
also in this case , however , the possibility of ` dynamical reconstruction ' rather then ` generation ' is upheld .
if dynamical reconstruction takes place , there is no conflict between the quark model assignment of ref . @xcite and
the above mentioned recent studies .
note , also , that dynamical reconstruction is in agreement with the discussion of ref .
@xcite .
the lagrangian @xmath64 with @xmath180 gev has been a key ingredient allover the present discussion but it has not been made explicit because unknown
. an improved knowledge of ( at least parts of ) @xmath181 would surely represent a progress in understanding the low - energy hadron system . as a last step we discuss which properties it might have .
clearly , it must reflect the symmetries of qcd , most notably spontaneous breakdown of chiral symmetry .
the ( pseudo- ) scalar meson matrix @xmath182 , the ( axial-)vector , tensor mesons and the scalar glueball are its basic building blocks . moreover , if additional non - dynamically generated scalar states such as tetraquarks exist , they shall be also included .
we thus have a complicated , general @xmath26-model lagrangian with many terms , in which operators of all orders can enter , because renormalization is not a property that an effective hadronic lagrangian should necessarily have .
the question is if it is possible to obtain a ( relatively ) simple form out of this complicated picture , see also @xcite . a possibility to substantially simplify
the situation is via dilation invariance ; let us consider the ( pseudo-)scalar meson matrix @xmath183 which transforms as @xmath184 ( @xmath185 @xmath186 ) under chiral transformation and the dilaton field @xmath187 , subject to the potential @xcite @xmath188 , where @xmath189 is a dimensional parameter of the order of @xmath190 ( the glueball emerges upon shifting @xmath187 around the minimum of its potential @xmath191 ) . consider @xmath192 where @xmath193 is the dynamical part and @xmath194 @xmath195 $ ] is the potential describing masses and interactions of the fields ( dots refer to other degrees of freedom , such as ( axial-)vector ones ) .
we assume that : ( i ) in the chiral limit the only term in @xmath196 which breaks dilation invariance -and thus mimics the trace anomaly of qcd- is encoded in @xmath197 ( via the dimensional parameter @xmath189 ) .
( ii ) the potential @xmath196 is finite for any finite value of the fields . as a consequence of ( i ) only operators of order ( exactly ) four can be included .
they have the form @xmath198,$ ] @xmath199,$ ] @xmath200^{2}, ... $ ] as a consequence of ( ii ) a huge set of operators are not admitted .
in fact , an operator of the kind @xmath201^{2},$ ] is excluded because , although of dimension 4 , it blows up for @xmath202 . in this way we are left with a sizably smaller number of terms , even smaller than what renormalizability alone would impose @xcite . work alone this direction , including ( axial-)vector degrees of freedom is ongoing @xcite and can constitute an important source of informations for spectroscopy and for future developments at nonzero temperature and densities , where in the chirally restored phase a degeneration of chiral partners is manifest . in conclusion , a way to implement these ideas and use the definition of dynamical generation can be sketched as follows : after writing a general chirally symmetric lagrangian up to fourth order including the glueball and the quark - antiquark fields as basic states , one should attempt without further inclusion of any other state to describe physical processes up to @xmath203 gev , as pion - pion scattering , decay widths , etc . in doing this one should of course include loops . if , for instance , we start with a basic scalar - isoscalar quark - antiquark field above 1 gev , do we correctly reproduce the resonance @xmath204 when solving the bethe - salpeter channel in the @xmath125 sector below 1 gev ? if the answer is positive , the latter resonance is dynamically generated and there is no need of any other additional state .
if , albeit including loops , the attraction among pions turns out to be too weak to generate the resonance @xmath205 we conclude that it is necessary to enlarge our model by explicitly introducing a field which describes it .
as argued previously , this field can be identified as a tetraquark state . if this ambitious program will lead to a successful result is matter of future research .
in this section we start from a toy lagrangian , in which two mesons @xmath53 and @xmath206 interact .
a large-@xmath0 dependence is introduced in such a way that both fields behave as quarkonium states .
then , the field @xmath206 , which is taken to be heavier than @xmath207 is integrated out and a low - energy lagrangian with the field @xmath53 only is obtained .
a bethe - salpeter study is applied to the latter lagrangian : the question is if the original state @xmath206 , which was previously integrated out , can be reobtained in this way .
the answer is generally positive , however care is needed concerning the large-@xmath0 limit . in a straight bs - approach
the quarkonium - like large-@xmath0 limit of the state @xmath206 can not be reproduced .
however , as it shall be shown , within a modified bs approach the large-@xmath0 limit can be correctly obtained . the toy lagrangian @xcite consisting of the two fields @xmath53 ( with mass @xmath208 ) and @xmath206 ( with bare mass @xmath209 ) reads@xmath210{c}% , \end{array } \label{ltoy}%\ ] ] which we assume to be valid up to @xmath211 the @xmath0 dependence of the effective lagrangian is encoded in @xmath212 only : @xmath213 in this way both masses behave like @xmath65 and the decay amplitude for @xmath214 scales as @xmath215 just as if @xmath53 and @xmath206 were quarkonia states .
@xmath216 is the analogous of @xmath163 in a simplified toy world .
for definiteness we refer to values in gev : @xmath217 , @xmath218 @xmath219 will be varied within @xmath220 and @xmath221 [ ptb ] fignew.eps the propagator of the field @xmath206 is modified via @xmath53-meson loops and takes the form ( at the resummed 1-loop level , see fig .
1):@xmath222 ^{-1}%\ ] ] where @xmath223 is the 1-loop contribution , which is regularized via a 3-d sharp cutoff @xmath224 @xcite .
the dressed mass can be defined via the zero of the real part of @xmath225 i.e. @xmath226 in the large-@xmath0 limit @xmath227 this is true whatever definition of the mass of the resonance is chosen . for finite @xmath0
one has in general @xmath228 due to the loop corrections ( see ref .
@xcite for details ) . the @xmath193 matrix for @xmath229 scattering in the @xmath230-channel upon 1-loop resummation
is depicted in fig . 1 and reads @xcite:@xmath231 note ,
the present interest is focused on the 1-particle pole of the @xmath206 resonance and its corresponding enhancement in the @xmath193-matrix .
thus , for simplicity we limit the study of two - body scattering to the exchange on one ( dressed ) meson @xmath206 .
other diagrams , as for instance the exchange of two ( or more ) @xmath206 mesons , are not considered here ( see , for instance , ref .
@xcite ) . a more refined and complete approach should also include the dressing of the @xmath53 propagator and of the @xmath232 vertex .
all these complications , while important in a realistic treatment , can be neglected at this illustrative level .
we now turn to development of a low - energy lagrangian which involves only the light meson field @xmath233 we assume that @xmath219 is not too large so that for @xmath71 the mass @xmath57 lies above the threshold @xmath234 in this way it is allowed to integrate out the field @xmath206 from @xmath235 and obtain a low - energy ( le ) effective lagrangian for the @xmath229 interaction valid up to @xmath236 @xcite:@xmath237@xmath238 the lagrangian @xmath239 contains only quartic term of the kind @xmath240 @xmath241 ... . @xmath239 is the analogue of chiral perturbation theory or , more in general , of a low - energy lagrangian in this simplified system . the fact that we know explicitly the form of @xmath242 allows us to calculate the ` low - energy constants ' @xmath243 of eq .
( [ lec ] ) .
if the form of @xmath242 were unknown , then also the @xmath243 would be such .
note , each @xmath243 scales as @xmath244 _ _ bs - inspired unitarization , way 1__@xmath245 as a first exercise let us consider the low - energy lagrangian @xmath246 up to a certain order @xmath152 by approximating the potential to @xmath247 . by performing a bethe - salpeter study with this approximate potential ,
see fig . 1
, we obtain the following @xmath193 matrix:@xmath248@xmath249 where @xmath250 is the bare tree - level amplitude corresponding to the sum of all the quartic terms up to order @xmath251 clearly , @xmath252 is an approximate form of @xmath253 of eq .
( [ fullt ] ) .
the larger @xmath152 , the better the approximation .
formally one has @xmath254 @xmath255 .
what we are doing is a ` dynamical reconstruction ' of the state @xmath206 via a bethe - salpeter analysis applied to the low - energy lagrangian @xmath246 : we reobtain the state @xmath206 which has been previously integrated out .
let us keep @xmath152 fixed and perform a large-@xmath0 study of @xmath256 do we obtain the correct result , that is @xmath257 the answer is no .
in fact , in the large-@xmath0 limit @xmath250 scales as @xmath258 due to dependence encoded in @xmath212 , while @xmath223 scales as @xmath65 ( we assume that the cutoff does not scale with @xmath0 @xcite ) . in the large-@xmath0 limit
we obtain @xmath259 but @xmath250 is a polynomial in @xmath260 and , for any finite @xmath261 does not admit poles for finite @xmath260 , but only for @xmath262 .
thus , we find the incorrect result that in the large-@xmath0 limit the mass of the dynamically reconstructed state is infinity .
this is shown in fig . 2 for a particular numerical choice . [ ptb ] fig1.eps
although our analysis has been applied to a simple toy model , the form of eq .
( [ bs ] ) is general .
one has a polynomial form for @xmath250 as function of @xmath260 and a mesonic loop @xmath223 which is independent on @xmath263 complications due to different quantum numbers do not alter the conclusion .
we also note that these results are in agreement with the discussion of ref .
@xcite , where the scalar @xmath26 meson is first integrated out and then ` reconstructed ' in the framework of the linear sigma model .
_ iam inspired unitarization _
if we would , instead , apply the iam unitarization scheme to the @xmath264 approximate form we would obtain the correct result in the large-@xmath0 expansion .
in fact , in this case one schematically has ( neglecting @xmath265 and @xmath266 channels ) : @xmath267 where @xmath268 in our notation .
being @xmath269 and @xmath270 one finds @xmath271 which represents a valid approximation of the full @xmath193 matrix if @xmath212 is not too large @xmath272 it is straightforward to see that the iam approximation delivers the correct large-@xmath0 result , namely @xmath273 and a width decreasing as @xmath274 clearly one could repeat this study for increasing @xmath152 finding a better and better approximation of @xmath193 .
_ _ bs - inspired unitarization , way 2__@xmath245 _ _ contrary to the bs - inspired unitarization described above ( way 1 ) , it is possible to follow a different bs - inspired approach which is in agreement with the large-@xmath0 limit . for simplicity
we discuss it in the explicit case @xmath264 @xcite .
one has : @xmath275 now , instead of plugging @xmath276 directly into eq .
( [ tappr ] ) , we first invert it obtaining the approximate form @xmath277 valid up to order @xmath278 : @xmath279 the next step is to write the @xmath193 matrix in terms of @xmath280 : @xmath281 thus , this new approximate form derived from the bs equation is now in agreement with the large-@xmath0 limit and is equivalent to the iam - inspired unitarization approach described above .
this shows an important fact in this discussion : it is not the bs method which fails in bs - way 1 , but rather the adopted perturbative expansion .
we could as well develop a second iam - inspired unitarization which fails to reproduce the correct large @xmath0 results and that is equivalent to bs - way1 . from this perspective
we can rearrange the unitarizations as ` large @xmath0 correct ' ( bs - way2 and iam ) and ` large @xmath0 violating ' ( bs - way1 and iam - way2 ) .
the reason why we associate to the the different unitarizations the names bs or iam is simply due to the way the equations are settled down in the different cases .
it offers a simple mnemonic to their development . by studying the large-@xmath0 limit one can see closer the relations between the two described bs - unitarizations : in the case @xmath264 and in the large-@xmath0 limit
, the @xmath193 matrix in the first bs form reads @xmath282 which obviously has no pole . in the second bs - unitarization one
has in the large-@xmath0 limit @xmath283 and the correct pole @xmath284 is recovered .
it is however important to notice that -just as in the iam case- at least two terms in the expansion of the amplitude @xmath285 are necessary to perform this second bs unitarization .
this is the reason why it can not be applied in the case studied in the next subsection ( sec .
iv.b ) , where only the lowest term of the amplitude is kept . in most studies employing the bs analysis only the lowest term of the effective low - energy lagrangian is kept . within the present toy model
it is not possible to reconstruct a resonance with mass @xmath286 with only the lowest term ( @xmath287 ) @xcite .
however , a simple modification of the model which allows for such a study is possible : @xmath288 in this way an extra - repulsion ( whose quartic form is assumed to be valid up to @xmath7 ) has been introduced .
the @xmath193 matrix takes the form : @xmath289 when deriving the low - energy lagrangian everything goes as before , but the @xmath290 term is now absent:@xmath291 note , in this case the @xmath229 scattering vanish at low momenta and in the chiral limit @xmath292 ( just as the @xmath125 scattering in reality ) . a study of the case @xmath264 ( corresponding to the first term only in the expansion ) is now possible .
we consider the following situation : let the original lagrangian @xmath293 of eq .
( [ lnew ] ) be unknown .
the low - energy potential at the lowest order reads @xmath294 @xmath295 , but the low - energy coefficient @xmath296 is also unknown .
moreover , from low energy informations only one does not know the value of the cutoff @xmath224 to be employed in mesonic loops : a new cutoff @xmath297 , not necessarily equal to the original @xmath298 is also introduced as a free parameter . from the perspective of low - energy phenomenology ,
one writes down the following approximate form for the @xmath193 matrix , which depend on two ` free parameters ' @xmath296 and @xmath297 : @xmath299 the question is if it is possible to vary @xmath296 and @xmath297 in such a way that the approximate @xmath193-matrix @xmath300 reproduces the ` full ' result @xmath253 of eq .
( [ fullt ] ) between , say , @xmath301 gev and @xmath302 gev for @xmath71 .
[ ptb ] fig2.eps the answer is that this is generally possible , but the results for @xmath296 and @xmath297 vary drastically with the coupling constant @xmath219 in the original lagrangian . in particular , if @xmath219 is small a good fit implies a very large and unnatural value of @xmath297 ( fig .
for instance , for @xmath303 gev the mass @xmath304 gev is only slightly shifted from the bare mass @xmath305 gev . in this case
the approximate form @xmath306 reproduces @xmath307 only if @xmath297 @xmath308 ( astronomically high and seemingly unnatural from the perspective of the low - energy theory ! ) .
the situation changes completely if @xmath219 is large : it is possible to find a satisfactory description in which @xmath309 .
for instance , for @xmath310 gev one has @xmath311 gev and the approximate @xmath306 reproduces well @xmath312 for @xmath313 ( fig .
3.d ) in the first case the failure of the dynamical reconstruction with a meaningful value of the cutoff @xmath297 is due to the quantitative inappropriate behavior of the bethe - salpeter approach when only the first term is kept . in the second case
a rather satisfactory description is possible for a meaningful value of the cutoff . on the light of the results of the low - energy lagrangian only
, one could also propose the interpretation that the obtained state @xmath206 is dynamically generated , and shall be regarded as a @xmath229 molecular state .
this is , however , _ not _ the correct interpretation in the present example . we know , in fact , that this state corresponds -by construction- to the original , preexisting , quarkonium - like state @xmath206 . in both cases ,
as soon as we increase the number of colors , the approximate @xmath193-matrix @xmath306 and the full @xmath307 show a completely different behavior ( fig .
3 , panels ( b , c ) and ( e , f ) ) : while the peak of @xmath314 approaches @xmath305 gev and becomes narrower according to the correct large-@xmath0 limit of the @xmath206 meson , the dynamically reconstructed state fades out , because of the _ incorrect _ behavior of bs unitarization with large-@xmath263 this is clearly visible from the interaction term @xmath315 because @xmath296scales as @xmath66 . however , although the interaction term disappears with large-@xmath316 the state @xmath206 is still the original quark - antiquark state .
this example shows that the reconstruction of the state @xmath206 is not possible in the large-@xmath0 limit , but does not mean that @xmath206 is a dynamically generate state of molecular type .
note , this is just a sub - case of the previous general discussion on large-@xmath0 dependence : the fact that only one term is kept generates a much faster ` fading out ' of the reconstructed state , compare fig . 2 and fig .
3 . in the previous subsection it was shown
that -while a straight application of the bs equation is at odds with the large-@xmath0 limit- a second bs - unitarization allows for a correct description of the large-@xmath0 limit .
the second bs - unitarization is however not applicable in the present case .
in fact , _ at least _ two terms in the expansion of @xmath250 are needed to follow it . if only the lowest term is kept , as done here with the term @xmath264 in eq .
( [ tappr2 ] ) , this is no longer feasible .
this is similar to the fact that also the iam method needs at least two terms in the expansion of the amplitude @xmath285 in order to be applicable @xcite .
the original toy - lagrangian @xmath317 of eq .
( [ ltoy ] ) is assumed to be valid up to an energy @xmath318 the corresponding low - energy lagrangian @xmath319 of eq .
( [ le ] ) -obtained by integrating out the @xmath206 field- is valid up to an energy @xmath320 when unitarizing @xmath239 , one can enlarge the validity of the low - energy theory up to @xmath321 and then infer the existence of the resonance @xmath206 with mass @xmath322 however , if no other input is known , the nature of the state @xmath206 can not be further studied , see the general discussion of the point ( d ) in sec .
this situation is similar to the example ( c ) in sec .
ii : the fermi lagrangian @xmath31 alone does not allow to deduce the nature of the @xmath29 meson , even if the existence of the latter is inferred by unitarization arguments applied to @xmath31 .
it is also similar to the cases studied in sec .
iii.c : when a resonance is obtained by unitarizing a low - energy mesonic lagrangian , ( at first ) no statement about its nature can be done .
further information is needed : in the case of the @xmath29 meson the full electroweak lagrangian @xmath34 is known and leads to the straightforward conclusion that the @xmath29 meson is not a dynamically generated state , but a fundamental field of the standard model . in the framework of the toy model
, this corresponds to the knowledge of the ` full lagrangian ' @xmath317 of eq .
( [ ltoy ] ) .
the ` quarkonium ' nature of @xmath206 can then be easily deduced . in the case of low - energy mesonic theories discussed in sec .
iii.c , the full hadronic lagrangian is not known .
the only additional knowledge is the large-@xmath0 scaling of the low - energy constants of the low - energy lagrangian(s ) . in the framework of the toy model
, this corresponds to the knowledge of low - energy lagrangian @xmath323 of eq .
( [ le ] ) ( up to a certain @xmath152 ) together with the scaling of the quantities @xmath243 in eq .
( [ lec ] ) .
the latter additional knowledge can lead to the correct conclusions about the nature of the @xmath206 meson , although -as discussed in sec .
iv.a- care is needed when the bs method is chosen . moreover , as further studied in section .
iv.b , when only _ the lowest term _ of the low - energy lagrangian is kept , it is not possible to reproduce the correct large-@xmath0 behavior of the resonance @xmath324 although the ` dynamical reconstruction ' of the state @xmath206 is possible , the state @xmath206 ` looks like ' a molecular state which fades out in the large-@xmath0 limit .
this however is not true : in fact , we know from the very beginning that the state @xmath206 corresponds ` by construction ' to a quark - antiquark state .
although this discussion is based on toy models and the real world is much more complicate than this , the same qualitative picture can hold in low - energy qcd .
in fact , the use of the bs equation in the literature is often limited to the lowest term of a low - energy lagrangians for @xmath325 @xmath326 ... interactions . in our view , such low - energy lagrangians emerge upon integrating out all the heavier fields in @xmath327 in which tensor , axial - vector and scalar quark - antiquark fields must exist below 2 gev .
then , the use of the bs equation , similarly to the dynamical reconstruction of @xmath206 in this simple example , leads to the dynamical reconstruction of the axial - vector , tensor and scalar mesons above 1 gev : they are preexisting , quark - antiquark states , which are reobtained from low - energy lagrangians via unitarization methods .
future unitarization studies , involving the leading and the next - to - leading - terms in the effective lagrangians , may shed light on this point .
in this work we studied the issue of dynamical generation both in a general context and in low - energy qcd . a dynamically generated resonance has been defined as a state which does not correspond to any of the fields of the original lagrangian describing the system up to a certain maximal energy @xmath7 , provided that its mass lies below this maximal energy .
this discussion also offered us the possibility to distinguish in principle tetraquark from mesonic molecular states in low - energy qcd : while the former are fundamental and shall be included as bare fields in the ( yet - unknown ) hadronic lagrangian @xmath87 , this is not the case for the latter .
note , @xmath74 represents the complete hadron theory valid up to @xmath60 gev . in the application to the hadronic world we also discussed dynamical reconstruction of resonances : these are resonances which are obtained via unitarization methods from low - energy effective lagrangians , but still represent fundamental fields ( such as quark - antiquark states ) in @xmath87 .
note , the low - energy effective lagrangians can be seen as the result of integrating out heavier ( quarkonia , glueballs , ... ) fields representing intrinsic , fundamental states in @xmath87 . in the scenario of dynamical reconstruction
, one reconstructs these heavier resonances by unitarizing the appropriate low - energy lagrangian . within a simple toy model
these issues have been examined .
this model consists of two fields , @xmath53 and @xmath328 with the latter being heavier and with a nonzero decay width into @xmath329 we introduced a large-@xmath0 dependence which mimics that of quarkonium states in qcd .
the field @xmath206 has been first integrated out and the emerging low - energy interaction lagrangian involving only the field @xmath53 has been derived . out of it
the state @xmath206 has been ` dynamically reconstructed ' . in order to do it
we have used a unitarization inspired by the bethe - salpeter equation and we have shown that the original , quarkonium - like large-@xmath0 behavior of @xmath206 can not be reproduced if only the lowest term in the effective lagrangian is kept .
( note , when more terms are kept this problem can be easily solved and the large @xmath0 result is correct also within the bs approach .
the problem is not the latter but the adopted perturbative expansion , see sec iv.a ) .
we then proposed that a similar , although more complicated , dynamical reconstruction mechanism takes place for tensor , axial - vector and scalar mesons above 1 gev : these resonances , studied in recent works , can be interpreted as fundamental quark - antiquark states , which are reobtained when unitarizing low - energy effective lagrangians . in this scenario
there is no conflict between the ` old ' quark model assignments and recent developments , because they would both represent a dual description of the same , preexisting quark - antiquark resonances .
this interpretation , although not yet conclusive , represents a possibility which deserves further study .
dynamical reconstruction can also hold for light scalar mesons below 1 gev , if the latter form a quarkonium ( quite improbable ) or a tetraquark nonet .
the situation in this case is , as discussed in the text , still unclear . in this work we limited the study to the light mesonic sector , but the present discussion about dynamical generation / reconstruction can also hold , with due changes , in the baryon and in the heavy quark sectors .
j. a. oller , e. oset and j. r. pelaez , phys .
* 80 * ( 1998 ) 3452 [ arxiv : hep - ph/9803242 ] . j. a. oller , e. oset and j. r. pelaez , phys .
d * 59 * ( 1999 ) 074001 [ erratum - ibid .
d * 60 * ( 1999 errat , d75,099903.2007 ) 099906 ] [ arxiv : hep - ph/9804209 ] .
a. gomez nicola and j. r. pelaez , phys .
d * 65 * ( 2002 ) 054009 [ arxiv : hep - ph/0109056 ] .
a. dobado and j. r. pelaez , phys .
d * 56 * ( 1997 ) 3057 [ arxiv : hep - ph/9604416 ] .
e. oset _ et al .
_ , arxiv:0806.0340 [ nucl - th ] . j. a. oller and e. oset , nucl . phys . a * 620 * ( 1997 ) 438 [ erratum - ibid .
a * 652 * ( 1999 ) 407 ] j. a. oller and e. oset , phys .
d * 60 * ( 1999 ) 074023 [ arxiv : hep - ph/9809337 ] .
m. f. m. lutz and e. e. kolomeitsev , nucl .
a * 730 * ( 2004 ) 392 [ arxiv : nucl - th/0307039 ] .
m. wagner and s. leupold , phys .
b * 670 * ( 2008 ) 22 [ arxiv:0708.2223 [ hep - ph ] ] .
m. wagner and s. leupold , phys .
d * 78 * ( 2008 ) 053001 [ arxiv:0801.0814 [ hep - ph ] ] .
s. leupold and m. wagner , arxiv:0807.2389 [ nucl - th ] .
e. van beveren , t. a. rijken , k. metzger , c. dullemond , g. rupp and j. e. ribeiro , z. phys .
c * 30 * ( 1986 ) 615 [ arxiv:0710.4067 [ hep - ph ] ] .
e. van beveren and g. rupp , eur .
j. c * 10 * ( 1999 ) 469 [ arxiv : hep - ph/9806246 ] .
e. van beveren and g. rupp , int .
j. theor .
group theor .
* 11 * ( 2006 ) 179 [ arxiv : hep - ph/0304105 ] .
o. krehl and j. speth , nucl .
a * 623 * ( 1997 ) 162c .
l. li , b. s. zou and g. l. li , phys .
d * 63 * ( 2001 ) 074003 [ arxiv : hep - ph/0010196 ] .
n. isgur and j. speth , phys .
* 77 * ( 1996 ) 2332 .
see , however , also m. harada , f. sannino and j. schechter , phys .
* 78 * ( 1997 ) 1603 [ arxiv : hep - ph/9609428 ] .
m. b. voloshin and l. b. okun , jetp lett . * 23 * , 333 ( 1976 ) [ pisma zh .
* 23 * , 369 ( 1976 ) ] .
k. maltman and n. isgur , phys .
lett . * 50 * , 1827 ( 1983 ) . k. maltman and n. isgur , phys .
d * 29 * , 952 ( 1984 )
. v. baru , j. haidenbauer , c. hanhart , yu . kalashnikova and a. e. kudryavtsev , phys .
b * 586 * ( 2004 ) 53 [ arxiv : hep - ph/0308129 ] .
t. branz , t. gutsche and v. e. lyubovitskij , arxiv:0712.0354 [ hep - ph ] . in both cases ( a ) and ( b ) one can write a ( nonlocal ) lagrangian in which the lightest positronium ( deuteron ) field @xmath330 @xmath331 directly couples to its constituents , the electron and the positron ( the nucleon and the proton ) .
in this case , the compositness condition should be introduced , see s. weinberg , phys .
rev . * 130 * , 776 ( 1963 ) ; a. salam , nuovo cim . *
25 * , 224 ( 1962 ) ; k. hayashi , m. hirayama , t. muta , n. seto and t. shirafuji , fortsch .
phys . * 15 * , 625 ( 1967 ) .
it ensures that the field @xmath330 @xmath331 is a composite object and thus not on the same footing as its elementary constituents .
for a recent application for the deuteron we also refer to y. b. dong , a. faessler , t. gutsche and v. e. lyubovitskij , phys .
c * 78 * ( 2008 ) 035205 [ arxiv:0806.3679 [ hep - ph ] ] , and for the positronium to f. giacosa , glueball phenomenology within a nonlocal approach ( ph.d .
thesis ) , tuebingen ( online at http://w210.ub.uni-tuebingen.de/dbt/volltexte/2005/1705/ ) .
for instance , in the positronium case the interaction lagrangian ( neglecting nonlocality ) has the form @xmath332 surely , this lagrangian is useful to study properties of the positronium state , but can describe the @xmath333-@xmath334 interaction only in a very small energy range close to @xmath335 where the positronium resonance dominates .
moreover , as soon as the photon is added to the lagrangian ( a obviously necessary step ) , one shall not include @xmath330 from the very beginning , otherwise it is double counted .
indeed , the compositness condition can be directly related to the bethe - salpeter approach applied to @xmath336 where the composite nature of the positronium is manifest .
similar arguments hold for the deuteron . the definition of sec .
ii does not allow to conclude that the resonance @xmath19 obtained in the framework of unitarized @xmath14 is dynamically generated because @xmath337 @xmath338 this is in agreement with the fact that at the level of the unitarized version of @xmath14 one can not deduce if the field @xmath19 appears explicitly or not in the more complete lagrangian @xmath41 ( i.e. one can not state if @xmath19 is dynamically generated or not ) .
note , even in the case @xmath339 the unitarized version of @xmath14 is not equivalent to @xmath48 ( therefore one should not use the definition of sec .
ii with the energy scale @xmath47 instead of @xmath7 ) . only some of the properties of @xmath41 can be inferred by the extrapolation to higher energy of the information contained in the lagrangian @xmath14 . while a _ local _ tetraquark current can be transformed into a molecular and a quarkonium currents , this does not hold when the tetraquark is regarded as a nonlocal object , where the diquarks are separated in space .
this idea can be extended : each hadron with an irreducible , nonlocal color substructure can not be dynamically generated .
most of these hypotethical hadrons do not find counterpart in the pdg because they do not bind or are too massive and broad to be seen .
the tetraquark sector below 1 gev can be an exception , due to the strong binding of the diquark in the color and flavor anitriplet channel and to the @xmath230-wave both in the diquark and in the tetraquark objects .
y. nambu and g. jona - lasinio , phys .
* 122 * ( 1961 ) 345 . for reviews : t. hatsuda and t. kunihiro , phys .
rept . * 247 * , 221 ( 1994 ) [ arxiv : hep - ph/9401310 ] .
s. p. klevansky , rev .
* 64 * ( 1992 ) 649 .
see also the recent devlopments in m. x. su , l. y. xiao and h. q. zheng , nucl .
a * 792 * ( 2007 ) 288 [ arxiv:0706.1823 [ hep - ph ] ] , m. k. volkov , m. nagy and v. l. yudichev , nuovo cim .
a * 112 * ( 1999 ) 225 [ arxiv : hep - ph/9804347 ] c. amsler and f. e. close , phys .
d * 53 * ( 1996 ) 295 [ arxiv : hep - ph/9507326 ] .
w. j. lee and d. weingarten , phys .
d * 61 * , 014015 ( 2000 ) .
[ arxiv : hep - lat/9910008 ] ; f. e. close and a. kirk , eur .
j. c * 21 * , 531 ( 2001 ) .
[ arxiv : hep - ph/0103173 ] .
f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
d * 72 * , 094006 ( 2005 ) .
[ arxiv : hep - ph/0509247 ] .
f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
b * 622 * , 277 ( 2005 ) .
[ arxiv : hep - ph/0504033 ] .
f. giacosa , t. gutsche and a. faessler , phys .
c * 71 * , 025202 ( 2005 ) [ arxiv : hep - ph/0408085 ] .
h. y. cheng , c. k. chua and k. f. liu , phys .
d * 74 * ( 2006 ) 094005 [ arxiv : hep - ph/0607206 ] .
d. parganlija , f. giacosa and d. h. rischke , aip conf .
proc . * 1030 * ( 2008 ) 160 [ arxiv:0804.3949 [ hep - ph ] ] .
d. parganlija , f. giacosa and d. h. rischke , arxiv:0812.2183 [ hep - ph ] .
see also : m. urban , m. buballa and j. wambach , nucl .
a * 697 * ( 2002 ) 338 [ arxiv : hep - ph/0102260 ] .
s. gallas , f. giacosa and d. h. rischke , arxiv:0907.5084 [ hep - ph ] .
the details of the regularization are not important , as long as momenta larger than @xmath224 are suppressed in the meson loop .
note that , although the toy model is formally renormalizable , we still work with a finite cutoff which directly affects the results .
this is namely also the situation in the true world : a cutoff of the order of 1 gev is an essential ingredient in many effective models of low - energy qcd .
the assumption that @xmath224 does not depend on @xmath0 can be justified by assuming that @xmath341 the latter being @xmath0 independent .
an alternative behavior of @xmath224 as function of @xmath0 have been discussed in ref .
@xcite , where the case @xmath342 leads to an increase of @xmath224 with @xmath343 here we note that in the context of the toy model the mass of the reconstructed state still diverges when admitting a power - like dependence @xmath344 with @xmath345 although the quantitative way how the limit is reached is strongly affected by the value of @xmath346 this is due to the fact that the lowest term has no @xmath260 dependence and to the particular form of @xmath347 , see ref .
the reconstructed mass is actually infinite .
the first term only , although inadequate in this case , could still work if the loops were so strong that @xmath348 ( even if @xmath209 ) , but we do not discuss this case here . if , instead of @xmath349 the case @xmath350 is considered , then again the second bs unitarization is possible .
in fact , being @xmath351 one has @xmath352 and @xmath353 , which is in agreement with the large-@xmath0 expectation . | a definition of ` dynamical generation ' , a hotly debated topic at present , is proposed and its implications are discussed . this definition , in turn , leads to a method allowing to distinguish in principle tetraquark and molecular states .
the different concept of ` dynamical reconstruction ' is also introduced and applies to the generation of preexisting mesons ( quark - antiquark , glueballs , ... ) via unitarization methods applied to low - energy effective lagrangians .
large @xmath0 arguments play an important role in all these investigations . a simple toy model with two scalar fields
is introduced to elucidate these concepts .
the large @xmath0 behavior of the parameters is chosen in order that the two scalar fields behave as quark - antiquark mesons . when the heavier field is integrated out , one is left with an effective lagrangian with the lighter field only . a unitarization method applied to
the latter allows to ` reconstruct ' the heavier ` quarkonium - like ' field , which was previously integrated out .
it is shown that a bethe - salpeter ( bs ) analysis is capable to reproduce the preformed quark - antiquark state , and that the corresponding large-@xmath0 behavior can be brought in agreement with the expected large @xmath0 limit ; this is a subtle and interesting issue on its own . however , when only the lowest term of the effective lagrangian is retained , the large @xmath0 limit of the reconstructed state is not reproduced : instead of the correct large @xmath0 quarkonium limit , it fades out as a molecular state would do .
implications of these results are presented : it is proposed that axial - vector , tensor and ( some ) scalar mesons just above 1 gev obtained via the bs approach from the corresponding low - energy , effective lagrangian in which only the lowest term is kept , are quarkonia states , in agreement with the constituent quark model , although they might fade away as molecular states in the large @xmath0 limit . |
early - type galaxies ( etgs hereafter ) are the dominant population in galaxy clusters .
their stellar population is thought to be homogeneously old and passively evolving as shown in the tight color - magnitude ( cm ) relation ( e.g. , bower et al .
1992 ; kodama & arimoto 1997 ) .
this picture of etgs consisting of homogeneous stellar population breaks down when we examine their properties at different wavelengths .
it is known from earlier infrared observations that some etgs exhibit excess far - infrared emission ( knapp et al .
1989 ) , and mid - infrared emission ( knapp et al . 1992 ; xilouris et al .
the observation in uv also confirms a diversity of etgs ( smail et al . 1998 ; yi et al . 2005 ; schawinski et al .
a particular interest can be placed on the mid - infrared ( mir ) emission .
recently , clemens et al .
( 2008 ) found that about 32% of etgs in the coma cluster have excess flux over photospheric emission at mir .
this mir - excess can be due to various mechanisms , such as emission from circumstellar dust around agb stars ( knapp et al .
1992 ) , star formation , and agn activities ( quillen et al .
recent spitzer observations indicate that the origin of the mir - excess is the dusty agb stars for the majority of etgs in the red sequence ( bressan et al .
according to the model where the mir - excess is caused by the dust emission from agb stars , the mir - excess is prominent if the luminosity weighted mean age of the stellar populations is young ( piovan et al .
2003 , hereafter p03 ) .
etgs with mir - excess could be used as a tracer of morphological transformation in the cluster environment .
for example , star - forming spiral galaxies infalling to a cluster have been suspected that they would be morphologically transformed into quiescent s0s through various physical processes ( e.g. , dressler et al .
1997 ; goto et al . 2003 ; boselli & gavazzi 2006 ; park & hwang 2008 ) leaving relatively young etgs showing mir - excess .
however , such an analysis has limited due to the paucity of complete mir ( 7 24 @xmath1 m ) imaging observation covering a large field around a cluster of galaxies . in this paper
, we examine the relation between the cluster environment and the mir - excess of etgs in a2218 , a galaxy cluster at @xmath7 ( e.g. , biviano et al .
2004 ) , using _
_ space telescope ( murakami et al . 2007 ) that offers a wide - field imaging of the galaxy cluster at 11 and 15 @xmath1 m for the first time .
a2218 is a rich cluster ( richness class = 4 ) with a virial radius of ( 1.5 2.3 ) mpc and a mass of ( 4.8 21)@xmath8 ( girardi & mezzetti 2001 ; pratt et al .
2005 ) .
a2218 has been observed by the infrared space observatory ( iso ) , and the previous iso study revealed the lack of mir - excess early - type galaxies at the central region of a2218 ( biviano et al .
2004 ) , while the analysis of optical / nir imaging and spectroscopy of this cluster have shown a wide dispersion in the stellar ages of faint etgs ( smail et al .
2001 ; ziegler et al . 2001 ; sanchez et al . 2007 ) .
the two results seem contradictory , and a wide - field ir imaging could offer a solution . throughout this paper , we use _
@xmath9 _ = 70 km s@xmath10mpc@xmath10 , _ @xmath11 _ = 0.3 and _ @xmath12 _ = 0.7 ( spergel et al . 2003 ) .
in this cosmology , the angular scale of 1 arcsec at the distance of this galaxy cluster corresponds to 2.97 kpc .
all magnitudes are given in the ab system .
the rich cluster a2218 is one of the targets selected for studying low redshift clusters by the _ clusters of galaxies evolution studies _ ( clevl , pi : h. m. lee ; i m et al .
2008 ) , an _ akari _ mission program aiming to understand the formation and evolution of galaxies in cluster environments .
the _ akari _
observation for the a2218 field was carried out using 6 broadband filters ( _ nir : n3 , n4 , and mir : s7 , s11 , l15 , l24 _ ) of the infrared camera ( irc ; onaka et al .
2007 ) onboard the _
akari_. we covered a 15@xmath13 arcmin@xmath14 field centered on the x - ray brightness peak of a2218 , using 4 irc tiles .
the image depth differs according to the locations , and the central 5@xmath15 arcmin@xmath14 region is the deepest .
table 1 lists the total exposure time and the 5@xmath16 point - source detection limit over 11@xmath17 diameter aperture in this region .
the flux limit at the outer region is @xmath18 or 2 times shallower than the central region .
the raw images were processed and stacked with the irc imaging data reduction pipeline version 070104 and then co - added into one final mosaic image using swarp ] .
cosmic rays were rejected during the stacking .
additional removal of cosmic rays was carried out using l.a.cosmic ( van dokkum 2001 ) when necessary .
we checked if this procedure causes flux loss using the north ecliptic pole survey data of _ akari _ ( matsuhara et al .
2006 ; wada et al . 2008
; lee et al .
2009 ) , and found no significant flux loss .
we used sextractor ( bertin & arnouts 1996 ) to detect sources .
the measured adus were converted to the jy unit using the conversion factors from the irc manual version 1.4 . when doing the flux calibration for extended sources , we additionally considered color correction since the irc flux calibration assumed @xmath19 .
we calculated the correction factor following the method provided in the irc manual with the model seds of p03 .
we find that the color correction can affect the flux value at @xmath20 level , but @xmath21 flux change occurs in extreme cases . since it is difficult to know the sed shape a priori , we leave the color correction as unknown systematic error .
we used the mag - auto to estimate the total magnitude . to check the mag - auto as the total magnitude for all band
, we estimated large aperture photometry of several isolated galaxies in the final image , and found that the difference between mag - auto and mag - aper was within typical measurement errors ( nir : @xmath22 3% , mir : @xmath22 20% ) .
however , especially in the nir band , flux - auto of sources with close neighbors are affected by nearby sources . to derive fluxes of such objects , we used small circular aperture of 5.@xmath235 diameter and applied aperture corrections that are derived from the growth curve of isolated galaxies with similar flux . for the central 10@xmath24 region , we also used a deeper l15 image taken for another _ akari _ program ( pi : serjeant ) to derive 15 @xmath1 m fluxes .
the 5@xmath16 detection limit of the deep l15 image is 20.1 mag , and the details on this data will be presented elsewhere ( hopwood et al . in preparation ) .
additionally , we used optical fluxes ( @xmath25 ) from ziegler et al .
( 2001 , hereafter z01 ) , and 24/70 @xmath1 m data taken by the @xmath26/mips ( gto program @xmath2783 , pi : g. rieke ) in the archive for the overlapping region with _
akari _ where the mips 24 @xmath1 m image is deeper than the _ akari _ image . for the sample selection , we started with 48 etgs of a2218 of the central [email protected] arcmin@xmath14 in z01 and additional cluster members whose redshifts were obtained from the nasa / ipac extragalactic database ( ned ) search of the central 15@xmath2815 arcmin@xmath14 area of a2218 ( z=0.165 0.185 ) .
the morphology of 9 sources added from the ned search is determined to be etgs by the visual inspection of the archival optical images ( cfht and subaru ) and sed shapes .
these galaxies were matched with the _ akari _ source catalog , and we checked each member whether it is blended or not in the _ akari _ image by comparing with the optical images . through this process , we excluded galaxies that are blended with neighbors in the _ akari _ image . finally , we collected 39 ( 30 from z01 ) non - blended etgs from the _ akari _ data . among these
we have selected a sample of 22 galaxies with @xmath29 mag for mir - excess study .
in figure 1 , we show the cm diagram of a2218 galaxies in three colors ; @xmath30 ( a ) , @xmath31 ( b ) , and @xmath4 ( c ) .
the @xmath5 flux is chosen to be on the x - axis , as a rough measure of the stellar mass ( e.g. gavazzi et al . 1996 ) .
the @xmath30 versus @xmath5 cm diagram is shown for galaxies with @xmath29 mag .
it shows a usual tight red sequence of etgs , with two galaxies ( @xmath32 and @xmath33 in z01 ) slightly bluer than the others .
this reflects the fact that z01 sample is made of galaxies belonging to the red sequence or close to it .
figure 1b shows the @xmath31 versus @xmath5 cm relation .
the plot shows a tight blue sequence since the flux decreases as the wavelength increases at @xmath34 @xmath1 m .
next , we have examined the mir ( @xmath35 ) properties of galaxies selected using a box shown in figure 1b , i.e. , galaxies with @xmath29 mag .
this magnitude cut is imposed due to the relatively shallower @xmath35 detection limit , so that we can construct an unbiased sample in @xmath4 colors .
figure 1c shows the @xmath4 versus @xmath5 cm relation of the cluster members with @xmath29 mag .
also plotted are the cm relations from various model seds ( see below ) . at the bright end ,
four galaxies tend to form a blue sequence indicating that they are a homogeneous population .
however , figure 1c reveals a stunningly wide dispersion in the @xmath4 colors of the optical red sequence galaxies with @xmath36 mag .
more than @xmath650% show mir - excess at @xmath36 mag above the model line considering photospheric emission only ( the `` old noagb '' line ) .
this is in clear contrast to the result of biviano et al .
( 2004 ) where a small fraction of etgs showed this kind of mir - excess . to understand what is causing the mir - excess , we compared the observed cm relation with the model predictions . for the model seds
, we used single stellar population ( ssp ) models with a spontaneous burst , the salpeter ( 1955 ) imf , and the padova library of stellar spectra from p03 which also incorporates the dust emission from circumstellar dust around agb stars .
as it has been recognized , the optical cm relation can be described well with a single age model assuming a metallicity gradient ( the dashed line in fig .
1a ; e.g. , kodama et al .
the same model fits the @xmath4 versus @xmath5 cm relation at the bright magnitude ( the solid line in fig .
1c ) , but it fails to reproduce the dispersion in the @xmath4 colors of the fainter etgs .
the dispersion requires the existence of stellar populations with younger ages , or some other mechanisms . in the top panel of figure 2
, we examine the mir - excess of etgs in more detail by plotting @xmath4 versus @xmath31 color - color diagram .
the thick solid line indicates the track of a ssp as described in the dashed line of figure 1a but with several different ages ( 1@xmath615 gyrs ) .
the other lines indicate the tracks of model seds with the agb circumstellar dust of p03 .
we define mir - excess galaxies to be those lying at the redward of the dotted line .
this corresponds to the mir - excess more than typical 1@xmath16 error of ( @xmath4 ) color ( @xmath37 mag ) over the thick solid line . among galaxies in our magnitude limited sample , more than 41% ( 9/22 ) can be classified as galaxies having mir - excess . in the lower panels of figure 2 , mir - excess galaxies have relatively low velocity dispersion .
although considering our flux - limited sample , this result suggests that less massive etgs are more likely to have mir - excess than massive ones . however , mir - excess does not show strong correlation with h@xmath38 line index which is sensitive to recent star formation activity .
it is very possible that h@xmath38 line and mir - excess trace different epoch in star formation history , but it seems difficult to draw a firm conclusion on this based on the current data alone .
figure 3 shows the observed optical - mir seds of nine @xmath35-detected galaxies .
we performed the least square fit of seds using three different models
spectral synthesis models with or without agb dust from p03 ( @xmath39 or @xmath40 model ) , and templates of star - forming galaxies from chary & elbaz ( 2001 ) .
the best - fit seds from each model are overplotted in figure 3 .
the ssp models of p03 have three different metallicities ( @xmath41 ) and a wide range of ages ( 0.1@xmath615 gyrs ) . in case of the fit to the sf model
, we do not use the optical data for the fit , since the empirical sf sed templates do not include the diverse optical seds with different ages , metallicities , and star formation histories .
based on the significance of mir - excess , we categorize the sample into three sub - groups ; strong- , mid- , and weak / no mir - excess galaxies .
first , we note that the p03 @xmath40 model fails to fit the strong / mid mir - excess galaxies , but provides reasonable fits to the weak / no ones .
next , we find that most etgs with mir - excess returned acceptable sed fits to the p03 @xmath39 model with either a solar or 40% solar metallicities . the lowest metallicity p03 model ( 20% solar )
can fit the mir - excess if we consider the ir portion of the seds alone , but such a model fails to fit the optical portion of the seds .
the results suggest that the mir - excess correlates with ages more strongly than metallicities , with a distinct trend that the younger galaxies have the stronger mir - excess ( temi et al .
for the mir - excess etgs , the derived mean stellar ages span a wide range of 2 to 11 gyr , which qualitatively agrees with the findings by smail et al .
( 2001 ) and biviano et al .
this suggests that the mir - excess is a good indicator of mean stellar ages of etgs , as also noted by temi et al .
( 2005 ) .
the agb dust emission is not the only possible cause for the mir - excess .
we caution that the mir - excess may arise from the star - forming activities and agns . in some cases , it is difficult to exclude such possibilities from the current data alone .
deeper 70 @xmath1 m data could offer an effective way to test these possibilities .
however , even if the mir - excess arises from the star - forming activity , the derived sfrs from the sed fits suggest sfr @xmath42 yr @xmath10 for all of the mir - excess galaxies .
the only exception is @xmath43 in z01 which shows an abnormally high 15 @xmath1 m flux .
it is difficult to explain such a flux with the p03 @xmath39 model .
this may be a piece of evidence for existence of pah emissions from star formation for this galaxy .
the morphology of this galaxy seems to show a disk .
the objects @xmath44 , and @xmath33 have spiky 11 @xmath1 m fluxes , which may be due to unusual pah features at 11.3 and 12.7 @xmath1 m found in some etgs ( e.g. kaneda et al . 2005 ) .
we conclude that more than half of cluster etgs with @xmath45 18 mag are likely to have the mir - excess over stellar photospheric emission , which we interpret as the emission from circumstellar dust around agb stars .
our result is different from the isocam study of biviano et al .
( 2004 ) where they did not find such mir - excess for most of etgs in a2218 .
this can be attributed partly to the deeper depth of our data , but mostly to our much wider area coverage . to investigate the mir properties of a2218 etgs as a function of the cluster environment
, we plot the spatial and velocity distributions of 39 _ akari _ cluster etgs with the contour maps of the @xmath46 x - ray emission in figure 4 .
mir - excess etgs seem not to have a preference for the velocity distribution , but the clustercentric - distance distribution of them is exceptional .
we expect to find younger , mir - excess galaxies in the outskirts of the cluster if such galaxies are formed via transformation from field spirals accreted to the cluster environment . to check this ,
we divide our samples into those closer to the center and those in the outer region depending on the cluster - centric distance ( at 381 kpc ) in the same manner of z01 .
we find that about 64% ( 7/11 ) of the outer galaxies have mir - excess , while only 18% ( 2/11 ) of the inner ones do so .
these mir - excess galaxies are also relatively faint ones in our sample .
we remark that there are etgs in the inner region as faint as the mir - excess etgs in the outer region , but they do not show clear mir - excess . from these results ,
we conclude that these mir - excess galaxies are fainter , and preferentially located in the outer region of the cluster , suggesting that they could be the descendants of the infalling field galaxies transformed into etgs .
our conclusion appear to be reconciled with the result of z01 where they found larger spread of h@xmath38 values of a2218 cluster etgs in the outskirts .
we are carrying out similar studies using 6 more nearby clusters to obtain a more conclusive answer .
we thank l. piovan for providing his sed model .
we also thank the referee , bodo ziegler , for his constructive comments which improved our paper .
this work is based on observations with _
akari _ , a jaxa project with the participation of esa .
we acknowledge the creative research initiatives program , ceou of mest / kosef .
rhh , sbgs , and irs also acknowledge support from stfc .
bertin , e. & arnouts , s. 1996 , , 117 , 393 biviano , a. et al .
2004 , , 425 , 33 boselli , a. , & gavazzi , g. 2006 , , 118 , 517 bower , r. g. , lucey , j. r. & ellis , r. s. 1992 , , 254 , 601 bressan , a. et al .
2006 , , 639 , l55 chary , r. & elbaz , d. 2001 , , 556 , 562 clemens , m. s. et al .
2008 , , in press ( astro - ph/0808.2899 ) dressler , a. et al .
1997 , , 490 , 577 gavazzi , g. , pierini , d. & boselli , a. 1996 , , 312 , 397 girardi , m. & mezzetti , m. 2001 , , 548 , 79 goto , t. et al .
2003 , , 346 , 601 i m , m. et al .
2008 , aspc , 399 , 382 knapp , g. r. et al .
1989 , , 70 , 329 knapp , g. r. , gunn , j. e. & wynn - williams , c. g. 1992 , , 399 , 76 kodama , t. & arimoto , n. 1997 , , 320 , 41 kaneda , h. , onaka , t. & sakon , i. 2005 , , 632 , l83 lee , h. m. et al .
2009 , , in press matsuhara , h. et al .
2006 , , 58 , 673 maughan , b. j. et al .
2008 , , 174 , 117 murakami , h. et al .
2007 , , 59 , s369 onaka , t. et al .
2007 , , 59 , s401 park , c. , & hwang , h. s. 2008 , , submitted ( astro - ph/0812.2088 ) piovan , l. , tantalo , r. & chiosi , c. 2003 , , 408 , 559 pratt , g. w. , bhringer , h. & finoguenov , a. 2005 , , 433 , 777 quillen , a. c. et al .
1999 , , 518 , 632 salpeter , e. e. 1995 , , 121 , 161 sanchez , s. f. et al .
2007 , , 376 , 125 schawinski , k. et al .
2007 , , 173 , 512 spergel , d. n. et al .
2003 , , 148 , 175 smail , i. et al .
2001 , , 323 , 839 smail , i. et al .
1998 , , 293 , 124 temi , p. , mathews , w. g. & brighenti , f. 2005 , , 622 , 235 van dokkum , p. g. 2001 , , 113 , 1420 wada , t. et al .
2008 , , 60 , s517 xilouris , e. m. et al .
2004 , , 416 , 41 yi , s. k. et al .
2005 , , 619 , l111 ziegler , b. l. et al .
2001 , , 325 , 1571 | we present the _ akari _ infrared camera ( irc ) imaging observation of early - type galaxies in a2218 at z @xmath0 0.175 .
mid - infrared ( mir ) emission from early - type galaxies traces circumstellar dust emission from agb stars or / and residual star formation . including the unique imaging capability at 11 and 15 @xmath1 m
, our _ akari _ data provide an effective way to investigate mir properties of early - type galaxies in the cluster environment . among our flux - limited sample of 22 red sequence early - type galaxies with precise dynamical and line strength measurements ( @xmath2 18 mag at 3 @xmath3 )
, we find that at least 41% have mir - excess emission .
the @xmath4 versus @xmath5 ( 3 and 11 @xmath1 m ) color - magnitude relation shows the expected blue sequence , but the mir - excess galaxies add a red wing to the relation especially at the fainter end .
a sed analysis reveals that the dust emission from agb stars is the most likely cause for the mir - excess , with low level of star formation being the next possible explanation .
the mir - excess galaxies show a wide spread of @xmath4 colors , implying a significant spread ( 211 gyr ) in the estimated mean ages of stellar populations .
we study the environmental dependence of mir - excess early - type galaxies over an area out to a half virial radius ( @xmath61 mpc ) .
we find that the mir - excess early - type galaxies are preferentially located in the outer region . from these evidences
, we suggest that the fainter , mir - excess early - type galaxies have just joined the red sequence , possibly due to the infall and subsequent morphological / spectral transformation induced by the cluster environment . |
the electron - photon collision mode of an @xmath1 linear collider @xcite provides us with the possibility to produce single selectrons in association with the lightest supersymmetric particle ( lsp ) , which is assumed to be the lightest neutralino @xmath2 .
thus selectrons can be produced with masses beyond the kinematical range for pair production at an @xmath1 linear collider .
also the production mechanism ( electron exchange in the s - channel and selectron exchange in the t - channel ) for associated selectron - neutralino production is simpler than that for selectron pair production in @xmath1 collisions . assuming a common scalar mass @xmath3 at the unification point the masses of the selectrons are related to the mssm parameters @xmath4 and @xmath5 , the @xmath6 gaugino mass parameter , by renormalization group equations @xcite . in the mssm
quite generally the right selectron @xmath7 is lighter than the left selectron @xmath8 . in extended susy models ,
however , @xmath7 could be heavier than @xmath8 @xcite .
we study in this paper the process @xmath9 with polarized beams .
since the cross section and the forward - backward asymmetry of the decay electron depend sensitively on the selectron masses this process is suitable for testing the renormalization group relation between @xmath10 and @xmath11 .
the associated production of selectrons and the lsp proceeds via @xmath12 exchange in the s - channel and @xmath13 exchange in the t - channel . in the narrow width approximation the total cross section @xmath14 for the combined process of @xmath15 production and the subsequent decay @xmath16 factorizes into the production cross section @xmath17 and the leptonic branching ratio : @xmath18 the measured cross section @xmath19 in the @xmath1 cms is obtained by folding @xmath14 with the energy spectrum @xmath20 of the compton backscattered laser beam taking into account the mean helicity of the photon beam @xcite : @xmath21 @xmath22 @xmath23 is the circular photon asymmetry and @xmath24 is the ratio of the photon energy and the energy of the converted electron beam .
the energy spectrum @xmath25 and the mean helicity @xmath26 of the high energy photon beam sensitively depend on the polarizations @xmath27 of the laser beam and @xmath28 of the converted electron beam . beyond the cross section @xmath29 , we study the forward - backward asymmetry of the decay electrons : @xmath30 the forward direction is defined by the electron beam .
apart from the kinematics the selectron masses enter the cross sections and the forward - backward asymmetries explicitly via the selectron propagator in the t - channel . assuming a common scalar mass @xmath3 at the unification point the masses of the selectrons are related to the mssm parameters @xmath4 and the gaugino mass parameter @xmath5 by renormalization group equations @xcite : @xmath31 @xmath32 in the mssm quite generally the right selectron @xmath7 is lighter than the left selectron @xmath8 . in extended susy models these relations
are changed as a consequence of additional d - terms in the scalar potential and the right selectron may be heavier than the left selectron . in chapter 3 we study the dependence of the cross section @xmath33 and the forward - backward asymmetry @xmath34 on @xmath10 and @xmath11 .
we shall see that this process is useful for testing the gut - relations equs .
( 5 ) , ( 6 ) .
we present numerical results for the mssm parameters @xmath35 gev , @xmath36 gev , @xmath37 gev and @xmath38 for the cms - energy @xmath39 gev .
the lsp is gaugino - like with @xmath40 gev . for @xmath41 gev and @xmath42 gev
this corresponds to one ecfa / desy reference scenario for the linear collider @xcite .
1 shows the total cross section @xmath33 and the forward - backward asymmetry @xmath34 for @xmath43 and @xmath44 .
this choice of @xmath28 and @xmath27 leads to a strongly marked high energetic peak in the energy spectrum @xmath20 @xcite and therefore to maximal cross sections . for the electron beam in the @xmath45 collision we choose in fig .
1a , b the polarization @xmath46 . then due to @xmath47 the cross section for @xmath7
is enhanced whereas that for @xmath8 is strongly suppressed so that @xmath33 is nearly independent of @xmath11 . the cross section for this polarization configuration ( fig .
1a ) allows to constrain @xmath10 up to 344 gev . in a region around 200 gev the dependence of @xmath33 on @xmath10 is rather weak . in this case
@xmath34 ( fig .
1b ) gives additional informations on the mass @xmath10 .
for @xmath48 gev the production of right selectrons becomes impossible and due to the suppression by the polarization factor @xmath49 the cross section is rather small : @xmath50 fb for @xmath51 gev .
then @xmath34 only depends on @xmath52 and is independent of @xmath10 ( see fig .
1b ) . for figs .
1c and 1d we choose @xmath53 .
now the production and decay of left selectrons is no longer neglectible .
1c gives the cross sections for three different masses @xmath54 gev , 127 gev , 200 gev .
in all three cases it should be possible to constrain @xmath11 up to 170 gev .
for masses larger than 170 gev the dependence of @xmath33 on @xmath11 is too weak so that one obtains only a lower limit on @xmath11 . for large values of @xmath10 the measurement of the asymmetry @xmath34 ( fig .
1d ) could be helpful to extend this mass range to somewhat higher values .
[ f1lrhoch4 ] ( 12.6,9.2 ) ( -0.9,2.2 ) ( 5.2,5.2)@xmath10/gev ( 0.0,9.0)@xmath33/pb ( 2.5,5.1)(a ) ( 5.5,2.2 ) ( 11.4,7.9)@xmath10/gev ( 6.5,9.0)@xmath34 ( 7.19,5.8 ) ( 9.2,5.1)(b ) ( -0.9,-2.6 ) ( 5.2,0.3)@xmath11/gev ( 0.0,4.2)@xmath33/pb ( 1.62,1.33 ) ( 2.3,0.1)(c ) ( 5.5,-2.6 ) ( 11.4,4.1)@xmath11/gev ( 6.5,4.2)@xmath34 ( 8.19,2.12 ) ( 9.2,0.1)(d ) if the renormalization group relations equs .
( 5 ) , ( 6 ) are satisfied then @xmath11 is larger than @xmath10 , @xmath55
. this relation can be tested with the total cross sections in figure 1a and 1c up to @xmath56 gev , complementary to the measurements at an @xmath1-collider .
1d shows that for higher masses of @xmath7 the asymmetry could allow a test of the renormalization group relations , equs .
( 5),(6 ) .
with a suitable choice of beam polarizations it is possible to constrain @xmath10 up to 344 gev and @xmath11 up to 170 gev in the process @xmath57 from a measurement of the total cross sections .
the forward - backward asymmetry @xmath34 gives additional information on these masses . especially one could measure masses @xmath58 gev if @xmath10 is high enough . the cross sections and the forward - backward asymmetries allow to test the equations for the selectron masses complementary to an @xmath1-collider .
we are grateful to gudrid moortgat - pick and stefan hesselbach for valuable discussions .
claus blchinger thanks the organizers of the xxiii .
school of theoretical physics for the friendly atmosphere during the conference .
this work was supported by the deutsche forschungsgemeinschaft under contract no .
fr 1064/4 - 1 , the bundesministerium fr bildung und forschung ( bmbf ) under contract number 05 ht9wwa 9 and the stiftung fr deutsch - polnisch zusammenarbeit , warszawa .
99 i. ginzburg , g. kotkin , v. serbo , v. telnov , _ nucl .
. meth . _ 205 ( 1983 ) 47 .
+ v. telnov , _ proceedings of the first arctic workshop on future physics and accelerators _ , saariselka , 1994 , eds .
m. chaichian , k. huitu , r. orava . l. j. hall , j. polchinski , _ phys .
_ 152b ( 1985 ) 335 .
m. drees , _ nucl
_ b298 ( 1988 ) 333 .
cheng , l. j. hall , _ phys .
_ d51 ( 1995 ) 5289 .
+ c. kolda , s. p. martin , _ phys .
_ d53 ( 1996 ) 3871 .
+ e. keith , e. ma , b. mukhopadhyaya , _ phys .
_ d55 ( 1997 ) 3111 .
+ h. baer , m. a. diaz , j. ferrandis , x. tata , hep - ph/9907211 .
+ s. hesselbach , ph.d .
thesis , university of wrzburg , 1999 .
i. f. ginzburg , g. l. kotkin , s. l. panfil , v. g. serbo , v. i. telnov , _ nucl .
methods phys .
_ a219 ( 1984 ) 5 .
+ d. l. borden , d. a. bauer , d. o. caldwell , _ phys .
_ d48 ( 1993 ) 4018 .
+ d. l. borden , d. a. bauer , d. o. caldwell , slac - pub-5715 , 1992 , ucsb - hep-92 - 01 , 1992 .
+ f. cuypers , g. j. van oldenborgh , r. rckl , _ nucl .
_ b383 ( 1992 ) 45 .
+ f. cuypers , g. j. van oldenborgh , r. rckl , in _
@xmath1 collisions at 500 gev : the physics potential , part b _ , proceedings of the workshop , munich , annecy , hamburg , germany , 1993 , edited by p. m. zerwas ( desy report no . 93 - 123c , hamburg,1993 ) , p. 475 .
s. ambrosanio , g. a. blair , p. zerwas , ecfa - desy lc - workshop , 1998 , http://www.desy.de/conferences/ecfa-desy-lc98.html . | with the process @xmath0 it is possible to constrain selectron masses above the kinematical limit of the pair production process in @xmath1 colliders .
we investigate these mass ranges and discuss the possibility to test the renormalization group equations for the selectron masses . |
the surface electroclinic effect ( sece ) is an experimentally observed phenomenon in the chiral smectic - a ( sm-@xmath0 ) phase whereby a coupling between molecular dipoles and a surface induces local tilt of the director @xmath2 , away from the smectic layer normal @xmath3 at the surface , as shown in fig .
[ seceschematic ] .
the sece has been analyzed , both experimentally and theoretically , for materials with a continuous sm-@xmath0
sm-@xmath1 transition .
the investigation of w415 was notable in that an unusually large sece was found and the effects of varying the enantiomeric excess were explored .
the sece has not however , been analyzed for materials with first order sm-@xmath0 sm-@xmath1 transitions @xcite .
one motivation for the investigation of the sece in materials with first order sm-@xmath0 sm-@xmath1 transitions is that the strong bulk electroclinic effect ( bece ) of some de vries materials can be attributed to the first order nature of the sm-@xmath0
sm-@xmath1 transition @xcite .
the strong bece with its fast analog electrooptic characteristics makes such materials technologically promising .
however , as will be discussed below , the first order nature of the transition may also lead to a strong sece , which has implications for the design of proper surface alignment layers and possibly also for the long - term stability of the alignment in , for instance , ferroelectric liquid crystal display devices based on de vries materials . experimental and theoretical work on the bece in materials with first order sm-@xmath0 sm-@xmath1 transitions demonstrated the existence of a critical point ( in field ( @xmath4 ) temperature ( @xmath5 ) parameter space ) which terminates a line of first order sm-@xmath1 sm-@xmath1 transitions . for the sece
the analogous parameters would be surface coupling and temperature , although it is not immediately obvious how one can explore such a parameter space experimentally . in sample cells in which one of the two polymer coated glass plates is rubbed to align the director @xmath2 at the surface ( as shown in fig .
[ seceschematic ] ) the sece makes the smectic layer normal @xmath3 deviate by an angle @xmath6 from the rubbing direction , i.e. from @xmath7 .
this angle @xmath6 can be measured using polarized light microscopy .
however , surface pinning prevents the once - formed layer structure from rotating , which means that the surface tilt angle @xmath6 is effectively stuck at the temperature , @xmath8 , where the layers first form in the sm-@xmath0 phase .
thus , for such sample cells one can not explore the variation of the sece with @xmath5 .
we note that @xmath8 could perhaps be varied by quenching the system into the sm-@xmath0 phase at lower temperatures .
another approach would be to establish an alignment direction for @xmath3 without doing so for @xmath2 , i.e. align the smectic layers in a homogeneous bookshelf structure without having a rubbing direction at the surfaces .
this would then allow @xmath2 to rotate , and thus @xmath6 to change , at the surface as temperature is varied .
one way to do this is to shear - align the sample , i.e. to slide the top substrate with respect to the lower substrate under an applied ac field , as was done with the first experiments on surface - stabilized ferroelectric liquid crystal cells @xcite . , away from the smectic layer normal @xmath3 at the surface .
conversely , on rubbed surfaces the sece makes the bulk smectic normal @xmath3 form at an angle to the rubbing direction @xmath7 . ]
the other option is to vary the surface coupling , which increases monotonically with enantiomeric excess @xmath9 .
this should be done by preparing separate cells with varying degrees of @xmath9 .
it could in principle also be achieved via a cell made up of a contact preparation between opposite - handed moleclules of the same compound .
keeping the cell at elevated temperatures ( i.e. , in the isotropic phase ) after filling will facilitate diffusion over the contact boundary until a gradient in @xmath9 develops over a suitable distance in the cell plane . on one side of such a cell @xmath10 and as one moves to the center of the cell ( where there is an equal number of left and right handed molecules , resulting in a racemic mixture ) @xmath11 .
the second half of the cell is the mirror image of the first half , with @xmath12 .
after a long enough time continued diffusion will make the gradient in @xmath9 gradually disappear , but immediately after the first cooling down from the isotropic phase the surface coupling varies continuously across the cell in the @xmath13 direction , where @xmath13 is perpendicular to both @xmath3 and @xmath14 .
provided the gradient in @xmath15 is known we could directly probe the @xmath16 for all values of @xmath9 . in such an experiment , care should be taken to avoid any influence on the structure from the anchoring on the opposing surface . moreover , with time the gradient in @xmath17 on the surface might differ from the one in the bulk , due to surface pinning effects . as part of the present analysis we obtain a theoretical state map , as shown in fig .
[ statemap](a ) , in enantiomeric excess - temperature space which features a critical point that terminates a line of first order discontinuities in the surface induced tilt .
this critical point ( @xmath18 , @xmath19 ) is analogous to that found for the phase diagram ( in electric field - @xmath5 space ) for the bece .
we will demonstrate that for sufficiently low temperature @xmath20 , there will be a jump in surface induced tilt as @xmath9 is varied , as shown in fig .
[ statemap](b ) .
this jump is analogous to the field - induced first order sm-@xmath1
sm-@xmath1 transition for the bece . as with the bece
, the sece is enhanced due to the first order nature of the zero - field sm-@xmath0 sm-@xmath1 transition @xcite .
however , we will also argue that the sece in de vries type materials is further enhanced due to the minimal layer spacing change at the sm-@xmath0
sm-@xmath1 transition .
recent work @xcite , indicates that first order sm-@xmath0 sm-@xmath1 transitions with minimal layer chance are hallmarks of de vries - type materials .
thus , an enhanced sece effect in a material ( such as that described here ) could be a signature of its de vries - like nature .
we also investigate how the surface induced tilt decays as one moves from the surface into the bulk . for materials with continuous sm-@xmath0
sm-@xmath1 transitions , such a decay is gentle , meaning that the slope of the decay decreases the further one moves into the bulk .
our analysis shows that for a system with a first order sm-@xmath0
sm-@xmath1 transition this will not be the case and that there will be a kink in the decay .
the kink would be a region in which the the slope becomes steeper and the decay becomes much more rapid .
this is shown in fig .
[ statemap](c ) .
in this article we consider a cell with fixed , uniform enantiomeric excess @xmath10 . we will present analysis for a spatially varying @xmath9 ( as in the contact preparation discussed above ) in a future publication .
we consider a standard , simplified , geometry ( shown in fig . [ seceschematic ] ) with the liquid crystal in contact with a single surface at @xmath21 and extending to @xmath22 .
this is valid if the two cell surfaces are separated by a distance @xmath23 , where @xmath24 is a correlation length to be defined below . with this geometry
the layer normal @xmath3 will tilt away from the rubbing direction @xmath7 , within the @xmath25 plane , by an amount @xmath26 .
this tilt is biggest at the surface , i.e. at @xmath21 , and decays to zero with increasing @xmath27 , i.e. @xmath28 . to analyze the sece in the sm-@xmath0 phase we employ a landau expansion of the bulk and surface free energies , @xmath29 and @xmath30 respectively : @xmath31\ ; \label{bulk free energy}\end{aligned}\ ] ] and @xmath32 where @xmath33 is the area of the surface , @xmath34 is the component of the average polarization perpendicular to the surface and @xmath35 is a @xmath36-@xmath37 coupling constant .
the strength of the @xmath36-@xmath37 coupling is proportional to @xmath9 .
the coefficient @xmath38 is a generalized d.c
. electric susceptibility and the coefficient @xmath39 is proportional to the strength of the polar surface anchoring . in keeping only terms of order @xmath40
we make the standard assumption that the sm-@xmath0 - sm-@xmath1 transition is primarily driven by @xmath36 , with @xmath37 playing a secondary role . the piece @xmath41 is given by : @xmath42 where @xmath43 and @xmath44 . for @xmath45 the racemic ,
i.e. @xmath46 , bulk sm-@xmath47 - sm-@xmath48 transition is continuous and takes place at @xmath49 while for @xmath50 the transition is first order and takes place at @xmath51 .
we consider a racemic bulk sm-@xmath47 - sm-@xmath48 transition that is first order , i.e. , @xmath50 .
it should be pointed out that the transition can also be driven first order by increasing the enantiomeric excess . for the sake of simplicity
, we do not consider that possibility here but will do so in a future publication .
to lowest order @xmath52 , the twist elastic constant , and controls the decay , over a length scale @xmath53 , of @xmath36 along @xmath27 into the bulk .
we do not include elastic energy contributions due to the spatial variation ( along @xmath27 ) of the layer spacing . for de vries - type materials , the tilt induced layer contraction will be minimal .
setting @xmath37 equal to its minimum value @xmath54 leads to an elimination of the last two terms of eq .
( [ bulk free energy ] ) and an @xmath9 dependent @xmath55 and @xmath56 .
there is a corresponding upward , @xmath9 dependent , renormalization of @xmath57 : @xmath58 the dependence of @xmath57 on @xmath9 means that if @xmath9 is sufficiently large then the system s bulk will actually be in the sm-@xmath1 phase .
alternatively , for the case of fixed temperature @xmath5 it is useful to express the value of @xmath9 above which the system s bulk will be in the sm-@xmath1 phase : @xmath59 it is useful to rescale both @xmath60 and @xmath61 .
doing so leads to the following rescaled bulk and surface free energies : @xmath62\ ; \label{rescaled free energy}\end{aligned}\ ] ] and @xmath63\ ; , \label{rescaled free energy}\end{aligned}\ ] ] where @xmath64 and @xmath65 are the rescaled bulk and surface free energies per unit area , @xmath66 , and @xmath67 is the rescaled tilt at the surface .
we next obtain the minimum @xmath68 by first obtaining the euler - lagrange equation : @xmath69 where @xmath70 .
the above equation , along with the fact that @xmath71 as @xmath72 , can then be integrated to yield @xmath73 where we have chosen the negative slope to ensure that @xmath28 .
using eq .
( [ first integral ] ) we can express the rescaled total free energy per unit area @xmath74 as @xmath75\ ; , \label{total free energy ito theta_0}\end{aligned}\ ] ] where @xmath76 is a dimensionless measure of the strength of the coupling of tilt to the surface .
minimizing the above @xmath77 with respect to @xmath6 we find the following implicit equation for the surface tilt @xmath6 : @xmath78 the state map in enantiomeric excess ( @xmath9 ) temperature ( @xmath5 ) parameter space can be obtained using eq .
( [ t_1st ] ) for @xmath79 along with eqs .
( [ total free energy ito theta_0 ] ) and ( [ theta_0 condition ] ) , and is shown schematically in fig .
[ statemap](a ) .
profiles of the surface tilt @xmath6 as a function of @xmath9 , are shown in fig .
[ statemap](b ) .
since the surface induced tilt generally changes very little with @xmath5 we do not show the surface tilt @xmath6 as a function of @xmath5 . ) temperature ( @xmath5 ) parameter space .
see text for a detailed explanation of the state map .
( b ) three fixed - temperature profiles of @xmath80 .
only @xmath9 values corresponding to the system s bulk being in the sm-@xmath0 phase are considered .
see text for a detailed explanation of these profiles .
( c ) decay of tilt from the surface into the bulk , i.e. along the @xmath27 direction for @xmath81 .
only values @xmath82 are considered , i.e. the bulk of the system is in the sm-@xmath0 phase , where the tilt is zero . for low surface tilt ( lst ) the decay of @xmath36 does not have a kink . for high surface tilt ( hst )
the decay of the tilt does have a kink in an intermediate range of @xmath27 .
] for sufficiently large enantiomeric excess @xmath83 the system s bulk is in the sm-@xmath1 phase .
this regime lies to the right of the solid line in fig .
[ statemap](a ) . for the remainder of our analysis
we will focus on the region where @xmath84 corresponding to the system s bulk being in the sm-@xmath0 phase .
for sufficiently large temperatures @xmath85 the surface tilt @xmath80 will increase continuously ( until @xmath86 is reached ) as @xmath9 is increased , as shown schematically in fig .
[ statemap](b ) . for sufficiently small @xmath81
there will be a range of @xmath9 in which @xmath80 is multivalued and the system can either be in a state of low surface tilt ( lst ) , in which @xmath87 , or high surface tilt ( hst ) . in figs .
[ statemap](a ) and ( b ) this coexistence region lies between the dotted lines , which represent the limits of metastability of the hst and lst states .
the width of the coexistence region shrinks to zero at the critical point ( @xmath88,@xmath89 ) .
while the system can exist in either the lst or hst state within the coexistence region , for smaller @xmath9 values the lst state will be energetically favorable while for larger @xmath9 values the hst will be energetically favorable .
thus , the system will jump from one state to the other at @xmath90 , which is defined as the @xmath9 value where the two states have equal free energies .
this first order boundary @xmath91 , shown as a dashed line in fig .
[ statemap](a ) , terminates at the critical point ( @xmath88,@xmath89 ) , where the size of the jump shrinks to zero .
the boundary does not intersect @xmath86 at @xmath46 and @xmath92 , but rather at @xmath93 and @xmath94 @xcite .
thus , only within the temperature range @xmath95 will one see a jump between lst and hst as @xmath9 is varied . for @xmath96
the system will remain in the lst state as @xmath9 increases until the first order bulk sm-@xmath0sm-@xmath1 boundary is reached .
the location of the critical point can be found analytically using the fact that at ( @xmath88,@xmath89 ) , both the slope and curvature of the tilt profile @xmath80 diverge , or equivalently , @xmath97 .
this yields @xmath98 and @xmath99 \;.
\label{epsion_c}\end{aligned}\ ] ] the surface induced tilt decays as one moves from the surface into the bulk , i.e. as @xmath27 increases from zero . for a given value of @xmath9 this decay
is governed by eq .
( [ e - l eqn ] ) , or equivalently eq .
( [ first integral ] ) .
the latter equation implies that @xmath26 decays with increasing @xmath27 .
the length scale for this decay is @xmath100 which , due to the first order nature of the bulk sm-@xmath0sm-@xmath1 transition , remains finite as @xmath101 . for @xmath102
the function @xmath103 is monotonically increasing , which implies that the curvature @xmath104 is positive and the slope of @xmath36 vs @xmath27 becomes shallower with increasing @xmath27 , as shown qualitatively in fig .
[ statemap](c ) . for @xmath20 , the nature of the decay can be qualitatively different .
this is due to the function @xmath103 being nonomonotonic , with negative slope for an intermediate range @xmath105 . for the range @xmath106 , where the surface tilt @xmath107
, the decay of the tilt will be as described above .
however , for the range @xmath108 where the surface tilt @xmath109 there will be a range , @xmath110 ( corresponding to the range @xmath111 ) over which the the curvature of the decay becomes negative .
this corresponds to a kink , i.e. a slope that becomes steeper with increasing @xmath27 .
this is shown qualitatively in fig .
[ statemap](c ) .
it can be shown that for @xmath112 the width of the kink , @xmath113 grows as @xmath114 . for @xmath96 ,
the low surface tilt state is energetically favored for all @xmath82 and there will be no kink .
we have analyzed the surface electroclinic effect ( sece ) in a material that exhibits a first order bulk smectic-@xmath0 ( sm-@xmath0 )
smectic-@xmath1 ( sm-@xmath1 ) transition , focussing in particular on the effect of varying enantiomeric excess .
we have shown that due to the first order nature of the bulk sm-@xmath0
sm-@xmath1 transition @xcite , the sece can be unusually strong and that as enantiomeric excess is varied , a jump in surface induced tilt is expected .
a theoretical state map , in enantiomeric excess - temperature space , features a critical point which terminates a line of first order discontinuities in the surface induced tilt .
the decay of the surface induced tilt , as one moves from surface into bulk shows that for sufficiently high surface tilt the decay will exhibit a well defined spatial kink within which it becomes especially rapid . in chiral smectic liquid crystal devices , like ferroelectric liquid crystal displays ,
the smectic layers should ideally be homogeneously aligned perpendicular to the cell substrates . in materials with the phase sequence : isotropic nematic ( @xmath115)sm-@xmath0sm-@xmath1 the rubbed polymer alignment layers aligns the director @xmath2 in the @xmath115 phase and the smectic layers are formed perpendicular to @xmath2 at the @xmath115sm-@xmath0 transition . however , when the director starts to tilt upon entry to the sm-@xmath1 phase there is a shrinkage in smectic layer thickness leading to an unwanted chevron structure @xcite . such chevrons and the related zig - zag defects could be avoided by using de vries smectics which have essentially no layer shrinkage . but as de vries materials do not exhibit a nematic phase , the initial alignment of the smectic layers is much more difficult to achieve , especially with the generally strong sece in these materials . in order to get a homogeneous smectic alignment in device cells the rubbing directions at the two surfaces have to be carefully matched to compensate for the sece , e.g. by cross - rubbing .
our analysis suggests that it should be possible to minimize the sece - related alignment issues , by careful tuning of the material and cell parameters to be in the lst regime , at the isotropic to sm - a * transition . from an applicational point of view
it would also be important to investigate the switching dynamics and the long term stability of both lst and hst cells , and to see whether the lst state is stable under electric fields and temperature variations .
it is possible that w415 has a sm-@xmath0 sm-@xmath1 that is tricritical or very weakly first order .
unfortunately , the nature of the bulk transition is not discussed in the article . however , that the transition is not first order can be inferred from the lack of an s - shaped character in the tilt versus field curves for the bulk electroclinic effect ( fig . 4 in ) .
it is important to note in saying that the electroclinic effect ( bulk or surface ) is enhanced , we mean that the electroclinic effect at _ finite _ field is enhanced . above a second order transition
, the zero - field electroclinic effect susceptibility will diverge as the transition is approached , while for a first order transition it will remain finite .
thus the zero field susceptibility is actually diminished above the first order transition as compared to a second order transition . however , above a second order transition the finite field susceptibility decreases with increasing field , i.e. the slope of the response ( i.e. @xmath36 vs @xmath4 ) curve gets smaller . above a first order transition the response curve is superlinear between @xmath4 is zero and @xmath116 , and thus the finite field susceptibility _ increases _ with increasing field in this range . at @xmath117 ( where the critical temperature @xmath89 is above the sm-@xmath0sm-@xmath1 transition temperature , @xmath118 )
the susceptibility becomes infinite at @xmath116 . for @xmath119
the response curve becomes s - shaped and @xmath36 will jump discontinuously at finite @xmath4 .
this is to be contrasted with the response curve above a second order transition which exhibits no superlinear or s - shaped features .
thus , in saying that the electroclinic effect is enhanced above a first order transition , we mean that ( unlike the effect above a second order transition ) the response will exhibit superlinear and s - shaped features . | we analyze the surface electroclinic effect ( sece ) in a material that exhibits a first order bulk smectic-@xmath0 ( sm-@xmath0 )
smectic-@xmath1 ( sm-@xmath1 ) transition .
the effect of a continuously varying degree of enantiomeric excess on the sece is also investigated .
we show that due to the first order nature of the bulk sm-@xmath0
sm-@xmath1 transition , the sece can be unusually strong and that as enantiomeric excess is varied , a jump in surface induced tilt is expected . a theoretical state map , in enantiomeric excess - temperature space , features a critical point which terminates a line of first order discontinuities in the surface induced tilt .
this critical point is analogous to that found for the phase diagram ( in electric field - temperature space ) for the bulk electroclinic effect .
analysis of the decay of the surface induced tilt , as one moves from surface into bulk shows that for sufficiently high surface tilt the decay will exhibit a well defined spatial kink within which it becomes especially rapid .
we also propose that the sece is additionally enhanced by the de vries nature ( i.e. small layer shrinkage at the bulk sm-@xmath0
sm-@xmath1 transition ) of the material .
as such the sece provides a new means to characterize the de vries nature of a material .
we discuss the implications for using these materials in device applications and propose ways to investigate the predicted features experimentally .
= 10000 |
the hard - core model is a model of a gas composed of particles of non - negligible size and consequently configurations of the model are independent sets @xcite . for a ( finite ) graph @xmath17 and an activity @xmath18 ( corresponding to the fugacity of the gas ) , configurations of the model are the set @xmath19 of independent sets of @xmath20 where @xmath21 has weight @xmath22 .
the gibbs measure is defined as @xmath23 where @xmath24 is the partition function .
a fundamental question for statistical physics models , such as the hard - core model , is whether there exists a unique or there are multiple infinite - volume gibbs measures on @xmath2 .
an equivalent question is whether the influence of the boundary on the origin decays in the limit .
more formally , for a box in @xmath2 of side length @xmath25 centered around the origin , let @xmath26 ( @xmath27 ) denote the marginal probability that the origin is unoccupied conditional on the even ( odd , respectively ) vertices on the boundary being occupied . if @xmath28 then there is a unique gibbs measure on @xmath2 , and if this limit is @xmath29 then there are multiple gibbs measures .
it is believed that there is a critical activity @xmath1 such that for @xmath30 uniqueness holds , and for @xmath31 non - uniqueness holds .
for the infinite , regular tree @xmath7 of degree @xmath5 it is easy to show that @xmath32 @xcite .
there are long - standing heuristic results which suggest that @xmath33 @xcite . for the upper bound on the critical activity ,
a classical peierls type argument implies @xmath34 @xcite , and blanca et al .
@xcite improved this upper bound to show @xmath35 .
our focus is on the lower bound .
weitz @xcite showed that @xmath36 .
his result followed from the algorithmic result .
for all graphs with constant maximum degree @xmath5 , @xmath6 , weitz @xcite presented an @xmath4 for approximating the partition function . a central step in his approach
is proving a certain decay of correlations property known as _ strong spatial mixing ( ssm ) _ on the graph @xmath20 .
ssm says that for every @xmath37 , every @xmath38 and @xmath39 , and pair of configurations @xmath40 on @xmath41 which only differ on @xmath42 ( i.e. , @xmath43 then the difference in the influence of @xmath14 and @xmath44 on the marginal probability of @xmath11 decays exponentially in the distance of @xmath11 from the difference set @xmath42 ( see section [ sec : prelim ] for formal definitions of these concepts ) .
in contrast , _ weak spatial mixing ( wsm ) _ only requires that the influence decays exponentially in the distance to the set @xmath41 .
for the hard - core model , since fixing a vertex to be unoccupied ( or occupied ) can be realized by removing the vertex ( or the vertex and its neighbors , respectively ) , it then follows that ssm on a graph @xmath20 is equivalent to wsm for all vertex induced subgraphs of @xmath20 . given @xmath45 where @xmath10 is an ordering of the neighbors of @xmath11 ,
then weitz constructs a version of the tree of self - avoiding walks from @xmath37 that we will denote as @xmath46 ; see section [ sec : saw ] for its definition .
the key property is that for every @xmath14 , if for all @xmath11 ssm holds on @xmath46 then ssm holds on @xmath20 .
his variant of the self - avoiding walk tree fixes the leaves of the tree ( corresponding to the walk completing a cycle in @xmath20 ) to be occupied or unoccupied based on the ordering @xmath10 for the last vertex @xmath11 in the corresponding cycle .
he then shows that ssm holds on the complete tree @xmath7 , and hence ssm holds on all trees of maximum degree @xmath5 when @xmath6 .
restrepo et al .
@xcite improve upon weitz s approach for @xmath2 by utilizing its structure to build a better `` bounding tree '' than @xmath7 .
they define a set of branching matrices @xmath47 for @xmath48 corresponding to walks in @xmath2 containing no cycles of length @xmath49 ( see section [ sec : bm ] for a more formal introduction to these notions ) .
the key point is that , for certain orderings @xmath14 , the tree @xmath8 is a subtree of the tree @xmath50 defined by @xmath47 .
they then present a decay of correlation proof by using a suitable message passing approach for proving ssm for @xmath50 , and hence for @xmath8 as well .
they show that ssm holds on @xmath51 for @xmath52 , and ssm holds on @xmath53 for @xmath54 .
consequently , they establish that @xmath13 .
in addition , a recent paper @xcite presents a simpler condition for establishing ssm based on the connective constant of @xmath2 , but the bounds obtained by that approach are currently weaker than @xcite .
our first result establishes a limit to these approaches by showing that ssm does not hold on @xmath55 .
as mentioned earlier , in the construction of @xmath55 , the assignment for leaves depends on the ordering @xmath14 which , for every vertex @xmath11 , specifies an ordering of the neighbors of @xmath11 .
since @xmath2 is vertex - transitive , it is natural to define an ordering that is identical for every vertex ( e.g. , based on an ordering of the directions @xmath56 , and @xmath57 ) , which we refer to as a _
homogenous ordering_. thus , a homogenous ordering @xmath14 is one where @xmath58 for all @xmath59 .
we prove the following result .
[ thm : counter ] for
all @xmath14 , all @xmath15 , ssm does not hold on @xmath8 .
moreover , for all homogenous @xmath14 , all @xmath60 , ssm does not hold on @xmath8 .
the theorem follows from considering a tree @xmath41 that is a subtree of @xmath8 and establishing the threshold for wsm on @xmath41 .
the tree @xmath41 that we consider in the homogenous ordering case is quite simple .
when @xmath61 is first in the ordering , the tree is simply the never - go - south tree ( see section [ sec : counter ] ) . for any @xmath8 that is based on an inhomogeneous ordering @xmath14 ,
we are able to find another general subtree for which the wsm does not hold when @xmath62 .
such an example gives a strong evidence that in order to prove the ssm for @xmath2 when @xmath0 is close to the conjectured threshold , the self - avoiding walk tree approach might not be appropriate .
there are subtrees of the saw tree of @xmath2 that have lower wsm threshold and hence one has to figure out an approach to exclude such trees .
we then present an improvement of the approach of restrepo et al .
@xcite for proving ssm for the trees @xmath50 .
they consider a particular statistic of the marginal distributions of the vertices , and prove the correlation decay property inductively on the height .
the statistics can be viewed as a message passing algorithm , a variant of belief propagation .
the messages they consider are a natural generalization of the message which is used to analyze the complete tree up to the tree threshold @xmath63 ( which thereby reproves weitz s result @xcite ) .
they establish a so - called dms condition as a sufficient condition for these messages to imply ssm holds on the tree under consideration .
some of the limitations of their approach are that to find the settings for the parameters in their messages and the dms condition , they use a heuristic hill - climbing algorithm which might become trapped in local optima . in addition , verifying their dms condition is non - trivial . in this paper , we consider piecewise linear functions for the messages . as a consequence , we can find these functions by solving a linear program .
this yields improved results and simpler proofs of the desired contraction property .
consequently , we prove ssm holds for @xmath51 when @xmath64 ( previously , @xmath65 by the dms condition ) and ssm holds for @xmath53 when @xmath66 ( previously , @xmath67 ) .
this establishes the following theorem .
[ thm : critical ] @xmath68 .
the rest of the paper is organized in the following way .
we formally define wsm and ssm in section [ sec : prelim ] and also present there the self - avoiding walk tree construction used by weitz @xcite . in section [ sec : bm ] , we will introduce branching matrices and present the framework of restrepo et al .
@xcite in a manner tailored to our work . in section [ sec : counter ] we will discuss limitations of weitz s approach by showing several counter - examples . finally , in section [
sec : lp ] we discuss our linear programming approach for proving ssm , which yields an improvement on the lower bound for the uniqueness threshold of the hard - core model on @xmath2 .
for a graph @xmath17 and @xmath69 , we define the boundary condition @xmath14 on @xmath42 to be a fixed configuration on @xmath42 . for a boundary condition @xmath14 ,
let @xmath70 be the unoccupied probability of vertex @xmath11 in the gibbs distribution @xmath71 on @xmath20 conditional on @xmath14 .
we now formally define wsm and ssm .
[ def : wsm ] for the hard - core model at activity @xmath0 , for finite graph @xmath17 , _ wsm _ holds if there exists @xmath72 such that for every @xmath37 , every @xmath69 , and every two configurations @xmath73 on @xmath42 , @xmath74 where @xmath75 is the graph distance ( i.e. , length of the shortest path ) between @xmath11 and ( the nearest point in ) the subset @xmath42 . for an infinite graph @xmath20
, we define the wsm threshold for @xmath20 as @xmath76 for the hard - core model at activity @xmath0 , for finite graph @xmath17 , _ ssm _ holds if there exists a @xmath72 such that for every @xmath37 , every @xmath69 , every @xmath77 , and every two configurations @xmath73 on @xmath42 where @xmath78 , @xmath79 finally , let @xmath80 denote the ssm threshold for @xmath20 , defined analogously to @xmath81 but with respect to ssm . to contrast the definitions of wsm and ssm , note that in wsm the influence decays exponentially in the distance to the boundary set @xmath42 , whereas in ssm it is exponentially in the distance to the subset @xmath82 of the boundary that they differ on .
an important observation that we repeat from the introduction to emphasize it , is that for the hard - core model , for a graph @xmath20 , ssm holds if and only if for all ( induced vertex ) subgraphs of @xmath20 wsm holds .
we now detail weitz s self - avoiding walk tree construction @xcite . given @xmath17 , we fix an arbitrary ordering @xmath83 on the neighbors of each vertex @xmath84 in @xmath20 .
let @xmath85 be the collection of these orderings . for each @xmath37 , the tree @xmath46 rooted at @xmath11
is constructed as follows .
consider the tree of self - avoiding walks originating from @xmath11 , including the vertices closing a cycle in the walks as leaves .
denote this tree by @xmath86 .
we assign a boundary condition to the leaves by the following rule .
each leaf closes a cycle in @xmath20 , so say the leaf corresponds to vertex @xmath84 in @xmath20 and the path leading to the leaf corresponds to the path @xmath87 in @xmath20 .
then if @xmath88 in the ordering @xmath83 we fix this leaf to be unoccupied , and if @xmath89 in the ordering @xmath83 we fix this leaf to be occupied .
since we are in the hard core model , if the leaf is fixed to be unoccupied we simply remove that vertex from the tree . and
if the leaf is fixed to be occupied , we remove that leaf and all of its neighbors from the tree , i.e. , we remove completely the subtree rooted at the parent of that leaf .
if a boundary condition @xmath90 is assigned to a subset @xmath42 of @xmath20 , then the self - avoiding walk tree can also be constructed consistently to the boundary condition , i.e. , for a vertex @xmath91 of @xmath20 , we assign @xmath92 to every occurrence of @xmath84 in @xmath46 .
weitz proves that , for any boundary condition on @xmath20 and any vertex @xmath11 , the marginal distribution of @xmath11 on @xmath20 is the same as the marginal distribution of the root of @xmath46 with the corresponding boundary condition .
this further implies the following .
[ thm : weitz ] for a specific @xmath0 , for any @xmath14 , if for all @xmath11 ssm holds for @xmath46 , then ssm holds for @xmath20 .
let us first recall the recurrence of the marginal distributions on trees for the hard - core model .
for now , we fix our infinite tree to be @xmath41 .
let @xmath11 be a vertex of @xmath41 , and let @xmath93 denote the subtree of @xmath41 rooted at @xmath11 .
let @xmath94 denote the children of @xmath11 in @xmath93 .
let @xmath95 be the unoccupied probability of vertex @xmath11 in the subtree @xmath93 rooted at @xmath11 with boundary condition @xmath90 .
it is straightforward to establish that @xmath95 satisfy the following recurrence : @xmath96 there are two special boundary conditions : one is called the odd boundary condition ( denoted as @xmath97 ) which occupies all the vertices at level @xmath98 when @xmath98 is odd ( and unoccupies when @xmath98 is even ) ; the other is called the even boundary condition ( denoted as @xmath99 ) which occupies all the vertices at level @xmath98 when @xmath98 is even ( and unoccupies when @xmath98 is odd ) .
these two boundary conditions are the extremal ones , meaning that for any other boundary condition @xmath90 for the vertices at distance @xmath98 from the root @xmath100 of @xmath41 , @xmath101 when @xmath98 is even ( and with the inequalities reversed when @xmath98 is odd ) . to see that wsm holds for the tree @xmath41 , it is enough to show that for the odd and even boundary conditions @xmath102 and @xmath103 , the difference of the marginal probabilities at the root @xmath104 decay exponentially in @xmath98 .
recall that in order to show that uniqueness holds for @xmath2 for a certain @xmath0 , it is enough to show that for the same @xmath0 , ssm holds on a certain tree which is a super - tree of @xmath8 . due to the regularity of @xmath2 , in @xcite ,
deterministic multi - type galton - watson trees are proposed to characterize the candidate super - trees .
the trees can be defined by branching matrices in the following way .
a branching matrix @xmath105 is simply a square matrix composed of non - negative integer entries .
[ def : branching - matrix ] given a @xmath106 branching matrix @xmath105 , @xmath107 is the family of trees which can be generated under the following restrictions : * each vertex in tree @xmath108 has its type @xmath109 .
* each vertex of type @xmath110 has at most @xmath111 children of type @xmath112 .
we use @xmath113 to refer to the tree that is generated by the matrix @xmath105 , specifically , we mean the largest tree in the family @xmath114 .
the simplest @xmath105 such that @xmath8 is in the family @xmath115 is @xmath116 . in this case
, @xmath117 is the complete , regular tree of degree @xmath118 . because of the regularity of @xmath2
, it is clear that @xmath8 is a subtree of @xmath117 . as shown in @xcite , a more sophisticated set of branching matrices @xmath119 which contain @xmath55 in their family are trees @xmath120 corresponding to all walks of @xmath2 truncated when closing a cycle of length less than or equal to a certain constant @xmath121 .
clearly , @xmath120 is a super - tree of @xmath122 , because any path in @xmath120 will only avoid cycles of length @xmath123 whereas paths in @xmath8 are avoiding all cycles .
when one tries to avoid a cycle of length @xmath118 , the matrix becomes @xmath124 where each type is simply representing the various stages of completing a cycle of length @xmath118 in a walk .
it is easy to verify that @xmath8 , for any @xmath14 , is in the family @xmath125 . in @xmath126 , we have not yet taken into consideration the effect of the assignments to leaves as detailed in the construction of @xmath127 in section [ sec : saw ] .
when we do that , we are able to construct much more sophisticated branching matrices which yield better bounds . therefore , for @xmath48 , let @xmath47 denote the branching matrix generating the tree containing all walks in @xmath2 truncated when completing a cycle of length @xmath123 , where these leaf vertices are occupied or unoccupied according to the definition in section [ sec : saw ] based on some fixed homogeneous ordering @xmath14 of the neighbors for every vertex . by taking into account the boundary condition
we obtain a smaller tree since when a walk closes a cycle with an occupied assignment to a vertex @xmath128 , this forces the parent of @xmath128 to be unoccupied , which further trims down the size of the tree .
these more sophisticated matrices yield a `` tighter '' bound on @xmath8 , however the number of types increase .
for example , for @xmath129 , whereas @xmath126 has 4 types , @xmath130 has 17 types ( after some simplifications ) , see @xcite for details of @xmath130 . for @xmath131
there are 132 types , and for @xmath132 there are @xmath133 types . for each @xmath134 by @xmath134 branching matrix @xmath105
, we would like to derive a condition such that ssm holds for the tree @xmath113 . throughout this paper , for each type @xmath110 , we treat the row @xmath135 of @xmath105 as a multi - set and each entry @xmath136 of the row denotes the number of elements the set @xmath135 has of type @xmath112 .
we use @xmath137 to denote the type of vertex @xmath138 .
the following lemma , which is re - stating lemma 1 from @xcite in a slightly simpler form that is more convenient for our work , provides a sufficient condition for ssm to hold for the tree @xmath113 .
the proof is in section [ sec : a ] .
[ lem : con ] let a branching matrix @xmath105 be given .
assume there is @xmath139 such that for each type @xmath110 , there is a positive integrable function @xmath140 where @xmath141 for @xmath142 in the range @xmath143 $ ] for each child @xmath84 and @xmath144 defined in as a function of @xmath142 s .
then ssm holds for @xmath145 , i.e. , wsm holds for all trees @xmath41 in the family @xmath146 with a fixed rate @xmath147 .
usually , when one applies various methods trying to solve the functional inequality , one has to face the fact that the dimension of the matrix @xmath105 is huge , e.g. , @xmath148 for @xmath149 in @xcite .
a natural way to generate @xmath105 is using a dfs program that enumerates all of the types by remembering the history of the self - avoiding walk .
however , there are many types in such a matrix that are essentially the `` same '' . here
we provide a rigorous definition of what types are the same and can be reduced , and a heuristic approach for efficiently finding those types that are the same .
let @xmath150 be a partition of the types in @xmath105 , i.e. , @xmath151 such that @xmath152 $ ] .
we define the partition to be consistent with @xmath105 , if for every @xmath153 $ ] , each pair of types @xmath154 , the rows @xmath155 and @xmath156 are the same with respect to @xmath150 , that is @xmath157.\ ] ] [ def : reducible ] given @xmath105 and a partition @xmath150 of size @xmath158 which is consistent , we define the @xmath158-by-@xmath158 matrix @xmath159 by , @xmath160 by the choice of @xmath161 does not matter .
we say @xmath105 is _ reducible _ to a @xmath158-by-@xmath158 matrix @xmath162 if there is a consistent partition @xmath150 such that @xmath163 .
[ lem : red ] for a branching matrix @xmath105 which is reducible to a matrix @xmath162 , @xmath164 note that since the trees @xmath165 and @xmath113 are the same ( when one ignores the types ) , wsm holds on one iff it holds on the other , and similarly for ssm .
hence , if we can find a small matrix @xmath162 that @xmath105 is reducible to then we can use @xmath162 to simplify proofs of associated spatial mixing properties . consider a tree @xmath166 .
for each vertex in @xmath41 relabel it by its corresponding type in @xmath150 .
in other words , if vertex @xmath167 has type @xmath168 $ ] then relabel it to type @xmath169 $ ] where @xmath161 . by and
, this tree @xmath41 can be generated by @xmath162 and hence @xmath170 .
for the reverse mapping , consider a tree @xmath171 .
first , for the root @xmath11 of type @xmath169 $ ] , reassign it an arbitrary type @xmath161 . for a vertex @xmath84 , given its new label @xmath168 $ ] , by and , the number of children of @xmath84 that are of type @xmath172 $ ] is @xmath173 .
hence , we can relabel these children with types in @xmath174 so that they are consistent with row @xmath175 of @xmath105 .
after fixing such a relabeling of the children of @xmath84 , then we continue to the children of @xmath84 .
this method relabels the vertices of @xmath41 to types in @xmath176 $ ] so that the new labeling can be generated by @xmath105 and hence @xmath166 .
this proves @xmath177 and an identical approach shows that @xmath178 .
now the question is how to find a small consistent partition @xmath150 .
for a specific value @xmath179 , let @xmath180 be the fixed points of the recurrences of the marginal distributions defined by @xmath105 .
our conjecture is the following .
[ conj : red ] let the partition @xmath181 be the sets of types that have the same value of the fixed points in @xmath180 , i.e. , for each @xmath182 , for all @xmath183 , @xmath184 are the same . if for all @xmath0 , the partitions @xmath181 are identical , then @xmath150 is a partition that is consistent of @xmath105 . using the intuition from conjecture [ conj : red ] we are able to find good partitions in practice .
we simply run a dynamic programming algorithm on the tree @xmath113 to calculate an approximation of the fixed points in @xmath180 . once the approximation is good enough ,
we simply make the partition according to this approximation .
we then check the consistency of the partition with @xmath105 , and therefore , we know whether the resulting matrix generates the same tree as the original one or not by lemma [ lem : red ] . applying this reduction to @xmath131 the number of types goes down from @xmath185 to @xmath186 , and for @xmath132 the number of types goes down from @xmath133 to @xmath187 .
this significant reduction in the size of the matrices greatly reduces the number of constraints and variables in our linear programming formulation .
we will use this technique to simplify the branching matrix @xmath188 considered in section [ sec : inhomog ] for proving theorem [ thm : counter ] , and reduce the matrix from 7 types to 3 types .
as described in the introduction , previous approaches for lower bounding @xmath1 are based on proving ssm for @xmath8 for some @xmath14 . to provide a bound on the strength of these approaches we upper bound the ssm threshold for @xmath8 .
we will show that for @xmath189 , for all @xmath14 , ssm does not hold for @xmath8 .
we also show that for all @xmath190 , for any homogeneous ordering @xmath14 , ssm does not hold for @xmath8 .
note that these results do not imply anything about wsm / ssm on @xmath2 , they simply show a limitation on the power of the current proof approaches . to prove that ssm does not hold on @xmath8 we define a tree @xmath41 that is a subtree of @xmath8 and
prove that wsm does not hold on @xmath41 for sufficiently large @xmath0 .
we define a branching matrix @xmath191 such that @xmath192 corresponds to the never - go - south tree , and prove that wsm does not hold on this tree when @xmath60 . since we are assuming a homogeneous ordering @xmath14 , without loss of generality assume that north is smallest in the ordering .
we construct @xmath191 by considering those walks on @xmath2 that only go north , east , and west .
the branching rules can be written in the following finite state machine way : @xmath193 where @xmath194 corresponds to the origin and is a transient state so can be ignored when analyzing the recurrence .
the branching matrix corresponding to the above rule is @xmath195 where rows / columns 1 , 2 , and 3 correspond to north , east , and west respectively .
[ prop : never - south ] let @xmath14 be homogenous ordering where @xmath196 is the smallest in the order .
the tree @xmath192 generated by the branching matrix @xmath191 is a subtree of @xmath8 . in weitz s construction ( as we presented in section [ sec : saw ] ) , recall that @xmath55 denotes the tree of self - avoiding walks of @xmath2 originating from the origin , including the vertices closing a cycle in the walks as leaves ( i.e. , in @xmath55 we have not yet fixed the leaves to be occupied or unoccupied based on the ordering @xmath14 ) .
the tree @xmath192 consists of all those self - avoiding walks that never go south , and thus , it is a subtree of @xmath55 .
now , in the second part of weitz s construction , some vertices are deleted from @xmath55 to obtain @xmath8 .
we need to show that no vertex from @xmath192 is deleted .
a vertex is deleted from @xmath55 because ( i ) it is an occupied leaf , or ( ii ) it is the parent of an occupied leaf . for a leaf @xmath197 in @xmath55 ,
the path to @xmath197 in @xmath55 corresponds to a walk in @xmath2 finishing in a cycle .
since there are no cycles in the walks corresponding to @xmath192 , we know that @xmath197 does not appear in @xmath192 .
this handles case ( i ) . to handle case ( ii ) , consider a vertex @xmath44 in @xmath192 which is the parent of a leaf @xmath197 in @xmath55 . as @xmath197 is a leaf , it corresponds to a path finishing a cycle in @xmath2 , say the path is @xmath198 . then , since @xmath44 is the parent of @xmath197 it corresponds to the path @xmath199 . in @xmath44
there are no south moves , thus , to close a cycle , the edge @xmath200 must be a south move . therefore in the boundary condition , @xmath197 is fixed to be unoccupied , as @xmath88 in the ordering @xmath83 , because @xmath201 is at the north of @xmath84 , and north is smallest in the ordering , by assumption . for the tree
@xmath192 we can establish its wsm threshold as stated in the following result .
[ thm : three ] @xmath202 the second part of theorem [ thm : counter ] concerning homogenous orderings follows as an immediate corollary of lemmas [ prop : never - south ] and [ thm : three ] . using the partition @xmath203 of @xmath204 the matrix @xmath191
is reduced , as defined in definition [ def : reducible ] , to the following @xmath205 matrix : @xmath206 from lemma [ lem : red ] the matrices @xmath162 and @xmath191 generate the same family of trees .
now the recurrences for the marginal distributions of both types derived from are @xmath207 using some algebra , we are able to determine the fixed points of @xmath208 for @xmath209 @xmath210 we just need to check the eigenvalues of the jacobian of the recurrences at the fixed point , see e.g. , @xcite : if the largest eigenvalue is greater than @xmath211 , then the function around the fixed point is repelling and hence it is impossible for the boundary conditions to converge to this unique fixed point .
if the largest eigenvalue is strictly less than @xmath211 , the function is contracting to the fixed point in its neighborhood .
the jacobian at the fixed point @xmath212 is the following : @xmath213 denote the trace of @xmath214 as @xmath215 and its determinant as @xmath216 .
the largest eigenvalue of @xmath214 is then @xmath217 it is easy to check that @xmath218 is increasing and that for @xmath219 , @xmath220 , @xmath221 and @xmath222 . in this section
, we will define a branching matrix @xmath188 such that the generated tree @xmath224 is a subtree of @xmath127 independently on the ordering @xmath83 of edges for each vertex @xmath84 .
this new tree @xmath224 never goes south , and in particular it is a subtree of the tree @xmath192 defined in the previous section .
this new tree has further structure to ensure that its leaves are at least distance two from the leaves of the self - avoiding walk tree @xmath55 , which implies that @xmath224 is a subtree of @xmath8 for every boundary condition @xmath14 . to achieve this property ,
we add to the never - go - south tree the rule that if the walk goes east there must be at least two north steps before it goes west ( and similarly , for west to east ) . to achieve this we need to remember the last two steps .
the tree is constructed by the following rules : @xmath225 here the state @xmath194 corresponds to the origin , while @xmath226 , @xmath57 and @xmath61 correspond to the first edges in the path .
then each of the states corresponds to the last two visited edges .
notice also that states @xmath194 , @xmath226 , @xmath57 and @xmath61 are transient states .
we denote the branching matrix for this tree as @xmath188 .
[ prop : general ] let @xmath224 be the tree generated by the branching matrix @xmath188 , and let @xmath55 be the self - avoiding walk tree of @xmath2 as defined in section [ sec : saw ] .
let @xmath14 be an arbitrarily chosen boundary condition for the leaves of @xmath55 and let @xmath8 be the reduced tree ( removing leaves unoccupied in @xmath14 and for occupied leaves , removing the leaves and their parents ) .
then , @xmath227 is a subtree of @xmath8 .
a leaf vertex @xmath197 in @xmath8 corresponds to a path @xmath228 which closes a cycle in @xmath2 .
thus , @xmath229 where @xmath230 is the origin , all @xmath231 for @xmath232 are distinct and @xmath233 for some @xmath110 .
in contrast , we claim that the following property ( @xmath234 ) holds for the tree @xmath224 : for any vertex @xmath44 in @xmath224 , @xmath44 corresponds to a path @xmath235 which is distance at least two from closing a cycle in @xmath2 ( where distance means the minimum number of edges ) . in other words , @xmath236 where @xmath237 is the origin , all @xmath231 for @xmath238 are distinct and , for all @xmath239 , @xmath240 .
let us first prove this property ( @xmath234 ) . consider a vertex @xmath44 in @xmath224 , and let @xmath241 be the corresponding path in @xmath2 .
suppose @xmath242 is distance 1 from some @xmath243 for @xmath244 , then @xmath245 is a cycle in @xmath2 .
this cycle must contain at least one south move , but by the construction of @xmath224 there are no south moves in @xmath235 .
thus @xmath246 is the unique south move in the cycle @xmath247 . in @xmath247
since there is only one south move , there must be exactly one north move , and the rest of moves should be east and west moves .
there must be at least one east and one west move , but this contradicts that by the definition of @xmath224 we know that in @xmath235 between any east and west moves there are at least two north moves .
this completes the proof of ( @xmath234 ) . now using ( @xmath234 ) we can complete the proof of the lemma .
recall that @xmath55 is the tree of self - avoiding walks on @xmath2 before we assign a boundary condition to its leaves based on @xmath14 .
note that @xmath224 is a subtree of @xmath55 .
consider an arbitrary vertex @xmath44 in @xmath224 . if @xmath248 , then @xmath44 is either a leaf or the parent of ( an occupied ) leaf in @xmath55 .
to finish the proof it is then enough to show that ( @xmath234 ) implies that neither a leaf nor the parent of a leaf in @xmath55 is in @xmath224 .
let @xmath197 be a leaf in @xmath55 .
then @xmath197 corresponds to a path @xmath249 .
then the path @xmath228 is distance 0 from closing a cycle in @xmath2 , which contradicts ( @xmath234 ) , and thus @xmath197 is not in @xmath224 .
now , let @xmath250 be the parent of @xmath197 .
the corresponding path is @xmath251 and @xmath252 .
this again contradicts ( @xmath234 ) and thus @xmath250 is not in @xmath224 .
we establish the following bounds on the wsm threshold for the tree @xmath224 .
[ lem : d1 ] for the tree @xmath224 , at @xmath253 wsm does not hold .
theorem [ prop : general ] implies that @xmath224 is a subtree for @xmath8 for any ordering @xmath14 .
hence , the first part of theorem [ thm : counter ] concerning arbitrary orderings @xmath14 follows as an immediate corollary of theorem [ prop : general ] and lemma [ lem : d1 ] . to show that the bound in lemma [ lem : d1 ] is reasonably close to
tight we also show that at @xmath254 wsm holds .
[ lem : lower33 ] let @xmath255 .
then wsm holds for @xmath224 . using the method introduced in section [ sec : red ] , with the partition @xmath256,@xmath257,@xmath258 of the set of states ,
the branching matrix @xmath188 is reduced to the three - state matrix , @xmath259 .
therefore , the recurrences for the marginal distributions of the 3 types are @xmath260 let @xmath261 and @xmath262 .
also , for any @xmath263 let @xmath264 ^ 3 : x \le u \le y\}$ ] be the _
( rectangular ) cuboid defined by _ @xmath265 and @xmath266 .
let @xmath267 and @xmath268 .
then @xmath269 ^ 3 $ ] .
we have that wsm holds for @xmath224 if and only @xmath270 ^ 3)$ ] is a singleton @xmath271 , which in particular implies that @xmath272 is the ( unique ) fix point of @xmath273 in @xmath274 ^ 3 $ ] .
we use numerical computations as part of our proof . in order to do exact computations we use @xmath275 and @xmath276 functions approximating @xmath273 to @xmath277 decimal digits . to do this
we define @xmath278 and @xmath279 by @xmath280 and @xmath281 .
we have then for any @xmath282 , @xmath283 using monotonicity , and induction it follows that for any @xmath284 , and @xmath265 @xmath285 notice also that for any @xmath282 , @xmath286 can be computed using exact arithmetic
. fix @xmath62 .
let @xmath287\qquad & x_{lr } = [ 0.5489575 , 0.4874566 , 0.4182131]\\ & x_{rl } = [ 0.6927559 , 0.5906225 , 0.5236103 ] & x_{rr } = [ 0.6933359 , 0.5910425 , 0.5240703].\end{aligned}\ ] ] let @xmath288 and @xmath289 .
we check numerically that @xmath290 using monotonicity and we obtain @xmath291 .
similarly @xmath292 . applying the brouwer fixed point theorem , @xmath293 has a fixed point @xmath294 in @xmath295 and a fixed point @xmath296 in @xmath297 . as @xmath298 , @xmath299 .
thus @xmath300 ^ 3 ) \supset \{x_l , x_r\}$ ] . therefore wsm does not hold for @xmath62 .
now we show that wsm holds for @xmath224 when @xmath255 ( lemma [ lem : lower33 ] ) .
the main idea of the proof is to find @xmath301 ^ 3 $ ] where @xmath273 is contracting and such that @xmath302 . to show that @xmath273 is contracting in @xmath303
we will show that @xmath304 , the spectral radius of the jacobian of @xmath273 is strictly smaller than @xmath211 .
[ lem : fixpoint ] assume @xmath305 and @xmath306 , for some @xmath307 and that @xmath308 for all @xmath309 .
then @xmath270 ^ 3)$ ] is a singleton . from the brouwer fixed - point theorem
@xmath270 ^ 3 ) \neq \emptyset$ ] .
let @xmath310 . from monotonicity , @xmath311 for any @xmath312 ^ 3 $ ] .
in particular @xmath313 . as @xmath273 is contracting in @xmath314 , @xmath315 is also contracting on @xmath314 and from banach s fixed point theorem , @xmath316 is a singleton .
now let @xmath317 ^ 3)$ ]
. then @xmath318 ^ 3)$ ] for any @xmath319 . by assumption , @xmath320 .
therefore , @xmath321 and thus @xmath322 . as @xmath273 is one - to - one then @xmath323 .
notice that the jacobian of @xmath273 is @xmath324 in order to bound the spectral radius of @xmath325 we will compute the spectral radius of the matrix @xmath326 [ lem : boundrho ] let @xmath327 ^ 3 $ ] be such that @xmath328 .
then for any @xmath329 we have @xmath330 . for any @xmath329 we have @xmath331 .
thus for any @xmath332 , @xmath333 . thus ( see @xcite ) , @xmath334 let @xmath335 .
let @xmath336 $ ] and @xmath337 $ ] .
we check numerically that @xmath338 and @xmath339 .
using we obtain @xmath340 from lemmas [ lem : fixpoint ] and [ lem : boundrho ] it is enough to show that @xmath341 .
let @xmath342 , then ( see @xcite ) @xmath343 , where the last inequality is checked numerically using exact arithmetic .
brightwell et al .
@xcite give an example of a tree for which wsm holds but ssm does not hold for the same activity @xmath0 . here , we present another example which is more closely related to @xmath8 .
we show a tree @xmath344 , which is a super - tree of @xmath192 and subtree of @xmath8 for homogenous @xmath14 , for which wsm holds for some @xmath60 . to construct the tree @xmath344 we allow some south moves in the tree in a certain context .
in particular , we only allow that a south move happens when the path contains the following substring : nneeseen , i.e. , a south move is allowed if and only if it is after a sequence of nnee moves and followed by een moves . also , after the substring nneeseen we only allow the next move to be a north or east move .
symmetrically we allow paths containing the substring nnwwswwn followed by a north or west move . before formally defining the branching matrix it is useful to discuss the important property of this tree @xmath345 . in the never - go - south tree @xmath192 to complete a cycle one needs at least one additional south move .
the tree @xmath345 includes south moves but it is defined in such a way that to complete a cycle one either needs an additional south move or at least 2 additional moves ( such as ww ) .
the tree @xmath345 is formally defined by the following branching matrix denoted as @xmath346 .
the tree family can be formalized in the following finite state machine way : [ cols= " < , < , < " , ]
initially , when we write down the linear programs ( lps ) for the @xmath132 matrix with the size of intervals around @xmath347 , the number of constraints and variables is huge , approximately @xmath348 billion constraints and 1 million variables . solving this lp
directly is not possible , as the data will not even fit in memory .
notice that the lps we create have high constraint - variable ratios .
one standard technique to solve such lps is to write the dual which has a high variable - constraint ratio and apply the column generation method @xcite . from the primal point of view
, we try to guess the set of tight constraints , by picking a set of primal constraints , solving a smaller lp and checking whether the rest of the constraints are satisfied . when there are violated constraints , several of the most violated constraints are added to the set and we iterate the procedure until the lp is solved .
using column generation we obtain an lp that can be solved , but running the method takes too long .
next we will present two of our major techniques to reduce the size of the lps so that we can solve them within a few days . in theorem [ thm : lp ] , we break the intervals into subintervals of the same size . the algorithm was designed to start with a very coarse set of the subintervals with a uniform length and if the lp has no solution , then the algorithm will try to decrease the length and re - solve the new lp .
usually , the algorithm has to make the length as small as @xmath347 for the lp to have a solution .
this creates lots of constraints .
notice that , the constraints are tight only in a very small range of the interval @xmath349 .
therefore , we can try to break the intervals into subintervals of different sizes .
the goal of breaking intervals is to change the primal constraints so that the objective function @xmath11 can be achieved at a smaller value . in column generation ,
shadow prices are used for this purpose . however , here deciding which intervals to break , affects the objective in a nonlinear fashion .
thus , we use a heuristic pricing scheme on the intervals to pick which ones to break .
the following briefly describes our heuristic approach . for each interval
, we know how many constraints are involved for that interval and how many of them are violated ( i.e. , @xmath350 ) .
we sum up the values of @xmath351 for how much each constraint is violated and then scale this by a factor of the size of the interval to obtain what we define as its price .
the algorithm will pick several intervals with the highest prices to break .
the reason why we scale by a factor which is a function of the size of the interval is that we do not want to break the intervals that are already very small . in figure [ figure - phii ] we show a step function @xmath352 for a type @xmath110 in @xmath353 found by the lps .
one can observe from the figure that most of the intervals have large lengths ; in fact , there are some small intervals in the middle as these are the intervals that create tight constraints .
found by the lps , width=480 ]
we fix a tree @xmath41 in the family @xmath354 . assuming condition holds , we want to show that wsm holds for @xmath41 .
note that condition is independent of the tree @xmath41 we choose . as in @xcite
, we view the boundary condition @xmath90 as a continuous parameter .
hence , throughout the remainder of the proof , @xmath355 $ ] .
since we are simply aiming to prove wsm on tree @xmath41 , we can view the boundary as all of the vertices a fixed distance @xmath98 from the root of @xmath41 .
therefore , given a boundary condition @xmath90 and for a fixed @xmath98 , we assign the boundary condition by each vertex at depth @xmath98 being fixed to be unoccupied with probability @xmath90 and fixed to occupied with probability @xmath356 .
note that for @xmath98 even , @xmath357 corresponds to the even boundary , and @xmath358 corresponds to the odd boundary .
let @xmath359 be the marginal unoccupied probability for a type @xmath110 vertex @xmath11 in the tree @xmath93 rooted at @xmath11 where the boundary condition @xmath90 is assigned to the vertices at depth @xmath98 in @xmath93 .
putting this notation into equation , for the tree @xmath41 we have : @xmath360 where @xmath361 equals to @xmath211 if the vertex @xmath84 is not in the tree @xmath41 , and otherwise is the marginal unoccupied probability of vertex @xmath84 in tree @xmath41 with the fractional boundary condition @xmath90 . by integrating over @xmath90
we can see that if @xmath362 then wsm holds for @xmath41 at the vertex @xmath11 of type @xmath110 since the even and odd boundaries correspond to @xmath357 and @xmath358 ( depending on the parity of @xmath98 ) . for a vertex @xmath11 of type @xmath110
, we have the following equation for the derivatives at @xmath359 with respect to the boundary : @xmath363 from it is sufficient to show for all @xmath110 and all @xmath364 $ ] , @xmath365 to obtain and hence wsm holds for @xmath41 , where in the inequality @xmath144 .
note that , from here we already obtain a condition that implies the wsm holds for all trees @xmath41 in the family @xmath366 . however , technically , it is hard to show the contraction of the above inequality due to the nonhomogeneous marginal distributions @xmath142 from different children vertices as well as the irregular structure of the trees .
we instead use a monotonic mapping @xmath367 ( the messages from a vertex of type @xmath110 to its parent ) for each type @xmath110 , and show that @xmath368 which also implies that wsm holds for all trees @xmath369 . setting @xmath370 , we have @xmath371 and thus @xmath372 notice that to obtain , from this last equation we just need condition to be true .
current techniques for proving lower bounds on @xmath1 analyze ssm on @xmath8 .
this paper shows that this approach will not be sufficient to reach the conjectured threshold of @xmath373 .
one problem in this approach is that boundary conditions obtainable on @xmath55 are not necessary realizable on @xmath2 .
some of the boundary conditions are more `` extremal '' than the one that is on @xmath2 which yields a lower weak spatial mixing threshold .
finding a way to exclude certain boundary conditions for @xmath55 would be an extremely interesting direction .
r. restrepo , j. shin , p. tetali , e. vigoda , and l. yang .
improved mixing condition on the grid for counting and sampling independent sets . _ probability theory and related fields _ , 156(1 - 2):75 - 99 , 2013 .
c. robinson .
_ dynamical systems : stability , symbolic dynamics , and chaos _ , second edition .
boca raton , fl , crc press , 1999 .
a. sinclair , p. srivastava , and y. yin .
spatial mixing and approximation algorithms for graphs with bounded connective constant in _ proceedings of the 54th annual ieee symposium on foundations of computer science _
( focs ) , 300 - 309 , 2013 . | for the hard - core lattice gas model defined on independent sets weighted by an activity @xmath0 , we study the critical activity @xmath1 for the uniqueness / non - uniqueness threshold on the 2-dimensional integer lattice @xmath2 .
the conjectured value of the critical activity is approximately @xmath3 .
until recently , the best lower bound followed from algorithmic results of weitz ( 2006 ) .
weitz presented an @xmath4 for approximating the partition function for graphs of constant maximum degree @xmath5 when @xmath6 where @xmath7 is the infinite , regular tree of degree @xmath5 .
his result established a certain decay of correlations property called strong spatial mixing ( ssm ) on @xmath2 by proving that ssm holds on its self - avoiding walk tree @xmath8 where @xmath9 and @xmath10 is an ordering on the neighbors of vertex @xmath11 .
as a consequence he obtained that @xmath12 .
restrepo et al .
( 2011 ) improved weitz s approach for the particular case of @xmath2 and obtained that @xmath13 . in this paper , we establish an upper bound for this approach , by showing that , for all @xmath14 , ssm does not hold on @xmath8 when @xmath15
. we also present a refinement of the approach of restrepo et al . which improves the lower bound to @xmath16 . |
plasmas in external trapping potentials have been attracting increasing interest over the last few years in many fields , including trapped ions , e.g. @xcite , dusty plasmas , e.g. @xcite and electrons and positrons in penning traps , see e.g. @xcite for an overview . among the main reasons
is that , in these systems , it is relatively easy to realize strong correlation effects in charged particle systems
. probably the most spectacular manifestation of these effects is coulomb liquid behavior and crystal formation which has been found in various geometries . in particular , the ion crystals and the recently observed spherical dust crystals or `` coulomb balls ''
@xcite have triggered intensive new experimental and theoretical work , e.g. @xcite .
the shell structure of these crystals , including details of the shell radii and the particle distribution over the shells has been very well explained theoretically by a simple model involving an isotropic yukawa - type pair repulsion and an harmonic external confinement potential @xcite .
still , it remains an open question , how the average particle distribution inside the trap looks like , if it is the same as in the case of coulomb interaction .
it is well known that in a parabolic potential , particles interacting via the coulomb potential establish a radially constant density profile . here , we extend this analysis to a plasma with yukawa interaction by solving a variational problem for the ground state density ( sec .
ii ) . then , in sec .
iii we demonstrate that screening has a dramatic effect on the density profile giving rise to a parabolic decrease away from the trap center .
there we demonstrate that the result for the density profile can be directly generalized to any anisotropic confinement potential .
while our analysis is based on a continuous plasma model on the mean - field level , we find ( sec . iv ) , by comparison with molecular dynamics simulations , that the results apply also to spherical crystals with a shell structure .
we consider a finite one - component plasma ( ocp ) containing @xmath0 identical particles with mass @xmath1 and charge @xmath2 in an external potential @xmath3 with pair interaction potential @xmath4 described by the hamiltonian @xmath5 the classical ground state energy follows from eq .
( [ eq : h ] ) for vanishing momenta and can be written as @xcite @xmath6=\int d^3r\ , u({\bf r}),\ ] ] with the potential energy density @xmath7 being a functional of the density profile @xmath8 , and we neglected correlation contributions .
the ground state corresponds to the minimum of the energy ( [ eq : energy ] ) with respect to the density profile with the restrictions that the density is non - negative everywhere and reproduces the total particle number , i.e. @xmath9 this gives rise to the variational problem @xmath10}{\delta n({\bf r})},\ ] ] where @xmath11=e[n]+\mu\biggl\{n-\int d^3r\ , n({\bf r})\biggr\},\ ] ] and we introduced a lagrange multiplier @xmath12 ( the chemical potential ) to fulfil condition ( [ eq : n ] ) .
the variation leads to @xmath13 which holds at any point where the density is non - zero . also , eq . ( [ eq : varproblem ] ) is equivalent to vanishing of the total force on the particles separately at any space point @xmath14 , cf .
[ force_s ] . equation ( [ eq : varproblem ] ) is completely general , applying to any pair interaction @xmath4 and confinement potentials of arbitrary form and symmetry , see sec .
[ anyconf_s ] . of particular interest is the case of an isotropic confinement @xmath15 , which leads to an isotropic density distribution @xmath16 the outer radius @xmath17 of which is being fixed by the normalization condition ( [ eq : n ] ) which now becomes @xmath18 .
we now consider the case of an isotropic yukawa pair potential , @xmath19 which trivially includes the coulomb case in the limit @xmath20 . carrying out the angle integration in the interaction energy in eq .
( [ eq : varproblem ] ) we obtain @xcite @xmath21.\end{gathered}\ ] ] this equation is the desired connection between the ground state density @xmath22 of the yukawa plasma with the external confinement @xmath23 .
this integral equation can be solved for the density by differentiating two times with respect to @xmath24 @xcite with the result @xmath25 the yet unknown lagrange multiplier can be obtained by inserting this explicit solution into eq .
( [ eq : varequation ] ) , which is then treated as an equation for @xmath12 , with the result @xmath26 for the frequently encountered case of a parabolic external potential , @xmath27 , we obtain for the density from eq .
( [ eq : nr ] ) @xmath28 where the constant @xmath29 is given by @xmath30 finally , the outer radius @xmath17 limiting the density profile is calculated from the normalization ( [ eq : n ] ) with the result @xmath31 this equation has four complex and two real solutions , only one of which is non - negative and thus constitutes the unique proper result entering eq .
( [ eq : c ] ) . in the coulomb limit ,
( [ eq : parabolicsolution ] ) reduces to the familiar result of a step profile , @xmath32 where the outer radius is given by @xmath33{\frac{q^2(n-1)}{\alpha}}=r_0\sqrt[3]{\frac{n-1}{2}},\ ] ] which is fixed by the number of particles and the constant density , the latter being controlled by the curvature @xmath34 of the potential . in the right part of eq .
( [ rc ] ) we introduced the length scale @xmath35{2q^2/\alpha}$ ] , which is the stable distance of two charged particles in the absence of screening @xcite and which will be used below as the proper unit for lengths , screening parameter and density .
note that eq .
( [ rc ] ) holds also for a weakly screened yukawa plasma with @xmath36 .
[ positive real solution of eq .
( [ eq : r ] ) ] for a parabolic confinement potential @xmath27 and yukawa interaction with screening parameter @xmath37.,height=226 ] in the other limiting case , @xmath38 , the radius has the asymptotics @xmath39^{1/5}-1 $ ] . in general eq .
( [ eq : r ] ) can not be solved for @xmath17 explicitly . however , a general analytical result can be found by noting that all parameters entering eq .
( [ eq : r ] ) combine into only two parameters , @xmath40 and @xmath41 . introducing these paramters into eq .
( [ eq : r ] ) , an explicit solution is found for the inverse function @xmath42 , which can be written as @xmath43 fig .
[ fig1 ] shows the result for the dimensionless radius @xmath44 of the plasma cloud , i.e. the solution of eq .
( [ eq : r ] ) which for all values of @xmath37 and @xmath0 is given by a single curve . with the result for
@xmath17 the constant @xmath29 , which is proportional to central density , can be computed from eq .
( [ eq : c ] ) , and the complete density profile , eq .
( [ eq : parabolicsolution ] ) , is found .
the results are shown in fig .
[ fig2 ] for four particle numbers between @xmath45 and @xmath46 .
one clearly recognizes the inverted parabola which terminates in a finite density value , i.e. in a discontinuity , at @xmath47 .
with increasing @xmath0 , the density increases continuously at every space point and , at the same time , extends to higher values @xmath17 .
thereby the density profile retains its shape . and a constant screening parameter @xmath48 and four different particle numbers @xmath0 shown in the figure . for comparison , also the result for @xmath49 and @xmath46 is shown by the dashed line.,height=226 ] on the other hand , when the plasma screening is increased , at constant @xmath0 , the density profile changes dramatically ,
compare the two curves for @xmath46 .
increase of @xmath37 leads to compression of the plasma : the radius @xmath17 decreases , and the absolute value of the density increases , most significantly in the center . this compressional behavior is shown in fig . [ fig : compression ] , cf .
the full green line showing the ratio of the inner to outer densities of the plasma .
( red dotted line ) , density at the boundary , @xmath50 , ( blue dashed line ) and compression @xmath51 of the plasma ( green full line ) as a function of particle number and screening parameter.,height=226 ] the dependence on @xmath37 is analyzed more in detail in fig .
[ fig3 ] below for a fixed particle number @xmath46 . in the case of coulomb interaction , @xmath52 , we recover the constant density profile ( [ eq : rc ] ) . on the other hand ,
in the case of a screened potential , the density decays parabolically with increasing distance from the trap center , cf .
( [ eq : parabolicsolution ] ) . also , the density discontinuity at @xmath47 is softened compared to the coulomb case , and the step height increases . ) , of a three - dimensional plasma of @xmath46 for four values of the screening parameter ( lines ) , from bottom to top : @xmath52 ( red ) , @xmath49 ( yellow ) , @xmath48 ( green ) , @xmath53 ( blue ) .
symbols denote molecular dynamics results of a plasma crystal for the same parameters where the average density at the positions of the shells is shown , for details see discussion in sec .
[ yb_s].,height=226 ] besides minimizing the total energy , cf .
( [ eq : varproblem ] ) , the ground state density profile can be obtained from the condition of a local equilibrium of the total force ( internal plus external ones ) at each point where the density is non - zero .
on the other hand , the shape of the radial density profile and its change with variation of @xmath37 is directly related to a change of the force equilibrium . here
, we demonstrate this for the special case of a spherically symmetric confinement potential @xmath23 .
the forces can be directly derived from eq .
( [ eq : varequation ] ) by taking the gradient in radial direction @xmath54 which means that , for any spherical layer at a distance @xmath24 from the center , the external force @xmath55 which acts towards the center is balanced by the internal force due to the yukawa repulsion between the particles .
the internal force contains two parts where @xmath56 arises from the action of all particles inside the given layer , @xmath57 , and acts outward , whereas @xmath58 results from the action of all particles located outside @xmath59 , and acts inward , @xmath60 this force balance can be used to obtain the ground state density profile .
alternatively , we can use the computed profile to analyze the two internal force contributions and their dependence on @xmath37 .
consider first the limit of weak screening , @xmath36 .
then the forces approach the coulomb case and , in the case of a constant density profile ( [ eq : rc ] ) , @xmath61 with @xmath62 being the particle number in the inner region .
this means , the force is repulsive and increases linearly with @xmath24 and exactly compensates the linear external force @xmath63 for all values @xmath64 . in the general case of finite screening the outer force
@xmath65 does not vanish , cf .
( [ eq : forceequation ] ) .
since its direction is always towards the center , the force @xmath66 has to increase simultaneously , in order to compensate the combined effect of @xmath67 and @xmath65 .
this effect increases continuously with increasing @xmath37 which is directly verified by evaluating the expressions in eq .
( [ eq : forceequation ] ) .
the result for the density profile in an isotropic confinement , eq .
( [ eq : nr ] ) , can be easily extended to arbitrary geometry . for this purpose
we use the text book result that the charge density corresponding to the yukawa potential is @xmath68 .
this allows us to rewrite the poisson equation as @xmath69 showing that the yukawa potential is the green s function of eq .
( [ eq : poisson ] ) .
this fact can be used in eq .
( [ eq : varproblem ] ) for the case of a confinement potential @xmath3 of arbitrary geometry @xmath70 to get the explicit result for the density profile @xmath71
so far we have considered the model of a continuous density distribution @xmath72 . on the other hand ,
the ground state of a confined spherically symmetric system of discrete point - like charged particles is known to have a shell structure as was demonstrated for dusty plasmas in ref .
@xcite . it is , therefore , of interest to verify if such a shell structure can be derived from our starting equation ( [ eq : energy ] ) for the total energy and to compare our results to the radial density distribution in such coulomb or yukawa balls .
the concentric shells observed in spherical trapped coulomb crystals have led to the proposal of simple analytical models , cf .
e.g. @xcite . such a model for a trapped one - component plasma
is trivially derived from the total energy expression ( [ eq : energy ] ) by inserting for the density the ansatz @xmath73 which describes @xmath74 concentric shells of zero thickness with @xmath75 particles on shell @xmath76 with radius @xmath77 and @xmath78 , where @xmath79 denotes the number of particles in the trap center ( zero or one ) @xcite . as a result
, we obtain for the total ground state energy of a yukawa plasma in an isotropic general confinement potential @xmath3 @xmath80 this is essentially the yukawa shell model of totsuji et al . , ref .
@xcite where , however , the finite size correction factor @xmath81 in the intrashell contribution and the term @xmath79 are missing . in the coulomb limit , @xmath82 , the result simplifies with @xmath83 and @xmath84 , and we immediately recover the coulomb shell model of hasse and avilov @xcite ( plus the additional correction factor ) .
a further improvement is possible by including intrashell correlations @xcite .
the simplest model is obtained by replacing @xmath85 , where @xmath86 is a fit parameter close to one which allows to achieve excellent agreement with the exact ground state @xcite . an alternative way to include correlations
was proposed by ref .
@xcite . in order to compare the density profile @xmath72 of our continuous model with the density of discrete spherical yukawa crystals , we performed molecular dynamics simulations of the ground state of a large number of coulomb balls , for details , see refs .
@xcite . as an example , the numerical results for a coulomb ball with @xmath46 which is large enough to exhibit macroscopic behavior @xcite are included in fig .
the symbols denote the average particle density around each of the shells .
the averaging was accomplished by substituting each particle by a small but finite sphere , so that a smooth radial density profile was obtained .
with increasing @xmath37 the crosses move towards the center confirming the compression of the coulomb balls observed before @xcite .
obviously , the simulation results are very well reproduced by the analytical density profile ( [ eq : parabolicsolution ] ) of a continuous plasma .
but there are also small discrepancies in the central part which grow slowly with @xmath37 .
one reason is that , for large @xmath37 , the width of the inner shells increases rapidly , making the comparison difficult .
another possible reason could be the effect of correlations .
in summary , we have presented a theoretical analysis of the ground state density profile of spatially confined one - component plasmas in dependence on the form of the pair interaction .
an explicit result for the density profile for an arbitrary confinement potential has been derived . in particular , for an isotropic confinement , we have found that screening of the coulomb interaction substantially modifies the radial density distribution .
in contrast to a bare coulomb interaction for which the density inside a parabolic external potential is constant , for a screened interaction , a quadratic decay away form the center is found .
interestingly , while our results were derived for a continuous density distribution ( a macroscopic system ) and with neglect of binary correlations , our analytical results agree very well also with first - principle simulation results for strongly correlated coulomb and yukawa clusters containing several thousands of particles for screening paramters @xmath87 .
this agreement is by no means trivial and deserves further analysis .
further , it is very interesting to investigate the reason for the deviations at large values of the screening parameter , and analyze the effect of binary correlations on the density distributions is .
these questions will be subject of forthcoming work
. 26 d.j .
wineland , j.c .
bergquist , w.m .
itano , j.j .
bollinger , and c.h .
manney , phys .
lett . * 59 * , 2935 ( 1987 ) m. drewsen , c. brodersen , l. hornekaer , j.s .
hangst , and j.p .
schiffer , phys .
lett . * 81 * , 2878 ( 1998 ) j.b .
pieper , j. goree , and r.a .
quinn , phys .
e * 54 * , 5636 ( 1996 ) .
m. zuzic et al .
lett . * 85 * , 4064 ( 2000 ) y. hayashi , phys .
lett . * 83 * , 4764 ( 1999 ) d.h.e .
dubin , and t.m .
oneill , rev .
phys . * 71 * , 87 ( 1999 ) o. arp , d. block , a. piel , and a. melzer , phys .
lett . * 93 * , 165004 ( 2004 ) h. totsuji , c. totsuji , t. ogawa , and k. tsuruta , phys .
e * 71 * , 045401(r ) ( 2005 ) p. ludwig , s. kosse , and m. bonitz , phys .
e * 71 * , 046403 ( 2005 ) m. bonitz , d. block , o. arp , v. golubnychiy , h. baumgartner , p. ludwig , a. piel , and a. filinov , phys .
* 96 * , 075001 ( 2006 ) w.d .
kraeft , and m. bonitz , j. phys .
series * 35 * , 78,(2006 ) the procedure is the same as for coulomb interaction , see m. bonitz and j.w .
dufty condensed matter physics * 7 * , 483 ( 2004 ) .
if @xmath3 is not two times differentiable , the problem has to be solved separately in space regions where the second derivative exists .
hasse , and v.v .
avilov , phys .
a * 44 * , 4506 ( 1991 ) k. tsuruta , and s. ichimaru , phys .
a * 48 * , 1339 ( 1993 ) h. totsuji , t. kishimoto , c. totsuji , k. tsuruta , phys .
rev . lett .
* 88 * , 125002 ( 2002 ) j.p .
schiffer , phys .
lett . * 88 * , 205003 ( 2002 ) o. arp et al . , j. phys .
series * 11 * , 234 ( 2005 ) h. totsuji , t. ogawa , c. totsuji , and k. tsuruta , phys .
e * 72 * , 036406 ( 2005 ) | the ground state of an externally confined one - component yukawa plasma is derived analytically . in particular ,
the radial density profile is computed .
the results agree very well with computer simulations on three - dimensional spherical coulomb crystals .
we conclude in presenting an exact equation for the density distribution for a confinement potential of arbitrary geometry . |
[ [ section ] ] the space - borne egret _ ( energetic gamma ray experiment telescope ) _ detector on board the _ compton gamma - ray observatory _ discovered strong mev and gev @xmath0-ray emission from 66 blazars , mainly from flat spectrum radio quasars and unidentified flat spectrum radio sources @xcite .
ground - based cherenkov telescopes discovered tev @xmath0-ray emission from seven blazars , five of which were not detected by egret @xcite . although @xmath0-ray emission from blazars have been studied for more than a decade now , it is still unclear where and how the emission originates . according to the most common paradigm , the emission originates close to a mass - accreting supermassive black hole , in a relativistically moving collimated plasma outflow ( jet ) that is aligned with the line of sight to within a few degrees .
the relativistic doppler effect can explain the intensity of the blazar emission , and its rapid variability at x - rays and @xmath0-ray energies on hour time scales : for emission originating from synchrotron or synchrotron self - compton ( ssc ) models , the apparent luminosity increases approximately as the fourth power of the relativistic doppler factor @xmath3^{-1}$ ] with @xmath4 , the bulk lorentz factor of the jet plasma , @xmath5 , the angle between jet axis and the line of sight , and @xmath6 , the plasma velocity , in units of the speed of light . ] @xmath7 , and the observed flux variability timescale is inversely proportional to @xmath7 .
blazars are powerful sources across the electromagnetic spectrum
. typical spectral energy distributions ( seds ) for the high energy peaked tev blazars show two broad peaks , one at infrared to x - ray energies and the other at x - ray to @xmath0-ray energies .
the low - energy peak is commonly believed to originate as synchrotron emission from a population of relativistic electrons gyrating in the magnetic field of the jet plasma .
the origin of the high - energy peak is unproven .
the commonly adopted and best studied models assume that the @xmath0-rays are produced in inverse compton processes by the same electrons that emit the synchrotron radiation at longer wavelengths ( for a recent review of observations and models , see @xcite ) . in so - called hadronic models ,
@xmath0-rays are emitted as synchrotron radiation of extremely energetic protons @xcite , as inverse compton and synchrotron emission from a proton induced cascade ( pic ) @xcite , or from @xmath8 @xmath9 decays following the interaction of high energy protons with some target material @xcite .
recent reviews on observations of blazars with tev emission and models developed to describe the data can be found in various review articles and books @xcite .
reviews focussing on observations and models of sources with mev / gev emission can be found in @xcite .
broader overviews of the field of tev @xmath0-ray astronomy are given in @xcite .
multiwavelength observations are key for understanding the blazar phenomenon .
the acquisition of good multiwavelength data sets has encountered substantial difficulties as the sensitivities of current tev observatories require flares for sampling the tev light curves on a time scale of hours .
some sources were observed with excellent multiwavelength coverage but during relatively unspectacular quiescent phases ; in other cases , the sources were flaring , but the fluxes were only poorly sampled in frequency space and in time .
the most remarkable result from the multiwavelength campaigns is that there is good evidence for a correlation between the x - ray fluxes and the tev @xmath0-ray fluxes for the two sources mrk 421 @xcite and mrk 501 @xcite . in this paper
we present results from a multiwavelength campaign on the tev blazar markarian ( mrk ) 421 .
the source is a nearby ( z = 0.031 ) high energy - peaked bl lac object , and was the first extragalactic source detected in the tev @xmath0-ray band @xcite . in november 2002 ,
@xmath0-ray observations with the whipple 10 m telescope revealed several mrk 421 flares with fluxes exceeding three times the steady flux from the crab nebula .
the all sky monitor instrument aboard _
rxte _ also showed extremely strong 2 - 12 kev fluxes reaching 100 milli - crab .
collectively these triggered a coordinated campaign .
we invoked radio , optical , and x - ray ( _ rxte _ ) observations to commence as soon as the waning moon would allow the cherenkov telescope to take data once more .
although the x - ray and tev @xmath0-ray fluxes had decreased substantially when the campaign started on december 4th , we acquired a data set with a high signal - to - noise - ratio x - ray light curve and x - ray energy spectra , and good signal - to - noise - ratio tev light curves and tev @xmath0-ray energy spectra .
the combined data allowed us to study the x - ray / tev @xmath0-ray flux correlation over several weeks .
following our previous study @xcite , this is the second campaign that measures the x - ray / tev @xmath0-ray flux correlation over several weekes .
for the first time , we use here simulated lightcurves to address the statistical significance of the x - ray / tev @xmath0-ray flux correlation and to constrain the time lag between the two light curves .
simulations are necessary as subsequent data points in the light curve are not independent of each other ( see e.g. , the discussion by @xcite ) .
the rest of the paper is organized as follows.after describing the data sets in section 2 , we explain the method that we used to reconstruct tev @xmath0-ray energy spectra from the whipple data , and give the results obtained with the method in section 3 .
subsequently , we present the results of the campaign in section 4 , and conclude with a summary and a discussion in section 5 . if not mentioned otherwise , errors are quoted on the level of one standard deviation , and upper limits are given on 90% confidence level .
[ [ section-1 ] ] we used the university of michigan equatorially mounted 26-meter paraboloid in its automatic observation mode to observe mrk 421 at 4.8 ghz , 8.0 ghz and 14.5 ghz between december 3 , 2002 and january 10 , 2003 . both linear polarization and total flux density measurements
were made , but only the total flux density measurements are reported here .
each observation consisted of a series of five - minute on - off type measurements over a 40-minute time period interleaved with observations of other program sources .
observations were restricted to within three hours of the local meridian to minimize instrumental errors .
the telescope pointing corrections are interpolated from position scans through sources stronger than 2-janskys .
the flux density measurements have been corrected to the baars flux density calibration scale @xcite using observations of a grid of calibrator sources , distributed around the sky , which were observed at approximately two - hour intervals .
the @xmath10 error bars include both the estimated measurement and calibration uncertainties .
the observation and calibration procedures have been described in more detail elsewhere @xcite .
[ [ section-2 ] ] in the following , we will discuss three optical data sets .
the first was taken at the boltwood observatory ( stittsville , ontario , canada ) with a 0.4 m telescope , a site 502a ccd chip , and a johnson - cousins r filter .
the data were collected for @xmath11 10 days between december 3 , 2002 and january 12 , 2003 .
relative aperture photometry was performed with an aperture of 10 arcsec and `` star 1 '' from @xcite as the comparison star .
the background was estimated using a concentric annulus with a diameter between 37 and 44 arcsec .
we did not subtract the light from the host - galaxy . usually , five two - minute exposures were integrated before deriving the photometric value .
the typical statistical error on the relative photometry is 0.02 mag .
the analysis is compromised by two very bright stars near mrk 421 that cause a varying level of light to spill into the source and background regions . based on previous optical results on the same source and stars in the field of view
, we estimate that photometric measurements have a systematic error of 0.08 mag per data point .
[ [ section-3 ] ] the second set of optical observations were made using a 35 cm celestron telescope installed on the tube of the 60 cm kva telescope ( la palma , spain ) .
the observations were made with the st-8 ccd using a standard kron / cousins r - filter .
the analysis used the reduction programs developed by kari nilsson ( tuorla observatory ) with the reference stars one and two from @xcite .
[ [ section-4 ] ] the final set of optical data were collected using the wiyn 0.9 m telescope at kpno with the s2 kb ccd imager using a harris v - filter .
the data were collected from 6 december 2002 to 15 january 2003 ( with , however , a large gap 9 dec 2002 to 3 jan 2003 ) .
relative aperature photometry was performed using standard iraf routines with an aperature of 6 arcsec , sky annulus ranging from 27 to 30 arcsec in diameter , and `` star 1 '' of @xcite as the comparison star .
again , we did not subtract light from the host galaxy .
typical statistical errors from the photometric fits were smaller than 0.005 mag . based on the spread in magnitude difference between two reference stars , we estimate the uncertainty for each data point to be 0.02 mag for the purpose of determining variability . with regard to the absolute flux , due to the presence of host - galaxy light , we expect the values reported to contain an undetermined systematic offset of as much as 0.1 mag
. optical magnitudes for all three data sets are converted to absolute fluxes according to @xcite .
[ [ section-5 ] ] we reduced the data from the _ rxte _ proportional counter array with the standard _ rxte _ data analysis software .
standard-2 mode pca data taken with the top layer of the operational proportional counter units ( pcus ) were analyzed .
the number of pcus operational during a pointing varied between 2 and 4 .
we restricted the spectral analysis to the energy range from 4 kev to 15 kev .
we excluded data below 4 kev , as the analysis of earlier _ rxte _ data showed corrupted behavior ( exceptionally high or low count rates of individual bins not compatible with the energy resolution of the instrument ) .
we find that the data of most pointings are dominated by background above 15 kev . after applying the standard screening criteria ( including visual inspection of the electron rate ) and removing abnormal data spikes , the net exposure in each good time interval ranged from 168 sec to 9.01 ksec .
light - curves were then extracted with ` ftools ` v5.3.2 .
background models were generated with the tool ` pcabackest ` , based on the _ _ rxte _ _ gof calibration files for a `` bright '' source ( more than 40 counts / sec / pcu ) .
response matrices for the pca data were created with the script ` pcarsp ` v.10.1 .
the pcu `` pcu0 '' was not excluded for analysis as the ftools version gives the proper background model .
we assume for all fits a galactic neutral hydrogen density of 1.31 @xmath12 @xmath13 @xcite . for each pointing , a power law model was fitted over the energy range from 4 kev to 15 kev .
[ [ section-6 ] ] we complement the data from the pointed rxte telescopes with data from the rxte all sky monitor ( asm ) @xcite taken between december 2 , 2002 ( ut ) and january 14 , 2003 ( ut ) .
we derived fluxes by averaging the `` summed band intensities '' acquired during one day .
tev observations were taken with the whipple 10 m cherenkov telescope ( mount hopkins , az ) and with the ct1 telescope of the high energy gamma - ray astronomy ( hegra ) collaboration ( la palma , spain ) . in the following paragraphs we describe the two data sets .
the whipple observations were taken between 4 december 2002 ( ut ) and 15 january 2003 ( ut ) .
a total of 44 hrs of data were acquired : 32 hrs on the source , and 12 hrs on an adjacent field for background estimation purposes .
the data were analyzed using the standard `` hillas '' 2@xmath14-moment - parameterization technique @xcite .
standard cuts ( @xmath15 ) were used to select @xmath0-ray events and to suppress background cosmic - ray events @xcite .
the fluxes were normalized to the flux from the crab nebula using a data set of 15 hrs of on - source data and matching background observations taken in december 2002 and january 2003 @xcite . using the zenith angle dependence of this crab data set we account for the zenith angle dependence of the @xmath0-ray excess rate by normalizing our measured mrk421 rate to the crab rate at a corresponding zenith angle .
significances and corresponding error bars were calculated using the method of @xcite . from monte carlo simulations , we fold the crab spectrum with the instrument response to obtain the peak energy ( energy threshold ) for the whipple 10 m data to be consistent with the value 400 gev derived elsewhere @xcite . more detailed descriptions of whipple observing modes and analysis procedures can be found elsewhere @xcite .
details about the whipple telescope including the granite - iii camera have been given in @xcite .
[ [ section-7 ] ] a second tev @xmath0-ray data set was acquired with the hegra ct1 telescpe ( see @xcite for a description of the ct1 instrument ) between 3 november 2002 and 12 december 2002 . the telescope was equipped with a 127 pixel camera with a 3 degree diameter field of view , and with all - aluminum mirrors giving a total of @xmath16 reflecting surface @xcite .
the analysis used 17 hrs of data with zenith angles between 12@xmath17 and 58@xmath17 .
the hegra ct1 data were normalized to the crab flux in a similar way as the whipple data , taking into account the zenith angle dependence of the excess rate .
we estimate a mean energy threshold for the ct1 data set of approximately 700 gev .
[ [ section-8 ] ] the normalization to the steady crab flux is a convenient way to combine data from different instruments to avoid systematic errors resulting from errors in the absolute flux calibration of each instrument , and to perform a first - order correction for variations in rate with zenith angle .
the drawback of the method is that the different energy thresholds of the whipple and hegra observations can introduce a normalization error if the source energy spectrum deviates from the crab energy spectrum . using previous measurements of the mrk 421 tev spectral index as a function of flux level
, we estimate that the whipple / hegra normalization error is always smaller than 30% .
the spectral analysis of the whipple tev @xmath0-ray data used a different set of gamma - hadron separation cuts that minimize the systematic error associated with uncertainties in the @xmath0-ray selection efficiency of the cuts while still giving a good sensitivity .
the `` extended zenith angle scaled cuts '' @xcite select primary @xmath0-rays with an efficiency that is largely independent of the zenith angle of the observation and the energy of the primary photon .
the analysis is based on the grinnell - isu ( grisu ) package that uses the kascade airshower simulation code @xcite , followed by the simulation of the cherenkov light emitted by the air shower and the simulation of the detector response . to calibrate the overall gain of the whipple 10 m telescope in the simulations , we compared simulated and observed muon events .
muons show up as bright arcs of cherenkov light in the camera and are useful for calibration because the light per unit arc length is nearly constant , regardless of the impact parameter and angle of the muon trajectory .
the overall gain of the telescope can be found by comparing the distribution of the signal per arc length in a simulated set of muon events and in an observed set .
we took the simulated muon events from a sample of simulated proton and helium showers .
muons are identified with a dedicated muon identification algorithm that extracts @xmath11200 muon arcs per 28 min data run .
we adjusted the overall gain factor in the simulations until they reproduced the observed signal per arc length distribution .
the overall gain factor agrees to within 15% with the value computed from measurements of the mirror reflectivity , photo - multiplier tube ( pmt ) quantum efficiency , and electronic gain .
we used the forward folding technique to fit the energy spectra .
although earlier tev @xmath0-ray analyses used similar methods , this is the first time that we describe the method in detail . for each cherenkov event that passed the gamma - hadron separation cuts , we computed an estimator @xmath18 of its primary energy , based on the image parameters ` _ size _ ' @xmath19 ( sum of counts in an image ) and ` _ distance _ ' @xmath20 ( distance of the image centroid from the center of the field of view ) : @xmath21 with @xmath22 , @xmath23 @xmath24 and @xmath25 for @xmath26 and @xmath27 for @xmath28 .
the constants @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath6 , and @xmath35 are given in table [ parameters ] .
the first term in the energy estimator reflects the fact that total intensity of an image ( _ size _ ) is roughly proportional to the energy ( @xmath18 ) of the inducing @xmath0-ray .
the second term corrects this relationship depending on the distance of the telescope from the shower axis ( @xmath20 is proportional to the latter ) . using extended zenith angle cuts
, the energy estimator gives an energy resolution of @xmath36 0.25 .
we limited the spectral analysis to on - source data with associated background data sets ( the so - called on - off data ) taken at zenith angles less than 30@xmath17 .
after histogramming the energy estimator for both the on - source and background regions , the background histogram was subtracted from the on - source histogram .
subsequently , an energy spectrum was fitted to this `` excess histogram '' using the forward folding approach ( see e.g. @xcite ) , making use of a simulated set of @xmath0-ray showers .
the simulated set of showers consisted of 67,500 showers simulated over an area @xmath37 @xmath38 in the energy range from 50 gev to 25.6 tev over nine energy intervals .
the first energy interval went from 50 to 100 gev , the second from 100 to 200 gev and so forth . in the @xmath39 energy interval
, showers were simulated according to a power law distribution : @xmath40 with @xmath41 2.5 .
simple integration of eq .
[ pl ] gives the normalization constant @xmath42 as function of the lower bound @xmath43 and upper bound @xmath44 of the @xmath39 energy interval and the number @xmath45 of showers simulated in that energy interval : @xmath46 we fit the data with two models , a power law model and a power law model with an exponential high - energy cutoff : @xmath47 and @xmath48 where @xmath49 is the flux normalization at 1 tev , @xmath4 is the photon index , and @xmath50 is the high - energy cutoff . for each trial parameter
set @xmath51 ( with @xmath52 or @xmath53 ) , another histogram is filled with weighted @xmath0-ray showers that pass the @xmath0-ray selection cuts .
the weights are computed according to : @xmath54 where @xmath55 t is the observation time , and @xmath56 is the area over which showers were simulated . the numerator in eq . [ e1 ] gives the model flux for the parameter combination @xmath51 at energy @xmath18 .
the denominator gives the simulated flux .
while the weights depend on the true energy of the simulated @xmath0-rays , the showers are filled into the histogram according to their reconstructed energy .
the weighting saves computational time in the fitting procedure , as only the weights have to be re - computed for each set of model parameters .
[ [ section-9 ] ] we performed a search in parameter space until the parameter combination @xmath57 is found that minimizes the @xmath58-difference between the observed and simulated histograms .
we determined the 1 @xmath59 error region from the condition @xcite @xmath60 the best fit model parameters and the associated errors are the main results of a spectral analysis . plotting individual data points in an energy spectrum is well known to be an ill - defined problem .
a very good discussion can be found in @xcite .
owing to the finite energy resolution of the telescopes , some information about the true energy spectrum is irrevocably lost .
we have experimented with `` deconvolution methods '' , as for example , the backus - gilbert method @xcite .
owing to a combination of almost gaussian distributions of @xmath61 and @xmath62 and the modest signal - to - noise ratios of the tev @xmath0-ray energy spectra , we find that deconvolution methods improve only very little the effective energy resolution .
we thus use the simplest method to plot flux estimates , based on the counts in the excess histogram ( see e.g. @xcite ) . for an energy bin stretching from @xmath63 to @xmath64
, we plot the flux value at the energy @xmath65 10@xmath66 .
the flux value is given by scaling the best - fit model according to the observed number of excess counts : @xmath67 here , @xmath68 is the number of excess counts in the i@xmath69 bin of the signal histogram , and @xmath70 is the sum of weights of the simulated events in the i@xmath69 bin .
figure [ spectra ] shows the crab spectrum from small zenith angle data ( @xmath71 30@xmath17 ) acquired between 14
september 2002 ( ut ) and 24 march 2003 ( ut ) , and the time averaged tev @xmath0-ray spectrum of mrk 421 for the data set of the multiwavelength campaign . both data sets are from the whipple 10 m telescope .
the power law fit to the crab data gives a flux normalization of @xmath72 @xmath73 10@xmath74 photons @xmath13 s@xmath75 tev@xmath75 , and a photon index of @xmath76 2.5 @xmath77 0.1 .
the @xmath58 is 5.27 for four degrees of freedom .
these results should be compared to previous results .
the whipple collaboration obtained a flux normalization @xmath78= ( [email protected]@xmath770.60 ) @xmath7310@xmath74 photons @xmath13 s@xmath75 , and a spectral index @xmath76 [email protected]@xmath770.04 @xcite .
the hegra collaboration published @xmath78= ( [email protected]@xmath770.60 ) @xmath7310@xmath74 photons @xmath13 s@xmath75 , and @xmath76 [email protected]@xmath770.05 @xcite .
finally , the cat collaboration obtained @xmath79 ( 2.21 @xmath77 0.05 @xmath77 0.60 ) @xmath7310@xmath74 photons @xmath13 s@xmath75 , and @xmath76 [email protected] @xmath77 0.06 @xcite .
the first error values are the statistical error values , and the second are the systematic error values ( both errors on @xmath10 confidence level ) .
the systematic errors on the absolute flux are about 25% and derive from the uncertainty of the energy threshold of the instruments combined with the steep spectrum of the crab nebula .
our result derived here lies below the previously published values of whipple and hegra and above the value published by cat .
the discrepancies are comparable to the one - sigma confidence levels .
it should be noted that the whipple results published in 1998 were taken with a significantly different hardware configuration .
furthermore , we relied here on a energy threshold calibration with muons , while @xcite used the comparison between the detection rate of simulated and observed cosmic rays to calibrate the energy threshold of the telescope .
each calibration method has its own systematic uncertainty , and it is difficult to decide which one is more reliable .
the discussion shows that one should consider the full systematic error when comparing the two whipple results with each other .
the three experiments quote a systematic error of @xmath800.05 on the photon index . here
, our result agrees well with the previous whipple measurements , the hegra spectrum is somewhat steeper and the cat spectrum is significantly steeper .
the comparison of all the four results shows that systematic errors are larger than estimated . in the case of the blazar observations discussed below ,
the uncertainty of the absolute energy threshold are not that important , as we are mostly interested in relative flux variations .
we correct our fluxes with a `` throughput factor '' derived from the cosmic ray detection rate measured during each data run to correct for variations in the atmospheric conditions . as
cosmic ray showers are not identical to air showers , we estimate that the systematic uncertainty on diurnal fluxes is 10% , and the systematic error on diurnal photon indices is 0.1 .
[ [ section-10 ] ] the power law fit to the mrk 421 data gives a flux normalization of @xmath81 @xmath73 10@xmath74 photons @xmath13 s@xmath75 tev@xmath75 , and a photon index of @xmath76 [email protected] .
the @xmath58 is 5.06 for 4 degrees of freedom . the flux and photon index lie in the range of previous observations @xcite .
we find that the photon statistics do not allow us to derive meaningful constraints on the high - energy cutoff @xmath82 .
[ [ section-11 ] ] using the best - fit parameters from the crab spectral analysis , we can weight the monte carlo events by the determined spectrum and compare several simulated distributions of the image parameters with the background subtracted distributions for the crab data ( fig .
[ comparison ] ) . the good agreement between simulated and experimental data verifies that the simulations describe the air showers and the detector response adequately .
figures [ lc1 ] and [ lc2 ] combine all the light - curves measured in december 2002 , and january 2003 , respectively . from top to bottom , the figures show the tev @xmath0-ray data , tev photon indices @xmath4 , ( where @xmath83 ) , _ rxte _ pcu 10 kev flux amplitudes ( from the 4 - 15 kev spectral fits ) , the 4 - 15 kev photon indices , the _ rxte _ asm 2 - 12 kev fluxes , the optical data , and the radio data .
the tev @xmath0-ray fluxes varied between 0 to @xmath84 times that of the crab nebula , with slightly higher fluxes observed during the second half of the campaign .
we determined tev photon indices on a night to night basis whenever the flux was sufficiently high to warrant a spectral analysis . for epochs of low fluxes ( december 6 and 7 , 2002 ( ut )
( mjd 52614 - 52615 ) , december 8 , 9 and 10 , 2002 ( ut ) ( mjd 52616 - 52618 ) , december 14 , 15 , and 16 , 2002 ( ut ) ( mjd 52622 - 52624 ) , and january 7 and 8 , 2003 ( ut ) ( mjd 52646 - 52647 ) ) we combined the data of several nights to determine an energy spectrum . a @xmath85 test of statistical variability was performed by fitting the entire tev photon index dataset to a constant function .
the best fit to the data , with @xmath86 errors , is @xmath87 , with a @xmath85 value of 46.5 for 20 degrees of freedom .
the probability of obtaining this value by chance is @xmath88 .
seven data points lie outside the @xmath86 confidence region .
the 4 - 15 kev _ rxte _ data is shown in figs .
[ lc1 ] and [ lc2 ] .
the x - ray fluxes varied between @xmath89 cts @xmath13 s@xmath75 kev@xmath75 and @xmath90 cts @xmath13 s@xmath75 kev@xmath75 .
strong flares were observed on december 3 , 2002 ( mjd 52611 ) , december 5 , 2002 ( mjd 52613 ) , january 10 , 2003 ( mjd 52649 ) , and on january 14 , 2003 ( mjd 52653 ) .
the 4 - 15 kev x - ray photon indices show a large range of values from @xmath91 to @xmath92 .
the values of @xmath93 indicate that the presumed synchrotron sed peaked at and below the 4 - 15 kev energy range covered by observations . in two cases ,
the x - ray index varies very rapidly : on mjd 52619 it changes by @xmath94 within 1.4 hrs , and on mjd 52650 it changes by @xmath95 within 1.6 hrs . over the duration of the campaign , the 4 - 15 kev photon index changed by @xmath96 1 .
synchrotron cooling of a power law distribution of electrons produces a spectral break equal to , or smaller than @xmath97 0.5 if they high - energy cutoff of the electron spectrum is outside the range sampled by the observations . if the power - law index of the accelerated particles does not change with time , the detection of spectral variations with @xmath98 0.5 thus implies that the 4 - 15 kev fluxes are influenced by the high - energy cutoff of the relativistic electron population .
alternatively , flares might be associated with changes of several parameters constraining the emitting plasma , as for example the jet magnetic field or the jet beaming angle .
[ [ section-12 ] ] the _ rxte _ pcu and asm fluxes seem to trace each other , although the sparse sampling of the pcu data and the large statistical error bars of the asm data do not allow us to draw definitive conclusions .
the optical and radio data show substantial variability by about @xmath7750% of the mean flux , but no detailed correlation is present .
this is not a suprise since for the optical data one expects sizeable contributions from the galaxy light and perhaps thermal emission from the accretion disk in addition to optical synchrotron .
furthermore , one expects longer variability timescales for the lower energy synchrotron radiation , since the lower energy electrons which produce this radiation cool more slowly .
we studied the x - ray and tev @xmath0-ray flux correlation with the help of the discrete correlation function ( dcf ) of @xcite .
the dcf gives the linear correlation coefficient for the two light curves as a function of a time lag between them .
the dcf is the standard tool used in the case of sparsely sampled data and gives fewer spurious results than a traditional correlation function analysis that interpolates between light - curve data points .
we determined the statistical significance of the correlation coefficients with the help of a set of simulated @xmath0-ray light curves , computing for each simulated light curve the dcf with the observed x - ray data @xcite .
these light - curves are generated by a superposition of triangular shots , with all shots having the same amplitude and the same rise and fall time .
figure [ dcf ] gives the dcf for the x - ray and tev @xmath0-ray data sets . for a time lag of zero days
, we find a dcf value of @xmath99 .
the simulated data sets show that the correlation is significant . for uncorrelated lightcurves consisting of triangular shots with the same structure function
, we calculate that the chance probability to get a larger dcf value at a time lag of zero days is 3.12% .
figure [ xgamma ] shows the x - ray / tev @xmath0-ray flux correlation for all overlapping observations .
the measurements entering this figure are shown by the circled data points in figs .
[ lc1 ] and [ lc2 ] .
each of these overlapping observations are simultaneous to within 5 minutes .
the correlation shows substantial scatter , with similar x - ray ( tev ) fluxes sometimes corresponding to tev ( x - ray ) fluxes that differ by more than a factor of 2 from each other .
the scatter may be an inherent property of the emission mechanism .
alternatively , it may be caused by a short time lag between the flares in the two bands , not resolved by the sparse sampling during the campaign .
figure [ xxcorr ] shows the correlation between the 10 kev fluxes and the 4 - 15 kev photon indices .
a correlation can clearly be recognized in the sense that higher fluxes are accompanied by harder energy spectra . a similar correlation ( harder energy spectra for higher fluxes )
have been reported for other bl lac - type objects , e.g. mrk 501 @xcite , 1es 1959 + 650 @xcite , h 1426 + 428 @xcite , and pks 2155 - 304 @xcite .
we further scrutinized the temporal evolution of the photon indices with so called x - ray `` hysteresis '' curves @xcite , plotting the x - ray photon index as a function of x - ray flux , and indicating the temporal sequence of the measurement throughout the evolution of individual flares . in the simplest model whereby flares are formed by short lived shocks ( e.g. internal shocks from colliding blobs )
, one expects the temporal evolution to be dictated by the interplay of the acceleration , cooling , and confinement times . for the fermi mechanism
, the particle energies reached depend on the allowed acceleration time .
thus , during the beginning of flares , the x - ray and @xmath0-ray energy spectra are expected to harden .
once synchrotron cooling starts to dominate the particle energy spectra , the emitted photon energy spectra are expected to soften .
we studied two x - ray flares , one occurring at the beginning of the campaign ( mjd 52612 to mjd 52615 ) , and the other at the end ( mjd 52651 to mjd 52653 ) . the first flare is shown in fig .
this flare coincided with an increase of the tev @xmath0-ray flux by a factor of 2.4 from mjd 52612 to mjd 52613 .
the x - ray spectrum hardens during the rising phase of the flare and softens during the decaying phase .
furthermore , the x - ray spectrum is softer during the falling phase than during the rising phase .
the `` clockwise '' evolution in the @xmath100 plane is consistent with the expectations from stochastic fermi acceleration and synchrotron cooling as described above .
the data of the second flare are shown in fig .
[ hyst2 ] .
the tev @xmath0-ray flux increased from mjd 52650 to mjd 52651 by a factor of 3.8 and remained roughly constant during the following two nights of observations . while the x - ray flux increased from mjd 52652 to 52653 by a factor of @xmath101 , the tev flux measured at the same time as the x - ray fluxes did not increase substantially .
the general trend is that the spectrum hardens as the flux increases , although at mjd 52652.25 , the spectrum softens temporarily while the flux is still increasing .
unfortunately , our observations did not cover the decaying phase of the flare .
[ [ section-13 ] ] the multiwavelength campaign showed mrk 421 in a level of intermediate activity . during the observational campaign ,
mrk 421 showed significant flux variability in the radio , optical , x - ray and @xmath0-ray bands and significant spectral variability at x - rays and tev @xmath0-rays . while we measured an average tev @xmath0-ray photon index of @xmath76 2.8 ,
the observations revealed evidence for spectral variability on a time scale of days .
in particular , the data suggest very soft energy spectra with @xmath102 during the first half of the observation campaign .
one of the most interesting results from this campaign is that the x - ray and tev @xmath0-ray fluxes are correlated on the @xmath10397% confidence level , but that we find widely different tev @xmath0-ray fluxes for a single x - ray flux and vice versa .
the most extreme example is the `` orphan x - ray flare '' , seen on january 13 , 2003 ( mjd 52653 ) . the loose x - ray / tev correlation may suggest that the model parameters ( e.g. the volume of the emission zone ) change with time , or that the commonly made assumption of a single synchrotron self - compton ( ssc ) emission zone is an over - simplification . our previous observations of mrk 421 had already shown a rather loose x - ray / tev @xmath0-ray correlation @xcite and the same applied for 1es 1959 + 650 @xcite . in the case of mrk 501
, a rather tight correlation has been reported @xcite .
the analysis of the correlation between the x - ray flux and photon index during a flare indicated a `` clockwise '' hysteresis . for mrk 421
, @xcite also reported clockwise evolution .
however , @xcite reported evidence for both , clockwise evolution during some flares and anti - clockwise evolution during other flares .
if the ssc model indeed applies , these results may imply that the characteristic times scales of the most important processes ( acceleration time , radiative cooling time , escape time ) change from flare to flare and thus yield the different observed hysteresis behaviours .
recently , @xcite emphasized that the geometry of the emitting region and its orientation relative to the line of sight influences the observed flux and spectral evolution and might thus further complicate the interpretation of the results . the x - ray and tev @xmath0-ray emission from mrk 421 data
have been modeled with synchrotron self - compton codes by many groups , see e.g. @xcite .
a crucial model parameter is the jet doppler factor @xmath7 .
the published models with doppler factors @xmath7 of 20 or lower generally predict tev energy spectra that are softer than the observed ones , especially if a correction for extragalactic absorption would be applied which steepens the energy spectra considerably .
models with doppler factors @xmath7 on the order of 50 give satisfactory model fits to both the x - ray ( synchrotron ) and the tev ( inverse compton ) emission ( see the self - consistent modeling of @xcite and the discussions by @xcite ) .
@xcite observed the mrk 421 parsec - scale radio jet with the very large baseline array ( vlba ) .
remarkably , they find apparent pattern speeds of only @xmath104 . as discussed by the authors , the highly relativistic motion inferred from tev observations
can be reconciled with the modestly relativistic flow calculated from vlba observations by postulating that the jet slows down between the sub - parsec ( tev ) and parsec ( vlba ) regimes .
it may be possible to describe the multi - wavelength data with a synchrotron - compton model and lower doppler factors by invoking additional seed photons .
two new model variants that combine ingredients of ssc and external compton models have been proposed by @xcite and by @xcite .
while the first authors assume that downstream emission regions provide seed photons , the second authors speculate that the jet is composed of a fast spine with a slow - moving envelope . in this model ,
the fast spine emits the x - ray and @xmath0-ray radiation .
modeling of the data with these two inhomogeneous models is outside the scope of this paper . in this discussion
we do not want to embark on comprehensive modeling of the data from the entire campaign .
our main aim is to draw the attention of the reader to a single remarkable fact : while it is difficult to model the data with doppler factors on the order of 20 and lower , much higher doppler factors can not be excluded right away . fig .
[ nufnu ] shows two synchrotron self - compton models based on the simple snapshot code of @xcite .
the code assumes a single spherical emission volume of radius @xmath105 relativistically approaching the observer with a jet doppler factor @xmath7 .
the emission volume is homogeneously filled with a tangled magnetic field of strength @xmath30 and a non - thermal electron population .
the electron energy spectrum follows @xmath106 @xmath107 with @xmath108 2 for electron lorentz factor @xmath0 between @xmath109 and @xmath110 and @xmath111 for lorentz factors between @xmath110 and @xmath112 . the code models extragalactic absorption owing to the @xmath113 pair - production processes of tev photons on photons from the cosmic infrared background ( cib ) using the cib model of @xcite .
we discuss two models .
we show the first model for illustrative purposes only .
it uses the `` conventional '' model parameters ( @xmath114 ) derived from the time - dependent self - consistent analysis of a different but similar data set @xcite .
the second model uses a very high doppler factor ( @xmath115 ) .
all the model parameters are given in table [ fits ] .
for both models , we assured that the model parameters were chosen self - consistently .
causality arguments require that the radius @xmath105 satisfies @xmath116 cm for @xmath117 and @xmath118 cm for @xmath119 for a flux variability time scale of @xmath120 30 min .
note that the flux variability time scale sets a lower limit on @xmath105 but no upper limit , if the flux variability time scale is not dominated by light travel time effects but by other effects ( e.g. by the stability of a strong shock front ) .
we checked that the sed ( i ) fits the x - ray and tev @xmath0-ray data , and ( ii ) is consistent with the expected spectral shape owing to radiative cooling . in the first model ,
the latter self - consistency is assured by our previous self - consistent time - dependent modeling . in the second model ,
we construct an electron spectrum based on the general results for electrons suffering synchrotron and inverse compton losses @xcite ) .
we assume that the electron energy spectrum breaks at @xmath121 @xmath122 . the laboratory - frame synchrotron cooling time for electrons at the break is @xmath123 @xmath124 @xmath125 @xmath126 @xmath127 @xmath128 @xmath110 @xmath129^{-1}~$ ] @xmath130 ( @xmath131 is the thomson cross section and @xmath132 is the electron mass ) .
an electron spectrum as the one used here could result from the radiative cooling of a @xmath133 electron energy spectrum that extends from @xmath109 to @xmath112 , resulting in a spectrum with @xmath133 and @xmath111 below and above @xmath110 , respectively . as shown in fig .
[ nufnu ] , both models fail to predict the observed radio fluxes as a consequence of synchrotron self - absorption .
we would like to emphasize that we do not regard the discrepancy as a shortcoming of the model .
electrons producing the radio emission cool on much longer time scales , and the radio emission will be dominated by an accumulation of downstream plasma which stopped contributing to the x - ray and tev emission long time ago .
we could model the radio emission with another emission component ( see e.g. @xcite ) .
however , doing so is arbitrary : for a small number of data points we would add an additional model component with many free model parameters . in table
[ fits ] we list for both models the magnetic field energy density @xmath134 , the energy density in electrons @xmath135 , the ratio @xmath136 , and the kinetic luminosity @xmath137 for @xmath76 @xmath7 @xcite .
the model with a low doppler factor is closer to equipartition between magnetic field and particles .
the kinetic luminosity is similar for the two models , with a high radiative efficiency of the low-@xmath7 model making up for the stronger boosting of the high-@xmath7 model .
if taken seriously , the model with @xmath138 1000 would imply that the x - ray and tev @xmath0-ray emission is produced by an ultra - relativistic particle dominated wind , very close to the supermassive black hole .
the fact that seven blazars have been detected at tev energies seems to argue against extremely relativistic outflows with bulk lorentz factors @xmath4 on the order of 1000 , as isotropic emission would be beamed into an opening angle of @xmath139 and would make the observation of the emission unlikely .
however , the argument only applies if the jet opening angle is equal or smaller than @xmath139 . having a larger jet opening angle would require a higher total jet - luminosity as some jet segments would not contribute to the observed emission .
however , the jet - luminosities listed in the table are several orders of magnitude below the eddington luminosity of a @xmath140 solar mass black hole that is suspected to be at the core of mrk 421 @xcite .
we acknowledge an anonymous referee for helpful comments .
we acknowledge the technical assistance of e. roache and j. melnick .
this research is supported by grants from the u.s .
department of energy , the national science foundation , the smithsonian institution , by nserc in canada , by science foundation ireland , and by pparc in the uk .
aharonian , f.a . ,
akhperjanian , a.g . , andronache , m. , et al .
1999 , , 350 , 757 aharonian , f.a .
2000 , , 5 , 377 aharonian , f. , akhperjanian , a. , beilicke , m. , et al .
2004 , , 614 , 897 aharonian , f.a .
2004 , `` very high energy cosmic gamma radiation : a crucial window on the extreme universe '' , published singapore ( hong kong ) , world scientific aharonian , f. et al .
2005 , , 436 , l17 allen , c.w . , astrophysical quantities , the athlone press , 1973 aller , h.d . ,
aller , m.f . ,
latimer , g.e .
, hodge , p.e .
1985 , , 59 , 513 baars , j.w.m . ,
genzel , r. , pauliny - toth i.i.k . , & witzel , a. 1977 , , 61 , 99 .
backus , g. & gilbert , f. , 1970 , phil .
london , a , 266 , 123 barth , a.j .
, ho , l.c . , & sargent , w.l.w.2003 , apj , 583 , 134 bednarek , w. 1997 , , 285 , 69 bednarek , w. & protheroe , r.j . , 1999 , , 310 , 577 begelman , m.c . , rees , m.j . & sikora , m. 1994 , , 429 , l57 blazejowski , m. , et al .
2005 , , in press .
bttcher m. , mause h. , schlickeiser r. , 1997 , , 324 , 395 buckley j.h .
1996 , , 472 , l9 buckley , j.h .
et al.2001 , in snowmass 2001 , `` the future of particle physics '' , [ astro - ph/0201160 ] catanese , m. , akerlof , c.w . ,
badran , h.m . , et al .
1998 , apj , 501 , 616 buckley j.h .
et al . 2005 , in preparation coppi , p.s . , 1999 , in `` relativistic jets in agns '' , eds .
m. ostrowski , m. sikora , g. madejski , and m. begelman , jagellonian university press , p.333 , [ astro - ph/9903162 ] de la calle perez , i. , et al .
2003 , , 599 , 909 dickey , j.m .
& lockman , f.j .
1990 , , 28 , 215 djannati - atai , a. , piron , f. , barrau , a. , et al . 1999 , , 350 , 17 edelson r.a .
1988 , , 333 , 646 falcone , a. d. , cui , w. , finley , j. p. 2004
, apj , 601 , 165 falomo , r. , kotilainen , j.k . , & treves , a. 2002 , apj , 569 , l35 fenimore , e.e . ,
laros , j.g . ,
klebsadel , r.w . ,
stockdale , r.e .
, 1982 , aip proceedings of gamma - ray transients and related astrophysical phenomena , ed .
r. e. lingenfelter , h. s. hudson , d. m. worral , 201 finley j.p .
2001 , proc .
icrc 2001 , 2827 fossati g. et al .
2000 , , 541 , 153 .
georganopoulos , m. & kazanas , d. 2003 , , 594 , l27 ghisellini , g. , tavecchio , f. & chiaberge , m. 2005 , , 432 401 ginzburg v. l. & syrovatskii s. i. , 1964 , origin of cosmic rays .
pergamon , london hartman r.c .
et al . , 1992 , , 385 , l1 hartman r.c .
et al . , 1999 , , 123 , 79 hillas a.m. 1985 , proc icrc 1985 hillas , a.m. , akerlof , c.w . , biller , s.d .
, et al . , 503,744 horan d. & weekes t.c .
2004 , , 48 , 527 inoue , s. , & takahara , f. , 1996 , apj , 463 , 555 kardashev n. s. 1962 , sva , 6 , 317 kataoka , j. , takahashi , t. , makino , f. , inoue , s. , madejski , g.m .
, tashiro , m. , urry , c.m .
, & kubo , h. 2000 , , 528 , 243 kertzman , m.p .
& sembroski , g.h .
1994 , nucl .
methods phys .
a343 , 629 kino , m. , takahara , f. , & kusunose , m. 2002 , , 564 , 97 kirk , j.g . & mastichiadis , a. , 1999 , astroparticle phys .
, 11 , 45 kneiske , t.m . , mannheim , k. , hartmann , d.h .
2002 , , 386 , 1 konopelko , a. , mastichiadis , a. , kirk , j. , de jager , o.c . , &
stecker , f.w .
2003 , , 597 , 851 kosack , k. , et al .
2004 , , 608 , l97 krawczynski , h. 2005 , procs . of the blazar
variability workshop ii : entering the glast era " , [ astro - ph/0508621 ] krawczynski , h. 2004 , , 48 , 367 krawczynski , h. , hughes , s. b. , horan , d. 2004 , , 601 , 151 krawczynski , h. , coppi , p.s .
& aharonian , f. 2002 , , 336 , 721 krawczynski , h. , sambruna , r.m . ,
kohnle , a. , et al .
2001 , , 559 , 187 krawczynski , h. , coppi , p. s. , maccarone , t. , aharonian , f. a. , 2000 , a&a , 353 , 97 krennrich f. , biller s.d . , bond i.h . ,
et al . 1999 , , 511 , 149 krennrich f. et al . , 2002 , , 575 , l9 krennrich f. & dwek e. 2003 , proc
icrc 2003 , 2667 levine a.m. et al .
1996 , , 469,l33 li t. & ma y. 1983 , , 272 , 317 loredo , t.j . & epstein , r.i .
1989 , , 336 , 896 mannheim , k. 1998 , science , 279 , 684 maraschi , l. , et al .
1999 , , 526 , l81 mastichiadis a. & kirk j.g . , 1997 , , 320 , 19 m " ucke , a. , protheroe , r.j . ,
engel , r. , rachen , j.p . , & stanev , t. 2003 , astroparticle phys . ,
18 , 593 mirzoyan r. et al .
1994 , nucl .
methods phys .
res . , a351 , 513 - 526 ong , r.a . , in theuniverseviewedingamma - rays
r. enomoto , m. mori , s. yanagita ( universal academy press , tokyo , 2003 ) , p. 587 ,
[ astro - ph/0304336 ] pacholczyk a. g. , 1970 , radio astrophysics .
freeman , san francisco petry , d. et al .
2002 , 580 , 104 piner , b.g . & edwards , p.g .
2005 , , 622 , 168 piron , f. for the cat collaboration , in the 36th rencontres de moriond on `` very high - energy phenomena in the universe '' , astro - ph/0106210 pohl , m. & schlickeiser , r. 2000 , , 354,395 press w.h .
, teukolsky s.a . , vetterling w.t . , & flannery b.p .
1992 , numerical recipes in c , ( 2nd ed.;cambridge univ . press ) , 694 punch , m. et al .
1991 , proc .
icrc 1991 , 464 punch , m. & fegan , d.j .
1991 , aip conf.proc.220 : high energy gamma ray astronomy , 220 , 321 punch , m. , akerlof , c.w . ,
cawley , m.f .
, et al . 1992 , , 358 , 477 rauterberg g. 1995 , proc .
icrc 1995 , 460 reynolds , p.t . ,
1993 , , 404 , 206 sambruna , r.m . ,
aharonian , f.a . ,
krawczynski , h. , et al .
2000 , , 538 , 127 sikora , m. & madejski , g. 2001 , aip conference series , 558 , 275 sokolov , a. , marscher , a. , mchardy , i. m. 2004 , , 613 , 725 syrovatskii s. i. , 1959 , sva , 3 , 22 takahashi , t. , et al .
1996 , , 470 , l89 takahashi t. et al .
2000 , , 542 , l105 tanihata c. , urry c.m . , takahashi t. , et al . , 2001 , , 563 , 569 tavecchio , f. 2004 , proc .
10th marcel grossman meeting on general relativity , in press villata , m. , raiteri , c.m . ,
lanteri , l. , sobrito , g. , cavallone , m. 1998 , a&as , 130 , 305 weekes , t. c. 2003 , `` very high energy gamma - ray astronomy '' , bristol ( philadelphia ) , institute of physics pub .
weekes , t.c .
1996 , , 75,1 zweerink j.a . , akerlof c.w . , biller s.d . ,
1997 , , 490 , l141 c c c c c c c c c c @xmath144 & b ( g ) & r ( m ) & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 + + 50 & 0.2 & @xmath152 & 3.3 & 5.3 & 4.8 & @xmath153 & @xmath154 & 41.5 & 2.12 + 1000 & 0.5 & @xmath155 & 2.8 & 4.4 & 3.26 & @xmath156 & 400 & @xmath157 & 10.9 + | we report on a multiwavelength campaign on the tev @xmath0-ray blazar markarian ( mrk ) 421 performed during december 2002 and january 2003 . these target of opportunity observations
were initiated by the detection of x - ray and tev @xmath0-ray flares with the all sky monitor ( asm ) on board the _ rossi x - ray timing explorer _ ( _ rxte _ ) and the 10 m whipple @xmath0-ray telescope .
the campaign included observational coverage in the radio ( university of michigan radio astronomy observatory ) , optical ( boltwood , la palma kva 0.6 m , wiyn 0.9 m ) , x - ray ( rxte pointed telescopes ) , and tev @xmath0-ray ( whipple and hegra ) bands . at tev energies , the observations revealed several flares at intermediate flux levels , peaking between 1 and 1.5 times the flux from the crab nebula . while the time averaged spectrum can be fitted with a single power law of photon index @xmath1 from @xmath2 , we find some evidence for spectral variability . confirming earlier results , the campaign reveals a rather loose correlation between the x - ray and tev @xmath0-ray fluxes . in one case ,
a very strong x - ray flare is not accompanied by a comparable tev @xmath0-ray flare .
although the source flux was variable in the optical and radio bands , the sparse sampling of the optical and radio light curves does not allow us to study the correlation properties in detail .
we present a simple analysis of the data with a synchrotron - self compton model , emphasizing that models with very high doppler factors and low magnetic fields can describe the data . |
loop corrections modify the coulomb potential : electron loop insertion into the photon propagator leads to the uehling serber correction to the electric potential of point - like nuclei @xcite . though it leads to an important phenomenon contributing to the lamb shift of the energies of atomic electrons numerically the shift being of the order of @xmath0 is small ( we are using gauss system of units , where @xmath1 and in all formulas @xmath2 is implied ) . analogous correction in the case of external magnetic field
qualitatively change the behavior of atomic energies : in particular the energy of the ground level remains finite in the limit @xmath3 ; also spontaneous production of positrons becomes possible only for nuclei with @xmath4 . without taking radiative corrections into account in the limit of infinite magnetic field energy of ground atomic level tends to minus infinity and point - like nucleus with any @xmath5 becomes critical at large enough @xmath6 . at magnetic fields
@xmath7 the characteristic size of the electron wave function in the transverse to the magnetic field direction @xmath8 ( the so - called landau radius ) becomes smaller than bohr radius @xmath9 making the coulomb problem essentially one - dimensional .
singularity of the coulomb potential in @xmath10 is stronger than in @xmath11 . in @xmath10
the energy of the ground level is unbounded from below : the `` fall to the center '' phenomenon occurs . in the case of external magnetic field
the singularity is cured by the finite value of @xmath12 : at @xmath13 the coulomb problem remains three dimensional .
this is the reason why ground level goes down when @xmath6 grows . at superstrong magnetic fields @xmath14 radiative corrections screen the coulomb potential at short distances @xmath15 and the freezing of a ground state energy occurs : it remains finite at @xmath3 ( @xcite , @xcite , @xcite ) .
for @xmath4 the value of freezing energy is below @xmath16 , so the ground level enters lower continuum when @xmath6 increases and spontaneous production of @xmath17 pair from vacuum becomes energetically possible and thus takes place . in this process electron occupies ground level while positron is emitted to infinity . for @xmath18 freezing energy
is above @xmath16 and spontaneous positron production does not occur @xcite .
there exists the direct correspondence between radiative corrections to the coulomb potential in @xmath11 case in strong magnetic field @xmath19 and radiative corrections to the coulomb potential in @xmath10 qed .
that is why we start our presentation ( in section 2 ) from the analysis of the coulomb potential in @xmath20 qed of massive fermions . when these fermions are light , @xmath21 , the exponential screening of the coulomb potential at short distances occurs . in the limit @xmath22 ( massless @xmath20 qed , the so - called schwinger model )
this exponential screening occurs at all distances because photon gets mass , @xmath23 @xcite . in section 3
we analyze radiative corrections to the coulomb potential in @xmath24 qed in external magnetic field .
the role of the coupling constant @xmath25 here plays the product @xmath26 , and for @xmath27 the screening of the coulomb potential occurs as well . in section 4 the structure of atomic levels on which the lowest landau level ( lll ) in the presence of atomic nucleus splits is determined . in section 5
the dirac equation for hydrogenlike ion at superstrong magnetic field will be derived and effect of screening will be studied for @xmath28 . in section 6
the influence of the screening of the coulomb potential on the values of critical nuclei charges is discussed . in section 7
the obtained results are summarized .
let us finish the introduction discussing the numerical values of magnetic fields we are dealing with in these lectures .
the magnetic field at which the bohr radius of a hydrogen atom becomes equal to landau radius is @xmath29 @xmath30 gauss , which is much larger than a magnetic field ever made artificially on earth : @xmath31 gauss .
an interest to the atomic spectrum in the magnetic fields @xmath32 was triggered by the experiments with semiconductors , where electron - hole bound system called exciton is formed .
both effective charge and mass of electrons in semiconductors are much lower than in vacuum making @xmath33 in kilogauss scale reachable .
the so - called schwinger magnetic field @xmath34 gauss and magnetic field at which the screening of the coulomb potential occurs @xmath35 gauss should be compared with the magnetic fields at pulsars @xmath36 gauss and magnetars @xmath37 gauss .
although the application of the results obtained in the condensed matter physics ( say graphen , where the mass of charge carrier can be arbitrary low while the value of charge approach one ) can not be excluded , our main interest in the problem considered is purely theoretical .
summing up diagrams shown in fig .
1 we get the following formula for the potential of point - like charge : @xmath38 where @xmath39 is the one - loop expression for the photon polarization operator .
it can be obtained from the textbook expression for the polarization operator calculated with the help of dimensional regularization @xcite in the limit @xmath40 : @xmath41 \equiv -4g^2 p(t ) \;\ ; , \label{2}\ ] ] where @xmath42 = $ ] mass .
fig . 1 . _ modification of the coulomb potential due to the dressing of the photon propagator .
_ let us note that @xmath39 is finite though corresponding integral @xmath43 is divergent in ultraviolet .
the point is that the trace of gamma matrices which multiplies divergent integral is zero . in dimensional
regularization the trace is proportional to @xmath44 while ultraviolet divergency of integral over virtual momentum produces the factor @xmath45 and the product of these two factors is finite . in order to obtain an expression for the coulomb potential in the coordinate representation we take @xmath46 and
make the fourier transformation : @xmath47 the potential energy for the charges @xmath48 and @xmath49 is @xmath50 the integral in ( [ 3 ] ) can not be expressed through elementary functions
. however it appears possible to find an interpolating formula for @xmath51 which has good accuracy and is simple enough for the fourier transformation to be performed analytically .
the asymptotics of @xmath51 are : @xmath52 let us take as an interpolating formula for @xmath51 the following expression : @xmath53 the accuracy of this approximation is not worse than 10% for the whole interval of @xmath54 variation , @xmath55 .
substituting an interpolating formula in ( [ 3 ] ) we get : @xmath56 e^{ik_\parallel z } \frac{dk_\parallel}{2\pi } = \\ & = & \frac{4\pi g}{1 + 2g^2/3m^2}\left[-\frac{1}{2}|z| + \frac{g^2/3m^2}{\sqrt{6m^2 + 4g^2 } } { \rm exp}(-\sqrt{6m^2 + 4g^2}|z|)\right ] \;\ ; .
\nonumber \label{7}\end{aligned}\ ] ] in the case of heavy fermions ( @xmath57 ) the potential is given by the tree level expression ; the corrections are suppressed as @xmath58 . in the case of light fermions ( @xmath59 ) : @xmath60 @xmath61 corresponds to schwinger model ; photon gets mass .
light fermions make transition from @xmath62 to @xmath61 continuous . in the case of light fermions the coulomb potential in @xmath20 qed
is screened at distances @xmath63 . in fig .
2 the potential energy for @xmath64 is shown .
it is normalized to @xmath65 . fig.2 .
_ potential energy of the charges @xmath48 and @xmath49 in @xmath20 .
the solid curve corresponds to @xmath66 ; the dashed curve corresponds to @xmath67 . _
in order to find the potential of a point - like charge we need an expression for photon polarization operator in the external magnetic field @xmath6 . long ago an expression for the electron propagator in constant and homogeneous external magnetic field was found by schwinger @xcite as a parametric integral . for @xmath68 integration
can be easily performed and compact expression for @xmath69 follows . using it one obtains an analytic expression for a photon polarization operator
( see for example @xcite ) . to understand the reason for great simplification of the expression for the electron propagator in the limit @xmath70 one should start from the spectral representation of the propagator .
the solutions of dirac equation in the homogeneous constant in time @xmath6 are known , so one can write the spectral representation of the electron green function .
the denominators contain @xmath71 , and for @xmath72 and @xmath73 in sum over levels the lowest landau level ( lll , @xmath74 ) dominates . in the coordinate representation
the transverse part of lll wave function is : @xmath75 which in the momentum representation gives @xmath76 ( we suppose that @xmath6 is directed along the @xmath77 axis ) . substituting the electron green functions we get the expression for the polarization operator in superstrong @xmath6 . for @xmath78 ,
@xmath79 the following expression is valid @xcite : @xmath80 with the help of it the following result was obtained in @xcite : @xmath81 @xmath82 \;\ ; . \label{11}\end{aligned}\ ] ] for the magnetic fields @xmath83 the potential is coulomb up to small power suppressed terms : @xmath84 \label{12}\ ] ] in full accordance with the @xmath20 case with the substitution @xmath85 . in the opposite case of the superstrong magnetic fields
@xmath86 we get : @xmath87 @xmath88 the coulomb potential is screened at short distances @xmath89 . in fig .
3 the plot of a modified by the superstrong magnetic field coulomb potential as well as its short- and long - distance asymptotics are presented .
3 . _ the modified coulomb potential at @xmath90 g ( blue , dark solid ) and its long distance ( green , pale solid ) and short distance ( red , dashed ) asymptotics .
_ let us find the 3-dimensional shape of the screened coulomb potential .
the behavior of the potential in the transverse plane ( @xmath91 ) can be found analytically in the limit @xmath92 from the general expression : @xmath93 neglecting exponent in the denominator , which is valid for @xmath94 : @xmath95 and the coulomb potential is screened at large @xmath96 in complete analogy with the @xmath20 case . for @xmath97 in the integral ( [ 141 ] )
the values @xmath98 dominate and we get : @xmath99 fig .
4 . _ the equipotential lines at @xmath100 . the dashed line corresponds to @xmath101 _ in fig .
4 the equipotential lines are shown .
the behavior of the screened coulomb potential in the transverse plane was found numerically in @xcite . finally for @xmath102 expanding ( [ 143 ] )
we get : @xmath103 which coincides with the result obtained in @xcite where the expression for the photon polarization operator at @xmath104 was obtained as well .
the expression for the screened coulomb potential was obtained from the one - loop contribution to the photon polarization operator in the external magnetic field @xmath70 . in momentum space
it looks like : @xmath105 if higher loops contain the terms @xmath106 they will drastically change the shape of the potential in the coordinate space . to calculate the radiative corrections one should use the electron propagator in an external homogeneous magnetic field @xmath6 .
the spectral representation of the electron propagator is a sum over landau levels and for @xmath107 the contribution of the lowest level dominates @xcite : @xmath108 where the projector @xmath109 selects the virtual electron state with its spin oriented opposite to the direction of the magnetic field @xmath110 .
the contributions of the excited landau levels to @xmath111 yield a term in the denominator proportional to @xmath112 and they produce a correction of order @xmath113 in the denominator of ( [ 188 ] ) .
two kinds of terms contribute to the polarization operator at the two - loop level .
first , there are terms in the electron propagators which represent the contributions of higher landau levels . just like
in the one - loop case they produce corrections suppressed as @xmath114 in the denominator of ( [ 188 ] ) , i.e. terms of the order @xmath115 which can be safely neglected in comparison with the leading @xmath116 term .
second , there is the contribution from the leading term in the electron propagator , given by ( [ 189 ] ) .
let us consider the simplest diagram : the photon dressing of the electron propagator . neglecting the electron mass we get : @xmath117= -2\hat k_{0,3}(1+i\gamma_1 \gamma_2 ) \;\ ; , \label{190}\ ] ] which gives zero when multiplying external propagator of electron , since @xmath118 .
this result is a manifestation of the following well - known fact : in massless qed in @xmath20 ( schwinger model ) all loop diagrams are zero except the one - loop term in the photon polarization operator ( see for example @xcite ) .
that is why the contributions of the second kind are of the order of @xmath119 and they are not important .
the generalization of the above arguments to higher loops is straightforward .
let us note that absence of higher loop corrections to polarization operator in schwinger model is related to the absence of renormalization of axial anomaly by higher loops . in @xmath20 anomaly
is given by correlator of two currents and axial current is proportional to vector current ( see for example @xcite ) .
we are interested in the spectrum of a hydrogen - like ion in a very strong magnetic field @xmath6 .
we will write all formulas for hydrogen since their generalization for @xmath120 is straightforward . in the absence of magnetic field the spatial size of the wave function of the ground state atomic electron
is characterized by the bohr radius @xmath121 , its energy equals @xmath122 , where @xmath123 is the rydberg constant .
the transverse ( with respect to @xmath6 ) size of the ground state of the electron wave function in an external magnetic field @xmath6 is characterized by the landau radius @xmath124 .
the larmour frequency of the electron precession is @xmath125 . for a magnetic field @xmath126 gauss called `` atomic magnetic field '' , these sizes and energies are close to each other : @xmath127 , @xmath128 .
we wish to study the spectrum of the hydrogen atom in magnetic fields much larger than @xmath33 . in this case
the motion of the electron is mainly controlled by the magnetic field : it makes many oscillations in this field before it makes one in the coulomb field of the nucleus .
this is the condition for applicability of the adiabatic approximation , used for this problem for the first time in @xcite .
the spectrum of a dirac electron in a pure magnetic field is well known @xcite ; it admits a continuum of energy levels due to the free motion along the field : @xmath129 where @xmath130 ; @xmath131 is the spin projection of the electron on @xmath77 axis multiplied by two . for magnetic fields larger than @xmath132 ,
the electrons are relativistic with only one exception : electrons belonging to the lowest landau level ( lll , @xmath74 , @xmath133 ) can be non - relativistic . in
what follows we will study the spectrum of electrons from lll in the coulomb field of the proton modified by the superstrong @xmath6 .
the solution can be found in @xcite of the schrdinger equation for an electron in a constant in time homogeneous magnetic field @xmath6 in the gauge in which @xmath134 in cylindrical coordinates ( @xmath135 ) .
the electron energies are : @xmath136 where @xmath137 is the number of nodal surfaces , @xmath138 is the electron orbital momentum projection on the @xmath77 axis ( direction of the magnetic field ) and @xmath131 . according to @xcite ,
the lll wave functions are : @xmath139^{-1/2 } \rho^{|m|}e^{\left(im\varphi - \rho^2/(4a_h^2)\right ) } \;\ ; , \label{17}\ ] ] @xmath140 we should now take into account the electric potential of the atomic nucleus located at @xmath141 . for @xmath142 the adiabatic approximation can be used and the wave function should be looked for in the following form : @xmath143 where @xmath144 is the solution of a schrdinger equation for an electron motion along the direction of the magnetic field : @xmath145 \chi_n(z ) = e_n \chi_n(z ) \;\ ; . \label{19}\ ] ] without screening the effective potential is given by the following formula : @xmath146 which becomes the coulomb potential for @xmath147 @xmath148 and is regular at @xmath91 @xmath149 to take screening into account we must use ( [ 11 ] ) to modify ( [ 20 ] ) ( see below ) . since @xmath150 ,
the wave functions are odd or even under reflection @xmath151 ; the ground states ( for @xmath152 ) are described by even wave functions . in fig .
5 the different scales important in the consideration of the hydrogen atom in strong magnetic field are shown .
_ landau radius @xmath12 versus magnetic field @xmath6 .
_ to calculate the ground state of hydrogen atom in @xcite the shallow - well approximation is used : @xmath153 ^ 2 = -(m_e e^4/2)ln^2(b/(m_e^2e^3 ) ) \label{27}\ ] ] let us derive this formula .
the starting point is the one - dimensional schrdinger equation : @xmath154 neglecting @xmath155 in comparison with @xmath156 and integrating ( [ 28 ] ) we get : @xmath157 where we assume @xmath158 , that is why @xmath159 is even .
the next assumptions are : 1 .
the finite range of the potential energy : @xmath160 for @xmath161 ; 2 .
@xmath159 undergoes very small variations inside the well . since outside the well @xmath162
, we readily obtain : @xmath163 ^ 2 \;\ ; . \label{30}\ ] ] for @xmath164 ( condition for the potential to form a shallow well which means that the absolute value of the energy of ground level is much smaller than the absolute value of the potential in the well ) we get that , indeed , @xmath165 and that the variation of @xmath159 inside the well is small , @xmath166 .
concerning the one - dimensional coulomb potential , it satisfies this condition only for @xmath167 .
this explains why the accuracy of @xmath168 formula ( [ 27 ] ) is very poor .
much more accurate equation for atomic energies in strong magnetic field was derived by b.m.karnakov and v.s.popov @xcite .
it provides a several percent accuracy for the energies of even states for @xmath169 ( @xmath170 ) .
main idea is to integrate shrdinger equation with effective potential from @xmath171 till @xmath172 , where @xmath173 and to equate obtained expression for @xmath174 to the logarithmic derivative of whittaker function - the solution of shrdinger equation with coulomb potential , which exponentially decreases at @xmath175 . in this way in @xcite
the following equation was obtained : @xmath176 @xmath177 where @xmath178 is the logarithmic derivative of the gamma function .
the energies of the odd states are : @xmath179 so , for superstrong magnetic fields @xmath180 the deviations of odd states energies from the balmer series are negligible . from ( [ 32 ] ) we get an equation for @xmath181 : @xmath182 where @xmath178 has simple poles at @xmath183 .
so to reproduce large number at left hand side @xmath181 should be large ( ground level ) or follow balmer series ( excited levels ) . when screening is taken into account an expression for effective potential transforms into @xmath184
\;\ ; \label{35}\ ] ] screening modifies the coulomb potential at the distances @xmath185 and since at these distances @xmath186 , the approach leading to ( [ 32 ] ) still works . the modified karnakov - popov equation , which takes screening into account looks like : @xmath187 we see that at @xmath188 freezing of the energies occur : left hand side of ( [ 37 ] ) approach constant when @xmath6 further grows . in particular , for a ground state at @xmath189 we obtain : @xmath190 , @xmath191 kev . energy levels on which lll is splitted in the hydrogen atom at @xmath189 are shown on fig .
_ spectrum of hydrogen levels in the limit @xmath189 .
energies are given in rydberg units , @xmath192_.
in the previous section the spectrum of energies on which the lowest landau level ( lll ) splits in the proton electric field was found by solving the corresponding schrdinger equation . since the ground state energy of hydrogen in the limit of infinite @xmath6 equals @xmath191 kev ,
the use of the nonrelativistic schrdinger equation is at least selfconsistent .
however , the size @xmath194 of the electron wave function for @xmath195 in the direction transverse to the magnetic field is much smaller than the electron compton wavelength , @xmath196 , which makes the nonrelativistic approach a bit suspicious ( @xmath197 for @xmath198 ) . that is why in this section we will study the ground state energy of the electron in a hydrogen - like ion in the presence of an external magnetic field by analyzing the dirac equation . without taking screening into account this problem
was considered in paper @xcite ( see also @xcite , were results obtained in @xcite were reproduced ) , soon after it was found that a hydrogen - like ion becomes critical at @xmath199 : the electron ground level sinks into the lower continuum ( @xmath200 ) and the vacuum becomes unstable by spontaneous @xmath201 pairs production .
these results were obtained by solving the dirac equation for an electron moving in the field of a nucleus of finite radius . that the phenomenon of criticality can be studied only in the framework of the dirac equation
is an additional motivation for us to go from schrdinger to dirac . from the numerical solution of the dirac equation for the ground electron level of a hydrogen atom in the coulomb potential we will find that the corrections to the nonrelativistic results are small and that the estimate @xmath202 works well .
let us parametrize bispinor which describes electron wave function in the following way : @xmath203 substituting @xmath204 in the dirac equation for the electron in an external electromagnetic field we obtain : @xmath205 taking vector potential which describes constant magnetic field @xmath6 directed along @xmath77 axis in the form @xmath206 , we get : @xmath207 where @xmath208 , @xmath209 .
analogously we obtain : @xmath210 substituting two last expressions in the dirac equation we get : @xmath211 axial symmetry of electromagnetic field allows to determine @xmath212 dependence of the functions @xmath213 and @xmath214 : @xmath215 where @xmath216 , ... is the projection of electron angular momentum on @xmath77 axis . substituting ( [ 106 ] ) in ( [ 105 ] ) we get four linear equations for four unknown functions @xmath213 and @xmath214 ( here and below @xmath217 , @xmath218 ... ): @xmath219 where @xmath220 , @xmath221 , ...
ground energy state has @xmath222 , @xmath223 .
taking @xmath224 we should look for solution of ( [ 107 ] ) with @xmath225 : @xmath226 @xmath227 the dependence on @xmath96 is determined by ( [ 108 ] ) : @xmath228 substituting the last expressions in ( [ 109 ] ) and averaging over fast motion in transverse to the magnetic field plane we obtain two first order differential equations which describes electron motion along magnetic field in an effective potential @xmath229 : @xmath230 @xmath231 at large distances @xmath232 the effective potential equals coulomb , and the solutions of the equations ( [ 111 ] ) exponentially decreasing at @xmath233 are linear combinations of the whittaker functions . at short distances the equations ( [ 111 ] ) can be easily integrated for @xmath234 ( as far as @xmath235 condition for @xmath236 will be for sure valid for @xmath237 , which is equivalent to the following inequality : @xmath238 ) , where they looks like : @xmath239 the result of the integration is : @xmath240 where @xmath241 and @xmath242 , @xmath243 are normalization constants .
the functions @xmath244 and @xmath245 have opposite parities ; for the ground state @xmath244 should be even , so @xmath246 , and matching logarithmic derivatives at the point @xmath247 we obtain : @xmath248 @xmath249 substituting proper combination of the whittaker functions for @xmath244 we obtain an algebraic equation for the ground state energy ( it coincides with eq .
( 22 ) in @xcite in the limit @xmath250 , where @xmath251 is the nucleus radius ) : @xmath252 where @xmath253 is the euler constant , and the argument of the gamma function is given by @xmath254 for the ground level at @xmath255 one should take @xmath74 , while for @xmath256 it should be changed to @xmath257 .
according to ( [ 114 ] ) when the magnetic field increases the ground state energy goes down and reaches the lower continuum .
a matching point exists only if @xmath258 ( see ( [ 113 ] ) ) and ( [ 114 ] ) is valid only for these values of the magnetic field .
thus , without taking screening into account , from ( [ 114 ] ) we can obtain the dependence of the ground state energy of a hydrogen atom on the magnetic field for @xmath259 .
screening modifies the coulomb potential at distances smaller than the electron compton wave length , and from the condition @xmath260 we get @xmath261 .
it means that at @xmath262 the phenomena of screening does not allow to find analytically the ground state energy . in order to find the ground state energy at @xmath263 and to take screening into account the equations ( [ 111 ] ) were solved numerically .
this system can be transformed into one second order differential equation for @xmath244 . by substituting @xmath264 a schrdinger - like equation for the function @xmath265
was obtained in @xcite : @xmath266 @xmath267 where @xmath268 is the energy eigenvalue of the dirac equation and @xmath229 is given in ( [ 112 ] ) . an equation ( [ 116 ] ) was integrated numerically .
let us note that , while for @xmath269 the last three terms in the expression for @xmath156 are much smaller than the first one ( the only one remaining in the nonrelativistic approximation ) , at @xmath270 the relativistic terms dominate and are very big for @xmath70 at @xmath271 which makes numerical calculations very complicated .
.[tab:1]values of @xmath181 for @xmath28 without screening obtained from the schrdinger and dirac equations .
they start to differ substantially at enormous values of the magnetic field . [ cols="^,^,^,^,^ " , ] textbooks @xcite contain detailed consideration of the phenomenon of critical charge .
an analytical formula for the coulomb potential @xmath272 in a superstrong magnetic field has been derived .
it reproduces the results of the numerical calculations made in @xcite with good accuracy . using it , an algebraic formula for the energy spectrum of the levels of a hydrogen atom originating from the lowest landau level in a superstrong
@xmath6 has been obtained .
the energies start to deviate from those obtained without taking the screening of the coulomb potential into account at @xmath273 gauss and the energy of ground state in the limit @xmath274 remains finite .
a magnetic field plays a double role in the critical charge phenomenon . by squeezing the electron wave function and putting it in the domain of a stronger coulomb potential it diminishes the value of the critical charge substantially @xcite .
however , for nuclei with @xmath18 to become critical such a strong @xmath6 is needed that the screening of the coulomb potential occurs and acts in the opposite direction : the electron ground state energy freezes and the nucleus remains subcritical in spite of growing @xmath6
. 99 v.b .
berestetskii , e.m .
lifshitz , and l.p .
pitaevsky , quantum electrodynamics , theoretical physics , v. iv , m. : fizmatlit , 2001 , p. 562 .
shabad , and v.v .
usov , phys .
* 98 * ( 2007 ) 180403 ; phys . rev .
* d77 * ( 2008 ) 025001 .
vysotsky , pisma v zhetf * 92 * ( 2010 ) 22 [ jetp lett .
92 ( 2010 ) 16 ] .
b. machet and m.i .
vysotsky , phys .
d * 83 * , 025022 ( 2011 ) .
godunov , b. machet , and m.i .
vysotsky , phys .
d * 85 * , 044058 ( 2012 ) .
j. schwinger , phys . rev . * 128 * ( 1962 ) 2425 .
akhiezer , and v.b .
berestetskii , quantum electrodynamics m. : nauka , 1981 , appendix * a2 * , p. 422 .
j. schwinger , phys .
* 82 * ( 1951 ) 664 .
g. calucci and r. ragazzon , j. phys .
a * 27 * ( 1994 ) 2161 .
skobelev , izvestiya vysshikh uchebnykh zavedenii , fizika * 10 * ( 1975 ) 142 [ russian physics journal , v.18 , n.10 ( 1975 ) 1481 ] ; + yu.m .
loskutov , and v.v .
skobelev , phys . lett .
* a56 * ( 1976 ) 151 .
a. chodos , k. everding , and d.a .
owen , phys .
d * 42 * ( 1990 ) 2881 ; + v.p .
gusynin , and a.v .
smilga , phys .
b * 450 * ( 1999 ) 267 . v.b .
berestetskii , proceedings of liyaf winter school * 9 , part 3 * , 95 ( 1974 ) .
peskin , and d.v .
schroeder , an introduction to quantum field theory , addison - wesley , 1995 .
shiff , and h. snyder , phys
* 55 * ( 1939 ) 59 .
berestetskii , e.m .
lifshitz , and l.p .
pitaevsky , quantum electrodynamics , theoretical physics , v. iv , m. : fizmatlit , 2001 , p. 147 .
landau , and e.m .
lifshitz , quantum mechanics , theoretical physics , v. iii , m : fizmatlit , 2001 , p. 556
, problem 1 .
landau , and e.m .
lifshitz , quantum mechanics , theoretical physics , v. iii , m : fizmatlit , 2001 , p. 558
, problem 3 .
karnakov , and v.s .
popov , zhetf * 124 * ( 2003 ) 996 [ j. exp .
physics * 97 * ( 2003 ) 890 ] ; zhetf * 141 * ( 2012 ) 5 .
oraevskii , a.i .
rez , and v.b .
semikoz , zh .
* 72 * ( 1977 ) 820 [ sov .
jetp * 45 * ( 1977 ) 428 ] .
p. schlter , g. soff , k .- h .
wietschorke , and w. greiner , j. phys .
b : at . mol .
phys . * 18 * ( 1985 ) 1685 .
popov , pisma zh .
* 11 * ( 1970 ) 254 ; + v.s .
popov , zh .
* 59 * ( 1970 ) 965 ; + ya.b .
zeldovich , and v.s .
popov , ufn * 105 * ( 1971 ) 403 ; + v.s .
popov , yad .
* 14 * ( 1971 ) 458 .
w. greiner , and j. reinhardt , `` quantum electrodynamics '' springer - verlag ( 1992 ) berlin , heidelberg ; + w. greiner , b. mller , and j. rafelski , `` quantum electrodynamics of strong fields '' springer - verlag ( 1985 ) berlin , heidelberg . | the expanded variant of the lectures delivered at the 39th itep winter school in 2011 . |
the large-@xmath0 limit @xcite , while sharing several qualitative characteristics with qcd , exhibits significant simplifications . for this reason ,
it is a very popular technique for qualitative and semi - quantitative estimates of crucial qcd observables .
for instance , the string - gauge correspondence is profoundly intertwined with the large @xmath0 limit , since one of the requirements for a calculable dual holographic description is large @xmath0 @xcite . yet
, the existence of a smooth limit between @xmath1 and @xmath2 is a conjecture which , for some observables , is _ known _ to be false . we know that , at zero chemical potential , @xmath3 and finite @xmath4 deconfinement has to be a phase transition . at @xmath5 lattice qcd seems to indicate it is a cross - over @xcite . at finite chemical potential , in the confining phase , nuclear matter is known to be a strongly bound crystal in the large @xmath0 limit @xcite .
it seems to be a liquid at @xmath1 .
this means that , before predictions made at large @xmath0 can be made qualitatively reliable , the phase diagram in @xmath6 space needs to be explored .
the most current phenomenological application for these ideas is the existence of the so - called `` quarkyonic matter '' @xcite .
this talk explores these ideas , summarizing @xcite and drawing connections between these works .
our first investigation of these issues @xcite is to use the van der waals ( vdw ) ansatz to see under what conditions can we obtain both the familiar nuclear matter phase diagram at @xmath1 and the commonly accepted large @xmath0 phase diagram at @xmath7 @xcite . while the van der waals gas is certainly a very rough description , it is part of a universality class , to which the nuclear liquid - gas phase transition seems to belong . in the large @xmath0 limit ,
the only @xmath0-invariant scale of the theory is @xmath8 , the scale at which the t hooft coupling constant becomes @xmath9 . while a precise value of this scale depends on the scheme used to calculate it , its roughly @xmath10 mev @xcite .
it is therefore natural to expect that any physical quantity is @xmath11 , a dimensionless function of @xmath0 times a power of @xmath8 set by the dimensionality @xmath12 of the quantity .
henceforward we shall adopt this assumption , and , for brevity , set @xmath8 to unity in the equations . in this notation ,
the van der waals parameters @xmath13 , @xmath14 and the curvature correction become dimensionless @xmath15,@xmath16,@xmath17 times the appropriate power of @xmath8 ( 3 for @xmath15,2 for @xmath16,4 for @xmath17 ) , and the vdw equation becomes @xmath18 we start by noting that , naturally , @xmath15 can only go to @xmath19 at @xmath7 , and is significantly larger than @xmath19 at @xmath1 .
therefore , @xmath20 where @xmath21 is a constant to be determined .
the coefficients @xmath22 should , according to @xcite go as @xmath0 .
recent work @xcite , however , has cast doubt on this assumption and proposed they go as @xmath23 or @xmath24 .
the chemical potential can be obtained by the textbook thermodynamic relation @xmath25 .
inverting , and writing in terms of @xmath26 we have @xmath27\ ] ] where the first term is the nucleon mass and @xmath28 @xmath29 and @xmath30 are the density at the phase transition , which could be liquid @xmath31 or gas @xmath32 ( if the calculation is performed correctly the same chemical potential should come out ) .
@xmath33 are in turn the solutions to the equation [ vdw ] in the region where this equation has two solutions ( of which one is thermodynamically unstable ) .
obtaining all such solutions is trivial at the mathematical level through algebraically cumbersome .
the reader can get the detailed results in @xcite . as can be seen ( fig .
[ phasemu ] ) , to interpolate between the currenlty accepted nuclear liquid phase diagram at @xmath1 ( @xmath34 ) and the `` quarkyonic '' phase diagram at @xmath7 ( @xmath35 ) , @xmath36 must be used .
how would we interpret physically such a value ?
above all , what is the physical meaning of @xmath21 ?
we conjecture that @xmath21 tracks another dimensionless scale relevant at high density : the number of neighbors a nucleon has in a tightly packed nuclear material .
@xmath21 , of course , is a function not of @xmath0 , but the ( fixed ) number of dimensions @xmath12 and `` packing scheme '' .
the sensitivity of the varying @xmath0 equation of state to @xmath21 is undestandable due to the quantum nature of the problem and the fermionic nature of the quarks .
the more neighbors , the more pauli blocking of valence quarks must be important , and the more the presence of neighbors will disturb the configuration space part of the quark wavefunction inside the nucleons . since , due to the uncertainty principle , any such disturbance of the nuclear wavefunction adds an energy of the order of the confinement scale @xmath37 , the nuclear repulsive core will be larger than the inverse of the nuclear separation up to the deconfinement temperature . if the number of colors is larger than @xmath21 , this problem will not exist since it will be possible to arrange the color part of the wavefunction so the nearest quarks of neighboring baryons will be of different colors . in this limit
baryons can be tightly packed ( interbaryonic separation @xmath37 ) without the configuration space part of the baryonic wavefunction being disturbed .
space ( left panel ) and @xmath38 space ( right panel ) as a function of @xmath0 ( @xmath39 , with increasing color corresponding to a line with higher @xmath40 ) top panels assume nuclear interactions @xmath23 or @xmath24 , bottom panels as @xmath41 , title="fig:",scaledwidth=40.0% ] space ( left panel ) and @xmath38 space ( right panel ) as a function of @xmath0 ( @xmath39 , with increasing color corresponding to a line with higher @xmath40 ) top panels assume nuclear interactions @xmath23 or @xmath24 , bottom panels as @xmath41 , title="fig : " ] space ( left panel ) and @xmath38 space ( right panel ) as a function of @xmath0 ( @xmath39 , with increasing color corresponding to a line with higher @xmath40 ) top panels assume nuclear interactions @xmath23 or @xmath24 , bottom panels as @xmath41 , title="fig:",scaledwidth=40.0% ] space ( left panel ) and @xmath38 space ( right panel ) as a function of @xmath0 ( @xmath39 , with increasing color corresponding to a line with higher @xmath40 ) top panels assume nuclear interactions @xmath23 or @xmath24 , bottom panels as @xmath41 , title="fig : " ] the number of neighbours scaling suggests that percolation phenomena might be relevant in the middle of the two regimes . in the next section
we investigate this possibility .
the key insight suggesting that interesting structures might be lurking in @xmath0 is that 3d bond - percolation exhibits a phase transition at comparatively low critical link probability : for instance , @xmath42 for simple - cubic , body - centered - cubic and hexagonal - close - packed lattice , respectively @xcite .
such values suggest that long - distance correlations on the quark level could occur even with a somewhat low percentage of quarks hopping between baryons , i.e. firmly in the confined phase . while below @xmath43 the characteristic correlation distance @xmath44 ( @xmath45 cluster size ) is @xmath46 , above the threshold this quantity explodes to the total system size in a comparable amount of time .
we leave the meaning of `` correlation '' vague , as it can be either a quark hop or a gluon exchange ; in our context , it implies exchange of colour degrees of freedom within a confined tightly packed medium .
we encode the likelihood of exchange between neighbouring baryons in a link probability @xmath47 , to be compared with the percolation threshold @xmath43 in order to assess the formation of large - scale structures .
two baryons will be correlated if _
at least two _ quarks are correlated .
one has to sum over all possible multi - quark configurations , resulting in a strong @xmath0-dependence of @xmath47 .
we determine the latter by calculating the probability @xmath48 that _ no _ exchanges happen between neighbouring baryons . assuming the quarks inside the nucleon are uncorrelated ( fermi motion dominates )
, this probability factorises into a geometric distribution @xmath49 for quarks to be at a certain ( vector ) position @xmath50 , and a `` squared propagator '' transition amplitude @xmath51 for them : @xmath52 we assume a `` hard - sphere '' distribution for @xmath53 ( since we keep @xmath54 fixed , the distance between centres of neighbouring baryons is always @xmath55 ) @xmath56 and a probability of exchange @xmath57 based on a range of `` reasonable '' propagators , compatible with confinement ( fast fall - out in configuration space at distances greater than @xmath58 in units of @xmath59 ) and with the large @xmath0 limit of qcd , the interaction is @xmath60 @xcite .
the propagators we use are the simple @xmath61-function in configuration space and the momentum - space @xmath61-function used in @xcite , all normalised so their area is @xmath62 where @xmath63 is the range of propagation ( @xmath46 ) . in configuration space
the transition amplitudes are , respectively , @xmath64 other transition amplitudes , such as a gaussian distribution in configuration space , were also tried with no significant modifications .
this is unsurprising since , due to the fact that 3d percolation has a second - order phase transition at a certain @xmath65 , the results we obtain below have some degree of universality : as long as the qualitative features of confinement are observed ( the transition amplitude @xmath66 drops sharply above the scale @xmath63 , and the hadron density profile @xmath49 has a central plateau of radius @xmath46 and a sharply decreasing tail outside ) , the results we show vary quantitatively but not qualitatively .
the crucial parameter left is @xmath67 in eq .
[ ptotlink ] .
one can easily see that @xmath68 is in contradiction with the skyrme crystal picture at large @xmath0 : in this picture , @xmath69 approaches a constant large-@xmath0 value _ from above _ : _
low @xmath0 _ nuclear matter would be more correlated ( and hence more strongly bound ) than _ high @xmath0 _ nuclear matter . comparing strongly coupled @xmath7 nuclear matter @xcite to the weakly bound nuclear liquid at @xmath1 @xcite ,
this is obviously not right .
we therefore assume @xmath70 henceforward , natural if the link is actually realised by a gluon exchange rather than a quark flip ( it is obvious this does indeed dominate at large @xmath0 ) . in the large @xmath0 limit for the case @xmath70 ,
@xmath47 asymptotically approaches unity .
it is reasonable that this is the point where the `` dense baryonic matter as a skyrme crystal '' , theorised in @xcite , is reached . if this is the case , however , one should remember that a percolation second - order phase transition occurs at a @xmath71 .
hence , keeping @xmath72 fixed but varying @xmath0 , the features of the skyrme crystal should manifest not with a continuous approach , rather as a second order transition at a not too high @xmath0 , whose order parameter can be though to be the `` giant cluster '' density . below the critical @xmath0
there is little correlation between quarks of different baryons , while above this threshold they can correlate , with the distance boundary given only by causality .
we reiterate that this is _ not _ deconfinement since @xmath73 independently of the number of colours , and the _ fraction _ of correlated quarks from different hadrons is still @xmath74 at the percolation transition . right above this transition , therefore , the baryonic wavefunction should _ not _ be too different from the large-@xmath0 baryonic wavefunction described in @xcite .
the _ correlation distance _ of quarks will however be much larger than the baryon size .
the features of this new phase are therefore similar to those of the quarkyonic matter @xcite . assuming the lowest 3d value @xmath75 , appropriate for a closely packed hexagonal lattice , the critical number of colours is shown in fig .
[ lottini ] as a function of @xmath76 and @xmath63 . as can be seen ,
the critical number of colours is significantly larger than three for @xmath77 . considering fig .
[ lottini ] is a _ lower limit _ since @xmath43 is at its minimum ( @xmath43 is significantly higher both in a skyrme cubic crystal and in a disordered fluid ) , we can say that @xmath1 is disfavoured , although it can not be excluded .
changing temperature and @xmath54 should further change the critical @xmath0 .
exploring this parameter space , and seeing how it relates to the confinement phase transition , is the subject of a forthcoming work .
what are the phenomenological consequences of percolation ? if by `` correlation '' we mean energy - momentum - exchange via quark tunneling between baryons , it is reasonable that pressure and entropy density @xmath41 above the percolation threshold , while below it they stay @xmath23 .
this is because above the threshold , where interbaryon tunnelling is significant , `` typical '' excitations of the fermi surface will be superpositions across baryons of baryon - localized quark - hole excitations , similar of conduction band electrons in a metal ; while the localized excitation energy @xmath78 , the superposition makes its energy @xmath37 even if color degeneracy remains .
thus , the degrees of freedom of the system above percolation will be delocalized weakly interacting quarks in a lattice of confining potentials , a picture compatible with @xcite . below the threshold , where tunnelling is negligible
, excitations are either color singlets or of energy @xmath79 , suppressed below deconfinement .
the transitions discussed here might also be visible with holographic techniques _ beyond _ the supergravity limit , since finite @xmath80 corresponds , in gauge / string duality , to the string coupling constant @xmath81 @xcite .
thus , a transition at finite @xmath0 will manifest itself as a transition between a classical supergravity regime and a quantum - gravitational regime .
since gravity to one loop still has not been solved for backgrounds dual to confining theories , we can not rigorously say anything further .
however , in @xcite , we have made a first step in that direction using the reasonable ansatz , developed earlier @xcite , that the effect of quantum gravity corrections is to smoothen the black hole singularity to a gaussian of with @xmath82 , the plank length .
the problem of the first order quantum black hole reduces , than , to the problem of the classical general relativity sourced by a gaussian distribution of energy density @xmath29 .
the top energy - momentum tensor is then @xmath83 with the other components of @xmath84 being fixed by isotropy , the classical einstein equation @xmath85 and conservation laws ( @xmath86 ) .
thermodynamics can be obtained by the usual relationship between the entropy @xmath87 and the horizon area @xmath88 , @xmath89 , as well as maxwell s relations @xcite .
we rely on @xcite for details on solving this problem , and connecting it to the hawking - page phase transition @xcite , and just give the result : @xmath90 where @xmath91 is the free - energy of a classical black hole in ads space @xcite and @xmath92 once we examine the minimum structure of this free energy , we find a critical point : for a `` critical '' @xmath93 there is a critical point after which the hawking - page phase transition becomes a _
cross - over _ ( fig .
[ nicolini ] ) . the resulting phase diagram ( in @xmath94 vs @xmath82 )
is shown to be in the universality class of the van der waals gas .
the relevance of this paper to the topic at hand becomes apparent given that the hawking - page transition is thought to be dual to the deconfinement transition @xcite .
we know that at finite @xmath95 , deconfinement becomes a cross - over rather than a transition .
thus , this calculation can be thought of as an estimate of the critical @xmath95 for the deconfinement critical point in two - dimensions ( since the gravitatational setup is in 3d ) .
while the @xmath96 case presented here is considerably different from the @xmath97 or the @xmath98 background examined in @xcite , our results fall neatly into the van der waals universality class . because of that
, and the structure of the calculations in the previous section , we expect that the general structure of this phase diagram will be maintained in @xmath97 or any other background where a deconfinement phase transition exists .
the critical @xmath99 will of course be different , but not the behavior around the critical point .
this critical point will translate itself in the behaviour of @xmath0 of the boundary theory , once again in a way that is universal around the critical point but sensitive to the exact nature of the background theory .
when the number of colors @xmath100 the number of flavors @xmath4 ( the so - called thooft limit ) , confinement is rigorously known to be a phase transition in all dimensions greater than one , since the symmetries of the system are expected to change .
the low temperature phase will have acquired an extra @xmath101 symmetry , signifying a zero expectation value of the polyakov loop @xcite . in the high - temperature phase
the polyakov loop will aquire a finite expectation value , and @xmath101 will be spontaneusly broken . in the gravity picture , this can be thought of as a manifestation of the widely - believed cosmic censorship conjecture @xcite , the idea that any singularity in spacetime is surrounded by an event horizon ( and hence a discontinuity in the energy density at the boundary theory ) .
since the existance of the black hole tracks the confining polyakov loop behaviour in @xcite , making confinement into a cross - over would mean smoothening the black hole singularity .
this can only be done , eg , by our ansatz , for @xmath102 . at finite number of flavors @xmath4 ,
the @xmath101 symmetry is no longer exact at a fundamental level .
hence , it is natural to expect that at some critical @xmath103 `` a critical point '' for deconfinement appears , where for higher @xmath0 deconfinement is a phase transition and for lower @xmath0 it is a crossover .
this is indeed what emerges from lattice simulations @xcite in three dimensions with a finite number of flavors .
the free energy with respect to the polyakov loop expectation value @xmath104 and energy @xmath105 is therefore of the form @xmath106 @xmath107 at the mixed phase is a function with two local minima , corresponding to the two phases .
for high - enough @xmath95 , however , @xmath108 could destroy one of the minima and make the phase transition into a crossover .
the van der waals form of the free energy eq .
[ freen ] exactly parallels eq .
[ freengauge ] , with the @xmath109 parameter playing the role of @xmath95 in a gauge theory in 2 + 1 dimensions ( since the bulk calculated here is 3 + 1 dimensions ) .
the free energy @xmath110 in units of the cosmological constant @xmath111 for the system in equilibrium as a function of the schwarzschild radius @xmath112 from top to bottom for , in terms of @xmath113 , @xmath114 and @xmath115 ( solid lines ) .
the dashed lines represent the critical case @xmath116 , while the dotted line is the classical hawking - page phase transition ( @xmath117 ) .
see @xcite for details , height=207 ]
the preceding three sections , at first sight , look like they deal with very different subjects .
indeed , the current status is that any correlation between them is conjectural .
when @xmath0 is varied in the dense matter strongly coupled regime , deviations from the `` classical '' skyrme crystal arise due to the interplay between the number of colors and the number of neighbours @xmath21 in the densely packed system .
this `` geometrical '' picture is explored rigorously , in terms of an analytical model , in section [ seclottini ] and shown to give a percolation - type phase transition .
a similar behaviour can be argued in terms of the pauli exclusion principle , and if _ it is assumed _ , as shown in section [ secmishustin ] , leads to the transition between the loosely - bound nuclear liquid at @xmath1 and the tightly bound nucelar crystal at @xmath118 . since the solid - liquid transition can be understood as a percolation of crystal bonds @xcite , it is therefore compelling to identify the percolation transition with the physics described in section [ secmishustin ] , with the critical @xmath0 playing the role of `` critical point '' in @xmath6 space ( the transition in section [ seclottini ] is necessarily second order ) .
the connection with the hawking - page transition described in section [ secnicolini ] is , at the moment , not so clear .
it is worth pointing out , however , that this transition is also in the van der waals universality class .
while section [ secnicolini ] was at zero chemical potential ( the black hole was not charged ) , the hawking - page phase diagram for classical ( @xmath119 ) charged black holes also falls into this class @xcite .
thus , the phenomena examined at finite chemical potential in section [ seclottini ] and [ secmishustin ] might be analogous to the dynamics examined in section [ secnicolini ] at zero chemical potential .
the closeness of the critical @xmath99 to the critical number of @xmath120 required for percolation in 2d @xcite ( 6 colors per flavor ) is in this sense intriguing .
physically , in @xmath96 , the hawking - page transition coincides with the transition from a gas of black holes to one large black hole .
it is plausible that the critical point to a hawking - page cross - over occurs because the black holes in the gas are actually linked by super - horizon quantum gravitational interactions , modeled in our approach by `` smearing '' the super - horizon black hole distance by a planck - sized width . in this scenario ,
the connection of percolation to the cross - over in deconfinement is straight - forward . before this connection
can be made more rigorous , however , a gauge / string setup where percolation is relevant must be examined with a `` smearing '' analogous to section [ secnicolini ] .
work of this kind is in progress .
i m acknowledges support provided by the dfg grant 436rus 113/711/0 - 2 ( germany ) and grant ns-7235.2010.2 ( russia ) .
g.t . acknowledges the financial support received from the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich - konomischer exzellenz ) launched by the state of hesse .
we acknowledge the organizers of the max born symposium for their generous support and for the very stimulating atmosphere at this workshop .
e. witten , nucl .
b * 160 * , 57 ( 1979 ) .
j. m. maldacena , adv .
* 2 * , 231 ( 1998 ) ; + witten , e. adv .
* 2 * 253 - 591 ( 1998 ) + witten , e. adv .
* 2 * 505 - 532 ( 1998 ) y. hidaka , t. kojo , l. mclerran and r. d. pisarski , nucl .
a * 852 * , 155 ( 2011 ) c. d. lorenz , r. may , r. m. ziff , j. stat .
* 98 * 34 , 961 ( 2000 ) ; c. d. lorenz , r. m. ziff , phys .
e * 57 * , 230 ( 1998 ) . | we examine the phase diagram of hadronic matter when the number of colors , as well as temperature and density , are varied .
we show that in this regime several phase transitions are possible , and we examine issues related to these transitions . |
the discovery of superconductivity in the layered cobalt oxide na@xmath10co@xmath11yh@xmath12o @xcite has stimulated further interest in the electronic states of the mother compound na@xmath1coo@xmath0 .
the specific features of the system are the geometrical fluctuation of the coo@xmath0 plane which consists of a triangular lattice of co atoms , and the orbital degeneracy of @xmath13 bands of co @xmath14 electrons .
the electron filling in the coo@xmath0 plane is controlled by changing the na content @xmath15 : the hole concentration of co @xmath13 bands is given by @xmath16 .
for @xmath17 , na@xmath1coo@xmath0 is a normal paramagnetic metal , while , for @xmath18 , an anomalous behavior is observed : the magnetic susceptibility @xmath19 is curie - weiss like although the resistivity is metallic @xcite , the electronic specific heat coefficient @xmath20 is large and increases with @xmath15 @xcite , the thermopower is unusually large @xcite . for @xmath21 , a weak magnetic order is observed below 20k @xcite , where the ferromagnetic ordered coo@xmath0 layers couple antiferromagnetically with each other @xcite .
furthermore , na@xmath22coo@xmath0 exhibits a remarkable successive phase transition at @xmath23 k and @xmath24 k @xcite : the in - plane antiferromagnetic order is realized below @xmath25 and the system becomes insulator below @xmath26 . from the nmr and the neutron measurements , yokoi _ et .
_ @xcite have proposed the magnetic structure of na@xmath22coo@xmath0 , where chains of co@xmath27 with larger staggered moment and co@xmath28 with smaller moment exist alternatively within the coo@xmath0 plane .
the charge ordering of the co sites into chains of co@xmath27 and co@xmath28 is closely related to the ordered pattern of na ions which form one - dimensional chains below room temperature @xcite . in spite of the intense effort devoted in the last few years , the origin of metal - insulator transition at @xmath26 is still an open question .
the purpose of this paper is to investigate the electronic state of the coo@xmath0 plane including the successive phase transition of na@xmath22coo@xmath0 , particularly focused on the effect of the one - dimensional na order on the antiferromagnetism . for this purpose , we take into account of an effective one - dimensional potential on the coo@xmath0 plane due to the effect of the na order .
the contents of this paper are as follows ; in sec .
2 , we describe the model hamiltonian for the coo@xmath12 plane . in sec . 3 ,
results of the calculation are presented . finally , we summarize the results of the present work in sec .
to investigate the electronic states of the coo@xmath0 plane in the layered cobalt oxides na@xmath10coo@xmath0 , we employ the two dimensional triangular lattice @xmath29-@xmath30 model which includes 11 orbitals : @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 of co and @xmath36 , @xmath37 , @xmath38 ( @xmath39 , @xmath40 , @xmath41 ) of o in the upper ( lower ) side of a co plane .
the hamiltonian is given by ; @xmath42 where @xmath43 ( @xmath44 ) is a creation operator for a cobalt @xmath14 electron with wave vector @xmath45 ( site @xmath46 ) , orbital @xmath47 ( @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 ) and spin @xmath53 , and @xmath54 is a creation operator for a oxygen @xmath55 electron with wave vector @xmath45 , site @xmath56 , orbital @xmath57 ( @xmath15 , @xmath58 , @xmath59 ) and spin @xmath53 , respectively . in eqs .
( [ eq - model2 ] ) , ( [ eq - model3 ] ) and ( [ eq - model4 ] ) , the transfer integrals @xmath60 , @xmath61 and @xmath62 , which are written by the slater - koster parameters , together with the atomic energies @xmath63 and @xmath64 are determined so as to fit the tight - binding energy bands to the lda bands for na@xmath22coo@xmath0 @xcite , and to be suitable for the result of arpes @xcite parameters are slightly adjusted . in eq .
( [ eq - model6 ] ) , the sum run over the nearest neighbor pairs @xmath65 over all lattice sites . in the hamiltonian eq .
( [ eq - model ] ) , we consider the effects of the intra - atomic coulomb interaction at a co site : the intra- and inter - orbital direct terms @xmath2 and @xmath3 , the exchange coupling @xmath4 and the pair - transfer @xmath5 @xcite , and effects of the inter - atomic coulomb interaction between the nearest neighbor co sites @xmath8 . here and hereafter , we assume the rotational invariance yielding the relations : @xmath66 and @xmath67 . the effect of the one - dimensional na order , in which na ions form one - dimensional chains at @xmath6 , is considered by taking into account an effective one - dimensional potential on the coo@xmath0 plane : @xmath68 with the effective potential @xmath7 due to the na order . because a part of the effect of the inter - atomic coulomb interaction @xmath8 enhances @xmath7 , @xmath8 is corrected so as not to influence @xmath7 .
hereafter , in order to consider the both na order and the antiferromagnetic order , we use an extended unit cell including four co atoms and eight o atoms , together with the magnetic brillouin zone which is @xmath69 of the original brillouin zone .
now we discuss possible ordered states including magnetic , charge and orbital ordered states in the case of @xmath6 within the hartree - fock approximation @xcite . in this study , we assume that the order parameters are diagonal with respect to the orbital @xmath47 and the spin @xmath53 .
figure [ fig - struct](a ) shows a possible order pattern of the antiferromagnetic state .
chains of co 3.5+@xmath70 and co 3.5-@xmath70 exist alternatingly within the coo@xmath0 plane because of the one - dimensional na order @xcite .
figure [ fig - struct](b ) shows an order pattern of the coexistence state of antiferromagnetic , charge and orbital order , where the co sites out of the na ordered line with the valency of 3.5+@xmath70 have the antiferromagnetic moment @xcite , while , the co sites on the na ordered line with the valency of 3.5-@xmath70 show the charge order which is accompanied by the orbital order out of the na ordered line .
the coexistence state of the antiferromagnetic and charge order without the orbital order has not been obtained in the present calculation .
figure [ fig - phase](a ) shows the sublattice magnetization @xmath47 , charge order @xmath71 , orbital order @xmath72 and the density of states at fermi level @xmath73 as functions of @xmath2 with @xmath74 ev , temperature @xmath75 k and the effective potential @xmath76 ev .
the phase transition of antiferromagnetic ordered state occurs at @xmath77 ev . with increasing @xmath2 , sublattice magnetization @xmath47 increases , while , the density of states @xmath73 is almost constant .
due to the effect of the one - dimensional potential , the band structure becomes quasi one - dimensional and the fermi surface nesting is enhanced as shown in fig .
[ fig - fermi0 ] .
then , the antiferromagnetic ordered state appears in the presence of the na order @xcite .
figure [ fig - phase](b ) shows @xmath47 , @xmath71 , @xmath72 and @xmath73 as functions of @xmath8 with @xmath78 ev , @xmath79 ev , @xmath75 k and @xmath76 ev .
the phase transition from the metallic antiferromagnetic ordered state to the insulating coexistence state is obtained at @xmath80 ev . at @xmath81 ,
the charge order at co sites on the na ordered line is caused by the effect of @xmath8 , and the anisotropy between @xmath32 and @xmath33 at co sites out of the na ordered line is caused by the charge order as shown in fig .
[ fig - struct ] .
, charge order @xmath71 , orbital order @xmath72 and the density of states @xmath73 are shown as functions of the intra - atomic coulomb interaction @xmath2 ( a ) and inter - atomic coulomb interaction @xmath8 ( b ) .
, title="fig:",height=151 ] , charge order @xmath71 , orbital order @xmath72 and the density of states @xmath73 are shown as functions of the intra - atomic coulomb interaction @xmath2 ( a ) and inter - atomic coulomb interaction @xmath8 ( b ) .
, title="fig:",height=151 ] in fig .
[ fig - fermi0 ] , the fermi surface and the renormalized band structure of the normal state are shown at @xmath82 , @xmath75 k and @xmath83 ev .
the effect of the one - dimensional potential @xmath7 due to the na order enhances the fermi surface nesting . in fig .
[ fig - fermi1 ] , we plot the fermi surface and the renormalized band structure in the antiferromagnetic ordered state at @xmath78 ev , @xmath79 ev , @xmath74 ev , @xmath75 k and @xmath83 ev .
this state is metallic because the electron like fermi surfaces around k ( k , k ) points and the hole like fermi surfaces around m point yield a large value of the density of states @xmath73 .
the fermi surface and the renormalized band structure of the coexistence state are shown in fig .
[ fig - fermi3 ] .
this state is insulating because the renormalized band has a finite energy gap .
we investigated the electronic state in the coo@xmath0 plane of the layered cobalt oxides na@xmath22coo@xmath0 by using the 11 band @xmath29-@xmath30 model on the two - dimensional triangular lattice within the hartree - fock approximation .
we obtained two ordered states : ( 1 ) the metallic antiferromagnetic ordered state for @xmath84 .
( 2 ) the insulating coexistence state between antiferromagnetic , charge and orbital ordered state for @xmath84 and @xmath85 .
it is found that the one - dimensional na order enhances the fermi surface nesting around m and m points resulting in the antiferromagnetic order which is suppressed due to the frustration effect in the case without the na order . in the present work ,
the antiferromagnetism and the orbital order according to the charge order due to the effect of @xmath8 are strengthened each other , and the coexistence state is obtained .
however , there is a possibility that the coexistence state is obtained only due to the effect of @xmath2 without @xmath8 , where the cooperation between the antiferromagnetic order and the orbital order is a crucial role .
explicit results for such state and detailed comparisons with experimental results will be reported in a subsequent paper .
the authors thank m. sato , y. kobayashi , m. yokoi and t. moyoshi for many useful comments and discussions .
this work was performed under the interuniversity cooperative research program of the institute for materials research , tohoku university , and was supported by the grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology .
99 k. takada , h. sakurai , e. takayama - muromachi , f. izumi , r. a. dilanian and t. sasaki , nature * 422 * , 53 ( 2003 ) m. l. foo , y. wang , s. watauchi , h. w. zandbergen , t. he , r. j. cava and n. p. ong , phys .
lett . * 92 * , 247001 ( 2004 ) m. yokoi , t. moyoshi , y. kobayashi , m. soda , y. yasui , m. sato , and k. kakurai , j. phys .
. jpn . * 74 * , 3046 ( 2005 ) i. terasaki , y. sasago and k. uchinokura , phys .
b * 56 * , ( 1997 ) r12685 .
t. motohashi , r. ueda , e. naujalis , t. tojo , i. terasaki , t. atake , m. karppinen , and h. yamauchi , phys .
b * 67 * , 064406 ( 2003 ) s. p. bayrakci , i. mirebeau , p. bourges , y. sidis , m. enderle , j. mesot , d. p. chen , c. t. lin and b. keimer , phys . rev . lett . *
94 * , 157205 ( 2005 ) q. huang , m. l. foo , j. w. lynn , h. w. zandbergen , g. lawes , yayu wang , b. h. toby , a. p. ramirez , n. p. ong and r. j. cava , j. phys
. condens . matter .
* 16 * , 5803 ( 2004 ) t. mizokawa and a. fujimori , phys .
b * 54 * , 5368 ( 1996 ) d. j. singh , phys . rev .
b * 61 * , 13397 ( 2000 ) h .- b .
yang , z .- h . pan , a. k. p. sekharan , t. sato , s. souma , t. takahashi , r. jin , b. c. sales , d. mandrus , a. v. fedorov , z. wang , and h. ding1 , phys .
lett . * 95 * , 146401 ( 2005 ) .
y. yamakawa and y. no , j. phys . : condens .
matter * 19 * , 145289 ( 2007 ) | we investigate the electronic state of a coo@xmath0 plane in the layered cobalt oxides na@xmath1coo@xmath0 using the 11 band d - p model on a two - dimensional triangular lattice , where the tight - binding parameters are determined so as to fit the lda band structure .
effects of the coulomb interaction at a co site : the intra- and inter - orbital direct terms @xmath2 and @xmath3 , the exchange coupling @xmath4 and the pair - transfer @xmath5 , are treated within the hartree - fock approximation .
we also consider the effect of the na order at @xmath6 , where na ions form one - dimensional chains , by taking into account of an effective one - dimensional potential @xmath7 .
it is found that the one - dimensional na order enhances the fermi surface nesting and antiferromagnetism is caused which is suppressed due to the frustration effect in the case without the na order .
furthermore , we consider the effect of the coulomb interaction between the nearest - neighbor co sites @xmath8 and find that a coexistence of the magnetic , charge and orbital ordered state takes place for @xmath9 where the system becomes insulator . and cobalt oxides , fermi surface , magnetic structure , charge order , orbital order , metal - insulator transitions |
classified optically as a seyfert 1 galaxy , markarian 231 is also the most luminous infrared galaxy in the local ( @xmath5 ) universe ( @xcite ) .
ultraluminous infrared galaxies like mrk 231 have total infrared luminosities well above @xmath6 , as measured by iras .
such galaxies are thought to be stages along a sequence in the evolution of merging spiral galaxies ; the mergers generate enormous bursts of star formation and the merged galaxies eventually turn into quasars ( e.g. , @xcite ) .
both the enormous infrared luminosity of @xmath7 ( @xcite ; @xcite ) and the seyfert / quasar properties of mrk 231 argue that it is well along in the merger sequence . using a velocity relative to the 3 k background of 12,447 km s@xmath8 ( @xcite ) and @xmath9 km s@xmath8 mpc@xmath8 ,
the distance to mrk 231 is 166 mpc , and 800 pc subtend 1 . on a galactic scale ,
mrk 231 shows tidal tails indicative of a recent interaction ( e.g. , @xcite ; @xcite ; @xcite ) .
optical emission lines from an apparent star - forming region are seen 1015 ( @xmath10 kpc ) south of the nucleus ( @xcite ; @xcite ) .
an extended radio continuum source roughly 60 ( 48 kpc ) in diameter is also present , predominantly to the south of the nucleus ( @xcite ; @xcite ; @xcite ) .
armus et al .
( 1994 ) reported a second nucleus about 3.5 south of the main nucleus , but _ hubble space telescope _
( hst ) imaging shows only a series of star - forming knots at this location ( @xcite ) . there is a general consensus that many of the observed properties of the galaxy result from a merger occurring @xmath11@xmath0 years ago ( e.g. , @xcite ; @xcite ; @xcite ) .
baum et al . ( 1993 ) imaged a continuum source at 1.4 ghz that extends for 150 ( 120 kpc ) perpendicular to the galaxy disk , and interpreted that source as a superwind driven by the intense star formation triggered by the merger .
the inner region of mrk 231 contains a co disk aligned almost east - west , with an inner diameter of about 1 ( 800 pc ) , a lower density region extending to 3 , and a total gas mass exceeding @xmath12 ( @xcite ; @xcite ) .
a compact 10-@xmath13 m source has a maximum diameter of 0.6 ( @xcite ) and is apparently associated with the active galactic nucleus ( agn ) . that nucleus shows broad balmer emission lines , characteristic of a seyfert 1 nucleus , rather weak forbidden lines ( e.g. , @xcite ) , and variable , low - ionization , and broad absorption lines at redshifts up to 7800 km s@xmath8 relative to the systemic velocity ( @xcite ; @xcite ) .
the nucleus shows strong fe ii emission , although its x - ray and low - energy @xmath14-ray emission are anomalously weak for an agn ( @xcite ; @xcite ; @xcite ) .
asca observations imply a ratio @xmath15 , more consistent with a starburst galaxy than with an agn , although there clearly is a hard power - law component associated with the active nucleus ( @xcite ) .
the asca data show an x - ray absorbing column of @xmath16 @xmath17 , which may be associated with the gas in the broad - line region .
oh maser emission with an isotropic luminosity of @xmath18 also is associated with the galaxy ( @xcite ) but no h@xmath19o megamaser emission has been detected from the nucleus , with an upper limit of @xmath20 ( @xcite ) .
the oh emission is spread over a scale of a few hundred parsecs , rather than being confined to a very compact region as in most other active galaxies ( @xcite ) .
numerous very large array ( vla ) images of mrk 231 are available at @xmath211 or better resolution ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
these images are dominated by an unresolved vla core with a typical flux density of 100300 mjy between 1 and 22 ghz ; based on the available published data , this core appears to vary at gigahertz frequencies by tens of percent on time scales of years .
there also is very weak extended emission on the sub - arcsecond scale ( @xcite ) .
the hi absorption against the vla core ( @xcite ) occurs against a radio halo or disk of emission 440 mas ( 350 pc ) in extent , with an elongation pa in the east - west direction ( carilli , wrobel , & ulvestad 1998 , hereafter cwu ) , similar to the elongation pa of the 800-mas co emission ( @xcite ) .
the radio continuum from this 350-pc disk is strongest at low frequencies and is apparently responsible for much of the extra @xmath22 mjy of flux density missed between measurements on vlbi scales and those on vla scales ( e.g. , @xcite ; @xcite ; @xcite ) .
the properties of the 350-pc disk imply a massive star formation rate of @xmath23 yr@xmath8 ( @xcite ) . on a smaller scale of 50 mas ( 40 pc ) , the core resolves into a north - south triple at 1.7 ghz , as imaged with the european vlbi network in the mid-1980s ( @xcite ) .
optical and ultraviolet spectropolarimetry give an electric - vector position angle ( pa ) of @xmath24 , suggesting scattering of radiation by dust clouds flowing outward along this north - south vlbi axis ( @xcite ; @xcite ) .
this paper presents and interprets new continuum observations , with the very long baseline array ( vlba ) and the vla , designed to probe structures in mrk 231 on scales ranging from parsecs to kiloparsecs .
preliminary results were reported by ulvestad , wrobel , & carilli ( 1998 ) .
this radio continuum study of mrk 231 provides evidence for outflow , for synchrotron self - absorption , and for free - free absorption .
these new radio results are related , wherever possible , to published results from the aforementioned studies of radio spectral lines and in shorter wavelength bands .
the continuum emission from mrk 231 was observed with the 10-element vlba ( @xcite ) at frequencies ranging from 1.4 to 22.2 ghz , during the three separate observing sessions summarized in table [ tab : vlbaobs ] . during 1995 november ,
11-minute scans on mrk 231 typically were preceded by 2-minute scans on the delay - rate check source j1219 + 4829 , with a total time of about 39 minutes to cycle through 1.4 , 2.3 , and 5.0 ghz .
occasional scans on da 193 ( j0555 + 3948 ) , 3c 345 ( j1642 + 3949 ) , and j1740 + 5212 were obtained for ancillary calibration ( fringe - finding , manual pulse calibration , and amplitude calibration checks ) . on 1996
december 8 , scans of 12 minutes duration on mrk 231 were interleaved with 24-minute scans on j1219 + 4829 , and the total cycle time for 5.0 , 8.4 , and 15 ghz was 44 minutes . a single 25-m vla antenna was included along with the vlba to provide improved short - spacing coverage .
ancillary calibrators were 4c 39.25 ( j0927 + 3902 ) , j1310 + 3220 , and 3c 345 . on 1996 december 27 , both hi line and 22-ghz continuum data were acquired using the vlba and the phased vla .
the line observations were reported by cwu and will not be described further here . at 22 ghz , 8-minute scans of mrk 231 were interleaved with 3-minute scans of j1219 + 4829 .
short scans of 4c 39.25 , j1310 + 3220 , and 3c 345 were included for ancillary calibration .
all vlba observing and correlation of mrk 231 adopted the j2000 position for source j1256 + 5652 from patnaik et al .
( 1992 ) . at the vla ,
additional scans of 3c286 ( j1331 + 3030 ) were obtained on 1996 december 27 to set the flux density scale , and scans of mrk 231 at 5 , 8.4 , and 15 ghz were also acquired to provide a contemporaneous spectrum of the galaxy .
initial calibration of the vlba data was carried out using the standard gain values of the vlba antennas together with system temperatures measured every 12 minutes during the observing sessions .
autocorrelations were used to correct for imperfect adjustment of the sampler levels for these 2-bit ( 4-level ) data . for the 1995 data ,
an additional amplitude adjustment was made using the vlba image of da 193 , whose total flux density was constrained to be equal to the value measured at the vla .
baseline gain corrections were less than 5% at 5 ghz , 510% at 2.3 ghz , and 520% at 1.4 ghz .
the substantial corrections at 1.4 ghz were expected because the observing frequency differed by nearly 300 mhz ( roughly 20% of the observing frequency ) from the frequency where standard gains are measured . since the vlba gains are quite stable , and accurate system temperatures were regularly measured , the estimated uncertainty in the vlba flux density scale is less than @xmath10% .
the nrao astronomical image processing system ( aips ) ( @xcite ) was used for all vlba data calibration . the continuum emission from mrk 231
was observed in dual circular polarizations with the vla ( @xcite ) during the five separate observing sessions summarized in table [ tab : vlaobs ] . in 1988 and 1989 ,
scaled - array observations were made at 1.5 , 4.8 , and 15.0 ghz , using the b , c , and d configurations , respectively .
these observations yielded intrinsic resolution near 4 , useful for studying the emission on kiloparsec scales .
scans of @xmath25 minutes were interleaved with scans of j1219 + 4829 , the local phase calibrator , and the amplitude scale was set to that of baars et al.(1977 ) using short observations of 3c 286 .
these scaled - array observations were not full syntheses , as only 1.53 hr were spent on mrk 231 in each case . during 1995 november , a short observation in the b configuration was made to determine the total flux and spectrum of mrk 231 contemporaneously with the 1995 vlba observations .
five frequencies between 1.4 and 22 ghz were used at the vla , with 11 minutes spent on mrk 231 at each frequency .
resolutions ranged from @xmath26 at 1.5 ghz to @xmath27 at 22 ghz .
j1219 + 4829 was observed for several minutes at each frequency as a phase calibrator .
also , short scans of 3c 286 were obtained to set the vla flux density scale and short scans of da 193 were acquired to check the vlba flux density scale . during 1996 december , observations of mrk 231 at 5 , 8.4 , and 15 ghz were only single 2-minute snapshots , while the 1.4- and 22-ghz observations made together with the vlba were much longer .
phase and amplitude calibration using j1219 + 4829 and 3c 286 , respectively , was similar to that carried out for earlier vla epochs , with the added complication that additional editing was required for calibrator scans acquired in phased - array mode .
vla flux densities typically have errors of @xmath28% but residual phase noise in 1996 led to larger errors at the higher frequencies ( 7.5% at 8.4 ghz , 10% at 15 ghz , and 15% at 22 ghz ) . at 1.4 ghz
, a polarization calibration was performed using j1219 + 4829 to deduce the antenna polarizations and using 3c 286 to fix the absolute polarization pa , and the 1.4-ghz data were corrected for these effects .
aips was used for all vla data calibration .
following the initial calibration , all vlba data were imaged in aips .
after initial images were made , an iterative self - calibration procedure was used to correct the complex gains of the individual antennas for atmospheric and instrumental effects .
this process halted when the image quality and the r.m.s .
noise stabilized .
a major aim of this study was to derive the spectra of different components in mrk 231 to search for absorption . to accomplish this goal ,
full - resolution images were made at each vlba frequency from 1.4 to 15 ghz .
then , the data at each frequency were re - imaged using only the range of projected baseline lengths sampled at each of the lower frequencies . in this process
, the weighting of the data was tapered in the aperture plane to give approximately the same resolution as that available at each lower frequency , with a restoring beam fixed to have the same parameters as the full resolution beam at the lower frequency .
for example , the 8.4-ghz data were re - imaged at three different resolutions , equivalent to the full resolution at 1.4 , 2.3 , and 5.0 ghz .
the subsections below present a subset of images that are the most important for the scientific analysis .
the interpretation of the results is deferred to later sections .
the new vlba image at 1.4 ghz , presented in figure [ fig : vmatch ] , shows the same ( nearly ) north - south triple structure known from previous 1.7-ghz observations with the european vlbi network ( @xcite ) .
the triple consists of an unresolved core , together with two resolved lobes .
the approximate size of the triple is 50 mas ( 40 pc ) .
the emission 20 mas to the east of the core is not apparent at other frequencies , and is likely to be an artifact .
the other images in figure [ fig : vmatch ] are the 2.3 , 5.0 , and 8.4-ghz images of the same triple at a resolution matching that of the 1.4-ghz image .
note that as the frequency increases , progressively less diffuse emission is seen ; only the outer portion of the vlba lobes is detected at 8.4 ghz .
the total flux densities in the north ( n ) , central ( c ) , and south ( s ) components were measured by integrating over the entire area of each component .
in addition , the position of the peak intensity in each component , relative to component c , was determined by fitting a parabola to a few pixels in each image surrounding the peak .
results of these measurements are given in table [ tab : nsflux ] .
also included in table [ tab : nsflux ] are the results for component c from the tapered data at 15 ghz ; components n and s are too weak to be detected at that frequency .
the estimated errors in the flux densities are 10% for component c and for the low - frequency data on component s , but rise to as much as 50% for components
n and s at 8.4 ghz due to the increasing uncertainty in the lobe strengths at frequencies where they are largely resolved out .
spectra of components n , c , and s are shown in figure [ fig : vspectra ] . although the data at all frequencies were not taken simultaneously , the well - resolved components , n and s , are unlikely to have varied over the 13 months between sessions .
component c appears to have been roughly constant from 1.4 to 5 ghz , based on ( 1 ) the vla core results ( see section [ yflux ] ) ; and ( 2 ) the small differences of only 510 mjy between 1995 and 1996 in the vlba strengths at 1.4 and 5 ghz ( table [ tab : nsflux ] ; @xcite ) .
the data between 5 and 22 ghz were taken within 20 days , so figure [ fig : vspectra ] should be a good snapshot of the spectrum of the vlba components at a single epoch , 1996 december .
vlba images , at full resolution , of the north - south triple at 2.3 and 5.0 ghz are shown in figures [ fig : vfull]a and [ fig : vfull]b .
( the full - resolution image at 8.4 ghz , discussed below , shows no emission from components n or s. ) both of these images show the structure at the ends of the vlba lobes more clearly , with substantial resolution perpendicular to the direction to the central source . neither the total component flux densities nor the peak positions differ significantly from those values derived from the images at matched resolution . the first installment of the caltech - jodrell vlbi survey presented a 5.0-ghz image of mrk 231 , under the alias 1254 + 571 ( @xcite ) . that survey included intercontinental baselines , and indicated that mrk 231 was slightly resolved on a scale near 1 mas , in a nearly east - west direction , markedly different from the pa of the larger , north - south triple described above .
new vlba images , at full resolution , of component c at 8.4 and 15.4 ghz are shown in figures [ fig : vfull]c and [ fig : vfull]d .
the 8.4-ghz image shows clear resolution in a pa between 60 and 65 , somewhat north of the pa of 92 quoted by taylor et al .
( 1994 ) for the resolved core at 5 ghz , although their image does show an extension slightly north of east , in pa @xmath29 .
the 15-ghz image shows that the central source appears to break up into three separate components at sub - parsec resolution .
single- and multi - component gaussian fits were made to component c at all frequencies between 1.4 and 15 ghz . at 1.4 and 2.3 ghz , the source is not resolved , and the total flux densities are indistinguishable from the values given in table [ tab : nsflux ] . however , at 5 , 8.4 , and 15.4 ghz , component c is significantly resolved with a size of 0.81.0 mas ( 0.60.8 pc ) in a pa near 63 . at the 8.4-ghz resolution ,
component c in both the 8.4 and 15-ghz images is much better fitted by a two - component model . finally , at the full 15-ghz resolution , there may be a third component to the southwest .
all the fits to component c at the three different resolutions are summarized in table [ tab : cflux ] .
it is possible that the southwestern component is an imaging artifact , but since its location and that of the northeastern component are not symmetric with respect to the central component , the southwestern component is most likely real . in the multiple component
fits , only the strongest source appears significantly resolved .
since the resolution is predominantly along the direction of the structure , this indicates the possible presence of more components that would be separated from the strongest source at yet higher spatial resolution .
although mrk 231 also was observed with the vlba and the phased vla at 22.2 ghz , the steepening high - frequency spectrum of the core at higher frequencies meant that fringes were detected only on the relatively short baselines in the southwestern united states .
imaging of these limited data yields a total flux density of 30 mjy for the core , and that datum is included in the spectral plot given in figure [ fig : vspectra ] .
phase - referencing observations are required to image component c at 22 ghz with the resolution and sensitivity needed to identify subcomponents .
during 1988 and 1989 , mrk 231 was observed at 1.5 , 4.9 , and 15 ghz using scaled arrays of the vla , as described in section [ vlaobs ] .
images were restored with a common ( circular ) gaussian beam of size 4 ( full width at half maximum ) , to enable spectral comparisons .
figure [ fig : ymatch ] shows these images at 1.5 ghz and 4.9 ghz .
the 15-ghz emission is completely unresolved , so that image is not shown .
flux densities in the unresolved vla core were derived by fitting a gaussian constrained to the beam size to the central pixels of each image .
in addition , the total flux density in each vla image was determined by integrating over the region in which significant emission is detected ; the extended flux density , then , is taken to be the difference between the total flux density and that in the unresolved vla core . at 15 ghz , there is no detection of extended emission in a single beam area , so this difference is taken to be an upper limit to the total extended flux density .
peak intensities in the extended emission were measured at 1.5 and 4.9 ghz , while an upper limit of 3 times the r.m.s .
noise per beam area was assumed at 15 ghz .
results of all these flux density measurements appear in table [ tab : vlascale ] .
the spectral index of the extended emission is steep , both in total flux density and in a point - by - point comparison .
the two - point spectral index , @xmath30 of the total emission is @xmath31 between 1.5 and 4.9 ghz ( @xmath32 , where @xmath33 is the flux density at frequency @xmath34 ) .
a spectral - index image between these frequencies indicates that all regions with significant emission at both frequencies have spectral indices ranging between @xmath4 and @xmath35 , with the total spectrum being somewhat steeper because of the areas detected at 1.5 ghz that are below the detection threshold at 4.9 ghz .
the total spectral index of the extended emission between 4.9 and 15 ghz is @xmath36 .
vla observations of mrk 231 were made in 1995 november and 1996 december , at frequencies ranging from 1.4 ghz to 22 ghz , as described previously .
the data were calibrated , imaged , and self - calibrated in the usual way , in order to measure the flux density of the unresolved vla core .
this flux density was taken to be the peak in the final self - calibrated images .
results are presented in table [ tab : vlaflux ] , with errors quoted as discussed in section [ vlaobs ] .
the measurements show that the flux density of the core appears to have been relatively constant at frequencies up to 5 ghz , but decreased significantly at 15 and 22 ghz from 1995 to 1996
. therefore , the spectrum of the central component of the vlba triple source ( see figure [ fig : vspectra ] ) should be valid as of the 1996 december epoch .
weak extended emission in the vicinity of the core , within the central arcsecond , is present in the highest resolution vla data , as discussed by taylor et al .
this emission has no significant impact on the flux - density measurements for the unresolved core .
a long observation of mrk 231 was made at 1.4 ghz with the phased vla in 1996 december , as part of a vlba observation of the hi ( cwu ) .
the phased - array data essentially undergo a real - time calibration of the vla phases , and then can be calibrated further in aips , as described previously .
mrk 231 was imaged in stokes i , q , and u ; the q and u images were further combined in the usual way to obtain images of the linearly polarized intensity , p , and electric - field pa , @xmath37 .
figure [ fig : ydeep ] shows a composite image , with lines representing p and @xmath37 superposed on contours of stokes i emission .
the stokes i emission in figure [ fig : ydeep ] strongly resembles that evident in fig .
[ fig : ymatch ] .
however , a major new discovery from figure [ fig : ydeep ] is that some regions of the diffuse stokes i emission to the south of the vla core are significantly linearly polarized , reaching a peak polarized intensity of 185 @xmath13jy beam@xmath8 about 26 ( 21 kpc ) south of the vla core .
at this polarization peak , the percentage polarization is about 57% and @xmath37 ( electric vector position angle ) is about 15 .
no rotation measure corrections have been made .
the vla core is @xmath38 0.1% linearly polarized .
figure [ fig : montage ] summarizes the overall structure of mrk 231 on a variety of scales .
the two left - hand panels ( figures [ fig : montage]a and [ fig : montage]b ) are an optical b - band image of the galaxy from hamilton & keel ( 1987 ) , and , at the same scale , the 1.5-ghz vla image from data taken in 1989 .
these panels show that the radio emission to the south of the nucleus actually extends well beyond the dominant emission from the optical galaxy .
the core of the vla image contains a north - south radio source imaged with the vlba ( figure [ fig : montage]c ) on very much smaller scales , with a total extent of @xmath39 pc .
finally , the nucleus of the galaxy shows additional structures on the 1-pc scale at the highest resolution available with the vlba at 15 ghz , as shown in figure [ fig : montage]d .
mrk 231 is an ultraluminous infrared galaxy with a total luminosity in excess of @xmath40 ( @xcite ) . for over 20 years , it has been known to contain radio emission about an arcminute in extent ( @xcite ) , somewhat larger than the optical galaxy .
the bulk of this emission comes from a diffuse region within about 30 to the south of the galaxy nucleus .
off - nuclear optical imaging and spectroscopy ( @xcite ; @xcite ) revealed h@xmath30 emission in an apparent region of star formation centered roughly 1015 to the south of the nucleus . from the new vla imaging at 1.4 ghz ,
the total flux density in the diffuse emission is 42 mjy .
this emission is primarily concentrated in a higher brightness feature extending due south of the nucleus for about 20 ( 16 kpc ) , then appearing to curve toward the west , as described previously by baum et al .
several different possibilities for the origin of the diffuse emission are considered below .
most of the diffuse emission south of the nucleus has a steep spectrum , with a spectral index near @xmath35 , as shown by the scaled - array observations from 1.5 to 15 ghz .
on spectral grounds alone , it seems unlikely that thermal processes can make a substantial contribution .
in addition , it is possible to use the h@xmath30 surface brightness measured by hamilton & keel ( 1987 ) to compute the expected amount of thermal emission from hii regions .
we estimate the total h@xmath41 flux ( assuming case b recombination ) in a 4 vla beam 12 south of the nucleus to be @xmath42 erg @xmath17 s@xmath8 . the predicted thermal radio brightness at 4.9 ghz ( e.g. , ulvestad et al .
1981 ; @xcite ) then would be only about 1 @xmath13jy beam@xmath8 .
in contrast , the peak in the diffuse source at 4.9 ghz is @xmath43 @xmath13jy beam@xmath8 .
furthermore , the overall spectrum sets an upper limit of @xmath22 @xmath13jy beam@xmath8 for the flat - spectrum contribution at any point , entirely consistent with the prediction from the optical spectroscopy .
the lack of discernible thermal radio emission implies that the intrinsic h@xmath41 flux can be no more than @xmath22 times higher than that observed , implying an upper limit of @xmath44 for the extinction in the star - forming region .
the flattest radio spectrum , with @xmath45 between 1.5 and 4.9 ghz , occurs near the peak of the diffuse emission in the 4.9-ghz image .
this is near the peak in the fractional polarization , and well beyond the region of significant optical continuum and line emission ( @xcite ; @xcite ) .
the apparent lack of optical emission from young stars implies little or no thermal contribution to the extended radio emission in this area , so the flattening of the spectrum must have another cause . assuming a total thermal radio flux density of 110 @xmath13jy in the emission in the southern diffuse lobe , plus a distance of 166 mpc , the formulae given by condon ( 1992 ) imply an ionizing flux of @xmath46 photons s@xmath8 and a star formation rate of @xmath47 yr@xmath8 in stars above @xmath48 .
this predicts a nonthermal radio luminosity , from supernova remnants , of @xmath49 w hz@xmath8 at 4.9 ghz . the corresponding radio flux density at the distance of mrk 231 would be @xmath10 @xmath13jy
this prediction falls three orders of magnitude short of the total flux density in the southern vla lobe , and 1.5 orders of magnitude short of the peak flux density per beam at the location of the putative hii region .
alternatively , using the standard radio / infrared relation for starbursts ( e.g. , @xcite ) , the diffuse 1.4-ghz flux density of 42 mjy would predict infrared emission of 510 jy , extended over @xmath2130 , at 60 and 100 @xmath13 m .
this is inconsistent with the small size ( @xmath3812 ) found for the near- , mid- , and far - infrared emission ( @xcite ; @xcite ; @xcite ) . furthermore , although the diffuse radio emission and the optical galaxy have somewhat similar shapes , figure [ fig : montage ] shows that much of the radio emission is actually located beyond the region of significant b - band emission in the galaxy , which would not be expected if the radio emission were associated with a large population of supernova remnants .
finally , the high fractional polarization implies a magnetic field that is ordered on a scale much larger than that expected from a collection of supernova remnants .
therefore , we rule out the possibility that the diffuse radio emission south of the nucleus is generated by star formation and supernovae . the remaining , favored , possibility is that the diffuse emission is excited by a jet from the galaxy nucleus .
this inference is supported by an apparent ridge of slightly higher surface brightness emission connecting back to the nucleus .
also , the extremely high polarization at the outer edge of the diffuse lobe implies a well - ordered magnetic field , rather than the chaotic field expected from a collection of supernova remnants .
in addition , the polarization vectors indicate that the magnetic field appears to wrap around the outer edge of the lobe , as would be natural for emission fed by a nuclear jet .
the total emission of 42 mjy in this southern lobe at 1.4 ghz corresponds to @xmath50 w hz@xmath8 , and the total luminosity between 10 mhz and 100 ghz ( assuming equal proton and electron energies , and a spectral index @xmath51 ) is @xmath52 erg s@xmath8 .
if this luminosity arises from relativistic particles that uniformly fill a lobe 10 kpc in diameter , then the physical conditions in the lobe can be estimated ( cf .
the minimum - energy magnetic field is @xmath53 @xmath13gauss , the total energy in the lobe is @xmath54 erg , and the synchrotron lifetime is @xmath55 yr . the lobe peaks only @xmath56 kpc from the nucleus , and therefore could be supplied by a jet with an advance speed of only @xmath57 km s@xmath8 .
this required velocity is much smaller than the speeds of up to 7800 km s@xmath8 seen on parsec scales in the broad - absorption - line clouds ( e.g. , forster et al .
the north - south vlba triple implies the presence of an energy - supplying jet on smaller scales , although its speed of advance is not presently known .
it is interesting to note that the kiloparsec - scale radio emission emerges from the galaxy core in roughly the same direction as the strong optical emission seen just south of the core , in a pa between @xmath58 and @xmath59 ( see figure [ fig : montage ] ) .
the secondary optical peak was shown by surace et al .
( 1998 ) to consist of a series of star - forming knots . therefore ,
if it is correct to assume that the kiloparsec - scale emission is energized by a radio jet , that jet would appear to be related to the star formation ; perhaps the jet has compressed thermal material along its path enough to trigger a burst of star formation .
the large - scale jet then appears to curve back to the west only beyond the main extent of the optical galaxy , but in the same general sense as the tidal tails imaged by hamilton & keel ( 1987 ) and surace et al .
( 1998 ) .
cwu discovered a continuum `` halo '' at 1.4 ghz , with a size of @xmath60 mas ( 350 pc ) and containing about 130 mjy .
comparison of tables [ tab : nsflux ] and [ tab : vlaflux ] indicates that the amount of flux density missing between vlba and vla scales ranges from @xmath61 mjy at 1.4 ghz to @xmath62 mjy at 15 ghz .
if this emission is dominated by the 350-pc disk , then the spectral index of that disk appears to be @xmath63 between 1.4 and 15 ghz .
taylor et al .
( 1998 ) have recently imaged weak extended emission on a 1 ( 800 pc ) scale , using the vlba at 0.3 and 0.6 ghz as well as archival vla observations at frequencies of 5 ghz and higher .
they suggest that this emission comes from a weak `` outer disk '' that may have a spectral turnover near 8 ghz .
it may be that the emission on 0.51.0 scales contains both an optically thin , steep - spectrum component , and some regions that are free - free absorbed at the lower frequencies .
the sub - kiloparsec `` milli - halo '' seen in ngc 1275 ( 3c 84 ) by silver , taylor , & vermeulen ( 1998 ) has a similar intrinsic strength and size to the radio disk in mrk 231 .
the milli - halo in ngc 1275 is reported to have a steeper spectrum , with @xmath64 .
different spectral shapes for ngc 1275 and mrk 231 may imply different emission processes , different electron energy distributions , or just varying amounts of free - free absorption .
the suggested model for ngc 1275 is that the milli - halo is caused by particles leaking out of the radio jet into the surrounding medium ( @xcite ) .
however , for mrk 231 , the elongation of the 350-pc emission perpendicular to the 40-pc vlba triple and parallel to the slightly larger scale co disk indicates , instead , a possible relationship to the disk of material thought to surround active galactic nuclei ( e.g. , @xcite ) .
cwu associate this elongated emission with a disk containing atomic , molecular , and dust components on a scale of a few hundred parsecs .
the short electron lifetimes ( @xmath65 yr ) in this disk imply local particle acceleration , requiring star formation or other shock processes in the inner kiloparsec .
in fact , the radio continuum emission from this disk or torus is likely to be powered by massive star formation , since it is consistent with the canonical radio - infrared relation .
this is in agreement with the conclusion of downes & solomon ( 1998 ) , who suggest that the bulk of the far - infrared emission in mrk 231 is powered by a starburst .
the slight flattening of the radio spectrum relative to the canonical starburst spectral index of @xmath66 could be accounted for by localized free - free absorption within the disk .
the north - south triple source in mrk 231 has a total extent of roughly 50 mas , or 40 pc . at 1.4 and 2.3
ghz , there is considerable emission detected from the central core out to the outer edges of the triple .
however , at 5.0 ghz , most of the detected extended emission is in the outermost parts of the lobes .
the high resolution images at 2.3 and 5.0 ghz ( figure [ fig : vfull ] ) show that the ends of the lobes are significantly resolved perpendicular to the direction to the core , with transverse extents of 1520 mas ( @xmath211216 pc ) .
this may indicate the presence of shocks at `` working surfaces '' where jets are attempting to burrow out of the nuclear regions .
the outer edges of the vlba triple source are at a distance similar to the inner scale of the diffuse radio emitter and the hi - absorption cloud imaged by cwu , so it seems plausible that these lobes are generated by a jet running into the inner surface of an hi shell or disk .
it is also interesting to note that the major axis of the triple source is similar to the axis of the diffuse radio source and the elongation of the optical galaxy on scales 23 orders of magnitude larger ( see figure [ fig : montage ] ) .
this suggests a long - term memory of the symmetry axis of this seyfert galaxy , possibly associated with an accretion disk in its center . at the resolution of the 1.4-ghz vlba image , the northern component of the 40-pc triple
has a steep spectrum between 1.4 and 8.4 ghz , with a spectral index of @xmath67 . within the errors ,
the spectrum is straight across the entire flux range , consistent with optically thin synchrotron emission from an ensemble of electrons with a steep power - law distribution in energy .
however , the peak of the northern component shifts systematically toward the southwest with increasing frequency ( see table [ tab : nsflux ] ) , implying the presence of spectral gradients within the lobe . at 1.4 ghz , this peak is located 23.3 mas ( 17 pc ) from the core , in pa 6.5 . at the same low resolution ,
the 8.4-ghz peak is located only 19.1 mas ( 14 pc ) from the core in pa @xmath68 .
since most of the shift in the peak of the northern lobe occurs between 1.4 and 2.3 ghz , a spectral index image was made between those two frequencies , using the 2.3-ghz image tapered to the 1.4-ghz resolution .
that image , displayed in figure [ fig : spectral ] , shows the presence of a region with an inverted spectrum , roughly 18 mas ( 14 pc ) north of the galaxy nucleus , although the point - spread function at 1.4-ghz is too large to cleanly resolve the region .
( the sharp edges in the spectral - index map are caused by blanking the individual input maps at 8 times their respective noise levels . ) a similar image of the spectral index between 2.3 and 5 ghz , at the 2.3-ghz resolution , shows a small region with a nearly flat spectrum between those two frequencies near the same location .
this implies that the northern lobe may contain a small component , with a diameter no larger than @xmath26 mas ( 3 pc ) , whose spectrum turns over near a frequency of 23 ghz .
the most logical cause for the apparent spectral turnover in the northern lobe is free - free absorption in an ionized region with a temperature near @xmath69 k. for a turnover frequency of 2 ghz , the emission measure would be @xmath70 @xmath71 pc .
both the large variation in spectral index and the shift of the peak as a function of frequency occur over 34 mas , implying that the absorbing medium has a size of @xmath72 pc or less on the plane of the sky .
if this dimension is also used as an estimate of the line - of - sight distance through the absorber , then the average density along a 3-pc line of sight would be @xmath73 @xmath3 .
such a density is , for instance , fairly typical for ionized clouds in seyfert narrow - line regions ( e.g. , @xcite ) .
the spectral - index image also indicates a spectral gradient in the southern vlba component between 1.4 and 2.3 ghz , including a portion at the outer edge of the lobe with an inverted spectrum .
the spectral index image between 2.3 and 5.0 ghz shows no such inverted component .
the presence of an inverted component between 1.4 and 2.3 ghz is further confirmed by the overall lobe spectrum , which has @xmath74 between 1.4 and 2.3 ghz and @xmath75 between 2.3 and 8.4 ghz ( cf .
table [ tab : nsflux ] and figure [ fig : vspectra ] ) .
if the steep spectrum above 2.3 ghz continued to 1.4 ghz , the flux density at that frequency would be 65 mjy instead of the measured value of 37 mjy , implying that a substantial fraction of the total flux of the lobe is absorbed at the lowest frequency .
hypothesizing a turnover frequency due to free - free absorption midway between our two lowest observing frequencies , near 1.8 ghz , the emission measure would be @xmath76 @xmath71pc .
the region of strongest absorption appears to be at the outer edge of the lobe , between 22 and 28 mas ( @xmath77 pc ) from the central component .
this implies an absorbing cloud with a size near 5 pc , and the derived particle density is then @xmath78 @xmath3 .
both the density and the size of this ionized cloud are consistent with those required for free - free absorption in the northern component .
the overall high - frequency spectrum could be artificially steepened by resolution effects , which would tend to decrease the flux density at the higher frequencies .
however , the spectral index between 2.3 and 5 ghz is still near @xmath79 in images made at the full 2.3-ghz resolution , and the mix of short and long spacings on the vlba implies that little 5-ghz emission should be missing on this scale .
therefore , the overall spectrum of the southern vlba lobe is indeed quite steep intrinsically .
the 40-pc vlba triple appears to be significantly affected by free - free absorption due to ionized gas . on a somewhat larger scale , a neutral absorbing component has been detected in our hi absorption study ( @xcite ) .
therefore , a possible inference is that the ionized gas is merely the inner part of the putative disk seen in hi , ionized by the central continuum source in mrk 231 .
that disk may be a larger version of the disks traced by h@xmath19o maser emission in active galaxies such as ngc 4258 ( @xcite ; @xcite ; @xcite ) .
however , the viewing angle for the inner disk in mrk 231 ( e.g. , @xcite ) would be such that no h@xmath19o maser emission is present . the much larger size of the apparent disk in mrk 231 could be related to its much more luminous central continuum source .
the north - south vlba source in mrk 231 bears a striking resemblance to that seen in 3c 84 ( @xcite ; @xcite ; @xcite ) .
ngc 1275 , the host galaxy of 3c 84 , is slightly more than twice as close to us as mrk 231 , and the size of its north - south radio source is 10 pc , about a quarter the size of the triple in mrk 231 . in addition
, the northern component of 3c 84 ( the `` counterjet '' ) also exhibits free - free absorption near 15 ghz , attributed to ionized gas with a path length of several parsecs and a density above @xmath69 @xmath3 ( vermeulen et al .
models of the absorption in 3c 84 indicate that the absorbing gas is not spherically distributed , but can be in a warped disk that could be kept ionized by a central continuum source ( @xcite )
. however , in 3c 84 , the disk appears to block only the northern vlba component , whereas in mrk 231 the apparent free - free absorption toward parts of both the southern and northern components indicate that the disk may block both components .
this blockage occurs despite the fact that the vlba lobes in mrk 231 are several times farther from the radio core than is the absorbed region of the jet in 3c 84 .
levinson et al . ( 1995 ) estimated a total bolometric luminosity of @xmath80 ( scaling their value to @xmath81 ) for 3c 84 , while the total luminosity of mrk 231 is @xmath56 times greater .
this fact , combined with the inference that the ionized density in mrk 231 is @xmath10 times smaller than that derived from the higher - frequency spectral turnover in 3c 84 ( vermeulen et al .
1994 ) , indicates that the volume of the ionized region could be hundreds of times larger in mrk 231 than in 3c 84 .
this may account for the fact that free - free absorption is detected over a much larger scale in mrk 231 than in 3c 84 .
an alternative explanation for the free - free absorption is connected with the possibility that much of the infrared emission in mrk 231 may be generated by star formation rather than by an active galactic nucleus ( @xcite ) . in this case ,
a local source of ionizing radiation could account for the free - free - absorbing clouds .
ionization of spherical clouds 5 pc in diameter , with densities of @xmath82 @xmath3 , requires only @xmath83 ionizing photons per second .
this ionization rate could be accounted for by just a few early o stars ( @xcite ) , easily consistent with the high infrared luminosity .
the central source of the 40-pc vlba triple undoubtedly contains the actual nucleus of mrk 231 .
this central source was shown in section [ central ] ( see table [ tab : cflux ] and figure [ fig : vfull ] ) to consist of two dominant components separated by 1.1 mas ( 0.9 pc ) in pa @xmath84 , with a possible weak third component located 0.85 mas ( 0.7 pc ) to the southwest . at 8.4 ghz , the two stronger components have brightness temperatures of @xmath85 k and @xmath86 k. the bulk of the flux density in the strongest component is unresolved , so it may have a peak brightness temperature considerably exceeding @xmath1 k. the spectrum of component c peaks between 5.0 and 8.4 ghz ( see figure [ fig : vspectra ] ) ; the spectral index between 8.4 and 15 ghz is @xmath87 , indicative of optically thin synchrotron radiation .
table [ tab : cflux ] indicates that the two dominant components of this source both have steep spectra between 8.4 and 15 ghz .
the high - frequency ( `` intrinsic , '' or un - absorbed ) spectrum of the stronger component may be steeper than that of the weaker component , but this is somewhat uncertain due to the blending of the two components at 8.4 ghz .
two possibilities for the spectral turnover are either synchrotron self - absorption or free - free absorption similar to that deduced for the northern and southern lobes .
the stronger component has an 8.4-ghz brightness temperature of @xmath88 k , indicating that synchrotron self - absorption is a possible cause for the spectral turnover .
if the turnover occurs near 6 ghz at a flux density near 150 mjy , the 8.4-ghz size of @xmath89 mas implies a magnetic field strength near 0.5 gauss . for comparison ,
the minimum - energy magnetic field calculated for a straight spectrum with a spectral index of @xmath90 between @xmath91 and @xmath92 hz , and with equal proton and electron energy densities , is similar , @xmath93 gauss .
the equipartition magnetic field is thus near the value required for synchrotron - self - absorption to occur in the stronger component , making it likely that this could account for the overall turnover of the spectrum .
alternatively , free - free absorption causing a turnover near 6 ghz , and over a 1-pc path length , would require an ionized density of @xmath94 @xmath3 .
attributing the turnover above 5 ghz to synchrotron self - absorption in the stronger central component does not eliminate the possibility that free - free absorption also occurs at a somewhat lower frequency .
for example , the northeastern component of the central source has a spectral index of @xmath95 between 8.4 and 15 ghz , as indicated by the two - component fit to the central source .
an extrapolation of this spectrum to lower frequencies would predict a flux density of @xmath2145 mjy at 2.3 ghz and @xmath2165 mjy at 1.4 ghz .
the latter prediction exceeds the total observed value of 53 mjy for all of component c. this contradiction can be removed by postulating the presence of either free - free or synchrotron absorption of the weaker component in the vicinity of 2 ghz .
such absorption is consistent with the results for the outer vlba lobes , again indicating the presence of a large quantity of ionized gas in the inner regions of mrk 231 .
however , strong variability of the northeastern component between 1996 and 1998 ( ulvestad et al .
, 1998b , and in preparation ) implies that it may well be the location of the actual nucleus ; its brightness temperature may be considerably greater than the lower limit of @xmath0 k found in this work , consistent with the possible presence of synchrotron self - absorption .
the major axis of the central vlba source is near 65 , very different from the position angle of about 05 for the larger - scale vlba triple .
if we assume ( based on our recent observations of variability ) that the northeastern component of the central source is the galaxy core , and the inner radio jet extends to the southwest in position angle 115 , the obvious hypothesis would be that this jet must twist within the inner few parsecs of the galaxy to feed the more distant vlba lobes . however , there is no evidence in any vlba images for a direct connection between the 1-pc - scale source and the larger scale vlbi lobes .
apparent large changes in vlba position angles within the inner few parsecs are also present in ngc 4151 ( @xcite ) , so this morphological trait is not unique to mrk 231 .
indeed , a similar circumstance was discovered recently in the nearest active galaxy , centaurus a , where the sub - parsec axis defined by the vlbi radio jet ( @xcite ; @xcite ) is misaligned by some 70 from the axis of the infrared disk imaged with hst on a 40-pc scale ( @xcite ) .
an alternative hypothesis to jet curvature would be that the northeastern component of the central source is part of an accretion disk or torus as claimed for component s1 in ngc 1068 ( @xcite ) .
however , this seems untenable for mrk 231 ; the brightness temperature in mrk 231 is more than 1000 times higher than that in ngc 1068 , so an interpretation as thermal bremsstrahlung or reflected synchrotron emission , while reasonable for ngc 1068 , becomes implausible for mrk 231 .
if the central source in mrk 231 is the inner portion of a jet , its symmetry axis would presumably represent the inner axis of the accretion disk around the black hole that powers the active galactic nucleus .
in contrast , the vlba triple , plus the hi and co disks , clearly indicate a very different axis on scales greater than a few parsecs .
the smaller scale is typical for the optical broad - line region , while the larger scale matches that for the optical narrow - line region in seyferts .
therefore , a possible interpretation is that the broad - line and narrow - line regions have symmetry axes that are related only over the long term .
the simplest unified schemes for seyfert galaxies ( @xcite ) suggest that seyfert 1 galaxies like mrk 231 are those in which the line of sight to the broad - line region misses the nuclear disk or torus , while seyfert 2 galaxies are those in which our viewing angle prevents a direct view of the broad - line region .
these models typically rely on a common symmetry axis for the broad - line and narrow - line regions , whereas the data presented here indicate that those axes may be very different , at least at the current epoch .
this is in accord with recent calculations ( pringle 1996 , 1997 ; @xcite ) , showing that the accretion disks in active galactic nuclei may be severely warped by the local radiation source , and also with observations of sub - parsec - scale warped disks traced out by vlbi imaging of water megamasers in ngc 4258 ( herrnstein et al .
one could imagine that the larger scale vlba and vla sources represent the `` average '' axis of the disk , while the small - scale vlba source shows the instantaneous direction of an inner disk that might be determined by the gas most recently added to that disk .
the amount of jet curvature , or disk precession or warping , required would be reduced considerably if we happen to be viewing the radio jet nearly end - on , so that a small change in angle could appear much larger in projection .
the vlba and vla have been used to image the continuum emission from mrk 231 on scales ranging from parsecs to kiloparsecs .
an asymmetric , diffuse radio source is traced for more than 25 kpc .
it exhibits linear polarization as high as 57% , has a ridge of modest brightness aligned with a starburst region several kiloparsecs south of the galaxy nucleus and ( roughly ) with the 40-pc vlba triple source , and appears to be powered by energy deposition from the jet .
this diffuse radio source extends beyond the bulk of the optical emission from mrk 231 . inside the diffuse radio emission
, a 350-pc disk of radio continuum and hi ( cwu ) , appears to be caused by synchrotron emission that may be free - free absorbed in some regions .
this is consistent with the inference that the disk is powered by a massive starburst .
the 40-pc vlba triple source exhibits free - free absorption in both its northern and southern components at frequencies of 23 ghz , implying the presence of ionized gas clouds several parsecs in diameter with densities of 12@xmath96 @xmath3 .
these clouds may reside in the ionized inner region of the same disk responsible for the hi absorption ( @xcite ) .
their ionization may be powered either by the active galactic nucleus or by a few o stars .
the central component of the 40-pc source has an elongation pa @xmath84from that of the 40-pc vlba triple .
this core contains at least two components , with brightness temperatures of @xmath88 k and @xmath97 k , each of which is absorbed at low frequencies .
the data are consistent with the presence of synchrotron self - absorption between 5 and 10 ghz in the stronger component , implying a magnetic field of @xmath98 gauss .
the weaker component , about 1 pc to the northeast , is absorbed below 23 ghz , due either to free - free absorption or synchrotron self - absorption .
all data support the presence of two symmetry axes in mrk 231 : ( 1 ) an inner axis in pa @xmath84 that determines the initial direction of the radio jet , and is likely to be associated with the current axis of a central black hole or its accretion disk ; and ( 2 ) an outer axis that is near pa 0 , perpendicular to a disk or torus that is ionized in its inner 1020 pc radius , neutral out to @xmath99 pc from the center , and then molecular out to @xmath100 pc radius .
the outer axis presumably represents the average long - term axis of the central engine .
these two axes are substantially misaligned unless the parsec - scale radio source is viewed nearly end - on , so that small curvature would be enhanced greatly by projection .
we are grateful to bill keel for supplying the optical image of mrk 231 , and to greg taylor for useful discussions .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
ulvestad , j. s. , wrobel , j. m. , & carilli , c. l. 1998 , in iau colloquium 164 : radio emission from galactic and extragalactic compact sources , asp conf .
144 , eds .
j. a. zensus , g. b. taylor , & j. m. wrobel ( san francisco : asp ) , 199 lclrccc 1995 nov 13
& 1123 & vlba & 1.367 & 32 & lcp & 154 & & & 2.271 & 32 & rcp & 143 & & & 4.987 & 32 & lcp & 132 1996 dec 8 & 0921 & vlba+y1 & 4.987 & 32 & lcp & 152 & & & 8.421 & 32 & rcp & 152 & & & 15.365 & 32 & lcp & 156 1996 dec 27 & 0820 & vlba+yp & 1.367 & 16 & dual & 380 & & & 22.233 & 32 & lcp & 96 [ tab : vlbaobs ] lcrcr 1988 aug 10 & d & 14.940 & 50 & 186 1989 apr 14 & b & 1.490 & 50 & 86 1989 sep 1 & c & 4.860 & 50 & 88 1995 nov 17 & b & 1.425 & 50 & 11 & & 4.860 & 50 & 11 & & 8.440 & 50 & 11 & & 14.940 & 50 & 11 & & 22.460 & 50 & 11 1996 dec 27 & a & 1.365 & 25 & 380 & & 4.985 & 50 & 2 & & 8.415 & 50 & 2 & & 15.365 & 50 & 2 & & 22.235 & 50 & 96 [ tab : vlaobs ] lrcrr
n & 1.4 & @xmath101 & 2.64 & 23.19 & 2.3 & @xmath102 & @xmath1030.17 & 20.56 & 5.0 & @xmath104 & @xmath1030.58 & 19.63 & 8.4 & @xmath105 & @xmath1031.45 & 19.07 c & 1.4 & @xmath106 & 0.00 & 0.00 & 2.3 & @xmath107 & 0.00 & 0.00 & 5.0 & @xmath108 & 0.00 & 0.00 & 8.4 & @xmath109 & 0.00 & 0.00 & 15.4 & @xmath110 & 0.00 & 0.00 s & 1.4 & @xmath111 & @xmath1035.34 & @xmath10326.50 & 2.3 & @xmath112 & @xmath1035.06 & @xmath10327.48 & 5.0 & @xmath113 & @xmath1034.42 & @xmath10328.28 & 8.4 & @xmath102 & @xmath1035.65 & @xmath10327.67 [ tab : nsflux ] lrccrr c & 5.0 & @xmath114 & @xmath115 , pa 63.0 & 0.00 & 0.00 & 8.4 & @xmath116 & @xmath117 , pa 61.8 & 0.00 & 0.00 & 15.4 & @xmath118 & & 0.00 & 0.00 c1
&8.4 &
@xmath119 & @xmath120 , pa 70.1 & 0.00 & 0.00 c2
&8.4 & @xmath121 & & 1.00 & 0.49 c1 & 15.4 & @xmath122 & @xmath123 , pa 70.8 & 0.00 & 0.00 c2 & 15.4 & @xmath124 & & 1.03 & 0.44 c1 & 15.4 & @xmath125 & @xmath126 , pa 92.0 & 0.00 & 0.00 c2 & 15.4 & @xmath127 & & 0.98 & 0.43 c3 & 15.4 & @xmath128 & & @xmath1030.74 & @xmath1030.42 [ tab : cflux ] rcrc 1.425 & @xmath136 & 1.365 & @xmath137 + 4.860 & @xmath138 & 4.985 & @xmath139 + 8.440 & @xmath140 & 8.415 & @xmath141 + 14.940 & @xmath142 & 15.365 & @xmath143 + 22.460 & @xmath144 & 22.235 & @xmath145 + [ tab : vlaflux ] | the very long baseline array ( vlba ) and the very large array ( vla ) have been used to image the continuum radio emission from mrk 231 , a seyfert 1 galaxy and the brightest infrared galaxy in the local universe .
the smallest vlba scales reveal a double , or possibly triple , source less than 2 pc in extent .
the components of this central source have minimum brightness temperatures of @xmath0@xmath1 k , spectral turnovers between 2 and 10 ghz , and appear to define the galaxy nucleus plus the inner regions of a jet .
the strongest component is probably synchrotron self - absorbed , while the weaker component to the northeast may be either free - free absorbed or synchrotron self - absorbed . on larger vlba scales ,
the images confirm a previously known north - south triple source extending 40 pc and elongated perpendicular to a 350-pc starburst disk traced by hi and co. both lobes of the triple show evidence for free - free absorption near 2 ghz , probably due to ionized gas with a density of 12@xmath2 @xmath3 in the innermost regions of the starburst disk .
this free - free absorption resembles that toward the counterjet of 3c 84 , but requires ionized gas at lower density located considerably farther from the central source .
the absorbing gas may be ionized by the active nucleus or by local regions of enhanced star formation , possibly in the inner part of the starburst disk .
the elongation position angle of the 40-pc triple differs by 65 from that of the 2-pc source .
unless the radio source is seen nearly end - on , the different symmetry axes on different scales in mrk 231 imply a dramatic curvature in the inner part of the mrk 231 radio jet .
a comparison of vlba and vla flux densities indicates that the radio continuum from the 350-pc disk has a spectral index near @xmath4 at frequencies above 1.4 ghz and is plausibly energized by a massive burst of star formation , with the overall spectrum flattened somewhat by a contribution from free - free absorption . on vla scales ,
asymmetric and diffuse emission extends for more than 25 kpc .
this emission has a steep spectrum , exhibits linear polarization exceeding 50% at some locations , and shares the symmetry axis of the 40-pc triple , but on a scale larger by three orders of magnitude .
the large - scale radio emission extends beyond the bulk of the optical galaxy , but has an initial axis similar to a series of optical star - forming knots several kiloparsecs from the nucleus .
this diffuse radio source is probably generated by energy deposition from a slow - moving nuclear jet , which conceivably could help energize the off - nuclear starburst as well . 2 |
subordination is a powerful tool in the study of random processes . in the present work ,
we investigate subordination of random trees , and we apply our results to properties of the random metric space called the brownian map , which has been proved to be the universal scaling limit of many different classes of random planar maps ( see in particular @xcite ) .
these applications have been motivated by the work of miller and sheffield @xcite , which is part of a program aiming at the construction of a conformal structure on the brownian map ( see @xcite for recent developments in this direction ) . to explain our starting point ,
let us consider a compact @xmath0-tree @xmath2 .
this means that @xmath2 is a compact metric space such that , for every @xmath3 , there exists a unique ( continuous injective ) path from @xmath4 to @xmath5 , up to reparameterization , and the range of this path , which is called the geodesic segment between @xmath4 and @xmath5 and denoted by @xmath6 , is isometric to a compact interval of the real line .
we assume that @xmath2 is rooted , so that there is a distinguished point @xmath7 in @xmath2 .
this allows us to define a generalogical order on @xmath2 , by saying that @xmath8 if and only if @xmath9 .
consider then a continuous function @xmath10 , such that @xmath11 and @xmath12 is nondecreasing for the genealogical order .
the basic idea of subordination is to identify @xmath4 and @xmath5 if @xmath12 is constant on the geodesic segment @xmath6 .
so , for every @xmath13 , the set of all points that are identified with @xmath4 is a closed connected subset of @xmath2 .
this glueing operation yields another compact @xmath0-tree @xmath14 , which is equipped with a metric such that the distance between @xmath7 and @xmath4 is @xmath15 and is called the subordinate tree of @xmath2 with respect to @xmath12 ( see fig.[subordina - tree ] for an illustration ) .
furthermore , if our initial tree @xmath2 was given as the tree coded by a continuous function @xmath16\la\r_+$ ] ( see @xcite or section [ sec : determin ] below ) , the subordinate tree @xmath14 is coded by @xmath17 , where @xmath18 is the canonical projection from @xmath19 $ ] onto @xmath2 . on the left side , the tree @xmath2 , with segments where @xmath12 is constant pictured in thin blue lines . on the right side ,
the subordinate tree , where each connected component of @xmath2 made of thin segments has been glued into a single point.,width=529 ] our main interest is in the case where @xmath2 is random , and more precisely @xmath20 is the `` brownian tree '' coded by a positive brownian excursion @xmath21 under the it excursion measure .
one may view @xmath22 as a variant of aldous brownian crt , for which the total mass is not finite , but is distributed according to an infinite measure on @xmath23 . as previously , we write @xmath24 for the canonical projection from @xmath19 $ ] onto @xmath22 .
next , to define the subordination function , we let @xmath25 be ( linear ) brownian motion indexed by @xmath22 , starting from @xmath26 at the root @xmath7 .
a simple way to construct this process is to use the brownian snake approach , letting @xmath27 if @xmath28 , where @xmath29 is the `` tip '' of the random path @xmath30 , which is the value at time @xmath31 of the brownian snake driven by @xmath32 .
since @xmath32 is distributed according to the it measure , @xmath33 follows the brownian snake excursion measure away from @xmath26 , which we denote by @xmath34 ( see ( * ? ? ?
* chapter iv ) ) .
we then set @xmath35 . in terms of the brownian snake , we have @xmath36 whenever @xmath28 .
we also use the notation @xmath37 .
[ subor - brown ] let @xmath38 stand for the subordinate tree of @xmath22 with respect to the ( continuous nondecreasing ) function @xmath39 . under the brownian snake excursion measure @xmath34 , the tree @xmath40 is a lvy tree with branching mechanism @xmath41 recall that lvy trees represent the genealogy of continuous - state branching processes @xcite , and can be characterized by a regenerative property analogous to the branching property of galton watson trees @xcite .
our identification of the distribution of @xmath38 is reminiscent of the classical result stating that the right - continuous inverse of the maximum process of a standard linear brownian motion is a stable subordinator with index @xmath42 . in view of our applications , it turns out that it is important to have more information than the mere identification of @xmath38 as a random compact @xmath0-tree .
as mentioned above , @xmath38 can be viewed as the tree coded by the random function @xmath43 .
this coding induces a `` lexicographical '' order structure on @xmath22 ( see @xcite for a thorough discussion of order structures on @xmath0-trees ) .
somewhat surprisingly , it is not immediately clear that the order structure on @xmath38 induced by the coding from the function @xmath44 coincides with the usual order structure on lvy trees , corresponding to the uniform random shuffling at every node in the terminology of @xcite . to obtain this property ,
we relate the function @xmath44 to the height process of @xcite .
we recall that , from a lvy process with laplace exponent @xmath45 , we can construct a continuous random process @xmath46 called the height process , which codes the @xmath45-lvy tree ( here and below we say @xmath45-lvy tree rather than lvy tree with branching mechanism @xmath45 for simplicity ) .
see section [ sec : levy - tree ] below for more details .
[ subordi - coding ] there exists a process @xmath47 , which is distributed under @xmath34 as the height process of a lvy tree with branching mechanism @xmath48 , and a continuous random process @xmath49 with nondecreasing sample paths , such that we have , @xmath34 a.e . for every @xmath50 ,
@xmath51 both @xmath47 and @xmath49 will be constructed in the proof of theorem [ subordi - coding ] , and are measurable functions of @xmath52 .
it is possible to identify the random process @xmath49 as a continuous additive functional of the brownian snake , but we do not need this fact in the subsequent applications , and we do not discuss this matter in the present work .
theorem [ subordi - coding ] implies that the tree coded by @xmath53 is isometric to the tree coded by @xmath54 , and we recover theorem [ subor - brown ] .
but theorem [ subordi - coding ] gives much more , namely that the order structure induced by the coding via @xmath55 is the same as the order structure induced by the usual height function of the lvy tree .
order structures are crucial for our applications to the brownian map .
similarly as in @xcite , we deal with a version of the brownian map with randomized volume , which is constructed as a quotient space of the tree @xmath22 : two points @xmath4 and @xmath5 of @xmath22 are identified if @xmath56 and if @xmath57 for every @xmath58 $ ] , where @xmath59 $ ] is the set of all points that are visited when going from @xmath4 to @xmath5 around the tree in `` clockwise '' order ( for this to make sense , it is essential that @xmath22 has been equipped with a lexicographical order structure ) .
we write @xmath60 for the resulting quotient space ( the brownian map ) and @xmath61 for the metric on the brownian map ( see @xcite for more details ) .
the space @xmath60 comes with two distinguished points , namely the root @xmath7 of @xmath22 and the unique point @xmath62 where @xmath63 attains its minimal value in a sense that can be made precise , these two points are independent and uniformly distributed over @xmath60 .
let @xmath64 be the closed ball of radius @xmath65 centered at @xmath62 in @xmath60 .
for every @xmath66 , define the _ hull _ @xmath67 as the complement of the connected component of the complement of @xmath64 that contains @xmath7 ( informally , @xmath67 is obtained by filling in all holes of @xmath64 except for the one containing @xmath7 ) .
following @xcite , we define the metric net @xmath68 as the closure of @xmath69 ( this definition is in fact a little different from @xcite which does not take the closure of the union in the last display . ) we can equip the set @xmath70 with an `` intrinsic '' metric @xmath71 derived from the brownian map metric @xmath61 .
it is not hard to verify that , in the construction of @xmath60 as a quotient space of @xmath22 , points of @xmath68 exactly correspond to vertices @xmath4 of @xmath22 such that @xmath72 ( proposition [ descri - net ] ) .
this suggests that the metric net @xmath68 is closely related to the subordinate tree of @xmath22 with respect to the function @xmath73 , which is a @xmath45-lvy tree by the preceding results . to make this relation precise ,
we need the notion of the looptree introduced by curien and kortchemski @xcite in a more general setting . informally , the looptree associated with a lvy tree is constructed by replacing each point @xmath4 of infinite multiplicity by a loop of `` length '' equal to the `` weight '' of @xmath4 , in such a way that the subtrees that are the connected components of the complement of @xmath4 branch along this loop in an order determined by the coding function ( see fig.[looptree ] below for an illustration ) . to give a more precise definition , first note that theorem [ subordi - coding ] and an obvious symmetry argument allow us to find a process @xmath74 distributed as the height process of a @xmath45-lvy tree , and a continuous random process @xmath75 with nondecreasing sample paths such that @xmath76 .
then let @xmath77 , and let @xmath78 be the equivalence relation on @xmath79 $ ] whose graph is the smallest closed symmetric subset of @xmath79 ^ 2 $ ] that contains all pairs @xmath80 with @xmath81 , @xmath82 and @xmath83 for every @xmath84 .
the looptree @xmath85 is defined as the quotient space @xmath79/\sim$ ] equipped with an appropriate metric ( see @xcite ) .
clearly , @xmath86 makes sense for any @xmath87 and is interpreted as the height of @xmath88 .
the metric on @xmath85 is in fact not relevant for us , since we consider instead the pseudo - metric @xmath89 defined for @xmath90 by @xmath91 } h'_\gamma , \max_{\gamma\in[\beta,\alpha ] } h'_\gamma\bigg ) - h'_\alpha -h'_\beta,\ ] ] where @xmath92 $ ] corresponds to the subset of @xmath85 visited when going from @xmath88 to @xmath93 `` around '' @xmath85 in `` clockwise order '' ( see section [ sec : br - map ] for a more formal definition ) .
we write @xmath94 if @xmath95 ( informally this means that @xmath88 and @xmath93 `` face each other '' in the tree , in the sense that they are at the same height , and that points `` between '' @xmath88 and @xmath93 are at a smaller height ) .
it turns out that this defines an equivalence relation on @xmath85 . finally , we let @xmath96 be the largest symmetric function on @xmath97 that is bounded above by @xmath89 and satisfies the triangle inequality .
[ identi - net ] the metric net @xmath98 is a.s .
isometric to the quotient space @xmath99 equipped with the metric induced by @xmath96 .
see theorem [ identi - metricnet ] in section [ sec : br - map ] for a more precise formulation .
theorem [ identi - net ] is closely related to proposition 4.4 in @xcite , where , however , the metric net is not identified as a metric space .
the description of the metric net is an important ingredient of the axiomatic characterization of the brownian map discussed in @xcite .
let us briefly comment on the motivations for studying the metric net .
roughly speaking , the brownian map @xmath60 can be recovered from the metric net @xmath68 by filling in the `` holes '' . to make this precise , we observe that the connected components of @xmath100 are bounded by jordan curves ( proposition [ connec - compo ] ) and that these components are in one - to - one correspondence with connected components of @xmath101 , where @xmath102 .
each of the latter components is associated with an excursion of the brownian snake above its minimum , in the terminology of @xcite , and the distribution of such an excursion only depends on its boundary size as defined in @xcite ( this boundary size can be interpreted as a generalized length of the jordan curve bounding the corresponding component of @xmath100 ) .
theorem 40 in @xcite shows that , conditionally on their boundary sizes , these excursions are independent and distributed according to a certain `` excursion measure '' . in the brownian map setting , this means that the holes in the metric net are filled in independently , conditionally on the lengths of their boundaries .
the paper is organized as follows .
section 2 gives a brief discussion of subordination for deterministic trees , and section 3 recalls the basic facts about lvy trees that we need .
after a short presentation of the brownian snake , section 4 gives the distribution of the subordinate tree @xmath38 ( theorem [ subor - brown ] ) . in view of identifying the order structure of this subordinate tree , section 5 provides a technical result showing that the height process coding a lvy tree is the limit in a strong sense of the discrete height functions coding embedded galton watson trees .
this result is related to the general limit theorems of ( * ? ? ?
* chapter 2 ) proving that lvy trees are weak limits of galton watson trees , but the fact that we get a strong approximation is crucial for our applications . in section 6 , we prove that the brownian snake maximum process @xmath103 is a time change of the height process associated with a @xmath45-lvy tree ( theorem [ subordi - coding ] ) .
this result is a key ingredient of the developments of section 7 , where we identify the metric net of the brownian map ( theorem [ identi - net ] ) .
section 8 discusses the connected components of the complement of the metric net , showing in particular that they are in one - to - one correspondence with the points of infinite multiplicity of the associated lvy tree , and that the boundary of each component is a jordan curve .
section 9 , which is mostly independent of the preceding sections , discusses more general subordinations of the brownian tree @xmath22 , which lead to stable lvy trees with an arbirary index .
this section is related to our previous article @xcite , which dealt with subordination for spatial branching processes , but the latter work did not consider the associated genealogical structures as we do here , and the subordination method , based on the so - called residual lifetime process , was also different .
finally , the appendix presents a more general and more precise version of the special markov property of the brownian snake ( first established in @xcite ) , which plays an important role in several proofs .
in this short section , we present a few elementary considerations about deterministic @xmath0-trees .
we refer to @xcite for the basic facts about @xmath0-trees that we will need , and to @xcite for a thorough study of the coding of compact @xmath0-trees by functions .
let us consider a compact @xmath0-tree @xmath104 rooted at @xmath7 .
if @xmath3 , the geodesic segment between @xmath4 and @xmath5 ( the range of the unique geodesic from @xmath4 to @xmath5 ) is denoted by @xmath6 .
the point @xmath105 is then defined by @xmath106 .
the genealogical partial order on @xmath2 is denoted by @xmath107 : we have @xmath8 if and only if @xmath9 , and we then say that @xmath4 is an ancestor of @xmath5 , or @xmath5 is a descendant of @xmath4 .
finally , the height of @xmath2 is defined by @xmath108 let @xmath109 be a nonnegative continuous function on @xmath2 .
assume that @xmath11 and that @xmath12 is nondecreasing with respect to the genealogical order ( @xmath8 implies that @xmath110 ) .
we then define , for every @xmath3 , @xmath111 notice that @xmath112 is a symmetric function of @xmath4 and @xmath5 and satisfies the triangle inequality .
we can thus consider the equivalence relation @xmath113 thus @xmath114 if and only if @xmath115 , and this is also equivalent to saying that @xmath116 for every @xmath117 .
write @xmath118 for the quotient @xmath119 , and @xmath120 for the canonical projection from @xmath2 onto @xmath118 .
[ tree - quotient ] the set @xmath118 equipped with the distance induced by @xmath121 is again a compact @xmath0-tree .
one immediately verifies that , for every @xmath122 , @xmath123 is a segment in @xmath124 with endpoints @xmath125 and @xmath126 . from lemma 3.36 in @xcite , it then suffices to check that the four - point condition @xmath127 holds for every @xmath128 .
this is straightforward and we omit the details .
we call @xmath124 the subordinate tree of @xmath104 with respect to the function @xmath12 . by convention , @xmath118 is rooted at @xmath125 . since @xmath129 for every @xmath13 ,
we have @xmath130 . consider now a continuous function @xmath16\la \r_+$ ] , where @xmath131 , such that @xmath132 , and assume that @xmath104 is the tree coded by @xmath133 in the sense of @xcite or @xcite .
this means that @xmath134 is the quotient space @xmath19/\!\sim_h$ ] , where the equivalence relation @xmath135 is defined on @xmath19 $ ] by @xmath136 and @xmath137 is the distance induced on the quotient space by @xmath138 notice that the topology of @xmath139 coincides with the quotient topology on @xmath140 .
the canonical projection from @xmath19 $ ] onto @xmath140 is denoted by @xmath18 , and @xmath140 is rooted at @xmath141 . for @xmath142 $ ] , the quantity @xmath143 is interpreted as the height of @xmath144 in the tree .
one easily verifies that , for every @xmath145 $ ] , the property @xmath146 holds if and only if @xmath147 .
the function @xmath133 is not determined by @xmath140 .
in particular , if @xmath148\to [ 0,\sigma]$ ] is continuous and nondecreasing , and such that @xmath149 and @xmath150 , the tree coded by @xmath151 is isometric to the tree coded by @xmath133 .
this simple observation will be useful later .
[ tree - coding ] under the preceding assumptions , if @xmath12 is a nonnegative continuous function on @xmath140 such that @xmath152 and @xmath12 is nondecreasing with respect to the genealogical order on @xmath140 , the subordinate tree @xmath153 of @xmath139 with respect to the function @xmath12 is isometric to the tree @xmath154 coded by the function @xmath155 .
note that the function @xmath156 is nonnegative and continuous on @xmath19 $ ] , and @xmath157 .
we can therefore make sense of the tree @xmath154 and as above we denote the canonical projection from @xmath19 $ ] onto @xmath158 by @xmath159 .
we first notice that , for every @xmath145 $ ] , the property @xmath160 implies @xmath161 . indeed ,
if @xmath160 , then , for every @xmath162 $ ] , we have @xmath163 and therefore @xmath164 , so that @xmath165 } g(r),\ ] ] and @xmath161 . we can thus write @xmath166 , where the function @xmath167 is continuous and onto
. then , let @xmath168 and write @xmath169 and @xmath170 , with @xmath145 $ ] .
we note that , for every @xmath171 $ ] , @xmath172 , and furthermore @xmath173 , if @xmath174 is any element of @xmath175 $ ] at which @xmath133 attains its minimum over @xmath175 $ ] .
it follows that @xmath176 } g(r).\ ] ] hence , @xmath177 } g(r)= d_g(s , t).\ ] ] if @xmath120 is the projection from @xmath140 onto the subordinate tree @xmath178 , we see in particular that the condition @xmath179 implies @xmath161 and therefore @xmath180 .
it follows that @xmath181 , where @xmath182 is onto .
it remains to verify that @xmath183 is isometric , but this is immediate from the identities in the last display .
it is known that any compact @xmath0-tree can be represented in the form @xmath140 for some function @xmath133 ( see ( * ? ? ?
* corollary 1.2 ) ) .
thus proposition [ tree - coding ] provides an alternative proof of proposition [ tree - quotient ] .
in the next sections , we will consider the case where @xmath2 is the ( random ) tree coded by a brownian excursion distributed under the it excursion measure @xmath184 , and we will identify certain subordinate trees as lvy trees . in this section ,
we recall the basic facts about lvy trees that will be needed later .
we refer to @xcite for more details .
we consider a nonnegative function @xmath185 defined on @xmath186 of the type @xmath187 where @xmath188 , and @xmath189 is a @xmath190-finite measure on @xmath23 such that @xmath191 . with any such function @xmath185
, we can associate a continuous - state branching process ( see @xcite and references therein ) , and @xmath185 is then called the branching mechanism function of this process .
notice that the conditions on @xmath185 are not the most general ones , because we restrict our attention to the critical or subcritical case .
additionally , we will assume that @xmath192 this condition , which implies that at least one of the two properties @xmath193 or @xmath194 holds , is equivalent to the a.s .
extinction of the continuous - state branching process with branching mechanism @xmath185 @xcite .
special cases include @xmath195 for @xmath196 . under the preceding assumptions
, one can make sense of the lvy tree that describes the genealogy of the continuous - state branching process with branching mechanism @xmath185 .
we consider , under a probability measure @xmath197 , a spectrally positive lvy process @xmath198 with laplace exponent @xmath185 , meaning that @xmath199=\exp(t\psi(\lambda))$ ] for every @xmath200 and @xmath201 .
we define the associated height process by setting , for every @xmath200 , @xmath202 where the limit holds in probability under @xmath197 .
then ( * ? ? ?
* theorem 1.4.3 ) the process @xmath203 has a continuous modification , which we consider from now on .
we have @xmath204 if and only if @xmath205 , where @xmath206 is the past minimum process of @xmath207 let @xmath208 stand for the ( infinite ) excursion measure of @xmath209 . here
the normalization of @xmath208 is fixed by saying that the local time at @xmath26 of @xmath209 is the process @xmath210 .
let @xmath211 stand for the duration of the excursion under @xmath208 .
the height process @xmath47 is well defined ( and has continuous paths ) under @xmath208 , and we have @xmath212 , @xmath208 a.e . to simplify notation , we will write @xmath213 under @xmath208 . by definition ( see ( * ? ? ?
* definition 4.1 ) ) , the lvy tree with branching mechanism @xmath185 ( or in short the @xmath185-lvy tree ) is the random compact @xmath0-tree @xmath214 coded by the function @xmath215 under @xmath208 , or more generally any random tree with the same distribution note that the distribution of the lvy tree is an infinite measure .
we refer to @xcite for several results explaining in which sense the lvy tree codes the genealogy of the continuous - state branching process with branching mechanism @xmath185 . in the special case @xmath216 ,
@xmath207 is just a standard linear brownian motion , @xmath217 is twice a reflected brownian motion , and the lvy tree is the tree coded by ( twice ) a positive brownian excursion under the ( suitably normalized ) it measure .
conditioning on @xmath218 then yields the brownian continuum random tree . when @xmath195 with @xmath219 , one gets the stable tree with index @xmath220 .
the distribution of the height of a lvy tree is given as follows .
for every @xmath221 , @xmath222 where the function @xmath223 is determined by @xmath224 in the preceding considerations , the normalization of the infinite measure @xmath208 is fixed by our choice of the local time at @xmath26 of @xmath209 , and we can recover @xmath185 from @xmath208 by the formulas for the distribution of @xmath225 under @xmath208 . what happens if we multiply @xmath208 by a constant @xmath201 ?
the tree @xmath214 under @xmath226 is still a lvy tree in the previous sense , but the associated branching mechanism is now @xmath227 . to see this , consider the lvy process @xmath228 , whose laplace exponent is @xmath229 .
it is not hard to verify that the height process corresponding to @xmath230 is @xmath231 .
furthermore , if @xmath232 is the excursion measure of @xmath230 above its past minimum process , one also checks that the distribution of @xmath233 under @xmath232 is the distribution of @xmath203 under @xmath226 .
however , the tree coded by @xmath233 is the same as the tree coded by @xmath234 .
this shows that @xmath214 under @xmath226 is a lvy tree with branching mechanism @xmath235 .
note that this is consistent with the formula for @xmath236 .
we now state two results that will be important for our purposes .
we first mention that , for every @xmath237 , one can define under @xmath197 a local time process of @xmath47 at level @xmath133 , which is denoted by @xmath238 and is such that the following approximation holds for every @xmath239 , @xmath240 = 0\,,\ ] ] and the latter convergence is uniform in @xmath133 ( see ( * ? ? ?
* proposition 1.3.3 ) ) .
when @xmath241 we have simply @xmath242 .
the definition of @xmath238 also makes sense under @xmath208 , with a similar approximation .
we fix @xmath221 , and let @xmath243 be the collection of all excursion intervals of @xmath47 above level @xmath133 : these are all intervals @xmath244 , with @xmath245 , such that @xmath246 and @xmath247 for every @xmath248 . for each such excursion interval @xmath249
, we define the corresponding excursion by @xmath250 , for every @xmath50 .
then @xmath251 is a random element of the space @xmath252 of all continuous functions from @xmath253 into @xmath253 .
we also let @xmath254 be the @xmath190-finite measure on @xmath252 which is the `` law '' of @xmath255 under @xmath208 .
[ excu - above ] ( i ) under the probability measure @xmath197 , the point measure @xmath256 is poisson with intensity @xmath257 .
\(ii ) under the probability measure @xmath258 and conditionally on @xmath259 , the point measure @xmath256 is poisson with intensity @xmath260}(\ell)\,\d \ell\,\mathbf{n}^\circ(\d \omega)$ ]
. see ( * ? ? ?
* proposition 3.1 , corollary 3.2 ) for a slightly more precise version of this proposition .
it follows from the preceding proposition that lvy trees satisfy a branching property analogous to the classical branching property of galton watson trees . to state this property
, we introduce some notation .
if @xmath104 is a ( deterministic ) compact @xmath0-tree and @xmath261 , we can consider the subtrees of @xmath2 above level @xmath133 . here
, a subtree above level @xmath133 is just the closure of a connected component of @xmath262 .
such a subtree is itself viewed as a rooted @xmath0-tree ( the root is obviously the unique point at height @xmath133 in the subtree ) .
[ branching - levy ] let @xmath2 be a random compact @xmath0-tree defined under an infinite measure @xmath263 , such that @xmath264 and @xmath265 for every @xmath221 . for every @xmath266 , write @xmath267 for the number of subtrees of @xmath2 above level @xmath133 with height greater than @xmath268 . 1 .
suppose that @xmath2 is a lvy tree .
then , for every @xmath266 , for every integer @xmath269 , the distribution under @xmath270 of the unordered collection formed by the @xmath271 subtrees of @xmath2 above level @xmath133 with height greater than @xmath268 is the same as that of the unordered collection of @xmath271 independent copies of @xmath2 under @xmath272 .
conversely , if the property stated in ( i ) holds , then @xmath2 is a lvy tree .
the property stated in ( i ) is called the branching property . the fact that it holds for lvy trees is a straightforward consequence of proposition [ excu - above ] ( ii ) . for the converse
, we refer to ( * ? ? ?
* theorem 1.1 ) .
note that the branching property remains valid if we multiply the underlying measure @xmath263 by a positive constant , which is consistent with the remark above . to conclude this section ,
let us briefly comment on points of infinite multiplicity of the lvy tree @xmath214 .
the multiplicity of a point @xmath4 of @xmath214 is the number of connected components of @xmath273 , and @xmath4 is called a leaf if it has multiplicity one .
suppose that there is no quadratic part in @xmath185 , meaning that the constant @xmath93 in is @xmath26 ( note that condition then implies that @xmath274 has infinite mass ) .
then ( * ? ? ?
* theorem 4.6 ) , all points of @xmath214 have multiplicity @xmath275 , @xmath276 or @xmath277 .
the set of all points of infinite multiplicity is a countable dense subset of @xmath214 , and these points are in one - to - one correspondence with local minima of @xmath47 , or with jump times of @xmath207 .
more precisely , let @xmath4 be a point of infinite multiplicity of @xmath214 . then ,
if @xmath278 and @xmath279 , we have @xmath280 , @xmath281 , and @xmath282 : h_s= h_{s_1}\}$ ] is a cantor set contained in @xmath283 $ ] . in terms of the lvy process
@xmath207 , @xmath284 is a jump time of @xmath207 and @xmath285 , and @xmath286 consists exactly of those @xmath287 $ ] such that @xmath288\}=x_s$ ] .
furthermore , if for every @xmath289 $ ] we set @xmath290 , the points of @xmath286 are all of the form @xmath291 or @xmath292 .
the quantity @xmath293 is the `` weight '' of the point of infinite multiplicity @xmath4 .
see @xcite for more details .
in this section , we prove theorem [ subor - brown ] . we start with a brief presentation of the brownian snake .
we refer to ( * ? ? ?
* chapter iv ) for more details .
we let @xmath294 be the space of all finite paths in @xmath0 . here
a finite path is simply a continuous mapping @xmath295\la \r$ ] , where @xmath296 is a nonnegative real number called the lifetime of @xmath297 .
the set @xmath294 is a polish space when equipped with the distance @xmath298 the endpoint ( or tip ) of the path @xmath297 is denoted by @xmath299 . for every @xmath300 , we set @xmath301 . we also identify the trivial path of @xmath302 with zero lifetime with the point @xmath303 . the standard ( one - dimensional ) brownian snake with initial point @xmath303 is the continuous markov process @xmath52 taking values in @xmath302 , whose distribution is characterized by the following properties : 1 .
the process @xmath304 is a reflected brownian motion in @xmath253 started from @xmath26 . to simplify notation , we write @xmath305 for every @xmath50 .
2 . conditionally on @xmath306 , the process @xmath52 is time - inhomogeneous markov , and its transition kernels are specified as follows . if @xmath307 , 1 .
@xmath308 for every @xmath309 ; 2 .
the random path @xmath310 is independent of @xmath30 and distributed as a real brownian motion started at @xmath26 and stopped at time @xmath311 .
informally , the value @xmath30 of the brownian snake at time @xmath31 is a one - dimensional brownian path started from @xmath303 , with lifetime @xmath312 .
as @xmath31 varies , the lifetime @xmath312 evolves like reflected brownian motion in @xmath253 .
when @xmath312 decreases , the path is erased from its tip , and when @xmath312 increases , the path is extended by adding `` little pieces '' of brownian paths at its tip . for every @xmath313 , we write @xmath314 for the probability measure under which @xmath315 , and @xmath316 for the ( infinite ) excursion measure of @xmath33 away from @xmath303 .
also @xmath317 stands for the duration of the excursion under @xmath316 . under @xmath316 ,
@xmath318 is distributed according to the it excursion measure @xmath184 , and the normalization is fixed by the formula @xmath319 the following property of the brownian snake will be used in several places below .
recall that @xmath320 is the tip of the path @xmath30 .
we say that @xmath321 is a time of right increase of @xmath322 , resp . of @xmath323 ,
if there exists @xmath324 $ ] such that @xmath325 , resp .
@xmath326 , for every @xmath327 $ ] .
we can similarly define points of left increase . then according to ( * ? ? ?
* lemma 3.2 ) , @xmath316 a.e .
, no time @xmath328 can be simultaneously a time of ( left or right ) increase of @xmath322 and a point of ( left or right ) increase of @xmath323 .
le us fix @xmath329 and argue under @xmath34 .
as previously , we write @xmath22 for the ( random ) tree coded by the function @xmath330 , @xmath331\la \t_\zeta$ ] for the canonical projection , and @xmath332 for the root of @xmath22 . properties of the brownian snake show that the condition @xmath333 implies @xmath334 , and thus we can define @xmath335 for every @xmath336 by setting @xmath337 if @xmath28 .
we note that , if @xmath28 and @xmath338 $ ] , @xmath339 coincides with @xmath340 where @xmath5 is the unique point of @xmath341 at distance @xmath342 from @xmath343 . for every @xmath344 , set @xmath345 then , the function @xmath346 is continuous and nondecreasing on @xmath22 ( if @xmath28 and @xmath347 , the condition @xmath8 implies that @xmath348 and that @xmath335 is the restriction of @xmath349 to the interval @xmath350 $ ] , so that obviously @xmath351 ) . as in theorem
[ subor - brown ] , we write @xmath38 for the subordinate tree of @xmath22 with respect to the function @xmath346 .
_ proof of theorem [ subor - brown ] .
_ we first verify that the branching property stated in proposition [ branching - levy ] holds for @xmath40 under @xmath34 , and to this end we rely on the special markov property of the brownian snake .
let @xmath221 , and , for every @xmath344 , set @xmath352:\w(t)\geq h\}$ ] .
let @xmath353 be the connected components of the open set @xmath354 .
for every such connected component @xmath355 , for every @xmath356 $ ] , the path @xmath30 coincides with @xmath357 up to time @xmath358 ( these assertions are straightforward consequences of the properties of the brownian snake , and we omit the details ) .
we then set , for every @xmath50 , @xmath359 we view @xmath360 as a random element of the space of all continuous functions from @xmath253 into @xmath361 , and the @xmath360 s are called the excursions of @xmath33 outside the domain @xmath362 ( see the appendix below for further details in a more general setting ) . by a compactness argument ,
only finitely many of the excursions @xmath360 hit @xmath363 .
let @xmath364 be the number of these excursions .
it follows from corollary [ smpusual ] in the appendix that , for every @xmath269 , conditionally on @xmath365 , the unordered collection formed by the excursions of @xmath33 outside @xmath362 that hit @xmath363 is distributed as the unordered collection of @xmath271 independent copies of @xmath33 under @xmath366 . on the other hand , noting that @xmath38 is the tree coded by the function @xmath19\ni s\mapsto \ov w_s$ ] ( by proposition [ tree - coding ] ) , we also see that subtrees of @xmath38 above level @xmath133 with height greater than @xmath268 are in one - to - one correspondence with excursions of @xmath33 outside @xmath362 that hit @xmath363 , and if a subtree @xmath367 corresponds to an excursion @xmath360 , @xmath367 is obtained from the excursion @xmath360 ( shifted so that it starts from @xmath26 ) by exactly the same procedure that allows us to construct @xmath38 from @xmath33 under @xmath34 : to be specific , @xmath367 is coded by the function @xmath368 just as @xmath38 is coded by @xmath103 .
the preceding considerations show that @xmath38 satisfies the branching property , and therefore is a lvy tree . to get the formula for @xmath45
, we note that the distribution of the height of @xmath40 is given by @xmath369 where the last equality can be found in ( * ? ? ?
* section vi.1 ) .
since we also know that the function @xmath370 solves @xmath371 , the formula for @xmath45 follows .
in this section , we come back to the general setting of section [ sec : levy - tree ] .
our goal is to prove that the lvy tree @xmath214 is ( under the probability measure @xmath373 for some @xmath221 ) the almost sure limit of a sequence of embedded galton watson trees , and that this limit is consistent with the order structure of the lvy tree .
we refer to @xcite for basic facts about galton watson trees .
a key property for us is the fact that galton watson trees are rooted ordered ( discrete ) trees , also called plane trees , so that there is a lexicographical ordering on vertices . in what follows ,
we argue under the probability measure @xmath197 .
recall that @xmath207 is under @xmath197 a lvy process with laplace exponent @xmath185 , and that @xmath47 is the associated height process .
we fix an integer @xmath374 , and , for every integer @xmath375 , we consider the sequence of all excursions of @xmath47 above level @xmath376 that hit level @xmath377 .
we let @xmath378 be the ordered sequence consisting of all the initial times of these excursions , for all values of the integer @xmath375 ( so , @xmath379 corresponds to the beginning of an excursion of @xmath47 above @xmath26 that hits @xmath380 , @xmath381 may be either the beginning of an excursion of @xmath47 above @xmath26 that hits @xmath380 or the beginning of an excursion of @xmath47 above @xmath380 that hits @xmath382 , and so on ) . for every @xmath375 , we also let @xmath383 be the terminal time of the excursion starting at time @xmath384 .
we then set , for every integer @xmath385 , @xmath386 [ discre - height ] the process @xmath387 is the discrete height process of a sequence of independent galton
watson trees with the same offspring distribution @xmath388 .
recall that the discrete height process of a sequence of galton watson trees gives the generation of the successive vertices in the trees , assuming that these vertices are listed in lexicographical order in each tree and one tree after another .
see @xcite or ( * ? ? ?
* section 0.2 ) .
the ( finite ) height sequence of a single tree is defined analogously . by construction
, @xmath379 is the initial time of the first excursion of @xmath47 above @xmath26 that hits @xmath380 .
notice that this excursion is distributed as @xmath47 under @xmath389 .
let @xmath390 be the ( random ) integer such that @xmath391 , so that @xmath392 is the initial time of the second excursion of @xmath47 above @xmath26 that hits @xmath380 . with the excursion of @xmath47 during interval @xmath393
$ ] , we can associate a ( plane ) tree @xmath394 constructed as follows .
the children of the ancestor correspond to the excursions of @xmath47 above level @xmath380 , during the time interval @xmath393 $ ] , that hit @xmath395 and the order on these children is obviously given by the chronological order .
equivalently , the children of the ancestor correspond to the indices @xmath396 such that @xmath397 . then , assuming that the ancestor has at least one child ( equivalently that @xmath398 ) , the children of the first child of the ancestor correspond to the excursions of @xmath47 above level @xmath395 , during the time interval @xmath399 $ ] , that hit @xmath400 , and so on .
see fig.[tree - exc ] for an illustration .
the sequence @xmath401 and the tree @xmath394 ( in thick lines).,width=529 ] write @xmath402 for the number of children of the ancestor in @xmath394 .
it follows from proposition [ excu - above ] ( ii ) that , conditionally on @xmath402 , the successive excursions of @xmath47 above level @xmath380 , during the time interval @xmath393 $ ] , that hit @xmath395 are independent and distributed as @xmath47 under @xmath389 ( recall that our definition shifts excursions above a level @xmath133 so that they start from @xmath26 ) .
recalling the construction of the tree @xmath394 , we now obtain that , conditionally on @xmath402 , the subtrees of @xmath394 originating from the children of the ancestor are independent and distributed according to @xmath394 .
this just means that @xmath394 is a galton watson tree , and its offspring distribution @xmath388 is the law under @xmath389 of the number of excursions of @xmath47 above level @xmath380 that hit @xmath395 . with the second excursion of @xmath47 above @xmath26 that hits @xmath380
, we can similarly associate a galton watson tree @xmath403 with offspring distribution @xmath388 , and so on .
the trees @xmath404 are independent as a consequence of the strong markov property of the lvy process @xmath207 . by construction ,
the process @xmath387 is the discrete height process of the sequence @xmath404 .
[ conv - discre ] for every @xmath374 , set @xmath405 .
then , for every @xmath406 , @xmath407 in probability under @xmath197 . recall that , for every @xmath237 , @xmath238 denotes the local time of @xmath47 at level @xmath133 .
it will be convenient to introduce , for every @xmath374 and every @xmath375 , the increasing process @xmath408 as a consequence of proposition [ excu - above](i ) applied with @xmath409 , we get that @xmath410 is a poisson process with parameter @xmath411 .
we claim that , for every @xmath406 , @xmath412 \bigg)=0.\ ] ] to see this , first observe that , for every @xmath413 , @xmath414 and then write @xmath415\\ & \qquad\leq \sum_{p=1}^\infty \mathbf{e}\big[\mathbf{1}_{\{p-1\leq l^{j2^{-n}}_a\leq p\ } } \sup_{t\leq p } |v(2^{-n})^{-1}\,\nn^n_{(j)}(t ) - t|\big]\\ & \qquad\leq \sum_{p=1}^\infty \mathbf{p}(p-1\leq l^{j2^{-n}}_a)^{1/2}\,\mathbf{e}\big[\sup_{t\leq p } |v(2^{-n})^{-1}\,\nn^n_{(j)}(t ) - t|^2\big]^{1/2}.\end{aligned}\ ] ] then , if @xmath416 stands for a standard poisson process , we have by a classical martingale inequality @xmath417&= v(2^{-n})^{-2}\mathbf{e}\big[\sup_{t\leq v(2^{-n})p } ( \nn(t)-t)^2\big]\\ & \leq 4v(2^{-n})^{-2}\,\mathbf{e}[(\nn(v(2^{-n})p)-v(2^{-n})p)^2]\\ & = 4v(2^{-n})^{-1}p.\end{aligned}\ ] ] it follows that , for every @xmath375 , @xmath418 \leq \big ( \sum_{p=1}^\infty ( p\,\mathbf{p}(p-1\leq l^{j2^{-n}}_a))^{1/2}\big ) \times 2v(2^{-n})^{-1/2},\ ] ] and the proof of is completed by noting that @xmath419 as @xmath420 , and that @xmath421 because @xmath422 is bounded above in distribution by @xmath423 ( cf definition 1.3.1 in @xcite ) , and we know that @xmath424 has exponential moments .. let @xmath425 be an integer . by summing the convergence in over possible choices of @xmath426 , we also obtain that @xmath427 = 0.\ ] ] on the other hand , we have , for every @xmath50 , @xmath428 and it follows from that @xmath429 = 0.\ ] ] by combining and , we get @xmath430 = 0,\ ] ] where @xmath431 .
since @xmath432 can be made arbitrarily small by choosing @xmath433 large , we have obtained that @xmath434 in probability .
elementary arguments show that this implies @xmath435 in probability , and therefore also @xmath436 in probability .
this completes the proof . in what follows
, we will need an analog of the preceding two propositions for the height process under its excursion measure @xmath437 .
let us fix @xmath374 . under the measure @xmath438
, we define @xmath439 as the ordered sequence consisting of the initial times of all excursions of @xmath47 above level @xmath376 that hit level @xmath377 , for all values of the integer @xmath375 .
the analog of proposition [ discre - height ] says that , under @xmath440 , the finite sequence @xmath441 is distributed as the height sequence of a galton watson tree with offspring distribution @xmath388 .
this is immediate from the fact that an excursion with distribution @xmath442 is obtained by taking ( under @xmath197 ) the first excursion of @xmath47 with height greater than @xmath380 . by convention
, we take @xmath443 and @xmath444 if @xmath445 .
we next fix a sequence @xmath446 such that both and the convergence of proposition [ conv - discre ] hold @xmath197 a.s . along this sequence , for each @xmath406 . from now on ,
we consider only values of @xmath447 belonging to this sequence .
we claim that we have then also @xmath448 to see this , note that it suffices to argue under @xmath449 for some @xmath450 , and then to consider the first excursion ( under @xmath197 ) of @xmath47 away from @xmath26 with height greater than @xmath451 .
we abuse notation by still writing @xmath452 for the finite sequence of times defined as explained above , now relative to this first excursion with height greater than @xmath451 .
we observe that , provided @xmath447 is large enough so that @xmath453 , we have @xmath454 where @xmath455 is the index such that @xmath456 is the initial time of the first excursion of @xmath47 away from @xmath26 with height greater than @xmath451 , and @xmath457 is the first index @xmath458 such that @xmath459 does not belong to the interval @xmath460 $ ] associated with this excursion . notice that @xmath461 decreases to @xmath462 as @xmath420 .
our claim then reduces to verifying that @xmath463 which follows from proposition [ conv - discre ] and , recalling that both these convergences hold a.s . on the sequence of values of @xmath447 that we consider .
in this section , we prove theorem [ subordi - coding ] .
we consider the brownian snake under its excursion measure @xmath34 , and we recall the notation @xmath464 , and @xmath465 if @xmath28 . as in theorem
[ subor - brown ] , we write @xmath38 for the subordinate tree of the brownian tree @xmath22 with respect to the function @xmath346
. _ proof of theorem [ subordi - coding ] .
_ we set @xmath466 . for fixed @xmath374 ,
we define a discrete plane tree @xmath467 under the probability measure @xmath468 in the following way .
the children of the root correspond to the excursions of @xmath33 outside @xmath469 that hit @xmath395 ( recall the definition of these excursions from the proof of theorem [ subor - brown ] ) . note that these excursions in turn correspond to the excursions of the real - valued process @xmath55 above level @xmath380 that hit @xmath395 ( the point is that , if @xmath244 is the time interval associated with an excursion of @xmath33 outside @xmath469 , the process @xmath103 remains strictly above @xmath380 on the whole interval @xmath244 ) .
we obviously order the children of the root according to the chronological order of the brownian snake . by the special markov property , in the form given in corollary [ smpexcursion ] in the appendix below , conditionally on the number of excursions of @xmath33 outside @xmath469 that hit @xmath395 , these excursions listed in chronological order are independent and distributed according to @xmath468 , modulo the obvious translation by @xmath470 .
we can thus continue the construction of the tree @xmath467 by induction , and this random plane tree is a galton watson tree since it satisfies the branching property at the first generation .
let @xmath388 be the offspring distribution found in proposition [ discre - height ] in the case where @xmath471 .
we claim that @xmath388 is also the offspring distribution of @xmath467 . to see this , observe that @xmath388 is , by definition , the distribution of the number of points of a @xmath45-lvy tree at height @xmath380 that have descendants at height @xmath395 ( conditionally on the event that the height of the tree is at least @xmath380 ) .
thanks to theorem [ subor - brown ] and to the fact that @xmath38 is the tree coded by the function @xmath44 , we know that this is the same as the conditional distribution of the number of excursions of @xmath103 above level @xmath380 that hit @xmath395 , under @xmath468
. let @xmath472 be the ordered sequence consisting of the initial times of all excursions of @xmath55 above level @xmath376 that hit level @xmath377 , for all values of the integer @xmath375 .
note that each such excursion corresponds to a vertex of the tree @xmath467 , and so @xmath473 is just the total progeny of @xmath467 . by convention , we also define @xmath474 .
set @xmath475 and @xmath476 if @xmath477 .
then @xmath478 is the height sequence of @xmath467 ( note that the lexicographical ordering on vertices of @xmath467 corrresponds to the chronological order on the associated excursion initial times ) .
hence @xmath479 has the same distribution as the sequence @xmath480 which was defined at the end of the preceding section from the height process @xmath47 under @xmath389 .
but in fact more in true : the whole collection of the discrete sequences @xmath479 for all @xmath374 has the same distribution under @xmath34 as the similar collection of sequences @xmath480 constructed from the height process @xmath47 under @xmath437 ( the reason is the fact that , in both constructions , the tree at step @xmath447 can be obtained from the tree at step @xmath481 by the deterministic operation consisting in keeping only those vertices at even generation that have at least one child , and viewing that set of vertices as a plane tree in the obvious manner ) .
the convergence now allows us to set , for every @xmath200 , @xmath482 and the process @xmath483 is distributed as the height process of the lvy tree with branching mechanism @xmath48 .
the limit in the preceding display holds uniformly in @xmath342 , @xmath34 a.e .
, provided we argue along the subsequence of values of @xmath447 introduced at the end of the preceding section .
we set @xmath484 .
we observe that the distribution of @xmath485 under @xmath34 is the same as that of @xmath486 under @xmath208 , and so we must have , for every @xmath374 and @xmath385 , @xmath487 where @xmath488 are the initial times of the excursions of @xmath489 above level @xmath376 that hit level @xmath377 , for all values of the integer @xmath375 , and @xmath490 if @xmath491 .
notice that @xmath492 because the height sequence of the tree @xmath467 is @xmath493 .
also , if @xmath494 and @xmath495 , @xmath496 , the property @xmath497 holds if and only if @xmath498 : indeed these properties hold if and only if the vertex with index @xmath499 in the ( lexicographical ) ordering of @xmath467 coincides with the vertex with index @xmath500 in the ordering of @xmath501 , modulo the identification of the vertex set of @xmath467 as a subset of the vertex set of @xmath501 , in the way explained above .
we need to verify that @xmath53 can be written as a time change of @xmath502 . as a first step
, we notice that , for every @xmath503 , @xmath504 and so @xmath505 .
this suggests that the process @xmath49 in the statement of the theorem should be such that @xmath506 , for every @xmath495 and every @xmath447 . at this point
, we observe that @xmath507 indeed , if this property fails , a compactness argument gives two times @xmath508 $ ] with @xmath509 such that @xmath510 is monotone nonincreasing on @xmath511 $ ] . to see that this can not occur , we may replace @xmath502 by the process @xmath47 constructed from a lvy process excursion @xmath207 as explained in section [ sec : levy - tree ] .
we then note that jumps of @xmath207 are dense in @xmath79 $ ] , and the strong markov property shows that , for any jump time @xmath31 of @xmath207 , for any @xmath512 , we can find @xmath513 $ ] , with @xmath514 , such that @xmath515 ( use formula ( 20 ) in @xcite , or see the comments at the end of section [ sec : levy - tree ] ) .
let @xmath516 and , for every integer @xmath374 , let @xmath517 be the unique integer such that @xmath518 .
we note that the sequence @xmath519 is monotone nondecreasing ( this is obvious since @xmath520 is a subset of @xmath521 ) .
it follows that the sequence @xmath522 is also monotone nondecreasing : indeed , if @xmath494 and @xmath495 , @xmath496 are such that @xmath523 , we have automatically @xmath524 since , writing @xmath525 , the fact that @xmath526 implies that @xmath527 .
we can now set @xmath528 note that this limit will exist simultaneously for all @xmath516 outside a set of @xmath34-measure @xmath26 .
we also take @xmath529 for all @xmath530 .
clearly @xmath531 is nondecreasing and , by construction , the property @xmath506 holds for every @xmath532 and every @xmath447 , and we have @xmath533 when @xmath31 is of the form @xmath534 .
we also note that @xmath535 is continuous as a consequence of the property . to check right - continuity , observe that , if @xmath516 is fixed , and @xmath536 is such that @xmath537 then , for every @xmath538 , the property @xmath539 forces @xmath540 , hence ( letting @xmath541 tend to @xmath277 ) @xmath542 , and use .
left - continuity is derived by a similar argument . for @xmath516 , set @xmath543 and @xmath544 .
note that @xmath545 . on one hand , by passing to the limit @xmath420 in the equality @xmath546 we obtain that @xmath547 . on the other hand , @xmath548 must be constant on the interval @xmath549 $ ] . to see this
, we first observe that @xmath548 must be nonincreasing on @xmath549 $ ] ( otherwise there would be some @xmath447 and some @xmath495 such that @xmath550 , which is absurd ) .
so we need to verify that there is no nontrivial interval @xmath283 $ ] such that @xmath103 is both nonincreasing and nonconstant on @xmath283 $ ] , and , to prove this , we may replace nonincreasing by nondecreasing thanks to the invariance of @xmath34 under time - reversal .
argue by contradiction , and suppose that @xmath280 are such that that the event where @xmath551 and @xmath103 is both nondecreasing and nonconstant on @xmath283 $ ] has positive @xmath34-measure .
we can then find a stopping time @xmath552 such that , with positive @xmath34-measure on the latter event , we have @xmath553 , @xmath554 and @xmath555 is attained only at @xmath556 ( take @xmath557 , with @xmath450 small enough ) . using the strong markov property of the brownian snake , we then find @xmath558 such that @xmath559 is the restriction of @xmath560 to @xmath561 $ ] , for some @xmath512 , which implies @xmath562 and gives a contradiction with the fact that @xmath103 is nondecreasing on @xmath283 $ ] . finally , since @xmath548 is constant on @xmath549 $ ] , we have @xmath563 , which was the desired result .
this completes the proof of theorem [ subordi - coding ] .
in this section , we discuss applications of the previous results to the brownian map .
analogously to @xcite , we consider a version of the brownian map with `` randomized volume '' , which may be constructed under the brownian snake excursion measure @xmath34 as follows .
recall that @xmath564 stands for the tree coded by @xmath330 and @xmath24 is the canonical projection from @xmath19 $ ] onto @xmath22 . for @xmath565 , the `` lexicographical interval '' @xmath59 $ ] stands for the image under @xmath24 of the smallest interval @xmath566 $ ] ( @xmath145 $ ] ) such that @xmath567 and @xmath568 ( here we make the convention that if @xmath569 the interval @xmath566 $ ] is equal to @xmath570\cup [ 0,t]$ ] ) . for every @xmath571 , we set @xmath27 , where @xmath31 is such that @xmath567 .
in particular @xmath572 .
the random mapping @xmath573 is interpreted as brownian motion indexed by the `` brownian tree '' @xmath22 .
we then define a mapping @xmath574 by setting @xmath575 } z_c , \min_{c\in[b , a ] } z_c\big\}.\ ] ] for @xmath576 , we set @xmath577 if and only if @xmath578 , or equivalently @xmath579 } z_c , \min_{c\in[b , a ] } z_c\big\}.\ ] ] one can verify that if @xmath580 then @xmath581 for any @xmath582 ( the point is that , if @xmath580 with @xmath583 , then necessarily @xmath4 and @xmath584 are leaves of @xmath22 , and the reals @xmath585 such that @xmath586 and @xmath587 are unique , which implies that @xmath59\subset [ a , a']\cup [ a',b]$ ] and @xmath588\subset [ b , a']\cup [ a',a]$ ] ) .
we also set @xmath589 where the infimum is over all choices of the integer @xmath269 and of the points @xmath590 of @xmath22 . if @xmath580 then @xmath591 for any @xmath582 .
furthermore one can also prove that @xmath592 if and only if @xmath577 . since @xmath61 satisfies the triangle inequality
, it follows that @xmath593 is an equivalence relation on @xmath22 .
the brownian map ( with randomized volume ) is the quotient space @xmath594 , which is equipped with the distance induced by the function @xmath61 ( with a slight abuse of notation , we still denote the induced distance by @xmath61 ) .
we write @xmath595 for the canonical projection from @xmath22 onto @xmath60 , and @xmath596 .
we also write @xmath597 if @xmath598 and @xmath599 . in the usual construction of the brownian map , one deals with the conditioned measure @xmath600 instead of @xmath34 , but
otherwise the construction is exactly the same and we refer to @xcite for more details . the brownian map @xmath60 comes with two distinguished points .
the first one @xmath601 corresponds to the root @xmath343 of @xmath22 .
the second distinguished point is @xmath602 , where @xmath603 is the ( unique ) point of @xmath22 at which @xmath63 attains its minimum : @xmath604 we will write @xmath605 to simplify notation . the reason for considering @xmath62 comes from the fact that distances from @xmath62 have a simple expression . for any @xmath571 , @xmath606 the following `` cactus bound '' ( * ? ? ?
* proposition 3.1 ) also plays an important role .
let @xmath565 , and let @xmath607\to \bm$ ] be a continuous path in @xmath60 such that @xmath608 and @xmath609 .
then , @xmath610 where we recall that @xmath6 is the geodesic segment between @xmath4 and @xmath5 in @xmath22 , not to be confused with the interval @xmath59 $ ]
. in other words , any continuous path from @xmath611 to @xmath612 must come at least as close to @xmath62 as the ( image under @xmath595 of the ) geodesic segment from @xmath4 to @xmath5 in @xmath22 .
we now introduce the metric net , in the terminology of @xcite . for every @xmath613
, we consider the ball @xmath64 defined by @xmath614 for @xmath615 , we define the hull @xmath67 as the complement of the connected component of @xmath64 that contains @xmath7 .
informally , @xmath67 is obtained from @xmath64 by `` filling in '' the holes of @xmath64 except for the one containing @xmath7 .
write @xmath616 for the topological boundary of @xmath67 .
we define the _ metric net _ @xmath68 as the closure in @xmath60 of the union @xmath617 our goal is to investigate the structure of @xmath68 .
if @xmath571 and @xmath142 $ ] are such that @xmath567 , we write @xmath337 ( as previously ) and we use the notation @xmath618 where the last equality holds because , as already mentioned , the quantities @xmath339 for @xmath619 correspond to the values of @xmath620 along the geodesic segment @xmath341 .
we then introduce the closed subset of @xmath22 defined by @xmath621 we note that points of @xmath622 have multiplicity either @xmath275 or @xmath276 in @xmath22 . indeed , there are only countably many points of multiplicity @xmath623 , and it is not hard to see that these points do not belong to @xmath622 .
[ descri - net ] let @xmath624 .
then @xmath625 if and only if @xmath598 for some @xmath626 . fix @xmath627 and @xmath624 .
we claim that @xmath628 if and only if we can write @xmath598 with both @xmath629 , and @xmath630 indeed , if these conditions hold , we have @xmath631 by , and the image under @xmath595 of the geodesic segment from @xmath4 to @xmath343 provides a path from @xmath303 to @xmath7 that stays outside @xmath64 except at the initial point @xmath303 .
it follows that @xmath611 belongs to @xmath616 .
conversely , if @xmath628 , then it is obvious that @xmath631 giving @xmath629 for any @xmath4 such that @xmath632 .
write @xmath633 , where @xmath634 , and , for every @xmath447 , let @xmath635 such that @xmath636 .
the fact that @xmath634 implies that , for every @xmath637 belonging to the geodesic segment between @xmath638 and @xmath343 , we have @xmath639 ( otherwise the cactus bound would imply that any path between @xmath640 and @xmath7 visits @xmath64 , which is a contradiction ) . by compactness , we may assume that @xmath638 converges to @xmath4 as @xmath420 , and we have @xmath632 .
we then get that the property @xmath639 holds for @xmath637 belonging to the geodesic segment between @xmath4 and @xmath343 , except possibly for @xmath641 .
this completes the proof of our claim .
it follows from the claim that the property @xmath598 for some @xmath571 such that @xmath642 holds for every @xmath643 ( this property holds if @xmath628 for some @xmath644 and is preserved under passage to the limit , using the compactness of @xmath22 ) .
conversely , suppose that this property holds , with @xmath645 to discard a trivial case .
if the path @xmath335 hits its minimum only at its terminal point , the first part of the proof shows that @xmath628 for @xmath646 .
if the path @xmath335 hits its minimum both at its terminal time and at another time , then lemma 16 in @xcite shows that @xmath647 , where @xmath648 for every @xmath447 .
then the image under @xmath595 of the first point @xmath649 on the ancestral line of @xmath638 such that @xmath650 belongs to @xmath651 , with @xmath652 .
noting that @xmath649 must lie on the geodesic segment between @xmath4 and @xmath638 in the tree @xmath22 , we see that we have also @xmath653 , so that we get that @xmath654 belongs to @xmath68 .
the preceding arguments are closely related to ( * ? ? ?
* section 3 ) ( see in particular formula ( 16 ) in @xcite ) , which deals with the slightly different setting of the brownian plane .
if @xmath625 and @xmath655 is such that @xmath598 , the image under @xmath595 of the geodesic segment from @xmath4 to @xmath343 provides a path in @xmath60 that stays in the complement of @xmath64 for every @xmath656 ( indeed the values of @xmath657 for @xmath5 belonging to this segment are the numbers @xmath658 ) .
it follows that all points belonging to a geodesic from @xmath303 to @xmath62 also belong to @xmath68 .
we note that we can define an `` intrinsic '' metric on @xmath68 by setting , for every @xmath659 , @xmath660 it is obvious that @xmath661 .
in particular , @xmath662 implies @xmath663 , and it follows that @xmath71 is a metric on @xmath68 .
the quantity @xmath664 corresponds to the infimum of the lengths ( computed with respect to @xmath61 ) of paths from @xmath303 to @xmath665 that are obtained by the concatenation of pieces of geodesics from points of @xmath68 to @xmath666 ( we already noticed that these geodesics stay in @xmath68 ) .
we have clearly @xmath667 for every @xmath625 , and @xmath71-geodesics from @xmath303 to @xmath62 coincide with @xmath61-geodesics from @xmath303 to @xmath62 ( if @xmath659 and @xmath668 belong to the same @xmath61-geodesic to @xmath62 , the results of @xcite imply that @xmath669 ) .
* the topology induced by @xmath71 on @xmath68 coincides with the topology induced by @xmath61 . since @xmath670 and @xmath671 is compact , it is enough to prove that @xmath672 is also compact . however , if @xmath673 is a sequence in @xmath68 , we may write @xmath674 , with @xmath675 , and then extract a subsequence @xmath676 that converges to @xmath677 in @xmath22 .
we have @xmath678 because @xmath622 is closed .
furthermore the fact that @xmath679 converges to @xmath677 implies that @xmath680 tends to @xmath26 , and therefore @xmath681 also tends to @xmath26 , showing that @xmath673 has a convergent subsequence in @xmath672 .
the preceding proposition shows that the metric net @xmath68 has close connections with the subset @xmath622 of @xmath22 .
the latter set is itself related to the subordinate tree of @xmath22 with respect to the function @xmath682 . by theorem [ subordi - coding ] ( and an obvious symmetry argument )
we can construct a process @xmath215 distributed as the height process of the lvy tree with branching mechanism @xmath48 , and a continuous random process @xmath683 with nondecreasing sample paths such that @xmath684 , @xmath685 , and for every @xmath142 $ ] , @xmath686 we define a random equivalence @xmath78 relation on @xmath79 $ ] , by requiring that the graph of @xmath78 is the smallest closed symmetric subset of @xmath79 ^ 2 $ ] that contains all pairs @xmath80 with @xmath81 , @xmath687 , and @xmath688 for all @xmath689 .
we leave it to the reader to check that this set is indeed the graph of an equivalence relation ( use the comments at the end of section [ sec : levy - tree ] ) .
in addition to the pairs @xmath80 satisfying the previous relation , the graph of @xmath78 contains a countable collection of pairs @xmath244 , each of them associated with a point of infinite multiplicity @xmath4 of the tree @xmath214 by the relations @xmath690 and @xmath691 . a simulation of a looptree ( simulation by igor kortchemski ) .
for technical reasons , some of the trees branching off a loop are pictured inside this loop , but , from the point of view of the present work , it is better to think of these trees as growing outside the loop , so that the space inside the loop may be `` filled in '' appropriately.,width=453 ] we denote the quotient space @xmath79/\!\sim$ ] by @xmath85 .
then @xmath85 can be identified with the `` looptree '' associated with @xmath47 . roughly speaking ( see @xcite for more details )
the looptree is obtained by replacing each point of infinite multiplicity @xmath4 of the tree @xmath214 by a loop of `` length '' equal to the weight of @xmath4 , so that the subtrees that are the connected components of the complement of @xmath4 in the tree branch along this loop in an order determined by the coding function @xmath47 .
note that the looptree associated with @xmath47 is equipped in @xcite with a particular metric . here
we avoid introducing this metric on @xmath85 , because it will be more relevant to our applications to introduce a pseudo - metric that will be described below .
let us introduce the right - continuous inverse of @xmath49 .
for every @xmath692 , we set @xmath693 by convention , we also set @xmath694 .
the left limit @xmath695 is equal to @xmath696 , and @xmath49 is constant on every interval @xmath697 $ ] . note that @xmath698 and thus @xmath699 .
we next consider the subset @xmath700 of @xmath622 defined as follows .
if @xmath655 , we say that @xmath701 if there exist @xmath516 and @xmath702 such that @xmath567 and the function @xmath703 is constant on the interval @xmath704 $ ] .
notice that only leaves ( points @xmath4 of multiplicity @xmath275 , for which there is a single value of @xmath31 with @xmath567 ) may belong to @xmath700 .
indeed , if @xmath626 has multiplicity two , and @xmath705 are the two elements of @xmath706 such that @xmath707 , then both @xmath284 and @xmath708 are times of ( left or right ) increase of @xmath32 , and , together with the property @xmath709 , this implies that @xmath710 takes values strictly less than @xmath711 immediately after @xmath284 , resp . immediately after @xmath708 ( see section [ sec : subord - maxi ] ) .
one can describe the elements of @xmath700 in the following way .
let @xmath712 such that @xmath5 has a strict descendant @xmath637 with @xmath713 ( in the terminology of @xcite , @xmath5 is an excursion debut above the minimum ) .
let @xmath283\subset [ 0,\sigma]$ ] be the interval whose image under @xmath24 gives all descendants of @xmath5 .
then the set @xmath714 : \un w_r = \un w_{s_1}\hbox { and } \wh w_r>\un w_{s_1}\}\ ] ] is an open subset of @xmath283 $ ] , and the ( image under @xmath24 of the ) left end of each of its connected components belongs to @xmath700 .
furthermore any element of @xmath700 can be obtained in this way .
we set @xmath715 .
[ tech - rep ] for @xmath716 $ ] , the property @xmath717 holds if and only if @xmath718 .
furthermore , the mapping @xmath719\ni u\mapsto p_\zeta(\tau_u)$ ] induces a bijection from @xmath720/\!\sim$ ] onto @xmath721 , which will be denoted by @xmath722 .
we first show that @xmath723 belongs to @xmath721 , for every @xmath724 . by the definition of @xmath725 , @xmath726 for every @xmath512 , and
this implies that @xmath727 is not constant on @xmath728 $ ] ( use and the fact that @xmath47 is not constant on any nontrivial interval ) . in particular , we must have @xmath729 and therefore @xmath730 . the fact that @xmath731 is then immediate from the definition of @xmath732 , recalling that points of @xmath700 are leaves of @xmath22 .
we then verify that , if @xmath733 , there exists @xmath724 with @xmath734 .
we can write @xmath28 where the mapping @xmath703 is not constant on @xmath704 $ ] for every @xmath735 . using the formula @xmath736
, it follows that @xmath737 for every @xmath735 , and thus @xmath738 .
finally @xmath739 with @xmath740 .
next let us prove that @xmath717 implies @xmath718 .
let @xmath717 and without loss of generality suppose that @xmath741 .
we first assume that @xmath742 and @xmath743 for every @xmath744 .
then , we must have @xmath745 for every @xmath746 note that @xmath727 is constant over any interval @xmath747 $ ] .
this implies that @xmath748 for every @xmath746 ( if there exists @xmath746 such that @xmath749 , then , for every @xmath450 small enough , the properties of the brownian snake allow us to find such an @xmath31 with the additional property that @xmath30 is the restriction of @xmath750 to @xmath751 $ ] , which contradicts @xmath752 and we can make a symmetric argument if there exists @xmath753 such that @xmath754 ) .
it follows that @xmath755 . moreover ,
since @xmath756 is a point of left increase of @xmath32 , @xmath756 can not be a point of right increase of @xmath757 , so that there are values of @xmath758 arbitrarily close to @xmath756 such that @xmath759 , and therefore @xmath760 .
it follows that we have @xmath761 giving @xmath718 as desired .
suppose then that @xmath717 but the property @xmath743 for every @xmath744 does not hold .
then @xmath244 is the limit of a sequence @xmath762 such that , for every @xmath447 , @xmath763 and @xmath764 for every @xmath765 .
we must have @xmath766 and @xmath767 . by the first part of the argument
, we have @xmath768 for @xmath769 $ ] , and , letting @xmath447 tend to @xmath277 , we get @xmath770 , for every @xmath771 $ ] . then the fact that @xmath695 is a point of right increase of @xmath32 implies that @xmath772 ( by the same argument as above ) and we conclude again that @xmath718 . finally , it remains to prove that the property @xmath718 implies @xmath717 .
note that @xmath773 is only possible if @xmath774 , so that we may assume that @xmath775 .
then @xmath776 is a point of multiplicity @xmath276 of @xmath22 ( since @xmath626 , @xmath4 can not have multiplicity @xmath623 in @xmath22 ) , and the points @xmath777 for @xmath778 are descendants of @xmath4 , so that @xmath779 for every @xmath778 .
it follows that @xmath780 for @xmath781 ( write @xmath782 ) .
if @xmath783 for @xmath784 , this means that @xmath717 and we are done .
otherwise there exists @xmath744 such that @xmath785 , and this means that @xmath4 has a strict descendant @xmath786 such that @xmath787 .
this implies that the path @xmath335 hits its minimal value only at its terminal time ( otherwise @xmath349 would have two equal local minima ) .
we know that , just before @xmath725 , there are values of @xmath31 such that @xmath788 ( otherwise @xmath725 would a time of local minimum of @xmath32 , but this is excluded since such times correspond to points of multiplicity @xmath623 of @xmath22 and thus never satisfy @xmath789 ) , and it follows that there are times @xmath790 arbitrarily close to @xmath725 such that @xmath791 , and thus @xmath792 .
hence , if we set @xmath793 , we have @xmath794 as @xmath420 . similarly , if @xmath795 we have @xmath796 as @xmath420 .
clearly @xmath797 so that we also get @xmath717
. we write @xmath798 for the canonical projection from @xmath79 $ ] onto @xmath799/\!\!\sim$ ] .
if @xmath800 and @xmath801 , we will also write @xmath802 . in a way similar to the definition of intervals in @xmath22 , we can define intervals in @xmath803 . if @xmath804 , we set @xmath92=\mathrm{p}_\ll([s , t])$ ] , where @xmath805 $ ] are such that @xmath806 and @xmath807 and @xmath566 $ ] is as small as possible ( here again we use the convention @xmath566=[s,\chi]\cup[0,t]$ ] if @xmath808 )
. we will identify the metric net @xmath672 with a quotient space of the looptree @xmath85 .
informally , the latter quotient space is obtained by identifying two points @xmath88 and @xmath93 if they face each other at the same height in @xmath85 : this means that we require that @xmath809 , and that vertices `` between '' @xmath88 and @xmath93 have a smaller height . to make this more precise ,
we define , for every @xmath804 , @xmath91 } h_\gamma , \max_{\gamma\in[\beta,\alpha ] } h_\gamma\bigg ) - h_\alpha -h_\beta,\ ] ] and @xmath810 where the infimum is over all possible choices of the integer @xmath811 and of @xmath812 .
the following statement is a reformulation , in a more precise form , of theorem [ identi - net ] stated in the introduction .
[ identi - metricnet ] for @xmath804 , set @xmath94 if and only if @xmath813
. then the property @xmath814 holds if and only if @xmath95 , or equivalently @xmath815 } h_\gamma , \max_{\gamma\in[\beta,\alpha ] } h_\gamma\bigg).\ ] ] furthermore , @xmath96 induces a metric on the quotient space @xmath99 .
if @xmath816 denotes the bijection of lemma [ tech - rep ] , @xmath817 induces an isometry from @xmath818 onto @xmath98 . * remark . *
it is not a priori obvious that defines an equivalence relation on @xmath85 .
this property follows from the fact that holds if and only if @xmath813 , which we derive in the following proof from the relations between @xmath85 and the brownian map .
we first verify that , if @xmath804 and @xmath819 , we have @xmath820 let @xmath821 $ ] be such that @xmath822 .
note that we have then @xmath823 by the definition of @xmath722 .
hence , @xmath824 similarly , we have @xmath825 .
so the proof of reduces to checking that @xmath826 } h_\gamma = \max_{c\in[a , b ] } ( -z_c).\ ] ] to get this equality , we write @xmath827 } h_\gamma&=\min_{\mathrm{p}_\ll(s)=\alpha,\mathrm{p}_\ll(t)=\beta } \big ( \max_{r\in[s , t ] } h_r\big)\\ & = \min_{\mathrm{p}_\ll(s)=\alpha,\mathrm{p}_\ll(t)=\beta } \big ( \max_{r\in[\tau_s,\tau_t ] } ( -z_r)\big)\\ & = \min_{p_\zeta(u)=a , p_\zeta(v)=b } \big ( \max_{r\in[u , v ] } ( -z_r)\big)\\ & = \max_{c\in[a , b ] } ( -z_c).\end{aligned}\ ] ] the second equality holds because @xmath828 , and @xmath727 stays constant on intervals @xmath747 $ ] . to justify the third equality
, we note that the elements @xmath829 of @xmath19 $ ] such that @xmath830 are exactly the reals @xmath831 where @xmath31 is such that @xmath806 : since @xmath733 , any @xmath832 $ ] such that @xmath830 must be of the form @xmath831 with @xmath821 $ ] , and if @xmath829 is of this form , the property @xmath830 is equivalent to @xmath806 by lemma [ tech - rep ] .
this completes the proof of .
we then claim that , for every @xmath643 , there exists @xmath833 such that @xmath632 .
let @xmath643 . by proposition [ descri - net ] , we know that @xmath598 where @xmath626
. it may happen that @xmath834 , but then we can write @xmath28 where @xmath835 for every @xmath836 $ ] , for some @xmath512 .
the latter property if only possible if @xmath30 hits its minimal value both at its terminal time @xmath312 and at another time @xmath837 .
if @xmath838 , @xmath839 is the restriction of @xmath30 to @xmath840 $ ] . by the results recalled in section [ sec : subord - maxi ] , the fact that @xmath536 is a time of left increase for @xmath32 implies that there are values of @xmath841 arbirarily close to @xmath536 such that @xmath842 .
if we set @xmath843 , we have @xmath844 and @xmath845 .
furthermore , @xmath846 for every @xmath847 $ ] , and it follows from the definition of the equivalence relation @xmath593 that @xmath848 , hence @xmath849 .
our claim is proved .
let @xmath850 , and @xmath851 such that @xmath632 and @xmath852 . from and the preceding claim ,
we may write @xmath853 we then use the bijection @xmath722 of lemma [ tech - rep ] to observe that , if @xmath854 and @xmath855 , we have also , thanks to , @xmath856 in particular , if @xmath804 and @xmath857 , @xmath858 , we see that the condition @xmath813 holds if and only if @xmath662 , and ( since @xmath661 ) the latter condition holds if and only if @xmath859 , or equivalently @xmath95 ( by ) .
this gives the first assertion of the theorem .
then @xmath96 is symmetric and satisfies the triangle inequality , hence induces a metric on the quotient space @xmath99 . from the property @xmath860 , we see that the relation @xmath94 implies @xmath861 , so that @xmath817 induces a mapping from @xmath99 to @xmath68 .
this mapping is onto since @xmath862 , and is an isometry by .
in this section , we continue our applications to the brownian map .
we keep the notation and assumptions of the preceding section . in particular , the process @xmath863 , which is distributed under @xmath34 as the height process of the @xmath45-lvy tree , was introduced so that the representation formula holds .
our goal is to discuss the connected components of the complement of the metric net in the brownian map ( these are called brownian disks in @xcite ) .
we again argue under the excursion measure @xmath34 . for every @xmath413 ,
we denote by @xmath864 the ( total mass of the ) exit measure of the brownian snake from @xmath865 , see ( * ? ? ?
* chapter v ) .
then @xmath866 has a cdlg modification that we consider from now on ( see the discussion in section 2.5 of @xcite ) . for every @xmath413
, we can also consider the total local time of @xmath47 at level @xmath31 , which we denote by @xmath867 in agreement with section [ sec : levy - tree ] .
the ray - knight theorem of ( * ? ? ?
* theorem 1.4.1 ) shows that @xmath868 is distributed under @xmath34 according to the excursion measure of the continuous - state branching process with branching mechanism @xmath45 , and therefore has also a cdlg modification .
[ identi - exit ]
we have @xmath869 for every @xmath413 , @xmath34 a.e . let @xmath413 and , for every @xmath512 , let @xmath870 be the number of excursions of the brownian snake outside @xmath865 that hit @xmath871 .
then an easy application of the special markov property gives @xmath872 where @xmath873 . on the other hand , shows that @xmath870 is also the number of excursions of @xmath47 above @xmath874 that hit @xmath875 . by comparing the preceding approximation of @xmath864 with (
* theorem 4.2 ) , we arrive at the stated result .
following closely @xcite , we say that @xmath655 is an excursion debut if @xmath4 has a strict descendant @xmath5 such that @xmath876 for every @xmath877 .
we also say that @xmath878 is a local minimum of @xmath47 if there exist @xmath879 and @xmath880 such that @xmath881 and @xmath882 for every @xmath883 .
we now claim that the following sets are in one - to - one correspondence : 1 .
the set of all connected components of @xmath100 .
the set of all connected components of @xmath884 .
the set of all excursion debuts .
the set of all jump times of the exit measure process @xmath885 .
the set of all points of infinite multiplicity of @xmath214 .
the set of all local minima of @xmath47 .
let us explain these correspondences .
first the fact that local minima of @xmath47 correspond to points of infinite multiplicity of @xmath214 was explained at the end of section [ sec : levy - tree ] .
recall that , for every point of infinite multiplicity of @xmath214 , there is a cantor set of local minimum times corresponding to the associated local minimum ( see the end of section [ sec : levy - tree ] ) .
then , by ( * ? ? ?
* theorem 4.7 ) , each branching point @xmath5 ( necessarily of infinite multiplicity ) of @xmath214 corresponds to a discontinuity of @xmath886 at time @xmath887 , and the corresponding jump @xmath888 is the weight of the branching point @xmath5 . by ( * ? ? ?
* proposition 36 ) , discontinuity times for @xmath864 are in one - to - one correspondence with excursion debuts , and a discontinuity time @xmath31 corresponds to an excursion debut @xmath4 such that @xmath889 .
the fact that excursion debuts are in one - to - one correspondence with connected components of @xmath884 is proposition 20 in @xcite : if @xmath4 is an excursion debut , the associated connected component @xmath890 is the collection of all strict descendants @xmath5 of @xmath4 such that @xmath876 for every @xmath877 , and the boundary @xmath891 consists of all descendants @xmath5 of @xmath4 such that @xmath892 and @xmath876 for every @xmath893 .
furthermore , the `` boundary size '' of @xmath890 may be defined as the quantity @xmath888 , if @xmath31 is the associated jump time of the process @xmath885 ( this is also the weight of the corresponding point of infinite multiplicity of @xmath214 ) . finally , the fact that the sets ( a ) and ( b ) are also in one - to - one correspondence is a consequence of the following lemma .
[ connec - compo ] let @xmath890 be a connected component of @xmath884 .
then @xmath894 is a connected component of @xmath100 , and @xmath895 .
let @xmath336 be the excursion debut such that @xmath890 is the collection of all strict descendants @xmath5 of @xmath4 such that @xmath876 for every @xmath877 .
we first observe that @xmath894 is an open subset of @xmath100 .
this follows from the fact that the topology of @xmath60 is the quotient topology and @xmath896 ( to derive the latter equality , note that , if @xmath897 and @xmath898 are such that @xmath899 , we have @xmath900 , and it follows that @xmath901 ) .
since @xmath894 is connected , in order to get the statement of the lemma , we need only verify that , if @xmath902 is such that there is a continuous path @xmath903 that stays in @xmath100 and connects @xmath303 to a point @xmath665 of @xmath894 , then @xmath904 .
we argue by contradiction and assume that @xmath905 .
we then set @xmath906:\gamma(t)\in \pi(\cc)\}$ ] .
clearly , @xmath907 belongs to the boundary @xmath908 of @xmath894 .
on the other hand , it is easy to verify that @xmath909 ( if @xmath910 , we can write @xmath911 where @xmath912 , and , by extracting a subsequence , we can assume that @xmath913 in @xmath22 , so that we have @xmath914 , and @xmath915 since @xmath916 would imply @xmath917 , contradicting @xmath910 ) .
so @xmath918 , which contradicts our assumption that @xmath220 stays in @xmath100 .
for the last assertion , it remains to see that @xmath919 .
this is straightforward : if @xmath920 , we have automatically @xmath892 and @xmath921 , so that @xmath922 , forcing @xmath923 .
it is worth giving a direct interpretation of the correspondence between sets ( c ) and ( f ) above . if @xmath4 is an excursion debut , we can write @xmath924 , where @xmath280 , and the image under @xmath24 of the interval @xmath283 $ ] corresponds to the exploration of descendants of @xmath4 in @xmath22 .
it follows that we have @xmath925 for every @xmath926)$ ] .
there are points @xmath927)$ ] such that @xmath928 ( in fact one can find such points arbitrarily close to @xmath4 , as a consequence of the fact that points of increase for @xmath32 can not be points of increase for @xmath757 ) . if @xmath5 is such a point we can consider the last ancestor @xmath637 of @xmath5 such that @xmath929 , noting that the definition of an excursion debut implies that @xmath637 is a strict descendant of @xmath4 . then if @xmath930 are the two times in @xmath931 such that @xmath932 , one verifies that both @xmath933 and @xmath934 are local minimum times of @xmath47 corresponding to the local minimum @xmath935 .
in fact the set of all these local minimum times consists of all @xmath936 for @xmath937 such that @xmath938 belongs to the boundary of the connected component of @xmath884 associated with @xmath4 .
we will now establish that the boundary of any connected component of @xmath100 is a simple loop .
to this end , it is convenient to introduce the lvy process excursion @xmath939 associated with @xmath47 ( see section [ sec : levy - tree ] ) . note that @xmath207 can be reconstructed as a measurable function of @xmath47 , and that any branching point of @xmath47 corresponds to a unique jump time of @xmath207 , such that the size of the jump is the weight of the branching point .
recall the notation @xmath940 for the mapping introduced in lemma [ tech - rep ] .
[ bdry - compo ] let @xmath941 be a connected component of @xmath100 , let @xmath4 be the associated excursion debut , and let @xmath879 be the associated jump time of @xmath207
. for every @xmath942 $ ] , set @xmath943 then the mapping @xmath944 defines a simple loop in @xmath60 , whose initial and end points are equal to @xmath611 .
furthermore , the range of @xmath220 is the boundary of @xmath941 .
write @xmath890 for the connected component of @xmath884 such that @xmath945 . recall that @xmath946 by lemma [ connec - compo ] .
for every @xmath947 , the fact that @xmath948 for some @xmath949 forces @xmath950 .
let @xmath283 $ ] be the interval corresponding to the descendants of @xmath4 in the coding of @xmath22 . since @xmath284 and @xmath708 are both ( left or right )
increase times of @xmath32 , they can not be increase times for @xmath757 , and this implies @xmath951 and @xmath952 .
we next observe that any point of @xmath953 is of the form @xmath954 or @xmath955 for some @xmath956 $ ] such that @xmath957 .
we just noticed that this is true for @xmath4 .
if @xmath958 and @xmath959 , we can write @xmath960 with @xmath961 , and lemma 16 in @xcite implies that there are values of @xmath962 arbitrarily close to @xmath829 such that @xmath963 , which implies @xmath964 or @xmath965 .
furthermore , we have @xmath966 for any @xmath967 $ ] : if @xmath968 , the fact that @xmath727 stays constant on the time interval @xmath747 $ ] , together with the properties @xmath969 and @xmath970 ( which follow from lemma [ tech - rep ] ) , implies that @xmath971 .
hence , any point of @xmath972 is of the form @xmath973 for some @xmath956 $ ] such that @xmath957 , and we have then @xmath974 .
next note that the condition @xmath957 holds if and only if @xmath975 , and that @xmath976 $ ] is the interval corresponding to the descendants of the branching point associated with @xmath941 , in the coding of the tree @xmath214 .
using the comments of the end of section [ sec : levy - tree ] , we see that this interval is the same as @xmath977 $ ] with the notation of the proposition .
set @xmath978 to simplify notation .
the preceding discussion shows that any point of @xmath972 is of the form @xmath979 , with @xmath980 $ ] and @xmath981 .
conversely , for any @xmath65 satisfying these conditions , we have @xmath982 : this follows from the fact that @xmath983 under these conditions .
indeed , we have then , recalling that @xmath969 , @xmath984 and the fact that @xmath954 is a descendant of @xmath4 , which belongs to @xmath622 and satisfies @xmath985 , implies that @xmath983 .
note that the mapping @xmath986 is right - continuous . from the results recalled at the end of section [ sec : levy - tree
] , the times @xmath980 $ ] such that @xmath981 are exactly all reals of the form @xmath987 or @xmath988 for some @xmath989 $ ] .
moreover , for all @xmath829 such that @xmath990 , we have @xmath991 for every @xmath992 , so that @xmath993 and ( by lemma [ tech - rep ] ) @xmath994 . we have thus obtained that the range of the mapping @xmath220 of the proposition coincides with @xmath972 .
we note that @xmath995 , by the fact that @xmath996 ( because @xmath997 and @xmath998 , if @xmath999 is the point of infinite multiplicity of @xmath214 associated with @xmath941 ) .
it remains to verify that @xmath220 is continuous and that the restriction of @xmath220 to @xmath1000 is one - to - one .
the latter property is easy because the values of @xmath987 for @xmath1001 correspond to distinct equivalence classes for @xmath78 , and we can use lemma [ tech - rep ] .
next we note that @xmath220 is right - continuous with left limits by construction , and that the left limit of @xmath220 at @xmath1002 $ ] is @xmath1003 we already noticed that @xmath966 for any @xmath967 $ ] , and thus we have , for any @xmath1002 $ ] , @xmath1004 where the second equality holds because @xmath994 as mentioned above .
this shows that @xmath1005 and completes the proof .
let us conclude this section with some comments .
recalling that the brownian map @xmath60 is homeomorphic to the two - dimensional sphere @xcite , we get from jordan s theorem that all connected components of @xmath100 are homeomorphic to the disk .
in fact these connected components are called _ brownian disks _ in @xcite . if @xmath941 is a given connected component of @xmath100 , the structure of @xmath941 in a sense that we do not make precise here is described by the associated component @xmath890 of @xmath101 , and the values of @xmath63 on @xmath890 ( shifted so that the boundary values vanish ) .
the preceding data correspond to what is called an excursion of the brownian snake above its minimum in @xcite .
one key result of @xcite states that conditionally on the exit measure process @xmath866 , the excursions above the minimum are independent , and the distribution of the excursion corresponding to a jump @xmath1006 is given by a certain `` excursion measure '' conditioned on the boundary size being equal to @xmath888 .
this suggests that one can reconstruct the brownian map by first considering the metric net @xmath68 ( which is a measurable function of @xmath47 ) and then glueing independently on each `` hole '' of the metric net ( associated with a point of infinite multiplicity of the tree @xmath214 ) a brownian disk corresponding to a brownian snake excursion whose boundary size is the weight of the point of infinite multiplicity .
we postpone a more precise version of the previous discussion to forthcoming work ( see also @xcite ) .
in this section , which is mostly independent of the previous ones , we generalize the subordination by the maximum discussed in section [ sec : subord - maxi ] . to this end
, we deal with the brownian snake associated with a more general spatial motion .
specifically , we consider a strong markov process @xmath1007 with continuous sample paths with values in @xmath253 , and we write @xmath1008 for a probability measure under which @xmath1007 starts from @xmath303 .
we assume that @xmath26 is a regular recurrent point for @xmath1007 , and that @xmath1009=0.\ ] ] we can then define the local time process @xmath1010 of @xmath1007 at @xmath26 ( up to a multiplicative constant ) .
we make the following continuity assumption : there exist two reals @xmath1011 and @xmath512 , and a constant @xmath941 such that , for every @xmath1012 $ ] and @xmath1013 , @xmath1014&\leq c\,t^{2+\ve},\\ \label{contlocal } e_0[l(t)^p]&\leq c\,t^{2+\ve}.\end{aligned}\ ] ] we write @xmath1015 for the excursion measure of @xmath1007 away from @xmath26 associated with the local time process @xmath1016 , and @xmath1017 for the duration of the excursion under @xmath1015 . under the preceding assumptions , the brownian snake whose spatial motion is the pair @xmath1018 is defined by a straightforward adaptation of properties ( a ) and ( b ) stated at the beginning of section [ sec : subord - maxi ] ( see ( * ? ? ?
* chapter iv ) for more details ) , and we denote this process by @xmath1019 , where for every @xmath50 , @xmath1020 for every @xmath1021 , let @xmath1022 denote the excursion measure of @xmath1023 away from @xmath1024 . under @xmath1022 , the `` lifetime process '' @xmath318 is distributed according to the it measure @xmath184 , and as above we let @xmath1025 stand for the duration of the excursion @xmath318 . as previously , @xmath22 denotes the tree coded by @xmath330 and @xmath1026\la \t_\zeta$ ] is the canonical projection .
we write @xmath1027 for the total mass of the exit measure of @xmath1023 from @xmath1028 ( see ( * ? ? ?
* chapter v ) or the appendix below for the definition of exit measures ) .
this makes sense under the excursion measures @xmath1022 for @xmath1029 .
let @xmath1030 be the total local time at @xmath26 accumulated by the path @xmath30 .
if @xmath28 , we write @xmath1031 ( this does not depend on the choice of @xmath31 ) .
then the function @xmath1032 is nondecreasing with respect to the genealogical order .
[ levy - subor ] under @xmath1033 , the subordinate tree @xmath1034 of @xmath22 with respect to the function @xmath1032 is a lvy tree whose branching mechanism @xmath185 can be described as follows : @xmath1035 where @xmath541 is the invariant measure of @xmath1007 defined by @xmath1036 and the function @xmath1037 is given by @xmath1038 as in the proof of theorem [ subor - brown ] , we make use of the special markov property of the brownian snake .
fix @xmath1039 , and consider the domain @xmath1040 . write @xmath1041 for the exit measure from @xmath1042 .
the first - moment formula for exit measures ( * ? ? ?
* proposition v.3 ) shows that @xmath1041 is @xmath1033 a.e .
supported on @xmath1043 , so that we can write @xmath1044 where @xmath1045 is a nonnegative random variable .
let @xmath1046 stand for the @xmath190-field generated by the paths @xmath1019 before they exit @xmath1042 .
by corollary [ smpusual ] in the appendix , under @xmath1033 and conditionally on @xmath1046 , the excursions of the brownian snake @xmath1023 `` outside '' @xmath1042 form a poisson measure with intensity @xmath1047 . now notice that , for every @xmath221 , subtrees of @xmath1034 above level @xmath65 that hit @xmath1048 correspond to those among these excursions that exit @xmath1049 ( we again use proposition [ tree - coding ] to obtain that @xmath1034 is the tree coded by @xmath1050 ) . as in the proof of theorem [ subor - brown ] ( we omit a few details here ) , it follows that the distribution of @xmath1034 under @xmath1033 satisfies the branching property of proposition [ branching - levy ] , and so @xmath1034 under @xmath1033 must be a lvy tree . to determine the branching mechanism of this lvy tree , we fix @xmath1051 , and , for @xmath1052 , we set @xmath1053 by ( * ? ? ?
* theorem v.4 ) , @xmath1054 satisfies the integral equation @xmath1055=\lambda\ ] ] where the markov process @xmath1018 starts from @xmath1024 under the probability measure @xmath1056 , and @xmath1057 is the exit time from @xmath1058 for the process @xmath1018 .
when @xmath329 , excursion theory for @xmath1007 gives @xmath1059= \int_r^r
\mathscr{n}\big(\int_0^\eta u_\lambda(\xi_t,\ell)^2\d t\big ) \d \ell = \int_r^r \d \ell \int m(\d y)\,u_\lambda(y,\ell)^2.\ ] ] set @xmath1060 for @xmath1061 . by a translation argument ,
@xmath1062 does not depend on our choice of @xmath1063 provided that @xmath1064 .
it follows from the preceding considerations that @xmath1065 on the other hand , by applying the special markov property ( corollary [ smpusual ] ) to the domain @xmath1028 , we have , for every @xmath1066 and @xmath1039 , @xmath1067 with the notation introduced in the theorem .
we conclude that @xmath1068 where @xmath185 is as in the statement of the theorem .
note that the functions @xmath1069 are monotone increasing , and so is @xmath185 .
then also implies that @xmath1070 is a continuous nonincreasing function of @xmath65 , that tends to @xmath1071 as @xmath1072 .
it follows that @xmath1073 for every @xmath1039 , and then by dominated convergence that @xmath185 is continuous on @xmath186 .
the unique solution of is given by @xmath1074 ( in particular , we must have @xmath1075 ) .
as @xmath1076 , @xmath1070 converges to @xmath1077 , which coincides with @xmath1078 , where @xmath1079 denotes the height of @xmath1034 .
hence , the function @xmath1080 is given by @xmath1081 and this suffices to establish that the branching mechanism of @xmath14 is @xmath185 .
the formula for @xmath185 that appears in theorem [ levy - subor ] is not explicit and in general does not allow the calculation of this function
. we will now argue that we can identify @xmath185 , up to a multiplicative constant , if @xmath1007 satisfies a scaling property . from now on until the end of the section , we assume ( in addition to the previous hypotheses ) that there exists a constant @xmath1082 such that , for every @xmath1083 and @xmath201 , the law of @xmath1084 under @xmath1008 coincides with the law of @xmath1085 under @xmath1086 . in other words ,
the process @xmath1007 is a self - similar markov process with values in @xmath186 , see the survey @xcite for more information on this class of processes .
a particular case ( with @xmath1087 ) is the bessel process of dimension @xmath1088 .
the excursion measure @xmath1015 must then satisfy a similar scaling invariance property .
more precisely , for every @xmath201 , the law of @xmath1084 under @xmath1015 must be equal to @xmath1089 times the law of @xmath1085 under @xmath1015 , for some constant @xmath1090 .
the fact that @xmath1091 is clear since the scaling property implies that @xmath1092 and we must have @xmath1093 .
the inverse local time of @xmath1007 at @xmath26 is then a stable subordinator of index @xmath93 , which is consistent with assumption .
[ branchmecastable ] under the preceding assumptions , there exists a constant @xmath1094 such that @xmath1095 .
we first observe that @xmath1096 for some positive constant @xmath1097 . to see this ,
we write , for every @xmath201 , @xmath1098 it follows that @xmath541 has the form stated above .
we then observe that , for every @xmath201 , we can also consider the following scaling transformation of the brownian snake : @xmath1099 and the `` law '' of @xmath1100 under @xmath1101 coincides with @xmath1102 times the `` law '' of @xmath33 under @xmath1103 .
furthermore the exit measure @xmath1104 associated with @xmath1100 is equal to @xmath1105 ( we leave the details as an exercise for the reader ) . with the notation of theorem [ levy - subor ] , it follows that , for every @xmath1029 and @xmath201 , @xmath1106 hence , for every @xmath1039 and @xmath1107 , @xmath1108 using the change of variables @xmath1109 .
this completes the proof .
if @xmath1110 is the absolute value of a linear brownian motion @xmath1111 , then a famous theorem of lvy asserts that the pair @xmath1018 has the same distribution as @xmath1112 , where @xmath1113 ( to be specific , this holds with a particular choice of the normalization of @xmath1114 ) .
we then see that theorem [ subor - brown ] is a special case of theorem [ levy - subor ] and proposition [ branchmecastable ] . in that case ,
@xmath1115 , and we recover the formula @xmath1116 .
in this appendix , we derive a more precise and more general form of the special markov property for the brownian snake , which was first stated in @xcite . this result
is closely related to the special markov property for superprocesses as stated by dynkin ( * ? ? ?
* theorem 1.6 ) , but the formulation in terms of the brownian snake , although less general , gives additional information that is crucial for our purposes .
we consider the setting of ( * ? ? ?
* chapter v ) .
we let @xmath1007 be a markov process with values in a polish space @xmath1117 with continuous sample paths . for every @xmath1118 ,
the process @xmath1007 starts from @xmath303 under the probability measure @xmath1008 .
analogously to , we assume that the following strong continuity assumption holds for every @xmath1118 and @xmath200 , @xmath1119\leq c\,t^{2+\ve},\ ] ] where @xmath1120 , @xmath1011 and @xmath512 are constants .
according to ( * ? ? ?
* section iv.4 ) , this continuity assumption allows us to construct the brownian snake @xmath52 with continuous sample paths with values in the space @xmath1121 of all finite continuous paths in @xmath1122 ( the set @xmath1121 is defined by the obvious generalization of the beginning of section [ sec : subord - maxi ] , and we keep the notation @xmath1123 for the tip of a path @xmath1124 ) .
the strong markov property holds for @xmath52 , even without assuming that it holds for the underlying spatial motion @xmath1007 .
we again write @xmath314 for the probability measure under which the brownian snake starts from ( the trivial path equal to ) @xmath303 , and @xmath316 for the excursion measure of the brownian snake away from @xmath303 .
it will also be useful to introduce conditional distributions of the brownian snake given its lifetime process .
if @xmath1125 is a continuous function such that @xmath1126 and @xmath12 is locally hlder with exponent @xmath1127 for every @xmath450 , we write @xmath1128 for the conditional distribution under @xmath314 of @xmath52 knowing that @xmath1129 for every @xmath50
. see ( * ? ? ?
* chapter iv ) , and note that these conditional distributions are easily defined using the analog in our general setting of property ( b ) stated at the beginning of section [ sec : subord - maxi ] .
we now fix a connected open subset @xmath1130 of @xmath1122 and @xmath1131 .
we use the notation @xmath1132 , and , for every @xmath1133 , @xmath1134 , where in both cases @xmath1135 .
we assume that @xmath1136 and note that this implies that @xmath1137 we set , for every @xmath50 , @xmath1138 this definition makes sense @xmath314 a.s .
we let @xmath1139 be the @xmath190-field generated by the process @xmath1140 and the collection of all @xmath314-negligible sets .
informally , @xmath1139 represents the information provided by the paths @xmath30 before they exit @xmath1130 .
[ mbreflected ] for every @xmath50 , set @xmath1141 , and @xmath1142 under the probability measure @xmath314 , we have @xmath1143 for every @xmath50 , a.s . , and the process @xmath1144 is distributed as a reflected brownian motion in @xmath253 and is independent of the @xmath190-field @xmath1139 .
this is essentially lemma v.2 in @xcite , except that the independence property is not stated in that lemma .
however a close look at the proof in @xcite shows that the process @xmath1145 is obtained as the limit of approximating processes ( denoted by @xmath1146 in @xcite ) which are independent of @xmath1139 thanks to the strong markov property of the brownian snake .
we write @xmath1147 for the local time at @xmath26 of the process @xmath49 , and define a process with continuous nondecrasing sample paths by setting @xmath1148 then ( see ( * ? ? ?
* section v.1 ) ) , @xmath1149 for every @xmath50 , @xmath314 a.s .
the process @xmath1150 is called the exit local time process from @xmath1130 .
notice that the measure @xmath1151 is supported on @xmath1152 .
this follows from the proof of ( * ? ? ?
* proposition 2.3 ) in the special case where @xmath1007 is @xmath1155-dimensional brownian motion .
the argument however can be adapted to our more general setting and we omit the details . before stating the special markov property , we need some additional notation . for every @xmath613 ,
we set @xmath1156 we note that the process @xmath1157 is @xmath1139-measurable by lemma [ measurability - exit ] .
we now define the excursions of the brownian snake outside @xmath1130 .
we observe that , @xmath314 a.e .
, the set @xmath1158 is a countable union of disjoint open intervals , which we enumerate as @xmath355 , @xmath1159 .
here we can fix the enumeration by saying that we enumerate first the excursion intervals with length at most @xmath1160 whose initial time is smaller than @xmath276 , then the excursion intervals with length at most @xmath1161 whose initial time is smaller than @xmath1162 which have not yet been listed , and so on . if we choose this enumeration procedure , the variables @xmath1163 are measurable with respect to the @xmath190-field generated by @xmath1164 , and the variables @xmath1165 and @xmath1166 are measurable with respect to the @xmath190-field generated by @xmath683 . from the properties of the brownian snake , one has , @xmath314 a.e . for every @xmath1159 and every @xmath356 $ ] , @xmath1167 and more precisely all paths @xmath30 , @xmath356 $ ]
coincide up to their exit time from @xmath1130 . for every @xmath1159
, we then define an element @xmath360 of the space of all continuous functions from @xmath253 into @xmath1121 by setting , for every @xmath50 , @xmath359 by definition , the random variables @xmath360 , @xmath1159 , are the excursions of the brownian snake outside @xmath1130 ( the word `` outside '' is a bit misleading here ) .
notice that , for every @xmath1168 , @xmath1169 is a measurable function of @xmath683 .
indeed the processes @xmath1169 , @xmath1168 are just the excursions of @xmath49 away from @xmath26 enumerated as explained above .
it is convenient to introduce the auxiliary markov process defined by @xmath1172 .
we observe that the brownian snake associated with @xmath1173 can be obtained by the formula @xmath1174 notice that @xmath1175 is a measurable function of @xmath1176 , and recall that the intervals @xmath355 are just the connected components of the complement of the zero set of @xmath220 .
consider then , independently for every @xmath1159 , a process @xmath1177 which conditionally on @xmath1176 is distributed according to the probability measure @xmath1178 where we recall our notation @xmath1128 for the conditional distribution under @xmath314 of @xmath52 knowing that @xmath1129 for every @xmath50 .
define @xmath1179 for every @xmath50 by setting @xmath1180 if @xmath1181 and , for every @xmath1159 , for every @xmath1182 , @xmath1183 a tedious but straightforward verification shows that the finite marginal distributions of the process @xmath1184 are the same as those of the process @xmath52 .
it follows that , conditionally on @xmath1185 , the `` excursions '' @xmath360 are independent and the conditional distribution of @xmath360 is @xmath1186 . at this point
, we claim that we have a.s . for every @xmath1159 , @xmath1187 the first equality in is immediate . to get the second one , set @xmath1188 to simplify notation .
we first note that @xmath1189 because @xmath1190 ( the strong markov property of the brownian snake ensures that in each interval @xmath1191 $ ] , @xmath512 , we can find a set of positive lebesgue measure of values of @xmath31 such that @xmath1192 ) and we know that @xmath357 .
thus our claim will follow if we can verify that @xmath1193 on one hand the condition @xmath1194 implies @xmath1195 and @xmath1196 .
it follows that @xmath1197 . on the other hand ,
suppose that @xmath1198 .
we first note that @xmath1199 where the last equality holds by the support property of @xmath1151 .
furthermore , the left limit of @xmath1200 at @xmath1201 is smaller than or equal to @xmath1202 by construction .
so the condition @xmath1198 means that @xmath1203 is a discontinuity point of @xmath1200 .
however , we noticed that the random variables @xmath1166 are measurable functions of @xmath683 and therefore independent of @xmath1139 . since @xmath1204 is measurable with respect to @xmath1139 , and since @xmath1205 has only countably many discontinuity times , the ( easy ) fact that the law of @xmath1166 has no atoms implies that , with @xmath314-probability one , @xmath1166 can not be a discontinuity time of @xmath1200 .
this contradiction completes the proof of our claim .
next , let @xmath1206 be a bounded @xmath1139-measurable real random variable , and let @xmath12 and @xmath156 be nonnegative random variables defined respectively on @xmath253 and on the space of continuous functions from @xmath253 into @xmath1121 . by conditioning first with respect to @xmath1176 , we get @xmath1207 = \e_x\bigg [ u\times \prod_{i\in \n } \q_{\wh w_{a_i}}^{(\zeta^{(i)})}\big(e^{-g(l^d_{a_i } ) g(\cdot ) } \big)\bigg],\ ] ] noting that an @xmath1139-measurable real variable coincides @xmath314 a.s . with a function of @xmath1176 . using , we see that , for every @xmath1159 , @xmath1208 is a measurable function ( that does not depend on @xmath1209 ) of the pair @xmath1210 and of the process @xmath1211 , which is @xmath1139-measurable . the point measure @xmath1212 which is just the point measure of excursions of the reflected brownian motion @xmath49 , is poisson with intensity @xmath1213 , where @xmath1214 is as previously the it excursion measure .
since the latter point measure is independent of @xmath1139 ( by lemma [ mbreflected ] ) , we can now condition with respect to @xmath1139 , applying the exponential formula for poisson measures , to get that the quantities in are equal to @xmath1215.\ ] ] the statement of the theorem follows . in the preceding sections , we use a version of theorem [ smpgeneral ] under the excursion measure @xmath316 , which we will now state as a corollary .
we set @xmath1216 and observe that @xmath1217 ( if the quantity @xmath1218 were infinite , excursion theory would give a contradiction with the fact that the brownian snake has continuous paths under @xmath314 ) .
then we note that the conditional probability measure @xmath1219 can be interpreted as the law under @xmath314 of the first brownian excursion away from @xmath303 that exits @xmath1130 .
thanks to this observation , we can make sense of the exit local time process @xmath1150 under @xmath1220 by formula .
the definition of @xmath1221 and @xmath1222 remains the same , and we can set @xmath1223 as previously .
the difference of course is the fact that @xmath1224 is now finite @xmath1220 a.s .
so we may define @xmath1225 only for @xmath1226 .
after these observations , we may state the version of the special markov property under @xmath1220 .
we slightly abuse notation by still writing @xmath1227 and @xmath1228 for the excursions outside @xmath1130 and the associated intervals , which are defined in exactly the same way as previously .
the @xmath190-field @xmath1139 is again the @xmath190-field generated by @xmath1222 but should be completed here with the @xmath316-negligible sets .
[ smpexcursion ] under @xmath1220 , conditionally on the @xmath190-field @xmath1139 , the point measure @xmath1170 is poisson with intensity @xmath1229}(\ell)\,\d \ell \,\n_{\wh w^d_{\theta_\ell}}(\d \omega).\ ] ] note that @xmath1230 is @xmath1139-measurable .
the corollary is a straightforward consequence of theorem [ smpgeneral ] , interpreting @xmath1220 as the law under @xmath314 of the first brownian excursion away from @xmath303 that exits @xmath1130 .
we omit the details .
if we are only interested in the point measure @xmath1231 , we can state the preceding corollary in a slightly different form , by introducing the notion of the exit measure : the exit measure from @xmath1130 is the random measure @xmath1232 on @xmath1233 defined ( under @xmath316 or under @xmath1220 ) by the formula @xmath1234 see ( * ? ? ?
* section v.1 ) .
note that the total mass of @xmath1232 is @xmath1235 .
a change of variables shows that we have as well @xmath1236 in particular , @xmath1232 is @xmath1139-measurable by lemma [ measurability - exit ] .
the following result is now an immediate consequence of corollary [ smpexcursion ] . | we discuss subordination of random compact @xmath0-trees .
we focus on the case of the brownian tree , where the subordination function is given by the past maximum process of brownian motion indexed by the tree . in that particular case ,
the subordinate tree is identified as a stable lvy tree with index @xmath1 . as a more precise alternative formulation ,
we show that the maximum process of the brownian snake is a time change of the height process coding the lvy tree .
we then apply our results to properties of the brownian map .
in particular , we recover , in a more precise form , a recent result of miller and sheffield identifying the metric net associated with the brownian map . |
in the public goods game , each player @xmath2 in the population @xmath3 chooses whether to contribute ( @xmath4 ) or not to contribute ( @xmath5 ) to the common pool . given a fixed _ rate of return _
@xmath6 , the resulting payoff of player @xmath2 is then @xmath7 .
we shall call @xmath8 the game s _ marginal per - capita rate of return _ and denote it as @xmath9 . note that for simplicity , but without loss of generality , we have assumed that the group is the whole population . in the absence of restrictions on the interaction range of players @xcite , i.e. , in well - mixed populations , the size of the groups and their formation can be shown to be of no relevance in our case , as long as @xmath9 rather than @xmath10 is considered as the effective rate of return .
the directional learning dynamics is implemented as follows .
suppose the above game is infinitely repeated at time steps @xmath11 , and suppose further that @xmath2 , at time @xmath12 , plays @xmath13 with probability @xmath14 $ ] and @xmath15 with probability @xmath16 .
let the vector of contribution probabilities @xmath17 describe the state of the game at time @xmath12 .
we initiate the game with all @xmath18 lying on the @xmath1-grid between @xmath19 and @xmath20 , while subsequently individual mixed strategies evolve randomly subject to the following three `` directional bias '' rules : upward : : : if @xmath21 and @xmath22 , or if @xmath23 and @xmath24 , then @xmath25 if @xmath26 ; otherwise , @xmath27 .
neutral : : : if @xmath28 and/or @xmath29 , then @xmath27 , @xmath30 , or @xmath31 with equal probability if @xmath32 ; otherwise , @xmath27 .
downward : : : if @xmath21 and @xmath24 , or if @xmath23 and @xmath22 , then @xmath33 if @xmath34 ; otherwise , @xmath27 .
note that the second , neutral rule above allows random deviations from any intermediate probability @xmath35 .
however , @xmath36 and @xmath37 for all @xmath2 are absorbing state candidates .
we therefore introduce perturbations to this directional learning dynamics and study the resulting stationary states . in particular , we consider perturbations of order @xmath38 such that , with probability @xmath39 , the dynamics is governed by the original three `` directional bias '' rules . however , with probability @xmath38 , either @xmath27 , @xmath33 or @xmath25 happens equally likely ( with probability @xmath40 ) but of course obeying the @xmath41 $ ] restriction .
we begin with a formal definition of the _
@xmath0strong equilibrium_. in particular , a pure strategy imputation @xmath42 is a @xmath43-strong equilibrium of our ( symmetric ) public goods game if , for all @xmath44 with @xmath45 , @xmath46 for all @xmath47 for any alternative pure strategy set @xmath48 for @xmath49 . as noted in the previous section ,
this definition bridges , one the one hand , the concept of the nash equilibrium in pure strategies @xcite in the sense that any @xmath0strong equilibrium with @xmath50 is also a nash equilibrium , and , on the other hand , that of the ( aumann-)strong equilibrium @xcite in the sense that any @xmath0strong equilibrium with @xmath51 is aumann strong .
equilibria in between ( for @xmath52 ) are `` more stable '' than a nash equilibrium , but `` less stable '' than an aumann - strong equilibrium .
the maximal @xmath43-strengths of the equilibria in our public goods game as a function of @xmath10 are depicted in fig .
[ ana ] for @xmath53 .
the cyan - shaded region indicates the `` public bad game '' region for @xmath54 ( @xmath55 ) , where the individual and the public motives in terms of the nash equilibrium of the game are aligned towards defection . here
@xmath5 for all @xmath2 is the unique aumann - strong equilibrium , or in terms of the definition of the @xmath0strong equilibrium , @xmath5 for all @xmath2 is @xmath0strong for all @xmath56 $ ] .
the magenta - shaded region indicates the typical public goods game for @xmath57 ( @xmath58 ) , where individual and public motives are conflicting .
here there exists no aumann - strong equilibria .
the outcome @xmath5 for all @xmath2 is the unique nash equilibrium , and that outcome is also @xmath43-strong equilibrium for some @xmath59 , where the size of @xmath43 depends on @xmath10 and @xmath60 in that @xmath61 while @xmath62 .
finally , the gray - shaded region indicates the unconflicted public goods game for @xmath63 ( @xmath64 ) , where individual and public motives are again aligned , but this time towards cooperation . here
@xmath4 for all @xmath2 abruptly becomes the unique nash and aumann - strong equilibrium , or equivalently the unique @xmath0strong equilibrium for all @xmath65 $ ] .
if we add perturbations of order @xmath38 to the unperturbed public goods game with directional learning that we have introduced in section 2 , there exist stationary distributions of @xmath66 and the following proposition can be proven . in the following ,
we denote by `` @xmath43 '' the maximal @xmath0strength of an equilibrium .
proposition : : : as @xmath67 , starting at any @xmath68 , the expectation with respect to the stationary distribution is @xmath69 > 1/2 $ ] if @xmath70 and @xmath69<1/2 $ ] if @xmath71 .
@xmath72/\partial \epsilon<0 $ ] if @xmath73 , and @xmath72/\partial \epsilon>0 $ ] if @xmath74
. moreover , @xmath72/\partial \delta>0 $ ] , and @xmath72/ \partial \delta<0 $ ] if @xmath73 .
finally , @xmath72/ \partial k<0 $ ] if @xmath71 .
we begin the proof by noting that the perturbed process given by our dynamics results in an irreducible and aperiodic markov chain , which has a unique stationary distribution .
when @xmath75 , any absorbing state must have @xmath76 or @xmath20 for all players .
this is clear from the positive probability paths to either extreme from intermediate states given by the unperturbed dynamics .
we shall now analyze whether @xmath77 or @xmath20 , given that @xmath78 or @xmath20 for all @xmath79 , has a larger attraction given the model s underlying parameters . if @xmath73 , the probability path for any player to move from @xmath80 to @xmath81 in some @xmath82 steps requires a single perturbation for that player and is therefore of the order of a single @xmath38 .
by contrast , the probability for any player to move from @xmath83 to @xmath84 in @xmath85 steps is of the order @xmath86 , because at least two other players must increase their contribution in order for that player to experience a payoff increase from his non - contribution . along any other path or if @xmath17 is such that there are not two players @xmath87 with @xmath78 to make this move , then the probability for @xmath2 to move from @xmath83 to @xmath84 in @xmath85 steps requires even more perturbations and is of higher order .
notice that , for any one player to move from @xmath76 to @xmath81 we need at least two players to move away from @xmath77 along the least - resistance paths . because contributing 1 is a best reply for all @xmath73 , those two players will also continue to increase if continuing to contribute 1 .
notice that the length of the path is @xmath88 steps , and that the path requires no perturbations along the way , which is less likely the smaller @xmath1 .
if @xmath71 , the probability for any player to move from @xmath83 to @xmath89 in some @xmath82 steps requires a single perturbation for that player and is therefore of the order of a single @xmath38 .
by contrast , the probability for any player to move from @xmath76 to @xmath90 in some @xmath85 steps is at least of the order @xmath91 , because at least @xmath43 players ( corresponding to the maximal @xmath43-strength of the equilibrium ) must contribute in order for all of these players to experience a payoff increase .
notice that @xmath43 decreases in @xmath9 .
again , the length of the path is @xmath82 steps , and that path requires no perturbations along the way , which is less likely the smaller @xmath1 . with this , we conclude the proof of the proposition .
however , it is also worth noting a direct corollary of the proposition ; namely , as @xmath92 , @xmath69\rightarrow 1 $ ] if @xmath70 , and @xmath69\rightarrow 0 $ ] if @xmath71 .
lastly , we simulate the perturbed public goods game with directional learning and determine the actual average contribution levels in the stationary state .
color encoded results in dependence on the normalized rate of return @xmath9 and the responsiveness of players to the success of their past actions @xmath1 ( alternatively , the sensitivity of the individual learning process ) are presented in fig .
[ simul ] for @xmath93 .
small values of @xmath1 lead to a close convergence to the respective nash equilibrium of the game , regardless of the value of @xmath9 . as the value of @xmath1 increases , the pure nash equilibria erode and give way to a mixed outcome .
it is important to emphasize that this is in agreement , or rather , this is in fact a consequence of the low @xmath0strengths of the non - contribution pure equilibria ( see fig [ ana ] ) . within intermediate to
large @xmath1 values the nash equilibria are implemented in a zonal rather than pinpoint way . when the nash equilibrium is such that all players contribute ( @xmath64 ) , then small values of @xmath1 lead to more efficient aggregate play ( recall any such equilibrium
is @xmath94strong ) .
conversely , by the same logic , when the nash equilibrium is characterized by universal free - riding , then larger values of @xmath1 lead to more efficient aggregate play .
moreover , the precision of implementation also depends on the rate of return in the sense that uncoordinated deviations of groups of players lead to more efficient outcomes the higher the rate of return .
in other words , the free - riding problem is mitigated if group deviations lead to higher payoffs for every member of an uncoordinated deviation group , the minimum size of which ( that in turn is related to the maximal @xmath0 strength of equilibrium ) is decreasing with the rate of return .
simulations also confirm that the evolutionary outcome is qualitatively invariant to : i ) the value of @xmath38 as long as the latter is bounded away from zero , although longer convergence times are an inevitable consequence of very small @xmath38 values ( see fig . [ eps ] ) ; ii ) the replication of the population ( i.e. , making the whole population a group ) and the random remixing between groups ; and iii ) the population size , although here again the convergence times are the shorter the smaller the population size .
while both ii and iii are a direct consequence of the fact that we have considered the public goods game in a well - mixed rather than a structured population ( where players would have a limited interaction range and where thus pattern formation could play a decisive role @xcite ) , the qualitative invariance to the value of @xmath38 is elucidated further in fig .
we would like to note that by `` qualitative invariance '' it is meant that , regardless of the value of @xmath95 , the population always diverges away from the nash equilibrium towards a stable mixed stationary state . but as can be observed in fig .
[ eps ] , the average contribution level and its variance both increase slightly as @xmath38 increases .
this is reasonable if one considers @xmath38 as an exploration or mutation rate .
more precisely , it can be observed that , the lower the value of @xmath38 , the longer it takes for the population to move away from the nash equilibrium where everybody contributes zero in the case that @xmath58 ( which was also the initial condition for clarity ) .
however , as soon as initial deviations ( from @xmath36 in this case ) emerge ( with probability proportional to @xmath96 , the neutral rule in the original learning dynamics takes over , and this drives the population towards a stable mixed stationary state .
importantly , even if the value of @xmath38 is extremely small , the random drift sooner or later gains momentum and eventually yields similar contribution levels as those attainable with larger values of @xmath38 .
most importantly , note that there is a discontinuous jump towards staying in the nash equilibrium , which occurs only if @xmath38 is exactly zero .
if @xmath38 is bounded away from zero , then the free - riding nash equilibrium erodes unless it is @xmath94strong ( for very low values of @xmath97 ) .
we have introduced a public goods game with directional learning , and we have studied how the level of contributions to the common pool depends on the rate of return and the responsiveness of individuals to the successes and failures of their own past actions .
we have shown that directional learning alone suffices to explain deviations from the nash equilibrium in the stationary state of the public goods game .
even though players have no strategically relevant information about the game and/ or about each others actions , the population could still end up in a mixed stationary state where some players contributed at least part of the time although the nash equilibrium would be full free - riding .
vice versa , defectors emerged where cooperation was clearly the best strategy to play .
we have explained these evolutionary outcomes by introducing the concept of @xmath0strong equilibria , which bridge the gap between nash and aumann - strong equilibria .
we have demonstrated that the lower the maximal @xmath0strength and the higher the responsiveness of individuals to the consequences of their own past strategy choices , the more likely it is for the population to ( mis)learn what is the objectively optimal unilateral ( nash ) play .
these results have some rather exciting implications .
foremost , the fact that the provisioning of public goods even under adverse conditions can be explained without any sophisticated and often lengthy arguments involving selflessness or social preference holds promise of significant simplifications of the rationale behind seemingly irrational individual behavior in sizable groups .
it is simply enough for a critical number ( depending on the size of the group and the rate of return ) of individuals to make a `` wrong choice '' at the same time once , and if only the learning process is sufficiently fast or naive , the whole subpopulation is likely to adopt this wrong choice as their own at least part of the time . in many real - world situations , where the rationality of decision making is often compromised due to stress , propaganda or peer pressure , such `` wrong choices ''
are likely to proliferate .
as we have shown in the context of public goods games , sometimes this means more prosocial behavior , but it can also mean more free - riding , depending only on the rate of return .
the power of directional ( mis)learning to stabilize unilaterally suboptimal game play of course takes nothing away from the more traditional and established explanations , but it does bring to the table an interesting option that might be appealing in many real - life situations , also those that extend beyond the provisioning of public goods . fashion trends or viral tweets and videos might all share a component of directional learning before acquiring mainstream success and recognition .
we hope that our study will be inspirational for further research in this direction .
the consideration of directional learning in structured populations @xcite , for example , appears to be a particularly exciting future venture .
for the characterization of the stationary states , we introduce the concept of @xmath0strong equilibria , which nests both the nash equilibrium @xcite and the aumann - strong equilibrium @xcite as two special cases .
while the nash equilibrium describes the robustness of an outcome against unilateral ( @xmath20-person ) deviations , the aumann - strong equilibrium describes the robustness of an outcome against the deviations of any subgroup of the population .
an equilibrium is said to be ( aumann-)strong if it is robust against deviations of the whole population or indeed of any conceivable subgroup of the population , which is indeed rare .
our definition of the @xmath0strong equilibrium bridges the two extreme cases , measuring the size of the group @xmath98 ( at or above nash ) and hence the degree to which an equilibrium is stable .
we note that our concept is related to coalition - proof equilibrium @xcite . in the public goods game , the free - riding nash equilibrium is typically also more than @xmath99strong but never @xmath94strong .
as we will show , the maximal strength @xmath43 of an equilibrium translates directly to the level of contributions in the stationary distribution of our process , which is additionally determined by the normalized rate of return @xmath9 and the responsiveness of players to the success of their past actions @xmath1 , i.e. , the sensitivity of the individual learning process .
this research was supported by the european commission through the erc advanced investigator grant ` momentum ' ( grant 324247 ) , by the slovenian research agency ( grant p5 - 0027 ) , and by the deanship of scientific research , king abdulaziz university ( grant 76 - 130 - 35-hici ) .
ledyard , j. o. public goods : a survey of experimental research . in the handbook of experimental economics , kagel , j. h. and roth , a. e. , editors , 111194 .
princeton university press , princeton , nj ( 1997 ) . | we consider an environment where players are involved in a public goods game and must decide repeatedly whether to make an individual contribution or not .
however , players lack strategically relevant information about the game and about the other players in the population .
the resulting behavior of players is completely uncoupled from such information , and the individual strategy adjustment dynamics are driven only by reinforcement feedbacks from each player s own past .
we show that the resulting `` directional learning '' is sufficient to explain cooperative deviations away from the nash equilibrium .
we introduce the concept of @xmath0strong equilibria , which nest both the nash equilibrium and the aumann - strong equilibrium as two special cases , and we show that , together with the parameters of the learning model , the maximal @xmath0strength of equilibrium determines the stationary distribution .
the provisioning of public goods can be secured even under adverse conditions , as long as players are sufficiently responsive to the changes in their own payoffs and adjust their actions accordingly .
substantial levels of public cooperation can thus be explained without arguments involving selflessness or social preferences , solely on the basis of uncoordinated directional ( mis)learning . cooperation in sizable groups has been identified as one of the pillars of our remarkable evolutionary success . while between - group conflicts and the necessity for alloparental care
are often cited as the likely sources of the other - regarding abilities of the genus _ homo _
@xcite , it is still debated what made us the `` supercooperators '' that we are today @xcite .
research in the realm of evolutionary game theory @xcite has identified a number of different mechanisms by means of which cooperation might be promoted @xcite , ranging from different types of reciprocity and group selection to positive interactions @xcite , risk of collective failure @xcite , and static network structure @xcite .
the public goods game @xcite , in particular , is established as an archetypical context that succinctly captures the social dilemma that may result from a conflict between group interest and individual interests @xcite . in its simplest form
, the game requires that players decide whether to contribute to a common pool or not .
regardless of the chosen strategy by the player himself , he receives an equal share of the public good which results from total contributions being multiplied by a fixed rate of return . for typical rates of return
it is the case that , while the individual temptation is to free - ride on the contributions of the other players , it is in the interest of the collective for everyone to contribute . without additional mechanisms such as punishment @xcite , contribution decisions in such situations @xcite
approach the free - riding nash equilibrium @xcite over time and thus lead to a `` tragedy of the commons '' @xcite .
nevertheless , there is rich experimental evidence that the contributions are sensitive to the rate of return @xcite and positive interactions @xcite , and there is evidence in favor of the fact that social preferences and beliefs about other players decisions are at the heart of individual decisions in public goods environments @xcite . in this paper
, however , we shall consider an environment where players have no strategically relevant information about the game and/ or about other players , and hence explanations in terms of social preferences and beliefs are not germane .
instead , we shall propose a simple learning model , where players may mutually reinforce learning off the equilibrium path .
as we will show , this phenomenon provides an alternative and simple explanation for why contributions rise with the rate of return , as well as why , even under adverse conditions , public cooperation may still prevail .
previous explanations of this experimental regularity @xcite are based on individual - level costs of ` error ' @xcite .
suppose each player knows neither who the other players are , nor what they earn , nor how many there are , nor what they do , nor what they did , nor what the rate of return of the underlying public goods game is .
players do not even know whether the underlying rate of return stays constant over time ( even though in reality it does ) because their own payoffs are changing due to the strategy adjustments of other players , about which they have no information . without any such knowledge , players are unable to determine ex ante
whether contributing or not contributing is the better strategy in any given period , i.e. , players have no strategically relevant information about how to respond best . as a result ,
the behavior of players has to be _ completely uncoupled _ @xcite , and their strategy adjustment dynamics are likely to follow a form of _ reinforcement _ @xcite feedback or , as we shall call it , _ directional learning _
@xcite .
we note that , in our model , due to the one - dimensionality of the strategy space , reinforcement and directional learning are both adequate terminologies for our learning model .
since reinforcement applies also to general strategy spaces and is therefore more general we will prefer the terminology of directional learning .
indeed , such directional learning behavior has been observed in recent public goods experiments @xcite .
the important question is how _
well _ will the population learn to play the public goods game despite the lack of strategically relevant information .
note that _
well _ here has two meanings due to the conflict between private and collective interests : on the one hand , how close will the population get to playing the nash equilibrium , and , on the other hand , how close will the population get to playing the socially desirable outcome .
the learning model considered in this paper is based on a particularly simple `` directional learning '' algorithm which we shall now explain .
suppose each player plays both cooperation ( contributing to the common pool ) and defection ( not contributing ) with a mixed strategy and updates the weights for the two strategies based on their relative performances in previous rounds of the game .
in particular , a player will increase its weight on contributing if a previous - round switch from not contributing to contributing led to a higher realized payoff or if a previous - round switch from contributing to not contributing led to a lower realized payoff .
similarly , a player will decrease its weight on contributing if a previous - round switch from contributing to not contributing led to a higher realized payoff or if a previous - round switch from not contributing to contributing led to a lower realized payoff . for simplicity
, we assume that players make these adjustments at a fixed incremental step size @xmath1 , even though this could easily be generalized .
in essence , each player adjusts its mixed strategy directionally depending on a markovian performance assessment of whether a previous - round contribution increase / decrease led to a higher / lower payoff .
since the mixed strategy weights represent a well - ordered strategy set , the resulting model is related to the directional learning/ aspiration adjustment models @xcite , and similar models have previously been proposed for bid adjustments in assignment games @xcite , as well as in two - player games @xcite .
in @xcite the dynamic leads to stable cooperative outcomes that maximize total payoffs , while nash equilibria are reached in @xcite . the crucial difference between these previous studies and our present study
is that our model involves more than two players in a voluntary contributions setting , and , as a result , that there can be interdependent directional adjustments of groups of players including more than one but not all the players .
this can lead to uncoordinated ( mis)learning of subpopulations in the game . consider the following example .
suppose all players in a large standard public goods game do not contribute to start with .
then suppose that a player in a subpopulation uncoordinatedly but by chance simultaneously decide to contribute .
if this group is sufficiently large ( the size of which depends on the rate of return ) , then this will result in higher payoffs for all players including the contributors , despite the fact that not contributing is the dominant strategy in terms of unilateral replies . in our model ,
if indeed this generates higher payoffs for all players including the freshly - turned contributors , then the freshly - turned contributors would continue to increase their probability to contribute and thus increase the probability to trigger a form of stampede or herding effect , which may thus lead away from the nash equilibrium and towards a socially more beneficial outcome . our model of uncoordinated but mutually reinforcing deviations away from nash provides an alternative explanation for the following regularity that has been noted in experiments on public goods provision @xcite .
namely , aggregate contribution levels are higher the higher the rate of return , despite the fact that the nash equilibrium remains unchanged ( at no - contribution ) .
this regularity has previously been explained only at an individual level , namely that ` errors ' are less costly and therefore more likely the higher the rate of return , following quantal - response equilibrium arguments @xcite . by contrast , we provide a group - dynamic argument .
note that the alternative explanation in terms of individual costs is not germane in our setting , because we have assumed that players have no information to make such assessments .
it is in this sense that our explanation perfectly complements the explanation in terms of costs . in what follows , we present the results , where we first set up the model and
then deliver our main conclusions .
we discuss the implications of our results in section 3 .
further details about the applied methodology are provided in the methods section . |
0.2 cm i would like to thank the organizers of the ustron school for their kind hospitality .
many thanks go to andrea brignole and fabio zwirner for the pleasant collaboration on which this talk is based .
99 s. ambrosanio , g.l .
kane , g.d .
kribs , s.p .
martin and s. mrenna , phys .
d54 ( 1996 ) 5395 and refs . therein .
p. fayet , phys .
b70 ( 1977 ) 461 and b86 ( 1979 ) 272 ; r. casalbuoni , s. de curtis , d. dominici , f. feruglio and r. gatto , phys .
b215 ( 1988 ) 313 and phys .
d39 ( 1989 ) 2281 .
a. brignole , f. feruglio and a. zwirner , preprint hep - ph/9703286 , nucl .
b501 ( 1997 ) 332 and refs . therein .
t. moroi , h. murayama and m. yamaguchi , phys .
b303 ( 1993 ) 289 ; t. gherghetta , nucl .
b485 ( 1997 ) 25 ; j. a. grifols , r. mohapatra and a. riotto , phys .
b401 ( 1997 ) 283 ; m. a. luty and e. ponton , preprint hep - ph/9706268 ; j.a .
grifols , e. mass and r. toldr , preprint hep - ph/9707536 ; d.a .
dicus , r.n .
mohapatra and v.l .
teplitz , preprint hep - ph/9708369 .
see ref . [ 25 ] in @xcite .
b. de wit and d.z .
freedman , phys .
d12 ( 1975 ) 2286 and phys .
35 ( 1975 ) 827 .
volkov and v.p .
akulov , phys .
b46 ( 1973 ) 109 ; e.a . ivanov and a.a .
kapustnikov , j. phys .
a11 ( 1978 ) 2375 ; t. uematsu and c.k .
zachos , nucl .
b201 ( 1982 ) 250 ; s. ferrara , l. maiani and p.c .
west , z. phys .
c19 ( 1983 ) 267 ; f. liebrand , phys . rev .
d42 ( 1990 ) 3457 ; t.e .
clark and s.t .
love , phys .
d54 ( 1996 ) 5723 .
s. samuel and j. wess , nucl .
b221 ( 1983 ) 153 ; j. wess , in _ quantum theory of particles and fields _ ( b. jancewicz and j. lukierski eds . ) , world scientific , 1983 , p. 223 .
a. brignole , f. feruglio and f. zwirner , hep - th/9709111 , to appear in jhep , and refs .
a. brignole , f. feruglio and f. zwirner , in preparation . | i discuss the low - energy limit of several processes involving only ordinary particles and gravitinos .
astrophysical and laboratory applications are briefly addressed .
/#10=#1 - 00=5pt 0 by-00 by00.50 to0_/ _ 0.2 cm it is well known that the solution of the hierarchy problem in supersymmetric extensions of the standard model ( sm ) requires a mass splitting @xmath0 tev between ordinary particles and their superpartners .
this requirement , however , leaves largely undetermined the supersymmetry - breaking scale @xmath1 , or , equivalenty , the gravitino mass @xmath2 , @xmath3 being the planck mass . the ratio @xmath4 is given by the coupling of the goldstino to the matter sector under consideration .
if this coupling is gravitational , of order @xmath5 , then @xmath6 is of order @xmath7 and supersymmetry breaking takes place at the intermediate scale @xmath8 gev . on the other hand ,
if the goldstino coupling to matter is of order 1 , then @xmath1 is comparable to the mass splitting @xmath6 , and the gravitino becomes superlight , with a mass of about @xmath9 ev . in the absence of a theory of supersymmetry breaking , @xmath10 should be treated as a free parameter .
if the gravitino is superlight , then one expects a substantially different phenomenology from that characterizing the minimal supersymmetric standard model ( mssm ) . in this case , only the @xmath11 gravitino helicity states can be safely omitted from the low - energy effective theory , when gravitational interactions are neglected .
the @xmath12 helicity states , essentially described by the goldstino field , should instead be accounted for at low energy , because of their non - negligible coupling to matter .
the lightest supersymmetric particle is the gravitino and peculiar experimental signatures can arise from the decay of the next - to - lightest supersymmetric particle into its ordinary partner plus a gravitino @xcite .
moreover , even when all superymmetric particles of the mssm are above the production threshold , interesting signals could come from those processes where only ordinary particles and gravitinos occur .
as soon as the typical energy of the process is larger than @xmath7 , a condition always fulfilled in the applications discussed below , one can approximate the physical amplitudes by replacing external gravitinos with goldstinos , as specified by the equivalence theorem @xcite . if the masses of the ordinary particles involved are negligible with respect to the energy of interest , these processes are controlled by just one dimensionful parameter , the supersymmetry - breaking scale @xmath1 , entering the amplitudes in the combination @xmath13 , @xmath14 denoting the goldstino wave function .
this class of processes includes @xmath15 , @xmath16 , which may influence primordial nucleosynthesis , stellar cooling and supernovae explosion @xcite .
of direct interest for lep2 and for the future linear colliders is the reaction @xmath17 .
partonic reactions such as @xmath18 , @xmath19 and @xmath20 can be indirectly probed at the tevatron collider or in future hadron facilities . in the absence of experimental signals
, one can use these processes to set absolute limits on the gravitino mass . at variance with other bounds on @xmath7 discussed in the literature @xcite ,
these limits have the advantage of not depending on detailed assumptions about the spectrum of supersymmetric particles .
finally , the study of these processes can reveal unexpected features of the low - energy theory , which were overlooked in the standard approach to goldstino low - energy interactions .
the natural tools to analyse the above processes are the so - called low - energy theorems @xcite . according to these , the low - energy amplitude for the scattering of a goldstino on a given target is controlled by the energy
momentum tensor @xmath21 of the target . to evaluate the physical amplitudes , it is more practical to make use of an effective lagrangian , containing the goldstino field and the matter fields involved in the reactions , and providing a non - linear realization of the supersymmetry algebra @xcite .
for instance , in the non - linear construction of @xcite , the goldstino field @xmath14 and the generic matter field @xmath22 are incorporated into the following superfields : [ bigl ] _ ( q + ) _
= _ f + _
+ i f ( ^ - ^ ) _ _ f + , ( q + ) = + i f ( ^ - ^ ) _ + .
[ mat ] the goldstino matter system is described by the supersymmetric lagrangian : d^2d^2| ^2 |^2 , [ inv1 ] where @xmath23 is the ordinary lagrangian for the matter system .
this non - linear realization automatically reproduces the results of the low - energy theorems , in particular the expected goldstino coupling to the energy
momentum tensor @xmath21 associated to @xmath22 .
an alternative approach consists in constructing a low - energy lagrangian , starting from a general supersymmetric theory defined , up to terms with more than two derivatives , in terms of a khler potential , a superpotential and a set of gauge kinetic functions .
the effective theory can be obtained by integrating out , in the low - energy limit , the heavy superpartners @xcite . when applied to the process @xmath15
, the two procedures yield the same result .
the only independent , non - vanishing , helicity amplitude for the process is : a(1,-1,1/2,-1/2)=8 ^ 22 e^4f^2 , [ aph ] where ( 1 , @xmath241 ) and ( 1/2 , @xmath241/2 ) are the helicities of the incoming and outgoing particles , respectively ; @xmath25 and @xmath26 are the goldstino energy and scattering angle in the centre - of - mass frame .
the total cross section is @xmath27 . in earlier cosmological and astrophysical applications , with a typical energy range from about 1 kev to 100 mev , a cross - section scaling as @xmath28 was assumed , giving rise to a lower bound on @xmath7 close to @xmath29 ev .
when the correct energy dependence is taken into account , this bound is reduced by at least a factor 10 , and becomes uninteresting compared to those obtainable at colliders .
when considering @xmath16 , in the limit of massless electron , one has to face an unexpected result @xcite .
on the one hand , by integrating out the heavy selectron fields , one finds the following helicity amplitude : a(1/2,-1/2,1/2,-1/2)=4 ( 1+)^2 e^4f^2 , [ ae1 ] all other non - vanishing amplitudes being related to this one . on the other hand , by using the non - linear realization of @xcite , one obtains : a(1/2,-1/2,1/2,-1/2)=4 ^ 2e^4f^2 .
[ ae2 ] the amplitudes of eqs .
( [ ae1 ] ) and ( [ ae2 ] ) scale in the same way with the energy , but have a different angular dependence .
we should conclude that the low - energy theorems of ref .
@xcite do not apply to the case of a massless fermion .
a particularly disturbing aspect is that the non - linear realization of eq .
( [ inv1 ] ) is supposed to provide the most general parametrization of the amplitude in question , independently of any considerations about the low - energy theorems . in the case at hand ,
the lagrangian of eq .
( [ inv1 ] ) reads : _
e= d^2d^2| ^2 |^2 , [ inv1bis ] where @xmath25 and @xmath30 are the superfields associated to the two weyl spinors @xmath31 and @xmath32 describing the electron , according to eq .
( [ mat ] ) .
the solution to this puzzle @xcite is provided by the existence of an independent supersymmetric invariant that has been neglected up to now in the literature : _ e = d^2d^2| ( e ||e+e^c ||e^c ) .
[ inv2 ] the amplitudes of eq .
( [ ae1 ] ) are reproduced by the combination @xmath33 . on the other hand
, there is no reason to prefer either the result of eq .
( [ ae1 ] ) or that of eq .
( [ ae2 ] ) .
the process @xmath34 does not have a universal low - energy behaviour . in the framework of non - linear realizations ,
this freedom can be described by the invariant lagrangian @xmath35 , where @xmath36 is a free parameter of the low - energy theory .
the process @xmath37 suffers from a similar ambiguity @xcite .
indeed the soft and collinear part of the cross - section , which is the dominant one , is associated to the initial - state radiation and hence is determined by the cross - section for @xmath34 .
the total cross - section , with appropriate cuts on the photon energy and scattering angle , scales as @xmath38 .
the photon energy and angular distributions are not universal , as for the case of the goldstino angular distribution in @xmath34 .
they could be completely determined only through a computation performed in the fundamental theory . from the non - observation of single - photon events above the sm background at lep2
, one can roughly estimate a lower bound on @xmath1 of the order of the machine energy @xcite .
the precise value of this limit would require the analysis of the relevant background , as well as the inclusion of the above mentioned theoretical ambiguity .
however , in view of the quite strong power dependence of the cross - section on the supersymmetry - breaking scale , one expects only a small correction to the limit obtained by a rough dimensional estimate .
the partonic processes @xmath18 , @xmath19 are also expected to have cross - sections scaling as @xmath38 and @xmath39 respectively .
the agreement between data and sm expectations in @xmath40 and @xmath41 at the tevatron collider could then be used to infer a lower limit on @xmath1 .
this is expected to be around the typical total energy of the partonic subprocess , about @xmath42 gev . |
in the standard scenario of structure formation galaxies reside in massive dark matter ( dm ) haloes , where baryons can shock , cool and eventually form stars .
haloes form by gravitational instability , starting from some density fluctuation field @xmath3 .
specifically , a dark matter halo with mass @xmath4 , forms when the linear density field , smoothed on scale @xmath5 , exceeds some critical threshold @xmath6 @xcite . in a cold dark matter ( cdm )
framework , the galaxy formation process is hierarchical along cosmic time : small systems collapse earlier , in a denser universe , and later merge to form larger haloes . in particular
, haloes often grow as a consequence of repeated merging events with smaller satellites .
present - day surviving satellites form the so - called subhalo ( or substructure ) population of a given host system @xcite .
dark matter haloes and subhaloes provide the environment in which galaxies form and evolve @xcite . therefore
, understanding the assembly histories of dark matter haloes is the first step towards the comprehension of the more complex processes involved in galaxy formation . in order to study the framework of structure formation
, two different approaches have commonly been used .
the first is to run @xmath7-body simulations @xcite .
these are very powerful tools , that can be used to reproduce the collapse of dark matter haloes with high mass and force resolution , both on galaxy and galaxy - cluster scales .
they allow to follow gravitational collapse up to its fully non - linear evolution . on the other hand , simulations are computationally expensive , and nonetheless can not cover the full spectrum of masses relevant to structure formation .
the second approach is to use some analytical modeling , which allows a detailed study of the merging history of haloes over an arbitrarily large mass range , under suitable simplification of the problem .
this is the case of the press & schechter formalism @xcite . as underlined before , haloes collapse on a certain scale once the linear density contrast smoothed on that scale exceeds some threshold value .
the nonlinearities introduced by these virialized objects do not affect the collapse of overdense regions on larger scales .
this simple assumption leads to the derivation of the global mass function of dark matter haloes that is in fair agreement with that found in @xmath7-body simulations @xcite . during the last decades ,
an extension of this theory was made by several authors @xcite with the aim of computing the probability that a halo of mass @xmath8 , at redshift @xmath9 , belongs to a given halo of mass @xmath10 at redshift @xmath11 @xcite .
this quantity was named `` conditional '' or `` progenitor '' mass function , @xmath12 .
this allowed different authors to estimate quantities such as the merger @xcite and creation rate @xcite , and the formation time distribution of dark matter haloes @xcite .
the extended press - schechter formalism can be numerically implemented to produce stochastic realizations of the merging history tree of haloes of any mass ; these monte carlo merging histories can have , theoretically , arbitrary resolution in mass and time , and their result is the full population of progenitor haloes at all times , for some final halo . the only thing they do not provide is spatial information on the haloes themselves .
therefore , when the focus is on the statistical properties of haloes and not on their position or internal structure , monte carlo merger trees have the advantage - over @xmath7-body simulation - of virtually unlimited mass resolution . implementing a code to reach
the required resolution is however not straightforward , and caution must be used in order to preserve consistently with the theory . in this paper
we will build a spherical collapse monte carlo merger tree with arbitrary mass and time - step resolution , and will consider a @xmath13-power spectrum .
we will populate a milky - way sized halo with subhaloes with masses as small as @xmath14 .
this value correspond to the typical jeans mass for the a cdm weakly interacting massive particle ( wimp ) particle with @xmath15 @xcite .
such a value for the minimum mass can actually vary between @xmath16 and @xmath17 depending on the underlying particle physics @xcite .
we will take into account the progenitor haloes accreted by the main branch of our milky - way like halo , in order to obtain a snapshot of the spatial distribution of substructures today . * in this way we will select only the first order of substructures which populate the halo today ( those which are orbiting in the host halo potential ) , in order to correctly model the radial dependence of subhalo properties . in other words , we will avoid the bias of our previous works ( see , e.g .
@xcite ) towards the small scale structures . in those works ,
the embedding of sub - subhalos within subhalos was not treated correctly .
each structure existing inside the main halo was indeed considered as orbiting inside the potential of the host halo itself . in this way ,
an analytical treatment of the effect of subhalos on the expected photon flux from dm annihilation was possible .
however , it was not possible to take into consideration the fact that a number of small scale structures are indeed sub - substructures , that is to say they are orbiting inside the potential of the subhalo they belong to , and not inside the potential of the host halo . in the present work we will be able to separate and study the effect of introducing sub - subhalos .
* with the same partition code we will thus compute the subhalo population of subhaloes ( i.e. subhaloes within subhaloes ) , and we will consider the special case of a draco - like satellite . in the second part of the paper we will estimate the @xmath1-ray emission from dark matter annihilation , in the subhalo population of the milky - way and of a draco - like subhalo , and will discuss the different contributions to @xmath1-ray emissions due to the smooth and clumpy components of the considered systems . the paper is organized as follows : in sec .
[ tree ] we describe the merger tree technique developed by @xcite and its generalization to a @xmath18 power spectrum .
[ partition ] is dedicated to the study of the merger tree of a milky - way sized halo , and to compute its mass accretion history along cosmic time and its satellite mass function . in sec .
[ secdraco ] we model the hierarchical growth of a sample of draco - like satellites until their merging time with the main milky - way progenitor halo . in sec .
[ subgamma ] we study the @xmath1-ray emission from dark matter annihilation in subhaloes and sub - subhaloes and estimate the possibility to be detected with the fermi - lat telescope . in sec.[deconc ] we discuss our results and conclude .
the simplest algorithm for a merger tree uses a binary split . in this scenario each halo is split in two haloes at an earlier epoch .
each of these is in turn divided in other two pieces , and so on until all halo masses fall below an arbitrary chosen and desired mass resolution @xmath19 @xcite .
however , @xcite showed the failure of using a binary merger tree . in this way the first halo which is chosen from the press & schetchter distribution
follows the correct probability , but the second one is chosen only in order to conserve mass and does not follow the theoretical model as expected .
this leads to a conditional mass function and formation redshift distribution in disagreement with the extended press & schechter predictions . to solve this problem , @xcite proposed a new algorithm able to reproduce the conditional mass function quite well , at the expenses of using a finely tuned grid of time - steps .
on the other hand , @xcite used the results of @xcite and realized that , for white - noise initial conditions , mutually disconnected regions are mutually independent . in this case
it is possible to split a halo into progenitors very efficiently .
the so obtained conditional mass function is in excellent agreement with the theoretical model of the spherical collapse .
the great advantage of this method is that it is possible to obtain arbitrary high mass resolution for any given time - step , generating progenitors in a very fast and efficient way . in this section
we will describe the way the method developed by @xcite can be generalized considering a @xmath18 power spectrum using this algorithm , we will follow the creation and the assembly history of a present - day milky - way sized halo . assuming that the dark matter is a wimp , the jeans mass of the smaller dm halo is given by its free - streaming scale @xcite , and for a dm particle mass of 100 gev this smallest halo mass has a typical mass @xmath20 .
we therefore extend our merger tree down to microsolar mass resolution , in order to study the present - day subhalo population down to the dark matter jeans mass .
let us start considering an initial poissonian distribution of identical particles .
@xcite and @xcite showed that the probability to find a clump containing @xmath7 particles is expressed by the borel distribution @xcite : @xmath21 where the variables @xmath7 and @xmath22 are such that @xmath23 and @xmath24 .
@xcite also showed that this equation can be extended to the continuous case , in order to describe dark matter halo clustering . in this case
the variable @xmath22 has a redshift dependence given by : @xmath25 where @xmath26 is the critical overdensity threshold predicted by the spherical collapse model , decreasing as the universe expands .
the probability that a randomly chosen particle belongs to an @xmath7-clump is given by @xmath27 .
for large values of @xmath7 and small values of @xmath6 , the factorial stirling s approximation gives : @xmath28 since all matter is in clumps , it holds the relation @xmath29 .
@xcite showed also that eq .
( [ s95e ] ) can be obtained starting from an initial gaussian density field with white noise initial conditions ( @xmath30 with @xmath31 ) .
this equivalence underlines that the poissonian distribution can be thought as the analogue discrete of the white noise gaussian power spectrum @xcite . the critical collapse overdensity decreases with the expansion of the universe , so that small systems collapse earlier than large ones , i.e. , dark matter halo clustering progresses hierarchically .
an important quantity that describes the hierarchical growth of the haloes is the conditional distribution .
it gives the probability that a particle , belonging to a clump with @xmath7 particles at time @xmath32 , is part of an @xmath33-clump at @xmath34 , and can be written as : @xmath35^{n - n-1}\,,\end{aligned}\ ] ] where @xmath36 and @xmath37 @xcite .
however , for a complete description of the merging history tree of an @xmath7-clump at @xmath32 , we also need to know the probability that , at @xmath38 , it is divided in a sample of @xmath39 @xmath40-clumps with @xmath41 subfamilies ( so that @xmath42 ) . recalling the conservation of the particle number , @xmath43
, the poissonian galton - watson branching process gives the following equation for this probability : @xmath44^{n-1 } \mathrm{e}^{-n(b_1-b_0)}}{\eta(n , b_0 ) } \nonumber \\ & \times & \prod_{j=1}^{k } \frac{\eta(j , b_1)^{n_j}}{n_j!}\ , , \label{gwbpeq}\end{aligned}\ ] ] where @xmath45 is the borel distribution with time parameter @xmath22 ( see @xcite for more details ) .
since @xmath32 can be related to a density , the volume of the @xmath7-clump can be written as : @xmath46 where @xmath47 denotes the average density of the universe .
( [ gwbpeq ] ) can be thought of as the probability that a region with @xmath7 particles , having density @xmath48 , contains @xmath41 subregions each at average density @xmath49 , where @xmath50 .
the merging history of a present - day halo is described using the extended - press & schechter formalism @xcite and formulated in terms of its conditional mass function along consecutive time steps .
let us now consider an @xmath33-subclump of the @xmath7-clump , at time @xmath38 ; in appendix a of @xcite it is shown that , if @xmath51 is its associated volume , the remaining particles @xmath52 will occupy a volume such that : @xmath53 where @xmath54 is the unknown quantity to be determined and represents the density in the volume @xmath55 .
the spherical collapse model predicts that a group of uncollisionless @xmath7 particles , with the same mass @xmath56 , collapses forming a dark matter @xmath57-halo ( @xmath58 ) , if the smoothed density fluctuation filed on scale @xmath59 exceeds its predicted critical virial value @xcite .
for each collapsed system we can define its associated mass variance as : @xmath60 where @xmath61 and @xmath62 represent respectively the fourier transform of the smoothing window function and of the density fluctuation field .
it is possible to show that in the continuous limit eq.s ( [ boreleq ] ) and ( [ gwbpeq ] ) give : @xmath63 taking @xmath64 , this equation describes , at a fixed redshift ( and so at a given @xmath6 ) , the number of collapsed systems with mass variance between @xmath65 and @xmath66 .
the factor @xmath67 takes into account the so - called _ cloud - in - cloud problem _ , which is the possibility that at a given instant some object , which is nonlinear on scale @xmath57 , can be later contained within another object , on a larger mass scale .
( [ gauss ] ) describes also the conditional mass function at time @xmath68 , considering a halo at time @xmath69 with mass variance @xmath70 , when @xmath71 . before taking into account a @xmath18 power spectrum ,
let us consider the simpler case of a white - noise power spectrum .
we recall that when @xmath72 , eq ( [ massv ] ) becomes @xmath73 , that for @xmath31 gives @xmath74 .
let us suppose we want to split an @xmath10-halo , at time @xmath69 , in progenitor haloes at @xmath68 . to generate the first progenitor we draw a random number @xmath75 from the gaussian distribution , eq ( [ gauss ] ) , and compute its associated mass variance ,
@xmath76 considering that @xmath77 where for this first progenitor @xmath78 ( @xmath79 ) and @xmath80 .
its physical mass , @xmath81 , can be directly computed as the inverse of its mass variance . since for a white - noise power spectrum
disconnected volumes are mutually independent , the overdensity of the remaining mass @xmath82 will be given by the volume conservation relation in the continuous limit : @xmath83 to generate the second halo we draw another number from the gaussian distribution , compute the mass variance from the eq .
( [ sleq1 ] ) , and hence derive the corresponding mass .
the remaining mass will now be @xmath84 , with the corresponding overdensity given always by eq .
( [ sleq2 ] ) . keep going with this procedure it is possible to generate a sample of progenitor haloes until the desired @xmath19 mass resolution . in the case of a general power spectrum
the algorithm should be modified .
the assumption that disconnected volumes are mutually independent does not hold anymore when the initial conditions differ from white - noise . despite this
, @xcite noticed that , when expressed as a function of the variance rather than of the mass , all excursion set quantities are independent of the power spectrum . in this framework , each chosen mass @xmath86 , can be treated not as a progenitor having a mass @xmath86 , but as a region of volume @xmath87 containing a mass @xmath86 , populated by @xmath88 objects having all the same mass @xmath89 , with @xmath90 .
the number of objects is obtained by requiring that they have the same mass variance , that is @xmath91 . for a scale - free power spectrum @xmath92 ( @xmath93 ) @xmath94 , and for @xmath31 we have @xmath95
, the region @xmath87 contains exactly one halo , as seen in the previous section . for @xmath96 and general power spectrum ,
@xmath88 is neither unity nor even integer .
however , we will show in the next section that considering @xmath97 ( i.e. considering the nearest integer to the mass ratio ) the progenitor mass functions are in excellent agreement with the theoretical prediction at all redshifts and down to @xmath98 .
we recall that the @xmath13 and the white - noise power spectrum should satisfy the relation : @xmath99 where @xmath10 is the initial mass to be split .
this guarantees that for the considered initial mass there is one halo in the volume @xmath100 for both power spectra .
let us now take into account the case of a @xmath0cdm power spectrum .
the density parameter and mass variance ( @xmath101 and @xmath102 ) have been chosen to agree with the recent 3-year wmap data release @xcite .
we have linearly extrapolated the mass variance down to the @xmath103 integrating the power spectrum using a top - hat filter in the real space @xcite . in order to have one physical mass both for the @xmath13 and white - noise power spectrum
, we should consider a white - noise power spectrum normalized such that eq .
( [ wnlcdmpower ] ) holds . for this reason , because we think in term of the mass variance than the fisical mass , the mass resolution @xmath19 , for a @xmath0cdm power spectrum , corresponds to the mass resolution @xmath104 for the white - noise one , such that @xmath105 , with @xmath106 .
let us consider a present - day milky - way sized halo ( @xmath107 ) and suppose we wish to generate a sample of progenitors at different redshifts down to @xmath108 , using a @xmath18 power spectrum .
such mass resolution corresponds to @xmath109 .
we proceed as described in the previous section and obtain the conditional mass function plotted in fig .
[ cmffig01 ] .
the number of progenitor haloes in the @xmath18 power spectrum , for a given value of the mass variance , was obtained by computing the nearest integer of the ratio @xmath110 , where @xmath111 . in each panel
the dotted histogram shows the result of averaging @xmath112 realizations , and the solid line shows the theoretical prediction from the spherical collapse model . at each redshift we generated progenitors both with a single and a multi - steps technique starting from @xmath113 .
since the sl99 method is independent on the time - step , the results from the two different ways of progress are in perfect agreement . in order to generate the merging history tree along the halo main branch , we ran a sample of @xmath112 white - noise tree realizations for an @xmath114 initial halo and a mass resolution of @xmath115 .
as stressed by sl99 ( and demonstrated in fig.[cmffig01 ] ) , this method is time - step independent : to obtain a fine enough description of the merging of satellite haloes we chose a time resolution @xmath116 .
we followed the main branch ( the most massive halo among the progenitors ) along consecutive time - steps , storing all the information about the accreted progenitor haloes ( termed `` satellites '' , as in @xcite ) , until the mass of the main branch becomes as small as the mass resolution .
we call merging redshift @xmath2 the most recent redshift before a satellite is incorporated in the main halo .
finally , the poissonian trees is converted in the @xmath117 one as explained in the previous section .
we recall that each progenitor halo in the first tree is converted into @xmath88 haloes in the second , in order to conserve the mass variance in both power spectra .
realizations of the merger history tree .
the mass of each satellite is rescaled in units of the final host halo mass .
the solid line shows the slope of the least squares fit to the histogram down to @xmath118 .
[ maccrmontc ] ] @xcite studied the substructure population of a present - day milky - way halo considering its progenitor mass functions at any redshift @xmath119 , and assuming that all progenitor haloes survived and retained their original virial mass until redshift zero .
this analytical distribution for the progenitor mass function was extrapolated down to microsolar mass resolution in order to study the @xmath1-ray emission from all substructures that populate the galactic halo .
integrating the conditional mass function at any redshift in different mass bins , the full hierarchy of substructures was automatically taken into account : subhaloes within subhaloes within subhaloes , and so on .
the @xmath1-ray emission process due to dark matter annihilation is proportional to the square of the dm density , and the possibility that subhaloes might boost the expected flux was taken into consideration ( see @xcite and references therein ) . in the present work we compute the @xmath1-ray emission by considering as substructure only the progenitor haloes accreted by the main branch of our merger tree .
* this procedure does not take directly into account the whole substructure population but only the first hierarchy of the population . progressing in * this way
, we can isolate the contribution to the boost from the single dm structure , including its own sub - substructures * subsequently , following its merging history tree*. moreover , in this way we can model correctly the spatial satellite distribution inside the galaxy , since many small mass halos will be in fact embedded inside larger ones . in fig .
[ maccrmontc ] we plot the satellite mass function , i.e. the mass function of progenitor haloes accreted directly by the main halo progenitor at any redshift .
satellite masses are expressed in units of the present - day mass of the host halo @xcite .
this figure shows that the mass distribution is well described by a single power law @xmath120 , with slope @xmath121 , plus an exponential cut off at large masses .
the slope was obtained computing the least squares fit to the data down to @xmath122 .
this slope is slightly steeper than that obtained by @xcite using numerical @xmath7-body simulations ( @xmath123 ) .
such a discrepancy is probably due to the fact that the present merger tree is based on the spherical collapse model , while numerical simulations are better described using ellipsoidal collapse @xcite . the satellite mass function in fig . [ maccrmontc ]
was obtained using the monte carlo code described in sec . [ tree ] .
we considered a halo of mass @xmath10 at redshift @xmath113 and followed the main branch of its merging history tree back in time ; we then : ( i ) generated a sample of progenitors at redshift @xmath124 ; ( ii ) identified the main ( most massive ) progenitor halo ; ( iii ) re - run the partition code computing its progenitors at redshift @xmath125 .
we iterated these steps until the mass of the main progenitor halo dropped below the mass resolution @xmath19 .
this produced the first merging history tree .
we repeated the whole procedure , starting from the same mass @xmath126 , to generate @xmath112 histories , to be averaged on .
after computing the satellite mass function ( also called _ unevolved _ subhalo mass function ) we associated to each halo a concentration parameter related to the quantity @xmath127 , where @xmath2 is the satellite merging redshift onto the host halo , and @xmath8 its virial mass at @xmath2 . .
the dotted histogram show @xmath112 realizations of the merger tree , while the solid curve is the eq .
( [ gauss]).[satnufun ] ] by definition , @xmath128 defines the rareness of the density peak the halo belonged to at the epoch of merging .
higher peaks of density fluctuations correspond to a collapse happened at higher redshifts
. such a rare halo will therefore be more concentrated than the bulk of halos with the same mass which formed at later epochs .
@xcite used @xmath7-body simulations to show that the present - day subhalo distribution preserves memory of their initial conditions . in detail ,
high density peaks are found to be more centrally concentrated and to move on more eccentric orbits than the overall mass distribution .
this correlation has been interpreted and parametrized by @xcite using the variable @xmath129 .
this variable ( i ) is related to the subhalo concentration - which in turn determines the @xmath1-ray emission , and ( ii ) enables one to compute the spatial distribution in the host halo ( see eq .
( 1 ) of @xcite ) .
the details of the models are given in sec .
[ rgammamerging ] in fig.[satnufun ] we plot the satellite mass function in terms of the universal variable @xmath128 .
the dotted histogram shows the result of @xmath112 monte carlo realizations of the partition algorithm ( as in fig.[maccrmontc ] ) , while the solid curve is the gaussian distribution , eq .
( [ gauss ] ) .
the figure shows that the satellite mass function is well described by a gaussian distribution for small and intermediate masses , with a cutoff at @xmath128=2 .
this fact can be qualitatively understood by recalling that the progenitor mass function has a gaussian distribution when expressed in the rescaled variable @xmath128 .
for each evolutionary step @xmath130 , the satellite mass function is obtained by removing from all progenitors the most massive one .
integrating over redshift , the total satellite mass function is effectively an integral over different gaussians deprived of their most massive progenitor .
the algorithm described in the previous sections can also be used to study and constrain some properties of the dark matter halos hosting the milk - way satellite galaxies . as an example
let us consider draco , a dwarf spheroidal galaxy with an old stellar population , dark matter dominated : draco s mass to light ratio is estimated to be @xmath131 @xcite .
this makes it a very interesting target to investigate the existence of dark matter ; specifically , we study the possibility of an observation of @xmath1-ray due to dm annihilation inside draco using the fermi large area telescope ( fermi - lat , formerly known as glast ) . to this extent
, we model the distribution of dark matter subclumps inside draco .
hierarchical models of galaxy formation predict that galaxies are hosted by dark matter haloes .
such haloes grow along cosmic time through repeated merging events . in this
scenario satellite galaxies reside in subhaloes accreted by the main halo progenitor .
satellite haloes grow themselves hierarchically , accreting mass and sub - progenitors up to the time when they merge onto the host halo .
after they fall in the gravitational potential well of the host , they start to loose mass due to gravitational heating , tidal stripping and close encounters with other subhaloes , so that only a fraction of their initial virial mass is still self - bound at redshift zero @xcite . in order to build the history tree for a subhalo hosting a satellite galaxy , we need to know its virial mass at merging time @xmath2 and use it as our starting point . from its merging redshift
we then can go backwards in time and reconstruct its subhalo population , in order to model its present - day sub - subhalo mass function and distribution .
_ unevolved _ and _ evolved _ subhalo mass function of draco candidates accreted by a milky - way halo .
the solid histogram shows the satellite distribution , the dotted one refers to the present day mass function evolved by using the average mass loss rate derived by @xcite .
the hatched region bounds the most likely values for the present - day mass of draco @xcite . ] using deep wide - field multicolor ccd photometry from the sloan digital sky survey , assuming a @xcite spherical model of equivalent size as a reference and adopting a line - of - sight velocity dispersion of @xmath132 km / s @xcite finds @xmath133 within 28 arcmin while @xcite , considering @xmath134 km/2 , finds @xmath135 within 40 arcmin .
differences in the experimental mass modeling determine the mass range we allow for the draco - like satellite in our analysis .
considering all the stars within the tidal radius , @xcite determined the total luminosity of the draco dwarf galaxy to be @xmath136 .
@xcite also studied the distribution of dark matter in draco by modeling the moments of the line - of - sight velocity distribution of its stars from the velocity dispersion data of @xcite , and obtained a best - fitting total mass equal to @xmath137 . the inferred mass - to - light ratio ( in the @xmath138-band ) was @xmath139 , almost constant with radius .
on the other hand , we lack a direct estimate for the merging redshift of draco onto the milky way . from its initial position and velocity
, the accretion time of draco can be derived indirectly by considering the one - to - one relation between virial radius and accretion time implied by the spherical secondary infall model @xcite .
@xcite showed that an upper limit for draco merging redshift is @xmath140 .
we have computed the draco virial mass from the previous mass and redshift estimates in the following way . from each one of
the @xmath112 monte carlo merging trees of a milky - way sized halo ( @xmath10 = @xmath141 ) we noted down all satellites with mass larger than @xmath142 accreted by the main branch between @xmath143 and @xmath144 , totaling @xmath145 draco candidates .
the redshift interval roughly corresponds to a time interval of @xmath146gyr from the upper limit of the accretion time computed by @xcite .
the lower limit on the draco mass guarantees that evolved masses will lie today within the experimentally estimated mass range . in fig .
[ dracop ] the solid histogram shows the mass function of these selected satellite haloes . once inside its host , each satellite will loose mass due to gravitational heating and tidal stripping effects .
we modeled these effects using the results in eq .
( 10 ) of @xcite . in that paper
the authors measured the subhalo mass loss rate in a sample of high resolution @xmath7-body haloes ranging from @xmath147 to @xmath148 .
following their recipe , the ratio of @xmath113 self bound mass to the original virial mass of the subhalo is uniquely determined by the amount of time the subhalo has spent inside its host . the dotted histogram in fig .
[ dracop ] shows the evolved subhalo draco - candidates mass function . considering that the measured fractional mass loss rate by @xcite is independent of the subhalo mass
, the evolved mass function has the same slope of the unevolved one . in order to compute the draco satellite mass function ,
each candidate of our sample has been further evolved back in time along its main branch using the partition code ( as it was done for the milky - way halo ) considering the same mass and time step resolution . in this case
each merger history tree starts at the corresponding satellite merging time . to increase the statistical significance of our result we run three realizations for each of the @xmath145 draco - like satellites , totaling @xmath149 monte carlo merger tree realizations . in fig .
[ dracoamf ] we show the _ unevolved _ sub - subhalo mass function accreted by the draco sample .
comparing fig .
[ dracoamf ] and fig .
[ maccrmontc ] , we see that the two accreted mass functions are indistinguishable : in fact , as found by @xcite , the accreted mass function is universal , independent both of final mass and observation redshift .
the slope of the power law is again @xmath150 . .
comparing the histogram with fig .
[ maccrmontc ] we notice that the slopes of the distributions are identical .
this is in agreement with what found by @xcite that is the _ unevolved _ subhalo mass function is independent both of final host halo mass and redshift.[dracoamf ] ]
the photon flux expected from dm annihilation in the population of galactic subhaloes can be modeled as @xmath151 where @xmath152 defines the direction of observation with respect to the galaxy center , and @xmath153 is the detector angular resolution . the particle physics dependence in eq . [ flussodef ]
is given by the annihilation spectrum and dm properties and is embedded in the term : @xmath154 where @xmath155 is the dm particle mass , @xmath156 the self annihilation cross section times the relative velocity of the two annihilating particles , and @xmath157 the differential photon spectrum for a given final state @xmath158 with branching ratio @xmath159 which we model after the results of @xcite . in this paper
we will set @xmath160 and refer to @xmath161 as a scaled photon flux . in the following subsections we review different ways to compute @xmath162 in the presence of substructures .
we will then show the computation relative to the merger tree technique presented in the previous sections and will show the comparison between the different predictions .
@xcite have considered the existence of a population of substructures inside a dm halo , described by the mass function @xmath163 and assumed that eq .
[ mass ] describes the mass function of all substructures in our galaxy as well as that of sub - substructures , at z=0 . in this case
all the halos that ever accreted onto the main one and survived are considered , including the higher level ones ( substructures within substructures ) . using this prescription
it is then impossible to discriminate a sub - halo from a sub - sub - halo , affecting in a wrong way the spatial distribution of substructures inside the main halo .
@xcite assumed that the substructure spatial distribution traces that of the underlying host mass from @xmath164 and down to a minimum radius , @xmath165 , within which disruption by galaxy tides and stellar encounters becomes relevant .
folding these indications together they modeled the number density of subhaloes per unit mass at a distance @xmath5 from the galactic center ( gc ) as : @xmath166 where @xmath167 represents the scale radius of the parent halo ( ph ) referred to a navarro , frenk and white ( nfw ) @xcite profile .
the effect of tidal disruption is accounted for by the heaviside step function @xmath168 . to determine the tidal radius , @xmath165
, they used the roche criterion and compute it as the minimum distance at which the subhalo self - gravity at @xmath169 equals the gravity pull of the host halo computed at the orbital radius of the subhalo . to normalize eq .
[ rho ] , they imposed that @xmath170 of the ph mass is distributed in subhaloes with masses in the range @xmath171 \ m_{ph}$ ] .
following the previous prescriptions , they found about @xmath172 of the mw mass ( @xmath173 , @xmath174 , @xmath175 = 9.8 ) condensed within @xmath176 subhaloes with masses in the range [ @xmath177 \msun / h$ ] .
@xcite repeated the calculation for the draco galaxy , for which @xmath178 40% of the halo mass ( @xmath179 , @xmath180 , @xmath175 = 21.2 ) condensed into @xmath181 halos with masses between [ @xmath182 \msun / h$ ] . for each substructure
, they used a nfw density profile whose concentration parameter @xmath175 ( referred to the radius enclosing a density equal to 200 times the critical density ) is given by the @xcite prescriptions for the subhaloes at redshift zero ( assuming the subhaloes are slightly more concentrated than the field halos ) .
the @xcite model holds down to masses of the order of @xmath183 . in order to cover the whole subhalo mass range
, they performed a double mass slope extrapolation down to the smallest masses , obtaining a concentration parameter of about 70 for a @xmath184 halo at redshift zero .
this value is in agreement with the extrapolation at redshift zero of what found in the numerical simulations by @xcite , who isolated @xmath185 halos at z=26 .
such a model for the concentration parameter has been called @xmath186 model .
the contribution of unresolved substructures to the annihilation signal along a cone of sight is given by @xmath187 @xmath188 @xmath189 \label{smoothphicosmo}\ ] ] where @xmath190 is the solid angle of observation pointing in the direction of observation @xmath152 and defined by the detector angular resolution @xmath191 ; @xmath192 the jacobian determinant ; @xmath5 the galactocentric distance , which , inside the cone , can be written as a function of the line of sight ( @xmath193 ) , the solid angle ( @xmath191 and @xmath194 ) and the pointing angle @xmath152 through the relation @xmath195 , where @xmath196 refers to the distance of the sun from the galactic center and @xmath197 ; @xmath198 is the radial coordinate inside the single subhalo located at distance @xmath193 from the observer along the line of sight defined by @xmath152 and contributing to the diffuse emission . finally , @xmath199 is the lognormal distribution of the values for the concentration parameters .
the expression @xmath200 @xmath201 \ , ; \label{singlehalophicosmo}\ ] ] describes the emission from each subhalo and @xmath202 is the dark matter density profile inside the halo .
numerical integration of eq .
[ smoothphicosmo ] gives the contribution to @xmath203 from unresolved clumps in a @xmath204 solid angle along the direction @xmath152 .
we show the results obtained in this theoretical framework with a dashed line in fig .
[ mw ] @xcite for the mw and in fig.[figdraco ] @xcite for the draco galaxy
. a further degree of detail was given by @xcite , who computed eq .
[ smoothphicosmo ] including the dependence of the subhalo spatial distribution from the initial conditions when the haloes accreted into the present - day milky way halo .
such a model is characterized by the universal variable @xmath205 . according to @xcite the dm density profile of our galaxy
can be written as @xmath206^{(\beta - \gamma)/\alpha } } \label{density}\ ] ] with @xmath207 .
subsequently , the following parameterization is used to reflect the fact that material accreted in areas with high density fluctuations is more concentrated toward the centre of the galaxy , and has a steeper outer slope : @xmath208 @xmath209 @xmath210 including a step function to take into account tidal disruption , as in eq .
[ rho ] , the number density of subhaloes per unit mass at a distance @xmath198 from the gc , for a given @xmath211 , becomes : @xmath212^{(\beta_\nu - \gamma)/\alpha } } , \label{rho}\ ] ] in units of @xmath213 .
the mass dependence in @xmath214 reflects the mass dependence of the virial parameter @xmath215 .
@xcite normalized the number of subhaloes such that 10% of the mw mass is distributed in subhaloes with masses in the range @xmath171 \ m_{mw}$ ] , ending up with @xmath216 subhaloes with masses between @xmath217 and @xmath218 , accounting for 74 % of the mw mass ( @xmath219 , @xmath220 ) .
the contribution to @xmath162 is given by : @xmath221 @xmath222 @xmath223 \label{smoothphicosmogpt}\ ] ] which accounts for the influence of cosmology in the flux computation .
@xmath224 is the probability distribution function for the peak rarity @xmath225 , calculated using the extended press - schechter formalism .
@xmath226 is the lognormal probability distribution for @xmath227 centered on @xmath228 .
while @xmath224 is determined by the merging history of each subhalo , @xmath226 describes the scatter in concentration for haloes of equal mass @xcite and the two probabilities may be assumed to be independent . as in the previous analytical estimate of the subhalo population
, no distinction can be made with this method between sub - halos and sub - sub - halos .
the result of this calculation for the mw is shown with the long - dashed line in fig .
[ mw ] . in this work
we have described a merger tree approach to infer the subhalo population of both the mw and draco . in this case
, we do nt need to use the @xmath229 and @xmath230 factors , nor to apply any normalization .
indeed , the output of the merger tree gives us directly the number of objects with a given mass and a given @xmath128 , which we call now @xmath231 .
the total number of sub(-sub)structures found with this method is @xmath232 for the mw and @xmath233 in the case of draco , at the merging epoch . in the case of the mw , and differently from the analytical methods described above , this number represents only the subhaloes , while sub - subhaloes must be treated separately , as it has been done for the draco - like subhalo . in order to compute the @xmath1-ray flux we can re -
write the subhalo distribution function as @xmath234 so that @xmath235 which can be rearranged into the following expression : @xmath236 let s define @xmath237 the expression of @xmath238 can be then written as : @xmath239 @xmath240 \,.\ ] ] with respect to eq .
[ smoothphicosmogpt ] we have only written @xmath241 in an explicit way and multiplied by @xmath242 = 1 . in this way we can recognize and use our merger tree output @xmath243 in the following way ( @xmath244 ) : @xmath245 @xmath246 \label{smoothphicosmomt}\ ] ] the result of this calculation is depicted with a dotted line in fig.[mw ] .
we note that the analytical methods predict more substructures , and therefore a higher signal , than the merger tree .
this is mainly due to the different mass function slope ( -2.0 versus -1.9 ) . in the case of the draco galaxy ,
the merger tree was computed at the draco merging epoch ( no sub - subhaloes accrete after merging ) , when its mass was @xmath247 .
although the sub - subhaloes evolve in redshift and loose mass as the parent halo does , for each sub - subhalo we considered a nfw profile whose concentration parameter has been computed for the sub - subhalo mass at the merging epoch , motivated by the fact that the inner profile of the structures should not or poorly be affected by evolution . from the observational point of view , we are interested in the draco - like galaxy as it is today , e.g. with a mass content reduced by a factor @xmath248 due to tidal interactions after merging @xcite . to estimate the number of sub - subhaloes in draco today we have retained all and only the subhaloes which , at the epoch of merging , were inside the radius containing the draco mass today , that is within @xmath249 kpc .
@xmath250 is also similar to the tidal radius obtained using the prescription given by @xcite to obtain the number of sub - subhaloes today .
the total number of subhaloes is @xmath251 smaller than the initial one .
this is the upper value for the number of subhaloes , since we are not considering here that half of the subhaloes exit the virial radius of the parent halo during their first orbit @xcite . in the case of field parent halos
they would then be re - attracted inside the halo , but if the parent halo is a subhalo itself ( like draco ) they would then be dispersed in the main halo ( that is the mw ) .
as for @xmath241 , we have used the scale quantities @xmath169 and @xmath252 computed for the evolved draco - like galaxy , that is for a @xmath253 halo .
the result of the computation of @xmath238 for draco is shown with a dotted line in fig.[figdraco ] .
for comparison we also show the result from the analytical calculation and the same quantity relative to the smooth nfw halo of draco . as in the case of the mw
, we note that the merger tree approach gives a comparable yet smaller result than the analytical one . as a function of the angle of view @xmath152 from the galactic centre , computed for the mw . ] as a function of the angle of view @xmath152 from the galactic centre , computed for the draco galaxy . ]
@xcite showed how the detectability of a @xmath1-ray flux from such a population of sub - subhaloes with the fermi - lat telescope serendipitously depends on a very boosted particle physics contribution .
the present and more accurate results give an even lower expected flux , thus furtherly reducing the hope for detection , unless an exotic boost from particle physics enters the play . in fact , @xcite have computed the map of the fermi - lat sensitivity to point sources of dm annihilations , by using the released fermi - lat response functions .
draco lies in a region of the sky where the @xmath254 detection flux above 100 mev in 1 year of data taking is @xmath255 . in order to obtain the afore - mentioned flux we have to multiply the particle physics contribution by the highest value of @xmath161 we have obtained , that is to say along the line of sight pointing towards the centre of draco .
our result is that we would need a boost factor of 600 ( 120 ) when using the already optimistic scenarios where @xmath256 ( 40 ) gev and @xmath257 . in the case of draco and for the @xmath186 model
, we can conclude that no higher order computations are necessary ( sub - sub - subhaloes and so on ) .
in fact , we have computed the boost factor due to substructures for any mass of the parent halo from @xmath258 to @xmath259 , that is to say the ratio of the integral of the density squared over the whole galaxy including subhaloes to the same quantity computed in the case of a smooth halo , which gives the boost to the @xmath1-ray flux due to substructures for point - like halos .
we found that its value is pretty flat and close to 1 , only slightly larger for larger mass haloes , which however does not appear point - like at the distance of draco .
this means that no boost is obtained when considering sub - substructures for point - like subhaloes .
as far as massive haloes ( with mass greater than @xmath260 ) are concerned , which are not point - like at the distance of draco , we have to consider the fact that , as in the case of draco , the boost must be distributed spatially .
indeed , this spatial distribution of the boost factor leads to an even smaller value of the expected flux along the line of sight toward the centre of the halo , where the flux is higher .
this means that including sub - substructures does not increase the total @xmath1-ray flux from the subhalo .
in this paper we used the sl99 merger tree technique to study the mass accretion history in satellite of a milky - way sized halo .
the partition code has been generalised for a @xmath18 power spectrum , considering a microsolar mass resolution of @xmath261 .
the mw - main progenitor halo has been followed along consecutive time steps from the present day down to when its mass dropped below the resolution limit . from the satellite haloes
we have identified a sample of draco - like systems and also followed them along their merger tree , starting from their accretion redshift into the mw - main halo progenitor . both the milky - way and the draco satellite mass function turned to be equivalent , testing the universality of this distribution .
the partition technique allowed us to build up models for the present - day subhalo mass function for both mw and draco considering only haloes that they accreted along the main branch .
we then computed the expected @xmath1-ray flux from dm annihilation in such a population of substructures .
we found that the prediction for the @xmath1-ray flux is indeed more pessimistic than the ones obtained in previous estimates * which were not correctly taking into account the embedding of small scale sub - subhaloes within subhaloes , and therefore adding their contribution to the host halo total flux instead of to the subhalo they belong to .
* detection with the fermi - lat telescope is probably out of the discovery range of the satellite , unless some exotic particle physics could boost the signal significantly .
while we were writing this paper , the results of the aquarius simulation came out @xcite , predicting a smaller number of subhaloes and a shallower internal slope for the subhaloes density profiles .
we will not repeat our analysis in their model since it would fourthly reduce the expected flux , but the reader should be aware that the results presented in this paper , though pessimistic , are upper limits for the expected signal from dm annihilation in galactic subhaloes .
cg thanks the partial support for this work given by the italian prin and asi . thanks also to r. k. sheth for very useful discussions during the time spent in philadelphia at the end of @xmath262 .
baltz , e. a. , berenji , b. , bertone , g. , bergstr@xmath263 m , l. , bloom , e. ; bringmann , t. , chiang , j. , cohen - tanugi , j. , conrad , j. , edmonds , y. , and 18 coauthors , 2008 , journal of cosmology and astro - particle physics , 7 , 13 borel , e. , 1942 , sur lemploi du theoreme de bernoulli pour faciliter le calcul dune infinite de coefficients .
application au probleme de lattente a un guichet .
comptes rendus , academie des sciences , paris , series a , 214 , 452 - 456 | the hierarchical growth of dark matter haloes , in which galaxies are hosted , has been studied and modeled using various approaches . in this paper
we use a modified version the sheth & lemson algorithm for a @xmath0 cold dark matter power spectrum , and model the growth of a milky - way sized halo with microsolar mass resolution , corresponding to the typical jeans mass for a dark matter weakly interacting massive particle with mass of 100 gev .
we then compute the _
unevolved _ subhalo mass function and build - up a milky - way halo placing and evolving its satellites .
this subhalo population is used to study the @xmath1-ray emission from dark matter annihilation . in this case , the subhaloes which populate the host halo have been computed considering only progenitor haloes accreted by the main branch of the tree , so as to correctly treat the embedding of sub - subhaloes inside subhaloes .
each subhalo will indeed host at the present - time sub - subhaloes accreted when it was an isolated system . in order to compute the sub - subhalo population of a milky - way dwarf galaxy , like draco , and to study its @xmath1-ray emission
, we first estimate the draco virial mass at merging redshift @xmath2 and then we run the merger tree from @xmath2 following the halo down to the dark matter jeans mass .
we then study the effect on the fermi - lat ( glast ) detectability for both subhaloes in the milky - way and in draco , and we show how subhaloes can not be responsible for the boost factor needed for detection .
[ firstpage ] galaxies : halo - cosmology : theory - dark matter - methods : analytical , numerical |
supercell . ( a ) h - bn honeycomb structure on which the adsorption energies are calculated .
nitrogen and boron atoms are represented by blue and green balls , respectively .
the most favorable binding site of carbon ad - atom is marked by the red star in the figure .
the path of diffusion of carbon ad - atom with the minimum energy barrier of @xmath1 is indicated by stars .
( b ) complete energy landscape of c ad - atom on h - bn structure .
light blue regions show favorable sites and the energy barrier further increases as the color goes to dark blue and purple .
( c ) energy variation of carbon ad - atom is shown along the path indicated by red arrows in ( a ) .
the energy difference between the most favorable site ( indicated by red star ) and the bridge(br ) , top of boron(tb ) , hollow(h ) , top of nitrogen(tn ) sites are calculated as @xmath2 , @xmath3 , @xmath4 and @xmath5 , respectively.,width=302 ] after the synthesis of graphene,@xcite a monolayer of @xmath6-bonded honeycomb structure of carbon atoms , interest has focused on two dimensional nanostructures having honeycomb structures .
planar hexagonal boron nitride(h - bn ) is an ionic honeycomb structure consisting of alternatively bonded boron and nitrogen atoms and is an analog of graphene.@xcite despite the structural similarity , h - bn differs from graphene with its wide band gap and dielectric properties.@xcite various boron nitride structures like nanosheets,@xcite nanotubes@xcite and nanowires@xcite have already been synthesized .
also , recent studies show that h - bn can be used to improve the current voltage properties of graphene transistors by improving the mobility of electrons in graphene as compared to graphene films on silicon substrates.@xcite these properties hold promise for novel technological applications of h - bn structures .
a thorough understanding of the properties of h - bn structure and investigating its functionalization is important for future applications of this wide band gap material .
carbon , being in the same row of the periodic table with boron and nitrogen , is one of those foreign atoms which can greatly change the physical and chemical properties of h - bn . in a recent study,@xcite
the effects of ad - atoms adsorbed on bn were investigated using first - principles calculations , and it was shown that high coverage of carbon ad - atoms can change the magnetic properties and band gap of the system , whereas ad - atoms induce localized states in the band gap at low coverage .
.,width=321 ] single carbon ad - atom and mono - atomic chains of carbon atom are interesting entities , which can functionalize h - bn .
a short segment of carbon atomic chain containing @xmath7 c atoms , indicated as cac(@xmath7 ) hereafter , is a truly one - dimensional carbon allotrope , and has recently drawn attention due to its linear geometry , high strength , size - dependent quantum ballistic conductance and interesting electronic properties .
these properties of cacs were both theoretically@xcite and experimentally@xcite investigated .
theoretical studies revealed the peierls distortion in cacs,@xcite half - metallic and spintronic properties,@xcite size dependent quantum conductance@xcite and other geometrical structures.@xcite chain structures of other group iv elements and group iii - v compounds were also treated.@xcite concomitantly , carbon atomic chains are synthesized.@xcite recently , it was also shown that cacs can be grown on graphene and modify its properties,@xcite which have also been justified by the images obtained earlier using high resolution tem.@xcite in this paper we studied the growth of short carbon and bn chains on single layer h - bn and graphene , respectively .
we first examined the adsorption of single carbon atom on h - bn by calculating its energy landscape and diffusion barrier .
this is followed by the investigation of the nucleation and growth processes of cacs on h - bn .
we performed both conjugate gradient calculations and molecular dynamics simulations in order to determine the stabilities and bonding properties of these cacs and show how they can grow on the plane of bn as new carbon atoms are introduced by one atom at a time at the close proximity of an existing cac . once these two materials , namely carbon atomic chains and monolayer h - bn , are combined fundamentally interesting properties are attained for promising future applications .
interestingly , the properties of the grown structures depend on the number of carbon atoms in the chains , such that they exhibit an even / odd disparity . in addition
, we showed that cacs grown on h - bn constitute chemically active sites for au , li and h@xmath0 , which are normally weakly bonded to h - bn .
for example , h@xmath0 approaching to the top of the cac(2 ) is dissociated and separated into two h atoms , each attached to the chain .
we also showed that cacs can attach to another h - bn from its free end to serve as pillars to increase the spacing between h - bn flakes , where molecules like h@xmath0 can be stored .
we presented the electronic energy band structure for various lengths of chains to see their variations with the number of chain atoms .
we finally studied the problem from the reverted point of view and presented the growth patterns , optimized structural and electronic structures of bn chains grown on graphene surface .
our results revealed the unique self - assembly character of carbon and bn chains on single layer honeycomb structures and interesting features attained thereof .
in our calculations we use the state - of - the - art first - principles plane - wave calculations within the density - functional theory@xcite combined with ab - initio , finite temperature molecular - dynamics calculations using projector augmented wave potentials.@xcite the exchange correlation potential is approximated by the generalized gradient approximation with van der waals correction.@xcite numerical computations have been carried out by using vasp software.@xcite a plane - wave basis set with energy cutoff of @xmath8 is used .
the brillouin zone is sampled in the * k*-space within the monkhorst - pack scheme,@xcite and the convergence of the total energy and magnetic moments with respect to the number of * k*-points are tested . the convergence for energy is chosen as @xmath9 between two consecutive steps . in the relaxation of structures and band structure calculations , the smearing value for all structures is taken as @xmath10 .
we consider adsorption of chains on @xmath11 supercells and treat the system using periodic boundary conditions .
the pressure on each system was kept smaller than @xmath12 per unit cell in the calculations . in the ab - initio md calculations
the time step was taken as @xmath13 and atomic velocities were renormalized to the temperature set at @xmath14 and @xmath15 at every @xmath16 time steps . in the md stability tests ,
the simulations were run for @xmath17 .
before going into detailed studies of carbon chains , we first investigate the adsorption and migration of single carbon ad - atom , which is the starting point of cac growth on h - bn substrate .
we use a @xmath11 supercell of h - bn that consists of @xmath18 boron and @xmath18 nitrogen atoms .
there is an energy difference of @xmath19 between the spin polarized and spin unpolarized energy values in favor of the spin polarized state , indicating that the system has a magnetic ground state .
therefore all of the calculations mentioned hereafter are performed using the spin polarized conditions .
the most favorable binding site of single carbon atom was determined by placing the ad - atom initially to various adsorption sites at a height of @xmath20 from the bn atomic plane and running fully self - consistent geometry optimization calculations by keeping the ad - atom fixed in @xmath21 and @xmath22 directions and letting the vertical @xmath23coordinate of the ad - atom , which is its height from the plane , free . meanwhile ,
the atoms in the bn supercell are relaxed in all directions except for one corner atom of the supercell , which is fixed in all directions to prevent h - bn from sliding . in fig .
[ fig1](a ) the nitrogen and boron atoms of the optimized h - bn structure are separated from each other by @xmath24 .
the most favorable bonding site of single carbon ad - atom , which turns out to be near the top site of nitrogen atom , is marked with a red star .
the migration(diffusion ) path of carbon ad - atom on h - bn with a minimum energy barrier is shown by black stars .
the minimum energy barrier is calculated as @xmath25.@xcite the energy landscape calculated over the whole bn hexagon also shows that the energy barrier to the diffusion further increases as the carbon atom moves away from the nitrogen atom as shown in fig . [
fig1](b ) .
the variation of energy calculated along the 2d path shown in fig .
[ fig1](a ) is presented in fig .
[ fig1](c ) .
as indicated in the figure , the most favorable site for the carbon atom is near the top nitrogen site , although not exactly on top of nitrogen .
the energy barrier between the most favorable site and the top bridge(br ) , top boron(tb ) , hollow(h ) and top nitrogen(tn ) sites were calculated as @xmath26 , @xmath3 , @xmath4 and @xmath5 , respectively .
in addition to the diffusion path analysis of a single carbon ad - atom , we next study the interaction between two carbon atoms on h - bn surface .
when the distance between the two ad - atoms becomes less than a threshold distance of @xmath20 , these two carbon atoms attract each other and form a cac(2 ) perpendicularly attached to h - bn .
this is indeed the nucleation for longer cacs .
the most favorable binding site cac(2 ) , is again near the top site of nitrogen .
a complete site analysis was also performed to confirm this result , by placing a cac(2 ) on various adsorption sites and comparing the total energy values .
) , and the heights(@xmath27 ) of odd and even numbered cac(@xmath7 ) s from the atomic plane of bn are shown in green , red and blue lines , respectively .
the @xmath27 values exhibit an even / odd family behavior depending on the number of carbon atoms in the chain .
the sudden peak in the binding energy arises from the change of the magnetic state of cac(2 ) from magnetic to nonmagnetic when it binds to hexagonal bn.,width=321 ] growth of the cac further continues when a third carbon ad - atom is introduced at the close proximity of cac(2 ) .
however , this time a complete site analysis of cac(3 ) shows that the most favorable bonding site is the top of boron atom , instead of the nitrogen site .
the formation of cac(3 ) happens as follow : cac(2 ) leaves its initial bonding position , moves higher from the bn plane and in the mean time it gets closer to the single ad - atom until they are bound to each other near the new energetic site , which is the top of boron site .
similar chain growth behavior is also seen during formation of cacs at different lengths .
this process is further investigated with ab - initio molecular dynamics simulations at 500k and the snapshots taken from the growth of cac(4 ) is presented in fig .
we initiate the md simulation by placing a carbon ad - atom and a cac(3 ) to their bonding sites as shown in fig .
[ fig2](i ) .
the simulation was run for 2000 time steps and snapshots taken from the initial , 20@xmath28 , 40@xmath28 , and 120@xmath28 time steps are shown . as the simulation proceeds to the 2000@xmath28 step , the chain stays in its position shown fig .
[ fig2](iv ) , which is an indication of its stability at that bonding site .
.most favorable binding sites , total binding energies(@xmath29 ) , magnetic moments(@xmath30 ) , heights(@xmath27 ) of cac(@xmath7 ) from the bn plane , and the distances of the lowest carbon atom of the chain from the nitrogen(@xmath31 ) and the boron(@xmath32 ) atom in the bn plane for different @xmath7 values of carbon chains . the bonding sites and magnetic properties of cacs on bn exhibit
an even / odd disparity . with the exception of the single carbon ad - atom , even numbered cacs bind to bn near the top of nitrogen(tn ) atom and the odd numbered cacs bind near the top of boron(tb ) atom .
additionally , the even and odd numbered chains grown on bn have magnetic and nonmagnetic(nm ) ground states , respectively , with the exception of cac(1 ) and cac(2 ) cases .
all calculations were performed on a @xmath33 supercell which contains 32 carbon atoms . [ cols="^,^,^,^,^,^,^",options="header " , ]
having shown the stable carbon chain structures grown on bn , we next investigate the growth of short bn chains on graphene to see if a similar self - assembly mechanism is also present for this reverted situation .
here we adopt the representation where bn(@xmath7)+graphene indicates a bn chain consisting of @xmath7 atoms that is attached to graphene through n atom .
for example ,
nb(3 ) + graphene ( or nb(3 ) only ) stands for b - n - b chain attached to graphene through b atom .
following the similar procedures described in sections iii and iv , we calculate most favorable binding sites , electronic properties , binding energies .
we first perform md simulations to see the self assembly mechanisms of bn chains . to this end
we start by placing single b and n ad - atoms on graphene supercell and run md simulations at 500k for 10ps . eventually , the two ad - atoms form a bn molecule perpendicularly attached to graphene . when a third ad - atom n is placed on the graphene at the close proximity of existing bn molecule ,
md simulations show that a chain comprising three atoms , b - n - b forms .
this procedure was repeated for different lengths and it was observed that b and n atoms are indeed self - assembled on graphene honeycomb structure to form short bn atomic chains .
snapshots taken from these md simulations are presented in fig .
[ fig9 ] .
atoms , the structure rearranges itself to form a chain comprising @xmath34 atoms .
the snapshots are taken from the quantum md simulations performed at 500k for 10ps .
blue , green and brown balls are , respectively , nitrogen , boron and carbon atoms.,width=302 ] the most favorable binding sites of short bn chains on graphene and their corresponding binding energies were calculated with self consistent conjugate gradient calculations .
it was found that , the binding site of a bn chain on the graphene depends on the type of its closest atom to the graphene .
a nb(@xmath7 ) chain is bound to the graphene layer at the bridge site , and a bn(@xmath7 ) is bound at the top site as shown in fig .
[ fig10 ] . however , both single b and single n ad - atoms are bound at the bridge site .
contrary to the cac case , no clear even / odd disparity was obtained in h - bn+graphene complexes owing to the ionic bonding between b - n atoms .
here there is another effect related with the number of electrons in the chain .
chains having odd number of atoms have odd number of electrons , while even numbered chains have even number of electrons .
the binding energies of these chains shown in in table [ table : 2 ] suggest that bn chains are bound with higher energies to the graphene as compared to the binding energies of cacs on bn , especially for longer chains . the binding energy of a boron ending chain ( i.e. nb ) is always higher than a nitrogen ending chain ( i.e. bn ) site . here
we note that the supercell size and adsorbate - adsorbate the even / odd disparity observed in the magnetic properties of cac(@xmath7)+h - bn are also not seen for bn chains grown on graphene .
single b ad - atom as well as single n ad - atom adsorbed on graphene have magnetic ground states .
however , while bn(2 ) and nb(2 ) chains on graphene are nonmagnetic , bn(3 ) and nb(3 ) on graphene have magnetic ground states .
this order is disturbed by the chain consisting of four atoms ; while bn(4)+graphene is nonmagnetic , nb(4)+graphene has magnetic ground state .
the electronic structures calculated for a single adsorbate , namely single boron and nitrogen ad - atoms , and single nb(2 ) and bn(2 ) molecules , perpendicularly attached to ( 4x4 ) supercell of graphene are shown in fig .
[ fig11 ] . owing to their odd number of electrons , single n and b ad - atoms shift the fermi level up and down relative to the dirac point of graphene .
this is an important feature to be used in electronic applications of single layer graphene .
for nb(2)+graphene the bands crossing at the fermi level attribute semimetallic character . on the other hand ,
bn(2)+graphene is a semiconductor with a narrow indirect band gap .
this situation shows the dramatic effect of the type of chain atom which is attached to graphene .
in addition , the supercell size , adsorbates - adsorbate coupling , as well as the symmetry of the adsorbate+supercell system are crucial for the resulting electronic structure for small supercells.@xcite for example , the semiconductor presented in fig .
[ fig11 ] ( d ) becomes metallic when bn+graphene system is treated in ( 6x6 ) supercell . in the case of
very large supercell having single adsorbate attached the band crossing is recovered and the states of adsorbate appear as localized ( resonance ) impurity states .
in conclusion , using first - principles calculations within the density functional theory , we revealed the growth mechanism of carbon and bn atomic chains on single layer honeycomb structures , namely hexagonal bn(or h - bn ) and graphene , respectively .
we found that with the inclusion of each new carbon atom , the existing atomic chain consisting of @xmath7 carbon atom leaves its previous position to join with the new carbon atom to from a chain of @xmath34 carbon atom .
similar , but more complex growth mechanisms were also found for bn chains grown on graphene layers depending on the type of chain atom(b or n ) which is attached to graphene .
these growth processes were simulated by ab - initio molecular dynamics calculations at various temperatures and the resulting chain structures were shown to be stable even though they are free to bend and slightly tilt .
nevertheless , these simulations are performed at high temperatures(1000k ) and the chains did not detach from the substrate plane for 10ps which is adequately long for ab - initio molecular dynamics simulations . therefore these simulations presents reliable stability results , especially at room temperature .
the growth of atomic chains on the single layer honeycomb structures heralds a self - assembly process , which may have fundamental and technological implications . the grown chains by themselves , exhibit interesting physical and chemical properties depending on the number of atoms forming the chain and the type of chain atom attaching to the substrate .
in particular , the physical properties of even numbered and odd numbered carbon chains can behave differently leading to interesting even - odd disparity .
we also showed that atomic chains grown on single layer substrates attribute useful functionalities to bare h - bn and graphene .
these properties are dependent on the number of carbon atoms in the chains .
apart from creating localized electronic states in the band gap and local magnetic moments it is demonstrated that carbon chains can also be used for increasing the interlayer spacing between bn flakes , where specific molecules can be stored .
atomic chains grown on graphene and h - bn create chemically active sites on bn surface for atoms such as au , li and h@xmath0 , which provides the connection of these materials . additionally , specific atoms forming strong bonds with atomic chains modify the electronic structure in the band gap and hence change the conductivity for possible sensor applications .
similarly , modifications of the electronic properties of graphene depending on the grown b or n ad - atoms and bn or nb chains are also worth emphasizing . moving the fermi level of graphene up or down by adsorbing n or b atoms ,
respectively can be a worthwhile feature in the electronic applications related with graphene .
the self - assembly in terms of atomic chains grown on semimetallic graphene and semiconducting h - bn and interesting functionalities achieved thereof can bring up new perspectives for further research on single layer honeycomb structures .
part of the computational resources has been provided by tubitak ulakbim , high performance and grid computing center ( tr - grid e - infrastructure ) and uybhm at istanbul technical university through grant no .
2 - 024 - 2007 .
this work was supported partially by the academy of sciences of turkey(tuba ) . in ref [ ]
the energy barrier for the migration of single carbon atom on h - bn was calculated @xmath35 0.3 ev . in those calculations
the van der waals correction was not included .
for the same reasons the present calculations with van der waals correction predicts the binding energy as 0.95 ev relative to the energy of the most favorable site , which is @xmath35 0.25 ev higher than that calculated without van der waals correction . | nucleation and growth mechanisms of short chains of carbon atoms on single - layer , hexagonal boron nitride ( h - bn ) , and short bn chains on graphene are investigated using first - principles plane wave calculations .
our analysis starts with the adsorption of a single carbon ad - atom and examines its migrations .
once a c@xmath0 nucleates on h - bn , the insertion of each additional carbon at its close proximity causes a short segment of carbon atomic chain to grow by one atom at at a time in a quaint way : the existing chain leaves its initial position and subsequently is attached from its bottom end to the top of the carbon ad - atom .
the electronic , magnetic and structural properties of these chains vertically adsorbed to h - bn depend on the number of carbon atoms in the chain , such that they exhibit an even - odd disparity .
an individual carbon chain can also modify the electronic structure with localized states in the wide band gap of h - bn . as a reverse situation we examined the growth of short bn atomic chains on graphene , which attribute diverse properties depending on whether b or
n is the atom bound to the substrate . these results together with ab - initio molecular dynamics simulations of the growth process
reveal the interesting self - assembly behavior of the grown chains .
furthermore , we find that these atomic chains enhance the chemical activity of h - bn and graphene sheets by creating active sites for the bonding of various ad - atoms and can act as pillars between two and multiple sheets of these honeycomb structures leaving wider spacing between them to achieve high capacity storage of specific molecules . |
magnetic relaxation is the process by which a magnetic field in a highly conducting fluid seeks a minimum energy state subject to pertinent topological constraints ( moffatt 1985 ) . in the perfect conductivity limit ,
the lines of force ( ` @xmath10-lines ' ) are frozen in the fluid , a topological constraint represented by the family of magnetic helicity invariants @xmath13 where @xmath14 , and @xmath15 is any lagrangian volume on whose surface @xmath16 ( with unit normal @xmath17 ) @xmath18 . these invariants represent the conserved degree of linkage of @xmath10-lines within @xmath15
. of particular importance is the global magnetic helicity @xmath19 , integrated over the whole domain @xmath20 of fluid .
the magnetic energy @xmath21 then has a lower bound ( arnold 1974 ) @xmath22 where @xmath23 is a constant that depends on the scale and geometry of @xmath20 .
the particular problem addressed in this paper concerns relaxation in a plasma of extremely low - density @xmath24 , in which the fluid pressure @xmath25 is negligible compared with the magnetic pressure @xmath26 , i.e. @xmath27 .
flow of the plasma is then driven solely by the lorentz force @xmath28 .
the low density implies further that inertia is negligible compared with the viscous force in the navier - stokes equation , the viscosity @xmath29 being essentially independent of @xmath24 in the limit @xmath30 .
these approximations have been adopted in a cartesian model by bajer & moffatt ( 2013 ) who treated relaxation of a single - component field , and by moffatt ( 2015 ) who considered the case of a two - component field with non - zero helicity . here
we shall consider the situation in a cylindrical geometry , for which the lorentz force includes the ` hoop stress ' associated with curvature of the @xmath10-lines .
we consider a two - component helical field in cylindrical polar coordinates @xmath32 of the form @xmath33 the associated current distribution is given by in the definition of @xmath34 . ]
@xmath35 the first objective is to determine how the relaxation of this field is constrained by the initial magnetic helicity distribution .
the component @xmath36 of the field is responsible for the classic ` pinch effect ' ( bennet 1934 ) . in this scenario , it is natural to suppose that the initial @xmath37-component of field is uniform , and that the initial @xmath38-component is concentrated near the outer cylindrical boundary , such a field then providing a radial lorentz force that tends to drive the plasma inwards .
we shall suppose that the resulting radial motion @xmath39 is controlled by viscosity , which , as indicated above , dominates over inertia when the plasma density is sufficiently small .
we further suppose that the fluid pressure is negligible compared with the magnetic pressure , i.e. this is a ` low-@xmath31 ' plasma .
of course , the fluid pressure increases in the inner region where the density increases ; the effect of this increase can be included without difficulty in the numerical treatment . in the perfect conductivity limit @xmath40 , the magnetic field evolves according to the ` frozen - field ' equation @xmath41 and , with the neglect of inertia and pressure gradient , the navier - stokes equation degenerates to @xmath42 where @xmath43 and @xmath44 are the shear and bulk viscosities . from ( [ induction0 ] ) and ( [ ns0 ] ) ,
an equation may easily be derived for the magnetic energy : @xmath45 \ , { \textnormal d}v.\ ] ] thus @xmath46 is monotonic decreasing and bounded below by ( [ bound ] ) .
equilibrium is attained only when @xmath47 , and then from ( [ ns0 ] ) , @xmath48 , i.e. the field is ` force - free ' .
hence @xmath49 for some pseudo - scalar function @xmath50 satisfying @xmath51 . for the particular one - dimensional geometry considered in this paper ,
it is therefore to be expected that , when the magnetic diffusivity @xmath1 is sufficiently weak , the field will relax rapidly during an initial stage to a force - free state ( with here @xmath52 ) that has minimum energy compatible with its initial ( conserved ) topology .
minimising energy subject to the _ single _ topological constraint of conserved global helicity yields a force - free field structure with @xmath53cst . , a condition that provides reversed axial field near the outer boundary ( taylor 1974 )
however , the dynamical process through which such a reversed field may spontaneously appear is not revealed by the simple process of seeking a minimum - energy state .
such a reversal can not in fact appear for so long as the pinching motion is purely radial .
however , it seems possible that instabilities of the basic relaxing field may lead to an @xmath2-effect , which could conceivably achieve reversal .
we shall explore this possibility in 8 ; first however , we treat simple radial relaxation neglecting any instabilities that may be present .
we suppose that the plasma is contained in the cylindrical annulus @xmath54 , where @xmath55 ; we shall scale all lengths so that , in effect , @xmath56 . the boundaries @xmath57 and @xmath58 are assumed to be thin perfectly conducting cylinders , separating the plasma from the internal and external regions ; these boundaries can therefore support current sheets with both @xmath37- and @xmath38-components .
the electric field is given by @xmath59 , and we suppose that @xmath60 in the internal and external regions , assumed insulating , i.e. for @xmath61 and @xmath62 .
since @xmath63 on both boundaries and both tangential components of @xmath64 are continuous across them , it follows that , with @xmath65 , @xmath66 i.e. that @xmath67 some type of boundary - layer behaviour is to be expected in the limit @xmath68 .
when the magnetic diffusivity @xmath1 is nonzero , the field evolution is described by the induction equation , @xmath69 and the fluid density @xmath24 satisfies the mass conservation equation @xmath70 these equations may be combined to give @xmath71 where @xmath72 , the lagrangian ( or ` material ' ) derivative . noting that , from ( [ b_field ] ) and ( [ u_field ] ) , @xmath73 when @xmath74eqn.([lagrangian_induction ] ) gives @xmath75 i.e. when following any material element of fluid , @xmath76 and @xmath77 are constant
. in eulerian form , eqn.([induction ] ) has components @xmath78 and @xmath79 the flux of @xmath80 between the two cylinders is @xmath81 and we note , using ( [ b_z_eqn ] ) and the conditions @xmath82 on both boundaries , that @xmath83_{\delta}^1\!=0.\ ] ] it follows that @xmath84 cst . similarly , the flux of @xmath85 in the @xmath38-direction , per unit axial length between the cylinders , is @xmath86 and it follows in the same way from ( [ b_theta_eqn ] ) , and the conditions @xmath87 on both boundaries , that @xmath88 cst . also .
the constraints @xmath89 provide an important check on the numerical computations that follow ( see fig .
[ fig_relaxation00](f ) ) .
the mass per unit axial length @xmath90 is of course similarly constant : @xmath91 the results ( [ flux_constraints ] ) are clearly compatible with ( [ br_and_btheta ] ) when @xmath40 .
we adopt as initial conditions for the magnetic field @xmath92 and @xmath93 \textnormal{e}^{-k ( 1 - r)^{2}}.\ ] ] these are chosen to be compatible with the boundary conditions ( [ bc2 ] ) for any values of the parameters @xmath94 , and to satisfy @xmath95 . fig . [ fig_bth0 ] shows the field @xmath96 for @xmath97 and for three values @xmath98 , together with corresponding values of the flux @xmath99 .
increasing @xmath100 leads to increasing concentration of @xmath96 near the outer boundary .
rather arbitrarily we choose @xmath101 in the computations that follow .
we further adopt initial conditions for the velocity and density fields , @xmath102 i.e. the plasma is initially at rest with uniform density @xmath103 .
the lorentz force in the cylindrical geometry considered here takes the form @xmath104 the navier - stokes equation , including this term , has only a radial component : @xmath105 where @xmath106 is an effective viscosity . as in moffatt ( 2015 ) ,
it is now convenient to introduce dimensionless variables @xmath107 and dimensionless parameters @xmath108 we assume that both these parameters are small : @xmath109 ( i.e. small diffusivity ) ; and @xmath110 ( i.e. low density ) . with the variables ( [ dimensionless_variables ] ) , and immediately dropping the hats , eqn .
( [ mass_conservation ] ) is unchanged , while in eqns .
( [ b_theta_eqn ] ) and ( [ b_z_eqn ] ) , @xmath1 is simply replaced by @xmath111 : @xmath112 and @xmath113 the momentum equation ( [ navier_stokes ] ) becomes @xmath114 here , the term @xmath115 is the gradient of magnetic pressure , and the term @xmath116 is the additional ` hoop stress ' that arises due to curvature of the @xmath117-lines .
we can now proceed to numerical integration of these equations , with boundary conditions as already stated .
when @xmath118 , these are @xmath119 and initial conditions as specified in the previous section .
results obtained with _
mathematica _ are summarised in fig .
[ fig_relaxation00](a - h ) , for the particular choice of parameters ( [ parameter_values ] ) , together with @xmath120 ( the behaviour for this choice is quite typical ) .
the panels of the figure show ( a ) the inward movement of the density field @xmath121 in response to the negative radial lorentz force ( b ) , which collapses rapidly to near zero ; ( c , d ) the corresponding evolution of the magnetic field components ; ( e ) the rise of both field components at the inner boundary ; ( f ) the decay of magnetic energy ; ( g ) the pseudo - scalar coefficient @xmath122 when the nearly force - free state has been established ; and finally ( h ) the fluxes @xmath123 and @xmath99 which , as expected , remain constant to within numerical error throughout the whole computational period @xmath124 .
the following points are particularly worth noting .
first , there are clearly two phases to the evolution : an initial phase ( here @xmath125 ) during which magnetic diffusion is negligible and the magnetic energy decreases relatively rapidly on the ( dimensional ) time - scale @xmath126 ; and a slow diffusive stage @xmath127 , during which @xmath80 slowly relaxes back to its initial uniform value ( its flux remaining constant ) ; during this phase , @xmath128 also slowly decays to a constant @xmath129 , implying ultimate concentration of axial current as a current sheet on the inner boundary @xmath57 .
second , although not shown in the figure , the energy @xmath130 of the @xmath80-field actually _ increases _ during the initial phase ( reaching a maximum at @xmath131 ) , but this increase is more than compensated by the decrease of energy of the @xmath85-field .
later , during the slow diffusive stage , both contributions to energy slowly decrease .
[ fig_relaxation00](f ) shows the decay of energy for @xmath132 as well as for @xmath133 ; as expected , the diffusive effect is less apparent in the former case , but the initial ( non - diffusive ) stage is quite similar .
third , as previously noted , when the lorentz force is effectively zero ( i.e. during the diffusive phase ) , @xmath134 ; the coefficient @xmath12 is then given by @xmath135 this coefficient , shown in fig .
[ fig_relaxation00](g ) , is far from uniform in @xmath136 , so this is certainly not a taylor state ( for which @xmath12 would necessarily be uniform ) .
the weak time - dependence of @xmath12 results from slow continuing evolution during the diffusive stage .
this stage is interesting because the field components continuously adjust themselves in such a way that the force - free condition is maintained ; in other words , this is not a pure diffusive process , but one that is still constrained to remain nearly force - free through the dynamics encapsulated in eqn .
( [ ns2 ] ) .
and @xmath132 : ( a ) @xmath137 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( b ) expanded view of the same curves in the region @xmath140 ; ( c ) @xmath141 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( d ) expanded view of the @xmath80-curves in the region @xmath140 ; , title="fig : " ] and @xmath132 : ( a ) @xmath137 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( b ) expanded view of the same curves in the region @xmath140 ; ( c ) @xmath141 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( d ) expanded view of the @xmath80-curves in the region @xmath140 ; , title="fig : " ] and @xmath132 : ( a ) @xmath137 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( b ) expanded view of the same curves in the region @xmath140 ; ( c ) @xmath141 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( d ) expanded view of the @xmath80-curves in the region @xmath140 ; , title="fig : " ] and @xmath132 : ( a ) @xmath137 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( b ) expanded view of the same curves in the region @xmath140 ; ( c ) @xmath141 at @xmath138 for @xmath139 ( blue ) and @xmath132 ( red ) ; ( d ) expanded view of the @xmath80-curves in the region @xmath140 ; , title="fig : " ] in the limit @xmath142 , i.e. @xmath139 , we drop the diffusion terms in ( [ b_theta_eqn_dim ] ) and ( [ b_z_eqn_dim ] ) , and only the boundary condition @xmath63 of ( [ bc3 ] ) survives , ( [ bc2 ] ) being then automatically satisfied .
the two fluxes @xmath123 and @xmath99 are still conserved , within numerical error , as in fig .
[ fig_relaxation00 ] .
comparison with the situation when @xmath132 is interesting .
the difference is in fact very slight up to about @xmath143 , but becomes visible , particularly near the boundary @xmath58 by @xmath138 , as shown in fig .
[ fig_limit_behaviour](a , b ) .
the expanded views in figs .
[ fig_limit_behaviour](b , d ) show that indeed when @xmath139 the boundary condition @xmath144 is _ not _ maintained at @xmath58 ; when @xmath132 a weak boundary layer is required , within which the solution adapts to this boundary condition . as time advances , the curve for @xmath132 continues to evolve due to weak diffusion towards the situation @xmath145 , whereas the curve for @xmath139 remains static . in this case of @xmath139
, there is only the initial phase of non - diffusive relaxation to a force - free minimum - energy field , which then remains essentially static .
it is obvious that purely radial flow can not lead to local reversal of an axial field that is initially uniform , even in conjunction with diffusion .
however , the situation considered by taylor ( 1974 ) in the context of the reversed - field pinch was that of a fully turbulent plasma in which more complex processes must be present .
such turbulence presumably results from persistent instability of the evolving mean field @xmath10 , which has non - zero helicity represented both by the integrals ( [ helicity ] ) and also by the non - zero pseudo - scalar @xmath146 .
such instabilities must inherit the helicity of this mean field , and may be expected to provide an @xmath2-effect , such that the ( non - dimensionalised ) mean electric field becomes @xmath147 where @xmath148 still represents the mean radial flow , and @xmath111 now includes turbulent as well as molecular diffusivity . a stability analysis following the approach of furth ,
killeen & rosenbluth ( 1963 ) has been carried out by mizerski ( 2017 ) , and a resulting anisotropic @xmath2-effect deduced . here
, we adopt the simpler isotropic prescription @xmath149 where @xmath5 is a pure scalar constant ( positive or negative ) ; the factor @xmath111 is included here in recognition of the diffusive origin of the @xmath2-effect .
the choice ( [ alpha ] ) of course ensures that @xmath2 has the same pseudo - scalar character as @xmath146 . with this prescription for @xmath2 ,
the boundary condition @xmath150 becomes @xmath151 and , in the cylindrical geometry , this is still obviously satisfied by @xmath152 on @xmath153 .
the only modification required is therefore the inclusion of a term @xmath154 in the induction equation . in exploratory vein , we include only the z - component ( mizerski 2017 , particularly eqn .
it is found that the @xmath38- and @xmath37- components of the curl of the resulting mean emf differ by a factor proportional to @xmath155 , and that , summing over all fourier modes , one of these components must vanish .
] , thus replacing ( [ b_z_eqn_dim ] ) by @xmath156,\ ] ] while leaving ( [ b_theta_eqn_dim ] ) unaltered . from ( [ b_field ] ) and ( [ current ] )
, we then have @xmath157,\ ] ] @xmath158 @xmath159 @xmath160 where , for convenience , we include the remaining unaltered evolution equations
. consider first whether and under what circumstances reversals of the axial field component @xmath80 may occur .
suppose that this does happen , and that the reversal is initiated at a critical time @xmath161 ; then @xmath162 must equal zero for some @xmath163 where @xmath164 may be an internal point or an end - point of the closed interval @xmath165 $ ] .
moreover , for @xmath166 , @xmath167 for all @xmath168 $ ] .
then , at @xmath169 , we have @xmath170 and @xmath171 since the curvature is necessarily positive at this point . from eqn.([b_z_eqn_final ] )
, we then have @xmath172}{\partial t}\right]_{t = t^*}=\kappa\left(1-q\ , [ b_{\theta}(r^*,t^*)]^2\right)\left[\frac{\partial^{2}b_z(r , t^*)}{\partial r^2}\right]_{r = r^*},\,\,\ ] ] all other terms vanishing by virtue of ( [ criticality1 ] ) .
hence it would appear that @xmath173 will indeed become negative in a neighbourhood of @xmath164 , provided @xmath174 ^ 2>1.\ ] ] this can therefore occur only if @xmath175 and @xmath85 is sufficiently strong ; and it is likely to occur first in the region where @xmath176 is maximal .
more generally , ( [ b_z_eqn_final ] ) may be written @xmath177 - q\,\kappa \frac{\partial b_z}{\partial r}\frac{\partial } { \partial r}\left(b_{\theta}^2\right ) .\end{aligned}\ ] ] the second term on the right - hand side has a diffusive character , but with negative diffusivity in any region where @xmath178 .
( the remaining terms of the right - hand side involve only @xmath80 and @xmath179 . )
it follows that @xmath80 can become negative only if this is ` triggered ' by a period of negative diffusivity in some @xmath136-interval .
we continue to use the parameter values ( [ parameter_values ] ) . with these values ,
@xmath180 is maximal at @xmath181 , with maximum value @xmath182 ; an interval of negative diffusivity therefore occurs for @xmath183 . , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] , @xmath185 , @xmath186 and @xmath187 between the case @xmath188 and the case @xmath8 for @xmath189 at times @xmath190 .
the left column shows , for reference , the evolution of @xmath191 , @xmath192 , @xmath121 and @xmath193 when @xmath8.,title="fig : " ] and @xmath194 for same parameter values as in fig .
[ fig_deltas_eta0d01 ] .
the left column shows , for reference , the evolution of @xmath195 and @xmath196 when @xmath8.,title="fig : " ] and @xmath194 for same parameter values as in fig .
[ fig_deltas_eta0d01 ] .
the left column shows , for reference , the evolution of @xmath195 and @xmath196 when @xmath8.,title="fig : " ] and @xmath194 for same parameter values as in fig .
[ fig_deltas_eta0d01 ] .
the left column shows , for reference , the evolution of @xmath195 and @xmath196 when @xmath8.,title="fig : " ] and @xmath194 for same parameter values as in fig .
[ fig_deltas_eta0d01 ] .
the left column shows , for reference , the evolution of @xmath195 and @xmath196 when @xmath8.,title="fig : " ] numerical integration for @xmath197 were quite regular , and not greatly different from the situation when @xmath8 .
however , as might be expected from the above discussion , numerical instabilities that are difficult to control appear when @xmath198 .
this however is the regime that must be investigated in seeking possible reversal of @xmath191 .
this led us to adopt a controllable numerical procedure , specifically 4th - order finite - differences in the radial direction and 2nd - order adams - bashforth time - stepping with crank - nicolson treatment of the diffusive terms .
the results for @xmath8 were as expected in complete agreement with those obtained using _
mathematica_. we focus first on the near - critical situation when @xmath188 .
let @xmath199 denote the density computed for any particular value of @xmath5 , and let @xmath200 similarly for @xmath201 and @xmath194 .
4 and 5 show numerical results for @xmath189 ; the left - hand columns show curves for @xmath8 , while the right - hand columns show the differences @xmath202 , etc .
note first from the last row of fig .
4 that the negative ( pinching ) velocity is initially decreased in magnitude by the @xmath2-effect when @xmath188 ; it actually becomes weakly positive near @xmath58 for @xmath203 and is positive over the whole range @xmath204 for @xmath205 .
this implies a corresponding net decrease in the inward transport of mass ; however , the decrease in transport for the magnetic field component @xmath173 is more than compensated by the direct action of the @xmath2-effect .
when @xmath188 , @xmath191 decreases much more rapidly than when @xmath8 ( by a factor of about 3 ) near @xmath58 in the early stage of relaxation ; @xmath206 increases more rapidly near @xmath58 , but by a more modest amount ( @xmath207 ) .
the function @xmath208 defined by eqn .
( [ gamma ] ) is also changed by @xmath209 when @xmath188 , but there is no apparent tendency for @xmath196 to become more uniform . as indicated above , we anticipated numerical problems for @xmath198 , and we did indeed run into these . typically , in the range of @xmath5 between 4 and 6 , a packet of oscillations in @xmath80 of very short wavelength appears at @xmath211 in the region of negative diffusivity near @xmath58 .
these oscillations move inwards , in tandem with @xmath128 which decreases till the local diffusivity @xmath212 becomes positive , at which stage the oscillations in @xmath80 are damped out , the subsequent evolution being quite smooth .
[ q_above_qc ] shows this subsequent evolution for @xmath210 .
note that already at the early time @xmath213 , @xmath80 is negative near @xmath58 , presumably a consequence of the early negative diffusivity in this region .
the field then relaxes back , becoming positive at @xmath58 by time @xmath214 .
the short period of reversed @xmath80 near @xmath215 is interesting in the context of the reversed - field pinch .
however , we ca nt be certain that this is a genuine physical effect rather than just a consequence of adopting a unphysical model for the @xmath2-effect yielding a period of negative diffusion .
the behaviour for @xmath211 is evidently non - analytic ; an asymptotic treatment of the behaviour as @xmath216 is presented in appendix a. for q=5.5 , @xmath189 , and other parameter values as in fig .
[ fig_deltas_eta0d01 ] . ]
we have investigated the relaxation of an axisymmetric magnetic field having both axial and toroidal components in a cylindrical geometry with perfectly conducting boundaries .
the density is assumed very small and fluid pressure is neglected compared with magnetic pressure .
a purely radial flow is driven by the lorentz force and energy is dissipated by viscosity . in the zero - resistivity limit , the field rapidly relaxes to a force - free state .
when weak resistivity is taken into account , the initial rapid relaxation is followed by slow decay of the field which is constrained to remain nearly force - free with @xmath134 .
however @xmath196 is quite strongly non - uniform , so this is not a taylor state .
in 8 , we have explored the possibility that an @xmath2-effect , with @xmath2 proportional to @xmath146 , might be capable of causing axial field reversal near the outer boundary where the toroidal field component @xmath85 is initially strong .
we have found that if this @xmath2-effect is sufficiently strong , it can produce a region of negative effective diffusivity of the @xmath80-field near the outer boundary ; this can instantaneously generate high - frequency short - wavelength oscillations which are rapidly damped as they move into the interior region of positive diffusivity .
a transitory reversal of @xmath217 occurs near the outer boundary during this process .
there are serious numerical difficulties in handling such a situation ; nevertheless , this work points to one possibility whereby a reversed axial field , as observed in the reversed field pinch ( taylor 1974 ) , can be dynamically generated from an initially uniform axial field , through the combined action of pinching by the @xmath85-field and a suitably contrived @xmath2-effect . of course
, it would be desirable to derive a correct form of this @xmath2-effect , through investigation of the turbulence that results from instabilities of the relaxing field .
work is ongoing on this aspect of the problem .
* acknowledgments*. the partial funding of the ministry of science and higher education of poland within the grant no ip 2014 031373 and statutory activities no 3841/e-41/s/2015 is gratefully acknowledged .
this work was also partially funded from the leading national research centre ( know ) received by the centre for polar studies in poland for the period 2014 - 2018 .
arnold , v.i . ( 1974 ) the asymptotic hopf invariant and its applications . _ proc .
summer school in diff .
_ erevan , armenia [ in russian ] .
229 - 256 .
bajer , k. & moffatt , h. k. ( 2013 ) magnetic relaxation , current sheets , and structure formation in an extremely tenuous fluid medium . _ astrophys .
j. _ * 779 * , 169 - 182 .
bender , c.m . &
orszag , s.a .
( 1978 ) _ advanced mathematical methods for scientists and engineers .
_ mcgraw?hill , new york .
bennett , w. h. ( 1934 ) magnetically self - focussing streams .
_ 45.12 : 890 .
furth , h. p. , killeen , j. & rosenbluth , m. n. ( 1963 ) finite - resistive instabilities of a sheet pinch _ phys . fluids _
, 459- mizerski , k. ( 2017 ) large scale emf in current sheets induced by tearing modes , _ fluid dyn .
_ , submitted .
moffatt , h. k. ( 1985 ) magnetostatic equilibria and analogous euler flows of arbitrarily complex topology .
. _ j. fluid mech . _ * 159 * 359 - 378 .
moffatt , h. k. ( 2015 ) magnetic relaxation and the taylor conjecture .
_ j.plasma phys .
_ 81.06 : 905810608 .
taylor , j. b. ( 1974 ) relaxation of toroidal plasma and generation of reverse magnetic fields , _ phys .
lett . _ * 33 * , 1139 - 1141 .
the short - time behaviour of @xmath191 is bound to be non - analytic , since even the linear diffusion term introduces short - time dependence of the sort @xmath218 .
although the problem is nonlinear , to get some insight into this short - time behaviour , we suppose that the controlling factor of @xmath80 , i.e. its most rapidly changing component ( bender & orszag 1978 ) , has the asymptotic wkb form @xmath219 where @xmath220 and @xmath221 are functions to be determined .
we further assume , that since the velocity is weak throughout the entire evolution , the terms involving @xmath193 in ( [ b_z_eqn_final ] ) and ( [ b_theta_eqn_final ] ) can be neglected at the initial stage leaving only the effect of diffusion in the @xmath85 component .
the behaviour at small @xmath222 of @xmath85 can therefore be expressed by a simple formula @xmath223 where @xmath224 as @xmath225 . substituting the asymptotic expressions for the magnetic field components into equation ( [ b_z_eqn_final ] ) we obtain , for small @xmath222 , @xmath226 and where the dot denotes a time derivative and the prime a derivative with respect to @xmath136 .
it follows that , @xmath227 the solution of the above equations is @xmath228^{-1/2}\mathrm{d}r + \mathrm{cst.}\right\}^2.\label{wkb_solution}\ ] ] we have seen that increasing the value of @xmath5 leads to negative diffusion in the evolution of @xmath191 .
as long as @xmath229 the short - time asymptotics of @xmath191 involve no significant irregularities . for @xmath5 above the critical value , the negative diffusion introduces serious irregularities and the point @xmath230 at which the diffusion coefficient @xmath231 changes sign becomes a singular ( critical ) point . in the vicinity of @xmath232 , the wkb solution ceases to be valid since gradients become infinite and a critical layer , @xmath233 being the critical layer thickness , one can expand @xmath234^{\prime}_{r = r_c}$ ] and the @xmath80 critical - layer equation is @xmath235 with @xmath236^{\prime}_{r = r_c}$ ] and @xmath237 note that as the time scale @xmath238 increases the critical layer thickens ; this equation may be solved in terms of a laplace transform in time with @xmath136-dependence of each laplace mode @xmath239 in the form @xmath240 , and matched to the wkb solutions on both sides , i.e for @xmath241 and @xmath242 .
] of width @xmath243 is required to match the @xmath80-derivatives across @xmath232 . a continuous wkb solution for @xmath244 for the controlling factor , with a jump in the first derivative at @xmath230 , satisfying the boundary conditions ( [ bc2 ] ) , can be found in the form @xmath245
^ 2\right\},\label{sol_region1}\ ] ] for @xmath241 , and @xmath246 @xmath247\right\}\cos\left
[ \frac{\sqrt{i_1 ^ 2+i_2 ^ 2}}{2\kappa t}\left(\mathcal{i}(r)-i_2\right ) \right],\label{sol_region2}\ ] ] for @xmath242 , where @xmath248 ( strictly , the upper limit in @xmath249 should be @xmath250 , and the lower limit in @xmath251 and @xmath252 should be @xmath253 , due to the presence of the critical layer at @xmath232 , which matches the derivatives of the wkb solutions on either side of the critical point @xmath232 , but this correction is negligibly small . ) -dependence of the controlling factor , i.e. the most rapidly changing component of @xmath254 at times @xmath255 - top row , and @xmath256 - bottom row , for @xmath257 @xmath258 and other parameter values as in the numerical simulations ( fig .
[ fig_deltas_eta0d01 ] ) .
the regions @xmath241 and @xmath242 are plotted separately in the left and right columns respectively .
note that in reality the sharp change in derivative at @xmath230 is smoothed out by a boundary layer , thickening with time.,title="fig:"]-dependence of the controlling factor , i.e. the most rapidly changing component of @xmath254 at times @xmath255 - top row , and @xmath256 - bottom row , for @xmath257 @xmath258 and other parameter values as in the numerical simulations ( fig .
[ fig_deltas_eta0d01 ] ) .
the regions @xmath241 and @xmath242 are plotted separately in the left and right columns respectively .
note that in reality the sharp change in derivative at @xmath230 is smoothed out by a boundary layer , thickening with time.,title="fig : " ] -dependence of the controlling factor , i.e. the most rapidly changing component of @xmath254 at times @xmath255 - top row , and @xmath256 - bottom row , for @xmath257 @xmath258 and other parameter values as in the numerical simulations ( fig .
[ fig_deltas_eta0d01 ] ) .
the regions @xmath241 and @xmath242 are plotted separately in the left and right columns respectively .
note that in reality the sharp change in derivative at @xmath230 is smoothed out by a boundary layer , thickening with time.,title="fig:"]-dependence of the controlling factor , i.e. the most rapidly changing component of @xmath254 at times @xmath255 - top row , and @xmath256 - bottom row , for @xmath257 @xmath258 and other parameter values as in the numerical simulations ( fig .
[ fig_deltas_eta0d01 ] ) .
the regions @xmath241 and @xmath242 are plotted separately in the left and right columns respectively .
note that in reality the sharp change in derivative at @xmath230 is smoothed out by a boundary layer , thickening with time.,title="fig : " ] it is clear , therefore , that for @xmath241 the solution is regular and controlled mainly by diffusion .
however , for @xmath242 oscillations of very short @xmath259 wavelength appear , which is smallest near @xmath232 and increases with @xmath136 ; since @xmath260 for any @xmath5 , the amplitude of these oscillations is exponentially small and decreases with increasing @xmath136 .
they appear instantaneously at @xmath211 and their wavelength increases with increasing @xmath222 . at very short times
they are strongly damped by the very small exponential term in ( [ sol_region2 ] ) .
the situation is depicted ( for @xmath257 ) in fig .
[ wkb_solutions ] .
these fast short - time oscillations are just as described in 8.3 for the run with @xmath210 . at the earliest stage of evolution
we observed a small drop in the axial flux @xmath123 of @xmath80 from @xmath261 to about @xmath262 , which subsequently remained constant ; the toroidal flux @xmath263 and the mass @xmath90 experienced much smaller jumps at the same moment as that of @xmath123 , but then also remained constant .
these jumps are an indication of unavoidable numerical inaccuracy at this earliest stage when the extremely short wavelength oscillations can not be adequately resolved by numerical procedure , however much refined . | the relaxation of a helical magnetic field @xmath0 in a high - conductivity plasma contained in the annulus between two perfectly conducting coaxial cylinders is considered .
the plasma is of low density and its pressure is negligible compared with the magnetic pressure ; the flow of the plasma is driven by the lorentz force and and energy is dissipated primarily by the viscosity of the medium .
the axial and toroidal fluxes of magnetic field are conserved in the perfect - conductivity limit , as is the mass per unit axial length .
the magnetic field relaxes during a rapid initial stage to a force - free state , and then decays slowly , due to the effect of weak resistivity @xmath1 , while constrained to remain approximately force - free .
interest centres on whether the relaxed field may attain a taylor state ; but under the assumed conditions with axial and toroidal flux conserved inside every cylindrical lagrangian surface , this is not possible .
the effect of an additional @xmath2-effect associated with instabilities and turbulence in the plasma is therefore investigated in exploratory manner .
an assumed pseudo - scalar form of @xmath2 proportional to @xmath3 is adopted , where @xmath4 and @xmath5 is an @xmath6 dimensionless parameter .
it is shown that , when @xmath5 is less that a critical value @xmath7 , the evolution remains smooth and similar to that for @xmath8 ; but that if @xmath9 , negative - diffusivity effects act on the axial component of @xmath10 , generating high - frequency rapidly damped oscillations and an associated transitory appearance of reversed axial field .
however , the scalar quantity @xmath11 remains highly non - uniform , so that again the field shows no sign of relaxing to a taylor state for which @xmath12 would have to be constant .
_ keywords _ : magnetic relaxation ; taylor conjecture ; @xmath2-effect ; reversed - field pinch |
roughly speaking , a combinatorial object is called _ quasirandom _ if it has properties that a random object has asymptotically almost surely .
this notion has been defined for various structures such as tournaments @xcite , set systems @xcite , subsets of @xmath10 @xcite , @xmath11-uniform hypergraphs @xcite , groups @xcite , etc . in particular , quasirandomness has been extensively studied for graphs . extending earlier results of rdl @xcite and thomason @xcite , chung ,
graham and wilson @xcite gave seven equivalent properties of graph sequences such that the sequence of random graphs @xmath12 possesses them with probability one .
these properties include densities of subgraphs , values of eigenvalues of the adjacency matrix or the typical size of the common neighborhood of two vertices .
in particular , it follows from the results in @xcite that if the density of @xmath8-vertex subgraphs in a large graph is asymptotically the same as in @xmath13 , then this is true for every fixed subgraph .
graham ( see ( * ? ? ?
* page 141 ) ) asked whether a similar phenomenon also occurs in the case of permutations .
let us state his question more precisely .
let @xmath14 consist of permutations on @xmath15:=\{1,\dots , k\}$ ] .
we view each @xmath16 as a bijection @xmath17\to [ k]$ ] and call @xmath18 its _
length_. for @xmath16 and @xmath19 with @xmath20 , let @xmath3 be the probability that a random @xmath11-point subset @xmath21 of @xmath22 $ ] induces a permutation _ isomorphic _ to @xmath0 ( that is , @xmath23 iff @xmath24 where @xmath21 consists of @xmath25 ) .
a sequence @xmath5 of permutations has _ property @xmath26 _
if @xmath6 and @xmath27 for every @xmath28 . it is easy to see that @xmath29 implies @xmath26 .
graham asked whether there exists an integer @xmath30 such that @xmath31 implies @xmath26 for every @xmath11 .
here we answer this question : [ th : main ] property @xmath32 implies property @xmath26 for every @xmath11 .
it is trivial to see that @xmath33 and an example that @xmath34 can be found in @xcite .
an unpublished manuscript of cooper and petrarka @xcite shows that @xmath35 and mentions that chung could also show this ( as early as 2001 ) .
being unaware of @xcite , we found yet another example that @xmath35 .
since it is quite different from the construction in @xcite , we present it in section [ sect - counter ] .
since these notions deal with properties of sequences of permutations , we find it convenient to operate with an appropriately defined `` limit object '' , analogous to that for graphs introduced by lovsz and szegedy @xcite .
here we use the analytic aspects of permutation limits that were studied by hoppen et al .
@xcite and we derive theorem [ th : main ] from its analytic analog ( theorem [ thm - main ] ) . let the ( normalized ) _ discrepancy _ @xmath36 of @xmath37 be the maximum over intervals @xmath38 $ ] of @xmath39 cooper @xcite calls a permutation sequence @xmath5 _ quasirandom _ if @xmath6 and @xmath40 .
he also gives other equivalent properties ( ( * ? ? ?
* theorem 3.1 ) ) and he discusses various applications of `` random - like '' permutations . using the results of @xcite , it is not hard to relate quasirandomness and properties @xmath26 : [ equiv ] a sequence @xmath5 of permutations is quasirandom if and only if it satisfies property @xmath26 for every @xmath11 .
the proof of proposition [ equiv ] can be found in section [ proofofprop ] .
thus our theorem [ th : main ] implies that @xmath32 alone is equivalent to quasirandomness .
finally , let us remark that mckay , morse and wilf ( * ? ? ?
* page 121 ) also defined a notion of quasirandomness for permutations . their definition ,
although related , is different from that of cooper as it deals with sequences of _ sets _ of permutations .
here we define convergence of permutation sequences and show how a convergent sequence can be associated with an analytic limit object .
we refer the reader to @xcite for more details .
let @xmath41 consist of probability measures @xmath42 on the borel @xmath43-algebra of @xmath44 ^ 2 $ ] that have _ uniform marginals _ , that is , @xmath45)=\mu([0,1]\times a)=\lambda(a)$ ] for every borel set @xmath46 $ ] , where @xmath47 is the lebesgue measure on @xmath44 $ ] .
fix some @xmath48 .
let @xmath49 for @xmath50 $ ] be independent random variables with @xmath51 ( that is , each @xmath52 has distribution @xmath42 ) .
we view an outcome @xmath53 as an element of @xmath44^{2k}$ ] . for permutations
@xmath54 , let @xmath55^{2k}$ ] correspond to the event that @xmath56 ( for example , the first statement above is equivalent to @xmath57 . )
since each of the vectors @xmath58 and @xmath59 is uniformly distributed over @xmath44^k$ ] , the probability of the _ degenerate event _
@xmath60^{2k}\ ] ] is zero .
note that the sets @xmath61 for @xmath54 partition @xmath44^{2k}\setminus d_k$ ] .
if we reorder the indices in an outcome @xmath62^{2k}\setminus d_k$ ] so that @xmath63 , then the new relative order on @xmath64 $ ] defines a random permutation @xmath65 . in other words , if we land in @xmath61 , then we set @xmath66 .
let the _ density @xmath67 _ of @xmath16 be the probability that @xmath68 .
equivalently , @xmath69 where the last equality uses the fact that @xmath70 does not depend on @xmath71 ( because @xmath72 are independent and identically distributed ) .
a sequence of permutations @xmath5 is _ convergent _ if @xmath6 and @xmath73 converges for every permutation @xmath0 .
this is the same definition of convergence as the one in @xcite except we additionally require that @xmath6 ; cf .
* claim 2.4 ) .
it is easy to show that every sequence of permutations whose lengths tend to infinity has a convergent subsequence ; see e.g. ( * ? ? ?
* lemma 2.11 ) .
furthermore , for every convergent sequence @xmath5 there is @xmath48 such that for every permutation @xmath0 we have @xmath74 for the reader s convenience , we sketch the proof from @xcite that @xmath42 exists . for @xmath16 ,
let @xmath75 be obtained by dividing the square @xmath44 ^ 2 $ ] into @xmath76 equal squares and distributing the mass uniformly on the squares with indices @xmath77 , @xmath78 . by prokhorov s theorem , @xmath79 has a subsequence that weakly converges to some measure @xmath42 .
we have @xmath48 as this set is closed in the weak topology .
finally , @xmath42 satisfies ( [ eq : limit ] ) because , for any fixed @xmath0 , the function @xmath80 is continuous in the weak topology and @xmath81 .
we remark that hoppen et al .
@xcite proposed a slightly different limit object : the regular conditional distribution function of @xmath82 with respect to @xmath21 , where @xmath83 .
lemma 2.2 and definition 2.3 in @xcite show how to switch back and forth between the two objects .
now , we are ready to state the analytic version of theorem [ th : main ] . let us call @xmath48
_ @xmath11-symmetric _ if @xmath84 for every @xmath16 .
[ thm - main ] every @xmath8-symmetric @xmath48 is the ( uniform ) lebesgue measure on @xmath44 ^ 2 $ ] . in particular , @xmath42 is @xmath11-symmetric for every @xmath11 .
let us show how theorem [ thm - main ] implies theorem [ th : main ] .
suppose on the contrary that some @xmath5 satisfies @xmath32 but not @xmath26 .
fix @xmath16 and a subsequence @xmath85 such that @xmath86 exists and is not equal to @xmath87 .
consider now a convergent subsequence @xmath88 of @xmath85 and let @xmath48 be its limit . by ( [ eq : limit ] )
, @xmath42 is 4-symmetric and , by theorem [ thm - main ] , @xmath42 is @xmath30-symmetric for every @xmath30 .
but then @xmath89 , which is the desired contradiction .
in this section , let @xmath48 be arbitrary with @xmath90 for every @xmath91 .
let @xmath92 denote the uniform measure on @xmath44 ^ 2 $ ] .
our objective is to show that @xmath93 .
let @xmath94 and @xmath95 be independent . for brevity ,
let us abbreviate @xmath96 ^ 2}$ ] to @xmath97 .
define a function @xmath98 ^ 2\to [ 0,1]$ ] by @xmath99\times [ 0,b])=\int_{v\le ( a , b ) } { \hspace{3pt}\mathrm{d}}v,\ ] ] where @xmath100 means that @xmath101 and @xmath102 .
since @xmath42 has uniform marginals , the function @xmath103 is continuous .
first , we show that the 4-symmetry of @xmath42 uniquely determines certain integrals .
[ lm - integrals ] @xmath104 let @xmath105 , for @xmath106 , be independent random variables distributed according to @xmath42 . by fubini s theorem , we have @xmath107 where @xmath108 ^ 6 $ ] . note that @xmath109 where @xmath110 is defined by ( [ eq : degenerate ] ) and the union is over @xmath111 such that @xmath112 .
the @xmath8-symmetry of @xmath42 and ( [ eq : density ] ) imply that @xmath113 for every @xmath114 and @xmath54 .
since @xmath115 , we have @xmath116 , as required .
likewise , @xmath117 where @xmath118 ^ 8 $ ] corresponds to the event that @xmath119 , @xmath120 and @xmath121 .
one can derive ( [ eq : int2 ] ) by replacing each factor by an integral ( for example , @xmath21 is replaced by @xmath122 ) and applying fubini s theorem .
the integral in the right - hand side of ( [ eq : int2 ] ) is equal to the @xmath123-measure of the union of @xmath61 over some ( explicit ) set of pairs @xmath124 .
the measure of this set is uniquely determined by the 4-symmetry of @xmath42 .
thus the integral does not change if we replace @xmath42 by any other 4-symmetric measure .
considering the uniform measure @xmath47 , we obtain @xmath125 , as required .
next , observe that @xmath126 is uniformly distributed in @xmath44 ^ 2 $ ] because @xmath127 and @xmath128 are independent and have uniform marginals .
again , the value of @xmath129 ^ 4 } f(x_1,y_2)^2 { \hspace{3pt}\mathrm{d}}(v_1,v_2)=\int_{v_3,v_4\le ( x_1,y_2 ) } { \hspace{3pt}\mathrm{d}}(v_1,\dots , v_4),\ ] ] does not depend on the choice of @xmath42 and can be easily computed by taking @xmath93 .
since @xmath21 is uniformly distributed in @xmath44 $ ] , we have @xmath130 . also , @xmath131 we use the above identities and apply the cauchy - schwartz inequality twice to get the following series of inequalities : @xmath132 thus we have equality throughout
. however , the last inequality is equality if and only if @xmath133 is equal to a fixed multiple of @xmath134 almost everywhere with respect to the uniform measure @xmath47 .
since @xmath103 is continuous and @xmath135 , we conclude that @xmath136 for all @xmath137 ^ 2 $ ] . thus the measures @xmath42 and @xmath47 coincide on all rectangles @xmath138\times [ 0,b]$ ] .
since these rectangles generate the borel @xmath43-algebra on @xmath44 ^ 2 $ ] , we have that @xmath93 by the uniqueness statement of the carathodory theorem .
this proves theorem [ thm - main ] .
our proof gives other sufficient conditions for @xmath93 .
for example , it suffices to require that each of the three integrals of lemma [ lm - integrals ] is @xmath139 .
the proof of the lemma shows that , if desired , these integrals can be expressed as linear combinations of densities @xmath67 for @xmath91 .
the single identity @xmath140 is also sufficient for proving that @xmath93
; however , if written as a polynomial in terms of permutation densities ( by mimicking the proof of lemma [ lm - integrals ] ) , it involves @xmath141-point permutations
. our method can give other sufficient conditions in this manner ; the choice of which one to use may depend on the available information about the sequence .
[ rm : unique ] also , the argument of lemma [ lm - integrals ] shows that , for every polynomial @xmath142 and @xmath48 , the value of @xmath143 can be expressed as a linear combination of permutation densities .
this observation combined with the stone - weierstrass theorem gives the _ uniqueness _ of a permutation limit : if @xmath144 have the same permutation densities , then @xmath145 ( cf .
* theorem 1.7 ) ) .
first , we construct a @xmath147-symmetric measure @xmath48 which is not 4-symmetric . for @xmath148 $ ] ,
let @xmath149 be the set of all the points @xmath150 ^ 2 $ ] such that @xmath151 or @xmath152 .
see figure [ fig - z ] for illustrations of this definition .
define @xmath153 for @xmath148 $ ] to be the permutation limit such that the mass is uniformly distributed on @xmath149 . because of the symmetries of @xmath154 ( invariance under the horizontal and vertical reflections ) , we have that @xmath155 for every @xmath156 if and only if @xmath157 , where @xmath158 is the identity 3-point permutation .
routine calculations show that @xmath159 and @xmath160 .
since @xmath161 is continuous in @xmath162 , there exists @xmath163 $ ] such that @xmath164 .
moreover , @xmath165 is not 4-symmetric .
this can be verified directly ; it also follows from theorem [ thm - main ] since @xmath165 is not the uniform measure .
take a sequence @xmath5 of permutations that converges to @xmath165 .
for example , the random sequence @xmath166 has this property with probability one , see ( * ? ? ?
* corollary 4.3 ) .
any such sequence @xmath5 satisfies @xmath146 but not @xmath32 .
[ rm : transform]there are other ways how one can get an example of a 3-symmetric non - uniform measure by transforming @xmath167 into @xmath168 . for example , for @xmath169 , let @xmath170 assign measure @xmath162 to @xmath167 and measure @xmath171 to @xmath168 with the conditional distributions being equal to @xmath172 and @xmath173 . again by continuity , there is @xmath162 such that @xmath174 is 3-symmetric .
[ rm : inflatable]let us call a permutation @xmath175 _ @xmath11-inflatable _ if @xmath176 and @xmath177 is @xmath11-symmetric , where @xmath178 is the measure associated with @xmath0 as is described after ( [ eq : limit ] ) .
cooper and petrarka @xcite discovered many @xmath147-inflatable permutations by computer search , thus giving examples that @xmath179 .
the results in @xcite show that a shortest 3-inflatable permutation has length @xmath180 and that @xmath181 has exactly four @xmath147-inflatable permutations : @xmath182 , @xmath183 , and their vertical reflections .
clearly , our theorem [ thm - main ] implies that no @xmath8-inflatable permutation can exist .
in particular , this proves ( in a stronger form ) conjecture 3 in @xcite that no 4-inflatable permutation with certain properties exists .
let @xmath5 be an arbitrary sequence of permutations with @xmath6 .
let @xmath184 be the measure associated with @xmath185 as is described after ( [ eq : limit ] ) .
it is straightforward to verify that @xmath186 , where @xmath187 denotes the _ discrepancy _ of @xmath48 , with the supremum ( in fact , it is maximum ) being taken over intervals @xmath188 $ ] . also , it is not hard to show ( cf .
remark [ rm : unique ] ) that @xmath5 converges to @xmath42 if and only if @xmath189 weakly converges to @xmath42 .
first , suppose that @xmath5 satisfies @xmath26 for each @xmath11 .
this means that @xmath5 converges to the uniform limit @xmath47 . for @xmath190
$ ] , let @xmath191\times [ 0,b])$ ] and @xmath192 . since @xmath193 and @xmath194\times[b_1,b_2])=f_j(a_2 , b_2)-f_j(a_1,b_2)-f_j(a_2,b_1)+f_j(a_1,b_1),\ ] ]
we conclude that @xmath195 .
the weak convergence @xmath196 of measures in @xmath41 gives that @xmath197 pointwise .
since @xmath103 and each function @xmath198 , defined on the compact space @xmath44 ^ 2 $ ] , are continuous and monotone in both coordinates , this implies that @xmath199 ( alternatively , ( [ eq : fn ] ) directly follows from ( * ? ? ?
* lemma 5.3 ) . )
thus @xmath200 and @xmath5 is quasirandom .
next suppose that @xmath40 .
one way to establish property @xmath26 is to use one of the equivalent definitions of quasirandomness from ( * ? ? ?
* theorem 3.1 ) ( namely property [ ms ] ) . alternatively ,
if @xmath26 fails , then ( by passing to a subsequence ) we can assume that @xmath5 converges to some @xmath48 with @xmath201 .
however , we have that @xmath202 , which implies @xmath93 , contradicting our assumption .
this finishes the proof of proposition [ equiv ] .
the theory of flag algebras developed by razborov @xcite can be applied to permutation limits : a permutation @xmath203 is viewed as two binary relations , each giving a linear order on @xmath204 .
for example , lemma [ lm - integrals ] can be stated and proved within the flag algebra framework .
this view has been helpful for us when developing our proof .
a graph can be associated with a permutation @xmath175 as follows : let @xmath205 be the graph on @xmath206 $ ] with vertices @xmath207 adjacent if @xmath208 .
fix @xmath48 and sample a random permutation @xmath209 .
define a function @xmath210 ^ 4\to\{0,1\}$ ] by @xmath211 if we have @xmath212 or @xmath213 componentwise and let @xmath214 otherwise .
in other words , @xmath215 is the indicator function of the event that @xmath216 is the identity 2-point permutation .
clearly , @xmath217 can be generated by sampling independently points @xmath218 ^ 2 $ ] , each with distribution @xmath42 , and connecting those @xmath219 $ ] for which @xmath220 . the latter procedure corresponds to generating a random sample @xmath221 , where @xmath210 ^ 2\times [ 0,1]^2\to[0,1]$ ] is viewed as a graphon represented on borel subsets of @xmath44 ^ 2 $ ] with measure @xmath42 , see ( * ? ? ?
* section 2.6 ) for details .
lovsz and ss @xcite and lovsz and szegedy @xcite presented various sufficient conditions for a graphon @xmath215 to be _ finitely forcible _ which , in the above notation , means that there is @xmath30 such that the distribution of @xmath222 uniquely determines that of @xmath223 for every @xmath11 .
as far as we can see , none of these conditions directly applies to the graphon associated with the uniform measure @xmath92 .
since we answered graham s question on quasirandom permutations by other means , we did not pursue this approach any further .
we also refer the reader to hoppen et al .
* section 5.3 ) who discuss finite forcibility for permutation limits , being motivated by some questions in parameter testing .
the authors thank carlos hoppen , andrs mth , and the anonymous referee for helpful comments .
99 f. r. k. chung , r. l. graham : quasi - random hypergraphs , random struct .
algorithms 1 ( 1990 ) 105124 .
f. r. k. chung , r. l. graham : quasi - random tournaments , j. graph theory 15 ( 1991 ) , 173198 .
f. r. k. chung , r. l. graham : quasi - random set systems , j. amer . math .
soc . 4 ( 1991 ) , 151196 . f. r. k. chung , r. l. graham : quasi - random subsets of @xmath224 , j. combin
. theory ser .
a 61 ( 1992 ) , 6486
. f. r. k. chung , r. l. graham , r. m. wilson : quasi - random graphs , combinatorica 9 ( 1989 ) , 345362
. j. n. cooper : quasirandom permutations , j. combin .
theory ser . a 106 ( 2004 ) , 123143 .
j. n. cooper , a. petrarka : symmetric and asymptotically symmetric permutations , available as arxiv:0801.4181 . w. t. gowers : quasirandomness , counting and regularity for 3-uniform hypergraphs , combin .
15 ( 2006 ) , 143184 .
w. t. gowers : hypergraph regularity and the multidimensional szemerdi theorem , annals of math .
( 2 ) 166 ( 2007 ) , 897946 .
w. t. gowers : quasirandom groups , combin .
17 ( 2008 ) , 363387 .
j. haviland , a. g. thomason : pseudo - random hypergraphs , discrete math . 75 ( 1989 ) 255278 .
c. hoppen , y. kohayakawa , c. g. moreira , b. rth , r. m. sampaio : limits of permutation sequences , j. combin .
theory ser .
b 103 ( 2012 ) 93113 .
c. hoppen , y. kohayakawa , c. g. moreira , r. m. sampaio : testing permutation properties through subpermutations , theoretical computer science 412 ( 2011 ) 35553567 . c. hoppen , y. kohayakawa , c. g. moreira , r. m. sampaio : limits of permutation sequences through permutation regularity , available as arxiv:1106.1663 .
y. kohayakawa , v. rdl , j. skokan : hypergraphs , quasi - randomness , and conditions for regularity , j. combin .
theory ser .
a 97 ( 2002 ) 307352 .
l. lovsz , v. ss : generalized quasirandom graphs , j. combin .
theory ser .
b 98 ( 2008 ) , 146163 .
l. lovsz , b. szegedy : limits of dense graph sequences , j. combin .
theory ser .
b 96 ( 2006 ) , 933957 .
l. lovsz , b. szegedy : finitely forcible graphons , j. combin .
theory ser .
b 101 ( 2011 ) , 269301 .
b. d. mckay , j. morse , h. s. wilf : the distributions of the entries of young tableaux , j. combin .
theory ser .
a 97 ( 2002 ) 117128 .
a. razborov : flag algebras , j. symbolic logic 72 ( 2007 ) , 12391282
. v. rdl : on the universality of graphs with uniformly distributed edges , discrete math . 59 ( 1986 ) , 125134 .
a. thomason : pseudo - random graphs , ann . discrete math . 33 ( 1987 ) , 307331 | for permutations @xmath0 and @xmath1 of lengths @xmath2 , let @xmath3 be the probability that the restriction of @xmath1 to a random @xmath4-point set is ( order ) isomorphic to @xmath0 .
we show that every sequence @xmath5 of permutations such that @xmath6 and @xmath7 for every @xmath8-point permutation @xmath0 is _ quasirandom _ ( that is , @xmath9 for every @xmath0 ) .
this answers a question posed by graham . |
[ chap : intro ] graphene , a one - atom - thick allotrope of carbon @xcite , has attracted much attention of theoretical physicists , since its electronic excitations are described by two - dimensional ( 2d ) _ massless dirac fermion _ ( see , for instance , @xcite ) .
let @xmath4 be the fermi velocity .
then , for a free fermion , the 2d massless dirac equation is given by @xmath5 where @xmath6 , @xmath7 , @xmath8 , and @xmath9 is a two - component wave function with the particle position @xmath10 .
for a wave - number vector @xmath11 , the eigenvalues of energy are given by @xmath12 , @xmath13 , where the @xmath14 signs correspond to two values of the helicity of fermion .
the wave functions are determined as @xcite @xmath15 where @xmath16 is the volume of the system and @xmath17 although the basic description of the system is very simple as shown above , physics of graphene is extremely rich , since the 2d dirac equation is much modified by external electric and magnetic fields applied to the graphene plane as well as by changing geometric and topological features of the 2d surface @xcite . as a matter of fact , an ultrathin membrane is unstable in the 3d real space and in suspended graphene and in graphene placed on a substrate nanometer - scale corrugations , which are simply called _ ripples _
@xcite , are observed @xcite . since spatial distribution of such microscopic structures on the surface can not be controlled , even when we apply uniform electric and/or magnetic fields , the rippled graphene will behave as a 2d system with randomly distributed scalar and vector potentials on it .
effect of ripples on electronic transport on graphene is an interesting subject both from theory and experiment @xcite .
recently , magnetotransport measurements on rippled graphene have been reported , where a magnetic field is applied parallel to the averaged 2d plane of the graphene @xcite ( see also @xcite for bilayer graphene ) .
if the surface is a perfect plane , such an in - plane magnetic field @xmath18 does not affect any motion of 2d electron , since electronic motion couples only to the component of magnetic field normal to the 2d plane .
but on real graphene , the magnetic field becomes to include a component normal to the surface depending on the local slope around ripples .
then the system has inhomogeneous out - of - plane magnetic fields .
with respect to electromagnetic property , we will be able to regard the rippled graphene with @xmath18 as a system of traveling 2d massless dirac fermions , which are randomly scattered by ripples @xcite .
in the present paper , we study a formula of lundeberg and folk for the magnetoresistance @xmath19 , which was derived from a model based on the boltzmann transport theory @xcite .
the relative height of ripple on the averaged 2d plane is denoted by @xmath20 for each 2d position @xmath10 .
we characterize the height distribution by the _ two - point height correlation _ @xmath21 which is defined as a statistical average of product of the height at @xmath22 and that at the position separated from @xmath22 by a displacement vector @xmath23 . for simplicity , we assume that the distribution of @xmath20 is translation invariant and isotropic in this paper . by this assumption , ( [ eqn : corr1 ] )
becomes independent of @xmath22 and it depends only on the distance of two points @xmath24 ; @xmath25 . as a function of wave - number @xmath13 of fermion
, we define @xmath26 where @xmath27 with the bessel function @xmath28 ( see appendix a for special functions @xcite ) .
the _ signed carrier - density _ of graphene is denoted by @xmath29 , which is the difference of the density of electrons @xmath30 and that of holes @xmath31 .
we assume that the temperature is low and the state is degenerated so that the wave - number of fermion @xmath32 is well - approximated by the fermi wave - number @xmath33 . with counting contributions from the two dirac points @xmath34 and @xmath35 and spin ,
@xmath36 is related with @xmath37 as ( see , for instance , sect .
ii.a in @xcite ) , @xmath38 where the absolute value of @xmath37 is used by the particle - hole symmetry ; @xmath39 .
let @xmath40 and @xmath41 be the angle between @xmath18 and the direction of the current @xmath42 along which the resistance @xmath43 is measured . set @xmath44 , the @xmath18-induced part of resistance , which we call _
magnetoresistance_. then the _ lundeberg - folk formula _ is given as @xmath45 remark that this formula is not given in the published paper @xcite of lundeberg and folk , but the corresponding equations are found in appendix of its arxiv version @xcite .
lundeberg and folk introduced a correlation length @xmath46 for ripple - height correlation @xmath1 and argued the asymptotic behavior of @xmath47 in @xmath48 and @xmath49 , and from the latter estimate the behavior of ( [ eqn : lf1 ] ) in high carrier - density region was predicted as @xmath50 .
this @xmath51-law was emphasized in their paper and has been used for analyzing experimental data @xcite , while it is an asymptotic law in @xmath52 for a special case with the gaussian height - correlation of ripples as clarified below in the present paper .
here we call the general equation ( [ eqn : lf1 ] ) the lundeberg - folk formula . the main purpose of the present paper is to show that , for the two cases @xmath53 with a variance of height @xmath54 , exact and explicit representations for the anisotropic magnetoresistance ( [ eqn : lf1 ] ) are obtained .
( for the gaussian correlated disorder , see @xcite . )
the results are the following .
let @xmath55 and @xmath56 be the gauss hypergeometric function and the confluent hypergeometric function , respectively .
( see appendix a for definitions and basic properties of special functions used in this paper . ) then , for ( i ) the exponential height - correlation , ( [ eqn : g1 ] ) is calculated as @xmath57 , \label{eqn : rhocase1}\end{aligned}\ ] ] where @xmath58 and @xmath59 are the complete elliptic integrals of the first kind and of the second kind , respectively ( see appendix a ) . for ( ii )
the gaussian height - correlation , ( [ eqn : g1 ] ) becomes @xmath60 e^{-(k r)^2/2 } , \label{eqn : rhocase2}\end{aligned}\ ] ] where @xmath61 is the modified bessel function of the first kind ( see appendix a ) . from these exact expressions , we can readily obtain the asymptotics of the magnetoresistance ( [ eqn : lf1 ] ) in high carrier - density region as follows . for @xmath52 , @xmath62 the asymptotic for the gaussian case ( ii ) shown in the second line in ( [ eqn : rhoasyma1 ] )
is exactly the same as eq.(3 ) reported in @xcite . on the other hand
, there is a logarithmic correction to the @xmath51-law in the case ( i ) with the exponential height - correlation of ripples . for the low carrier - density region ,
we have found the following behavior for the magnetoresistance ; when @xmath63 , @xmath64 that is , @xmath65 shows a cusp at the charge neutrality point @xmath66 and the dependence on @xmath37 around this point is linear @xmath67 . in the present paper
we apply our analytic results to the experimental data in the whole region of @xmath37 .
the result seems to be excellent if we assume the gaussian height - correlation of ripples .
see figures [ fig : fig2 ] and [ fig : fig3 ] in section [ chap : remark ] . in the vicinity of charge neutrality point
, electronic transport will be affected by strong density inhomogeneity in rippled graphene , which is called _ electron - hole puddle _ @xcite . moreover , in low carrier - density region , @xmath68 , we have @xmath69 and the semi - classical boltzmann transport theory is afraid to be broken - down .
to the present analysis corrections by more elaborate approximations or full quantum theory shall be added @xcite .
the paper is organized as follows . in section [ chap : lf ]
we give a derivation of the lundeberg - folk formula ( [ eqn : lf1 ] ) from a semi - classical model using the boltzmann transport theory .
section [ chap : proof ] is devoted to the proofs of our exact expressions ( [ eqn : rhocase1 ] ) and ( [ eqn : rhocase2 ] ) and their asymptotics ( [ eqn : rhoasyma1 ] ) and ( [ eqn : rhoasymb1 ] ) . in section [ chap : remark ] , we discuss comparison of the present analytic results with the experimental data measured by wakabayashi and his coworkers . appendix [ chap : appendixa ] is prepared for giving the mathematical formulas for special functions used in the text . in appendix [ chap :
appendixb ] an explanation is given for the linear relation between resistance and inverse of relaxation time based on the boltzmann transport theory .
[ chap : lf ] as remarked just below eq.([eqn : lf1 ] ) , the general expression of @xmath19 as a functional of the height - correlation function @xmath1 of ripples , ( [ eqn : lf1 ] ) with ( [ eqn : g1 ] ) and ( [ eqn : w1 ] ) , is found only in @xcite .
although the essence of its derivation is given there , the description is very brief . for convenience , here we would like to complete the derivation of the lundeberg - folk formula ( [ eqn : lf1 ] ) following @xcite .
we take the averaged 2d plane of graphene as the @xmath70-@xmath71 plane .
let @xmath20 be the height of ripple at @xmath10 . under the in - plane magnetic field in the @xmath70-direction ,
@xmath72 the _ geometric vector potential _ of berry @xcite , @xmath73 , is calculated as @xcite @xmath74 it gives a potential expressed by a @xmath75 complex matrix for dirac fermions with electric charge @xmath76 and the fermi velocity @xmath4 , @xmath77 where @xmath78 are pauli matrices .
the scattering matrix for an interaction operator @xmath79 is calculated by the fermi golden rule as @xmath80 where @xmath81 .
we note that @xmath82 with @xmath83 where the potential @xmath84 is given by ( [ eqn : u1 ] ) , @xmath85 , and @xmath86 . by using the two - component plane wave solution ( [ eqn : dirac2 ] ) for @xmath87 , ( [ eqn : u2 ] )
becomes @xmath88 with @xmath89 and @xmath90 the square of ( [ eqn : hq1 ] ) , @xmath91 , is then calculated as @xmath92 where we have renamed the integration variable as @xmath93 and @xmath94 .
the quantity involving the second integral in the second line of ( [ eqn : hq2 ] ) , @xmath95 , is a volume average of a product @xmath96 with respect to @xmath22 over the system .
if the distribution of the ripple height @xmath20 on graphene is translation invariant , this volume average will converge to the height correlation function ( [ eqn : corr1 ] ) as @xmath97 , by the self - averaging property of usual random systems .
moreover , it can be written as @xmath1 , if the system is also isotropic .
therefore , provided that @xmath16 is large enough , ( [ eqn : hq2 ] ) is equal to the fourier transform of @xmath1 @xmath98 multiplied by the volume @xmath16 of the system ; @xmath99 . inserting the above results into ( [ eqn : s1 ] ) , the volume factor @xmath16 is cancelled out as desired , and we obtain the following expression of scattering matrix , @xmath100 which was given as eq.(a.1 ) in @xcite . by definition of @xmath101 for @xmath11 given by ( [ eqn : dirac3 ] ) and its analogue @xmath102 for @xmath103 , under the energy conservation @xmath104 , @xmath105 , @xmath106 , and thus @xmath107 then , when @xmath89 with the condition @xmath104 , ( [ eqn : cq1 ] ) becomes @xmath108 .
\label{eqn : cq2}\end{aligned}\ ] ] lundeberg and folk introduced a semi - classical model based on the boltzmann transport theory for the system , in which the 2d electronic state with a wave - number vector @xmath109 is described by probability distribution @xmath110 .
the time evolution of @xmath110 follows the boltzmann equation ( see , for example , @xcite ) @xmath111_{\rm s } , \label{eqn : boltzmann0}\ ] ] in the external electric field @xmath112 , where we have assumed the spatial homogeneity of distribution @xmath113 and omitted the term @xmath114 in the lhs .
the ` collision term ' in the rhs , which describes the scattering of particles or holes by random vector potentials caused by ripples in the present system , is given by @xmath115_{\rm s } = d[s , f](\k ) \label{eqn : boltzmann}\ ] ] with the operator @xmath116(\k ) = \int \frac{d^2 k'}{(2\pi)^2 } s(\k , \k ' ) \big [ f(\k')-f(\k ) \big ] \label{eqn : d1}\ ] ] where @xmath117 is the total scattering matrix . here
@xmath118 is the zero field scattering matrix assumed to be in the isotropic form , @xmath119 , and @xmath120 is given by ( [ eqn : s2 ] ) .
we set @xmath121 and rewrite ( [ eqn : d1 ] ) as @xmath116(\k ) = \frac{k}{(2\pi)^2 } \int_{-\pi}^{\pi } d \theta_{k ' } \ , \widetilde{s}(\k , \k ' ) \big [ \widetilde{f}(\theta_{k ' } ) -\widetilde{f}(\theta_k ) \big ] , \label{eqn : d2}\ ] ] where @xmath122 shows the dependence of @xmath110 on the angular component @xmath101 of @xmath109 , when @xmath13 is given . in order to study @xmath101-dependence of scattering induced by @xmath18 , which is imposed in the @xmath70-direction
, we introduce the following orthogonal bases , @xmath123 satisfying @xmath124 the @xmath18-induced parts of scattering rates , which give the inverses of transport times enhanced by @xmath18 , are obtained as @xmath125(\theta_k ) \label{eqn : tau1}\ ] ] for @xmath126 , by the orthogonality relation ( [ eqn : orth ] ) of the two bases ( [ eqn : f1 ] ) . now combining ( [ eqn : tau1 ] ) and ( [ eqn : d2 ] ) with ( [ eqn : s2 ] ) and ( [ eqn : cq2 ] ) , we have multiple - integral expressions for @xmath127 such that integrals are performed with respect to @xmath128 and @xmath129 .
change the integral variables as @xmath130 , @xmath131 , and let @xmath132 \sin^2 \frac{\theta_+}{2 } \nonumber\\ & & \times \widetilde{f}_{\alpha}\left ( \frac{\theta_+ + \theta_-}{2 } \right ) \left [ \widetilde{f}_{\beta}\left ( \frac{\theta_+ - \theta_-}{2 } \right ) -\widetilde{f}_{\beta } \left ( \frac{\theta_+ + \theta_-}{2 } \right ) \right ] , \label{eqn : l1}\end{aligned}\ ] ] @xmath133 .
then @xmath134 by the boltzmann transport theory ( see appendix b for an explanation ) , the resistances induced by @xmath18 are related with @xmath135 as @xmath136 we performed the integrals ( [ eqn : l1 ] ) and obtained the results @xmath137 where @xmath28 is the bessel function with index @xmath138 ( see appendix a ) . then ( [ eqn : rhob1 ] ) with ( [ eqn : tau2 ] ) gives @xmath139 where @xmath47 is given by ( [ eqn : g1 ] ) with ( [ eqn : w1 ] ) .
the ` threefold relation ' between @xmath140 and @xmath141 is found in ( [ eqn : rhob2a ] ) @xcite .
let @xmath142 be the in - plane electric field applied in order to measure resistance .
if the current is written as @xmath143 , then the resistance coefficients @xmath144 , @xmath145 , are defined as @xmath146 in the isotropic system , @xmath147 as in ( [ eqn : rhob2b ] ) , then @xmath148 , @xmath149 .
we have chosen the direction of @xmath18 in the @xmath70-direction as ( [ eqn : b1 ] ) . then
, if we denote the angle between @xmath18 and @xmath42 as @xmath41 , @xmath150 with @xmath151 and the @xmath18-induced resistance @xmath19 for the current @xmath42 is given by @xmath152 then by the threefold relation ( [ eqn : rhob2a ] ) between @xmath140 and @xmath141 , the lundeberg - folk formula ( [ eqn : lf1 ] ) is obtained .
[ chap : proof ] let @xmath153 and use the formula ( [ eqn : jnu1 ] ) for @xmath138 . then ( [ eqn : g1 ] ) with ( [ eqn : w1 ] ) is written as @xmath154 with @xmath155 we apply the duplication formula ( [ eqn : gamma2 ] ) of gamma functions and use the pochhammer symbol ( [ eqn : poch ] ) , then we have @xmath156 which prove the first equality of ( [ eqn : rhocase1 ] ) by the definition of the gauss hypergeometric function ( [ eqn : f1 ] ) . from the recurrence relations ( [ eqn : g_rec1 ] ) and
( [ eqn : g_rec2 ] ) , we obtain the following two equalities , @xmath157 , \nonumber\\ & & f \left ( \frac{1}{2 } , \frac{3}{2},1 ; -(2kr)^2 \right ) = \frac{1}{1+(2kr)^2 } f \left ( -\frac{1}{2 } , \frac{1}{2},1 ; -(2kr)^2 \right ) .
\label{eqn : f1rec1}\end{aligned}\ ] ] combining them gives @xmath158 .
\label{eqn : f1rec2}\end{aligned}\ ] ] by the formulas ( [ eqn : ke2 ] ) , which express the complete elliptic integrals @xmath58 and @xmath59 by using the gauss hypergeometric functions , the second equality of ( [ eqn : rhocase1 ] ) is proved .
let @xmath159 and use the formula ( [ eqn : jnu1 ] ) for @xmath138 .
then ( [ eqn : g1 ] ) with ( [ eqn : w1 ] ) is written as @xmath160 where @xmath161 and @xmath162 is given as in ( [ eqn : ms ] ) . following the similar procedure mentioned above
, we have @xmath163 which proves the first equality of ( [ eqn : rhocase2 ] ) by the definition of the confluent hypergeometric function ( [ eqn : f2 ] ) . from the recurrence relations ( [ eqn : c_rec1 ] ) and
( [ eqn : c_rec2 ] ) , we obtain the following equality , @xmath164 by using this equality for @xmath165 and @xmath166 , we can rewrite the first line of ( [ eqn : rhocase2 ] ) as @xmath167 .
\label{eqn : g_b3}\ ] ] by the formula ( [ eqn : inu2 ] ) , which expresses the modified bessel function @xmath61 by using the confluent hypergeometric functions , the second equality of ( [ eqn : rhocase2 ] ) is proved .
_ case ( i ) : the exponential height - correlation _ 0.3 cm in gauss integral representation ( [ eqn : gaussint ] ) , if we change the variable as @xmath168 , we have @xmath169 then we obtain the equality @xmath170 .
\label{eqn : asymf1}\end{aligned}\ ] ] by the asymptotics of the complete elliptic integrals
@xcite @xmath171 we have the behavior @xmath172 , \quad \mbox{as $ k \to \infty$}. \label{eqn : asymf2}\ ] ] through the relation ( [ eqn : kn ] ) between @xmath32 and @xmath37 , the lundeberg - folk formula ( [ eqn : lf1 ] ) with the present result ( [ eqn : rhocase1 ] ) gives the first line of the rhs of ( [ eqn : rhoasyma1 ] ) . 0.3 cm _ case ( ii ) : the gaussian height - correlation _ 0.3 cm in this case
, we use the asymptotic expansion formula of the confluent hypergeometric function , @xmath173 it gives @xmath174 and we obtain the second line of the rhs of ( [ eqn : rhoasyma1 ] ) .
this result is exactly the same as eq.(3 ) reported in @xcite .
_ case ( i ) : the exponential height - correlation _ 0.3 cm by using the first two terms of the series ( [ eqn : f1 ] ) defining the gauss hypergeometric function , we obtain from ( [ eqn : rhocase1 ] ) @xmath175 .
\label{eqn : g_exp1}\ ] ] the first line of the rhs of ( [ eqn : rhoasymb1 ] ) is concluded from ( [ eqn : g_exp1 ] )
. 0.3 cm _ case ( ii ) : the gaussian height - correlation _ 0.3 cm by using the first two terms of the series ( [ eqn : f2 ] ) defining the confluent hypergeometric function , we obtain from ( [ eqn : rhocase2 ] ) @xmath176 .
\label{eqn : g_exp2}\ ] ] it gives the second line of the rhs of ( [ eqn : rhoasymb1 ] ) .
[ chap : remark ] the lundeberg - folk formula ( [ eqn : lf1 ] ) shows that the magnetoresistance induced by @xmath18 is proportional to @xmath0 , @xmath177 the coefficient @xmath178 is given by @xmath179 in which the function @xmath180 depends on the height - correlation function @xmath1 of ripples on graphene . in the present paper , we have clarified the dependence of @xmath19 on @xmath1 for the two cases , ( i ) @xmath153 ( exponential ) and ( ii ) @xmath181 ( gaussian ) , where @xmath182 is the variance of height of ripples and @xmath3 indicates the correlation length of ripples . in the low carrier - density region ,
the so - called electron - hole puddles appear on the rippled graphene and the homogeneity assumption of system will be invalid . in order to handle the inhomogeneity of the carrier density around the dirac point
, the present boltzmann approach should be corrected including higher - level approximations @xcite .
away from the charge neutrality point , however , the semi - classical modeling based on the boltzmann transport theory works well to describe the electronic transport properties of graphene .
here we consider our analytic expressions for @xmath19 as ` trial interpolation formulas ' connecting two high - carrier - density regimes , @xmath52 ; the positive - charge regime ( @xmath183 ) and the negative one ( @xmath184 ) .
our expressions for @xmath19 reported in this paper include two parameters @xmath2 and @xmath3 .
if experimental data are provided , we can evaluate the standard deviation of ripple height @xmath185 and the correlation length of ripples @xmath3 by numerical fitting of the data to our analytic expressions ( [ eqn : rhocase1 ] ) or ( [ eqn : rhocase2 ] ) . of the lundeberg - folk formula ( [ eqn : lfb2 ] ) for @xmath186 with the present exact solution ( [ eqn : rhocase1 ] ) for the exponential height - correlation of ripples
the parameters are chosen as @xmath187 [ nm ] and @xmath188 [ nm ] by the best fitting to the experimental data given by wakabayashi and his coworkers ( indicated by dots ) .
[ fig : fig1 ] ] of the lundeberg - folk formula ( [ eqn : lfb2 ] ) for @xmath186 with the present exact solution ( [ eqn : rhocase2 ] ) for the gaussian height - correlation of ripples is drawn .
the parameters are chosen as @xmath189 [ nm ] and @xmath190 [ nm ] by the best fitting to the experimental data given by wakabayashi and his coworkers ( indicated by dots ) .
the curve fits the experimental data very well in the whole region of carrier - density @xmath37 .
[ fig : fig2 ] ] figures [ fig : fig1 ] and [ fig : fig2 ] show results of numerical fitting of the lundeberg - folk formula ( [ eqn : lf1 ] ) with our exact solutions ( [ eqn : rhocase1 ] ) and ( [ eqn : rhocase2 ] ) for the exponential and the gaussian height - correlations to the experimental data , respectively , when @xmath186 in the both cases .
the experimental data are given by wakabayashi and his coworkers @xcite .
as we can see in these figures , the fitting in the high carrier - density region is more excellent in the gaussian case ( fig.[fig : fig2 ] ) than in the exponential case ( fig.[fig : fig1 ] ) . in the vicinity of the charge neutrality point ( @xmath66 ) .
dots are the experimental data given by wakabayashi and his coworkers .
the bold curve showing a cusp at @xmath66 is the present analytic solution ( [ eqn : rhocase2 ] ) for the gaussian height - correlation with @xmath189 [ nm ] and @xmath190 [ nm ] . in this low carrier - density region , the @xmath51-law shown by dotted curves is invalid , since it diverges as @xmath191 . in the very vicinity of the point @xmath66 ,
the simple expression given in the second line of the rhs of ( [ eqn : rhoasymb1 ] ) will work well , which is shown by thin black lines .
[ fig : fig3 ] ] for the vicinity of the charge neutrality point , @xmath63 , the fitting result for the gaussian height - correlation is shown by fig .
[ fig : fig3 ] , in which the @xmath51-law for the high carrier - density region @xmath52 ( the second line of ( [ eqn : rhoasyma1 ] ) ) and the expression for @xmath63 ( the second line of ( [ eqn : rhoasymb1 ] ) ) are also plotted for comparison .
the obtained values of parameters by the present numerical fitting are @xmath189 [ nm ] and @xmath190 [ nm ] , respectively .
these values should be compared with the direct experimental measurements @xcite . in particular , it was reported in @xcite the afm measurements gave @xmath192 [ nm ] and @xmath193 [ nm ] .
see also experimental values cited in @xcite from @xcite and others .
( we note that recent high - resolution measurements and analysis have revealed smaller - sized roughness of the substrate and graphene surfaces @xcite . )
we hope that the our expressions will be used for systematic analysis of experimental data also for different values of @xmath41 . in the present paper ,
we have assumed translation invariance and isotropy for distribution of ripple height . even in the high carrier - density regions ,
apart from the effect of electron - hole puddles , further calculation for general setting of ripple - height - correlations will be an interesting future problem .
* acknowledgements * the present authors would like to thank junichi wakabayashi for providing them experimental data of magnetoresistance measurements and for very useful discussion on the present work .
they also thank tohru kawarabayashi and mark lundeberg for helpful comments on the present study .
this work is supported in part by the grant - in - aid for scientific research ( c ) ( no.21540397 ) of japan society for the promotion of science .
[ chap : appendixa ] here we list the definitions and formulas of special functions used in the text ( see , for instance , @xcite ) .
the gamma function @xmath194 is defined by @xmath195 when @xmath196 , @xmath197 ( we set @xmath198 ) .
the following equality is called the duplication formula , @xmath199 the bessel function @xmath200 with index @xmath201 is given by @xmath202 for a complex number @xmath203 which is not negative real
. then the modified bessel function of the first kind is defined as @xmath204 for @xmath205 , the pochhammer symbol is defined as @xmath206 the gauss hypergeometric function is defined by @xmath207 and the confluent hypergeometric function is defined as @xmath208 the following recurrence relations for @xmath209 are known , @xmath210 = \alpha z f(\alpha+1 , \beta+1 , \gamma+1 ; z ) , \\ & & \alpha(1-z ) f(\alpha+1 , \beta , \gamma;z ) + [ \gamma-2 \alpha+(\alpha-\beta)z ] f(\alpha , \beta , \gamma;z ) \nonumber\\ \label{eqn : g_rec2 } & & \qquad \qquad \qquad \qquad \qquad + ( \alpha-\gamma ) f(\alpha-1 , \beta , \gamma;z)=0.\end{aligned}\ ] ] for @xmath211 , @xmath212 , \\ \label{eqn : c_rec2 } & & \alpha f(\alpha+1 , \gamma+1 ; z ) = ( \alpha-\gamma ) f(\alpha , \gamma+1 ; z ) + \gamma f(\alpha , \gamma ; z)\end{aligned}\ ] ] are satisfied .
the gauss hypergeometric function has the following integral expression , @xmath213 the complete elliptic integrals of the first kind , @xmath58 , and of the second kind , @xmath59 , are defined as @xmath214 respectively .
they are expressed by using the gauss hypergeometric functions as @xmath215 the modified bessel function ( [ eqn : inu1 ] ) is expressed by using the confluent hypergeometric functions as @xmath216
[ chap : appendixb ] let @xmath218^{-1}$ ] be the equilibrium fermi distribution function at temperature @xmath219 and chemical potential @xmath220 .
the relaxation time @xmath221 is phenomenologically introduced as @xmath222_{\rm s } = -\frac{\delta f(k)}{\tau(k ) } \quad \mbox{with } \quad \delta f(k)=f(k)-f_0(k ) .
\label{eqn : taua1}\ ] ] when we consider the long - term limit , @xmath223-term in the lhs is zero , and we have @xmath224 in low temperature limit , @xmath225 and @xmath226 , where @xmath33 denotes the fermi wave - number .
then in the 2d plane the current is calculated as follows , provided @xmath42 and @xmath112 are parallel ; @xmath227 since resistance @xmath43 is defined as @xmath228 , we have the linear relation between @xmath43 and @xmath217 as @xmath229 with the coefficient @xmath230 . in eq.([eqn : rhob1 ] ) @xmath33 is written as @xmath32 . 99 novoselov , k. s. , geim , a. k. , morozov , s. v. , jiang , d. , zhang , y. , dubonos , s. v. , grigorieva , i. v. , and firsov , a. a. : _ electric field effect in atomically thin carbon films_. science * 306 * , 666 - 669 ( 2004 ) stolyarova , e. , rim , k. t. , ryu , s. , maultzsch , j. , kim , p. , brus , l. e. , heinz , t. f. , hybertsen , m. s. , and flynn , g. w. : _ high - resolution scanning tunneling microscopy imaging of mesoscopic graphene sheets on an insulating surface_. proc . natl
104 * , 9209 - 9212 ( 2007 ) geringer , v. , liebmann , m. , echtermeyer , t. , runte , s. , schmidt , m. , rckamp , r. , lemme , m. c. , and morgenstern , m. : _ intrinsic and extrinsic corrugation of monolayer graphene deposited on sio@xmath231_. phys .
* 102 * , 076102 ( 2009 ) morozov , s. v. , novoselov , k. s. , katsnelson , m. i. , schedin , f. , ponomarenko , l. a. , jiang , d. , and geim , a. k. : _ strong suppression of weak localization in graphene_. phys .
lett . * 97 * , 016801 ( 2006 ) morozov , s. v. , novoselov , k. s. , katsnelson , m. i. , schedin , f. , elias , d. c. , jaszczak , j. a. , and geim , a. k. : _ giant intrinsic carrier mobilities in graphene and its bilayer_. phys .
* 100 * , 016602 ( 2008 ) rushforth , a. w. , gallagher , b. l. , main , p. c. , neumann , a. c. , henini , m. , marrows , c. h. , and hickey , b. j. : _ anisotropic magnetoresistance in a two - dimensional electron gas in a quasirandom magnetic field_. phys .
b * 70 * , 193313 ( 2004 ) cullen , w. g. , yamamoto , m. , burson , k. m. , chen , j. h. , jang , c. , li , l. , fuhrer , m. s. , and williams , e. d. : _ high - fidelity conformation of graphene to sio@xmath231 topographic features_. phys
. rev . lett . * 105 * , 215504 ( 2010 ) | application of an in - plane magnetic field to rippled graphene will make the system be a plane with randomly distributed vector potentials .
massless dirac fermions carrying charges on graphene are scattered by the vector potentials and magnetoresistance is induced proportional to the square of amplitude of in - plane magnetic field @xmath0 . recently ,
lundeberg and folk proposed a formula showing dependence of the magnetoresistance on carrier density , in which the coefficient of @xmath0 is given by a functional of the height - correlation function @xmath1 of ripples . in the present paper ,
we give exact and explicit expressions of the coefficient for the two cases such that @xmath1 is ( i ) exponential and ( ii ) gaussian .
the results are given using well - known special functions .
numerical fitting of our solutions to experimental data were performed .
it is shown that the experimental data are well - described by the formula for the gaussian height - correlation of ripples in the whole region of carrier density . the standard deviation @xmath2 of ripple height and the correlation length @xmath3 of ripples
are evaluated , which can be compared with direct experimental measurements . |
this note refers to the work @xcite , in which i have defined , via skein relation , an invariant of colored links .
let me recall some definitions .
an oriented link with @xmath0 components is a set of @xmath0 disjoint smooth oriented closed curves embedded in @xmath1 .
an oriented link is _ colored _
, if each component is provided with a color . in other words , a link is colored if a function @xmath2 is defined on the set of components to a finite set @xmath3 of colors .
let s call _ coloration _ such a function @xmath2 .
every coloration introduces an equivalence relation in the set of components : two components are equivalent if they have the same color .
let @xmath4 be the set of components of a link @xmath5 .
two colorations @xmath2 and @xmath6 of @xmath5 are said _ equivalent _ if there is a bijection in @xmath3 between the images @xmath7 and @xmath8 .
two colorations define the same partition of the set @xmath4 into classes of equivalence if and only if they are equivalent .
an invariant of colored links takes the same value on isotopic links with the same coloration , and may take different values on the same link with different colorations .
the invariant @xmath9 of colored links defined in @xcite takes the same value on isotopic links with equivalent colorations . in section 1 , the statement of the theorem defining the invariant @xmath9 is reported .
the invariant @xmath9 has three variables , say @xmath10 , @xmath11 and @xmath12 .
there are two natural ways to define an invariant of classical links by using @xmath9 .
the first way , is to consider a link as a colored link , whose components have all the same color . in this case
the invariant @xmath9 becomes independent of @xmath10 , and is in fact another instance of the homflypt polynomial .
in particular , it becomes the jones polynomial in the variable @xmath12 setting @xmath13 . given a classical link @xmath5 , we will denote @xmath14 the invariant @xmath9 on the colored link obtained by coloring @xmath5 with a sole color .
the second way , is to consider a link as a colored link , whose components have all different colors . in this case
@xmath9 yields an invariant which is stronger than the homflypt polynomial .
given a classical link @xmath15 with @xmath0 components , we will denote @xmath16 the invariant @xmath9 on the colored link obtained by coloring @xmath15 with n colors . among the 89 pairs of non - isotopic links with at most 11 crossings ,
non distinguished by the homflypt polynomial , 49 pairs contain links with three components .
i have computed @xmath17 on 20 pairs of such pairs : six of these pairs are distinguished by @xmath17 ( see proposition [ p2 ] in section 3 )
. however , there are other ways to define an invariant of classical links by using @xmath9 .
indeed , we can calculate @xmath9 on the whole set of colored links obtained from a link @xmath5 with @xmath0 components by coloring it by all the possible non - equivalent colorations .
then , for every integer partition @xmath18 of @xmath0 in @xmath19 parts ( @xmath20 ) , we take all partitions of the set of @xmath0 components in @xmath19 blocks , whose cardinalities correspond to the partition @xmath18
. these partitions of the set of components are said to be of type @xmath18 .
we thus define the invariant @xmath21 as the set of values taken by @xmath9 on the colored links obtained by coloring @xmath5 with @xmath19 colors according to each one of the type @xmath18 partitions , i.e. , by coloring the components with the same color if and only if they belong to the same block .
example : the link below has three components , that are colored according to the partitions indicated .
the invariant @xmath22 distinguishes pairs of the above mentioned pairs of non isotopic links , which are not distinguished by @xmath17 .
as we have said in @xcite , the definition of @xmath9 is a byproduct of the study of another class of links , the _ tied links _ ( see @xcite ) .
in fact , the tied links are links whose components may be connected in several points by ties . but these ties behave exactly as an equivalence relation between the link components , so that isotopic tied links are isotopic colored links with equivalent colorations ( see section 1 for more precise definitions ) .
tied links can be seen also as the closure of tied braids , that constitute a diagrammatic representation of the elements of an abstract algebra , the so - called nowadays bt - algebra ( algebra of braids and ties ) . in @xcite
we have defined , via skein relation , a polynomial @xmath23 , invariant of tied links , and we have shown that @xmath23 can be realized also by the jones recipe , using the trace on the bt - algebra established in @xcite .
the properties of the trace allowed us to make a computer program to calculate @xmath23 .
we may define a tied link @xmath24 from a colored link @xmath5 so that @xmath25 is recovered from @xmath26 by a variable change .
this has allowed me to verify the values of @xmath9 obtained on colored links by the skein relation , presented in this paper . finally , i show some unexpected identities satisfied from the values forming the invariants @xmath27 for the pairs distinguished by such invariants , that allows to formulate a conjecture .
the calculation of @xmath27 on all pairs of three - component links not distinguished by the homflypt polynomial has to be completed in order to understand these identities or to disprove the conjecture .
[ isotop ] two oriented colored links are _ c - isotopic _ if they are ambient isotopic , and their colorations are equivalent .
a colored link diagram is like the diagram of a link , where each component is colored .
let @xmath28 be a commutative ring and @xmath29 be the set of colored oriented link diagrams . by _ invariant of colored links _
we mean a function @xmath30 , which is constant on each class of c - isotopic links .
[ rem1 ] in the sequel , we denote by @xmath5 an oriented colored link as well as its diagram , if there is no risk of confusion .
[ the1 ] there exists a rational function in the variables @xmath31 , @xmath32 , invariant of oriented colored links , uniquely defined by the following three conditions on colored - link diagrams : * the value of @xmath9 is equal to 1 on the unknotted circle .
* let @xmath5 be a colored link , with @xmath0 components and @xmath19 colors .
by @xmath33 $ ] we denote the colored link with @xmath34 components consisting of @xmath5 and the unknotted and unlinked circle , colored with the @xmath35-th color .
then + @xmath36)= \frac { 1}{wx } f(l).\ ] ] * ( skein relation ) let @xmath37 and @xmath38 be the diagrams of two colored links , that are the same outside a small disc into which two strands enter , and inside this disc look as shown in figure [ linv1]a .
black and gray indicate any two colors , non necessarily distinct .
similarly , let @xmath39 and @xmath40 be two links that inside the disc look as shown in figure [ linv1]b , while outside the disc coincide with @xmath37 and @xmath38 , except -possibly- for the colors of all the components of the link having the same colors as the strands entering the disc : all these components in @xmath39 and @xmath40 are colored by a sole color .
then the following identity holds : @xmath41 and @xmath38 are not c - isotopic .
b ) the discs where @xmath42 , @xmath40 , and @xmath43 are not c - isotopic .
blue and green indicate any two colors . ]
[ rem2]the skein relation iii holds for _ any _ two colors of the involved strands . in particular , if the two colors coincide , then @xmath44 and @xmath45 , so that relation iii is reduced to the relation * @xmath46
relation iii also implies following two relations ; they will be used in the sequel .
* @xmath47 + \left(t-1\right ) f ( l_\sim).\ ] ] * @xmath48 + ( { t^{-1}}-1 ) f ( l_,\sim).\ ] ] [ rem3 ] as an example of using relations i , ii and iii , let s calculate the value of @xmath9 on the unlink @xmath49 , consisting of @xmath0 circles and @xmath19 colors .
if @xmath50 , then , using ii , @xmath51 if @xmath52 , then there are two circles colored by the same color .
regard them as belonging to @xmath53
. then @xmath54 ( see figure ) .
[ linv2 ] by iv , @xmath55 .
so @xmath56 where @xmath57 by induction , @xmath58 . thus @xmath59 [ rem4 ] let @xmath5 be a knot or a link whose components have all the same color .
then @xmath25 is defined by i and iv .
recall that the homflypt polynomial @xmath60 is defined by the condition that @xmath61 and the skein relation @xmath62 where @xmath63 and @xmath64 are two complex variables . comparing the skein relation iv , multiplied by @xmath65 , with the equation above
, we see that @xmath9 is a homflypt polynomial with @xmath66 in particular , for @xmath13 it becomes the jones polynomial .
[ rem5 ] the polynomial @xmath9 is multiplicative with respect to the connected sum of colored links @xmath67 and @xmath68 in the following sense . the components @xmath69 and @xmath70 of the two links that merge must have the same color , while the colors of the components in @xmath67 not equal the color of @xmath69 must be different from the colors of the components in @xmath68 not equal to the color of @xmath70 .
let @xmath5 be a classical link with @xmath0 components .
we denote by @xmath71 the ordered set of the components of @xmath5 .
a set - partition @xmath72 of @xmath73 , contains @xmath74 non void blocks . for every set - partition @xmath75 of @xmath73
, we define a colored link @xmath76 in this way : we color the components of @xmath5 with @xmath77 colors , in such a way that the components entering the same block have the same color .
observe that the number @xmath78 of non - equivalent colorations of the link @xmath5 satisfies @xmath79 , where @xmath80 is the @xmath0-th bell number .
the equality holds in general except in particular cases of symmetries among the components of the link .
given a set partition @xmath81 of @xmath73 , let @xmath82 , so that @xmath83 .
we call the integer partition @xmath18 of @xmath0 in @xmath19 part , @xmath84 the _ type _ of the partition @xmath85 . for instance , if @xmath86 , among the following partitions of @xmath87 in two blocks : @xmath88 the partitions @xmath89 and @xmath90 have the same type @xmath91 , and @xmath92 has type @xmath93 . for every partition @xmath18 of @xmath0 in @xmath19 parts ( @xmath94 ) , let @xmath95 be the number of partitions of @xmath73 of type @xmath18 which define non - equivalent colorations of @xmath5 with @xmath19 colors , and @xmath96 be the corresponding colored links .
the invariant @xmath97 of @xmath5 is the non - ordered set @xmath98 [ rem6]it is evident that for a link @xmath5 wit @xmath0 components , if @xmath99 , @xmath100 , so @xmath101 , and @xmath102 , i.e. , the value of @xmath9 on the link @xmath5 with all component colored with the same color .
furthermore , for @xmath103 the invariant @xmath104 .
[ rem7 ] in the following figures , two arrows going from a link @xmath5 with a marked crossing , to two colored links @xmath67 and @xmath68 , indicate that @xmath105 , where @xmath106 and @xmath107 are given by relation iv ( remark [ rem2 ] ) ; three arrows going from a colored link @xmath5 with a marked crossing , to three links @xmath67 , @xmath68 and @xmath108 , indicate that @xmath109 , where @xmath2 , @xmath110 and @xmath111 are given by relations va or vb , depending on the sign of the marked crossing .
we give here the experimental evidence that for @xmath112 , @xmath97 is more powerful than the homflypt polynomial .
consider , for instance , the links @xmath113 and @xmath114 , shown in [ linv2 ] .
they have four components , six positive crossings , and are colored with a sole color . @xmath113 and @xmath114 are evidently non isotopic .
the calculation of @xmath115 and @xmath115 shows that @xmath113 and @xmath114 are not distinguished by the homflypt polynomial ( see remark [ rem4 ] ) .
indeed , applying the skein relation iv to the crossings marked in figure [ linv3 ] , we obtain @xmath116 @xmath117 but @xmath118 and @xmath119 .
therefore @xmath9 coincides on @xmath113 and on @xmath114 . and
@xmath120 coincide ] consider now the colored links obtained by coloring @xmath113 and @xmath114 with two colors , more precisely three components with one color and the remaining component by another color ( i.e. , corresponding to the partitions of type @xmath121 ) .
there are two of such non - equivalent colorations of @xmath113 and of @xmath114 , giving the colored links @xmath122 , @xmath123 and @xmath124 , @xmath125 respectively ( see figure [ linv4 ] ) and @xmath126 the links @xmath113 and @xmath114 are distinguished by the invariant @xmath127 .
we have @xmath128 and @xmath129 from figure [ linv4 ] we see that @xmath130 .
we calculate now @xmath131 and @xmath132 .
we have : + @xmath133 , + @xmath134 , @xmath135 , + @xmath136 , @xmath137 , + @xmath138 + @xmath139 + @xmath140 , @xmath141 , @xmath142 .
+ we get + @xmath143 on the other hand , we have + @xmath144 + @xmath145 @xmath146 @xmath147 , + @xmath148 , @xmath149 .
+ therefore , + @xmath150 since @xmath151 , the links @xmath113 and @xmath114 are distinguished by @xmath152 .
in this section we show some pairs of links which are not distinguished by the homflypt polynomial , but that are distinguished by @xmath97 .
the calculation of @xmath9 by the skein relation is shown in the figures for some pairs .
for the meaning of the arrows , see remark [ rem7 ] .
the skein - trees terminate in some simple basic colored links , belonging to a list @xmath153 , shown in figure [ list ] .
the values of @xmath9 on these colored links are given in appendix 1 .
the links are named according to linkinfo , see @xcite .
[ p1 ] the invariant @xmath154 distinguishes the following pairs of links : @xmath155 these links have three components .
the values of @xmath9 are different on the colored links , obtained by coloring each of these links with three different colors . in appendix 2 ,
the values of @xmath17 are listed for the pairs , except for the pair ( @xmath156 , @xmath157 ) , for which the invariants are calculated in section 4 and are shown in appendix 4 .
[ p2 ] the invariant @xmath22 distinguishes the following pairs of links , which are not distinguished by @xmath154 : @xmath158 ] for every one of the above pairs , we proceed in this way : we firstly calculate @xmath17 on the links of the pair ( all links have three components ) .
these values coincide in each pair , so we obtain that @xmath154 does not distinguish these links
. then we calculate @xmath22 .
we show here the calculation of @xmath22 on the links of the pair ( @xmath159 )
. there are three non - equivalent colorations of @xmath160 and @xmath161 corresponding to the partitions of type @xmath162 of the set of components .
the non c - isotopic colored links obtained by such colorations are called respectively @xmath163 , and @xmath164 , see figure [ pairs ] . ] ] from figures [ lk ] and [ lh ] we obtain the values @xmath165 , see appendix 3 .
we calculate now @xmath166 , see figure [ kabc ] .
then we compare this set of values with @xmath167 , obtained from figure [ habc ] .
these sets of values , shown in appendix 3 , are different .
for a link with three components , one could define , instead of @xmath22 , the invariant defined as the _ sum _ of the values belonging to the set @xmath22 .
surprisingly , despite the fact that the values are all different for the pairs of proposition [ p2 ] , their sums coincide .
we have indeed : @xmath168 @xmath169 furthermore , i have found that , in the pairs of proposition [ p2 ] , the links are distinguished not only by @xmath170 , but by @xmath171 as well . naming @xmath172 and @xmath173 the colored links obtained from a pair by coloring the three component with three different colors , and @xmath174 , @xmath175 , @xmath176 , @xmath177 , @xmath178 , @xmath179 , the colored links obtained from that pair by coloring the components with two different colors , the corresponding values of @xmath9 satisfy : @xmath180 therefore , in the case of three components , we formulate the following conjecture . observe that in all the pairs of links here considered , the homflypt polynomial , and then @xmath181 , coincide .
denote by @xmath182 the sum of the elements of the set @xmath97 .
let @xmath67 and @xmath68 be two links with three components . if @xmath183 , then @xmath184 the reasons of the above identities are not evident to me and deserve further investigations . also , the invariants @xmath185 have to be calculated on all pairs of non isotopic links with the same value of @xmath186 . * example .
* in figures [ rs ] , [ rabc ] and [ sabc ] the trees for the calculations of @xmath154 and @xmath22 on the links @xmath187 and @xmath188 are shown .
observe that the values of @xmath9 on some leaves of the skein - trees for the colored links obtained from @xmath189 and @xmath190 , are obtained by using remark [ rem3 ] , e.g. @xmath191 , @xmath192 .
also , remark [ rem5 ] is used to calculate some values of @xmath9 , e.g. @xmath193 , @xmath194 .
the values of @xmath195 and @xmath22 are shown in appendix 4
. * acknowledgement . * a particular thank to konstantinos karvounis for having providing me with the list of pairs of non isotopic links that are not distinguished by the homflypt polynomial . ,
@xmath196 and @xmath197 ] , @xmath198 and @xmath199 ] on the links @xmath189 and @xmath190 ] on the link @xmath189 ] on the link @xmath190 ]
@xmath202 @xmath203 1 cm * @xmath204 * .5 cm @xmath205 @xmath206 @xmath207 @xmath208 1 cm @xmath209 @xmath210 @xmath211 @xmath212 @xmath213 @xmath214 1 cm * @xmath215 * .5 cm @xmath216 @xmath217 @xmath218 @xmath219 1 cm @xmath220 @xmath221 @xmath222 @xmath223
@xmath154 and @xmath22 on the links @xmath187 and @xmath188 . | new invariants of links are constructed by using the skein invariant polynomial of colored links defined in @xcite .
these invariants are stronger than the homflypt polynomial . |
in recent years there has been a continuous interest in the study of excited heavy mesons composed of a heavy quark and a light anti - quark .
this interest arises from several reasons . some of these excited states have been observed in experiments .
they will be objects of further study in future experiments with b - factories .
in particular , they are useful for tagging in cp experiments @xcite .
theoretically , the relative simplicity of the dynamical condition makes it of interest for exploring the internal dynamics of systems containing a light quark .
important progresses in the theoretical description of such systems have been achieved with the development of the heavy quark effective theory ( hqet ) @xcite .
based on the spin - flavor symmetry of qcd , valid in the limit of infinite heavy quark mass @xmath5 , this framework provides a systematic expansion of heavy hadron spectra and transition amplitudes in terms of the leading contribution , plus corrections decreasing as powers of @xmath6 . however , to obtain some detailed predictions one needs to combine it with some non - perturbative methods .
the spectra and decay widths of heavy meson excited states have been studied with the @xmath6 expansion in the relativistic bethe - salpeter equations in @xcite .
they can also be studied with the qcd sum rules in hqet which have been used both for ground states of heavy mesons @xcite and for lowest excited heavy meson doublets @xmath3 and @xmath4 at leading order of the @xmath6 expansion@xcite . for related results ,
see @xcite .
the effective lagrangian of the hqet , up to order @xmath6 , can be written as @xmath7 where @xmath8 is the velocity - dependent field related to the original heavy - quark field @xmath9 by @xmath10 @xmath11 is the operator of nonrelativistic kinetic energy with a negative sign defined as @xmath12 where @xmath13 , with @xmath14 is the gauge - covariant derivative , and @xmath2 is the pauli term , describing the chromomagnetic interaction : @xmath15 where @xmath16 , @xmath17 . apart from leading contribution
, the lagrangian density contains to @xmath0 accuracy two additional operators @xmath1 and @xmath2 .
the matrix elements of these two operators over heavy meson states are fundamental parameters in the @xmath6 expansion with hqet that should be either determined phenomenologically from experimental data or estimated using a nonperturbative theoretical approach .
the matrix element of the chromomagnetic interaction operator is the leading contribution to the mass splitting among the states belonging to the same heavy meson doublet . in our previous work @xcite , the masses of lowest excited heavy meson doublets @xmath18 and @xmath19
are calculated at the leading order of @xmath6 expansion by using the qcd sum rule approach in hqet .
as an extension of this work , in the present article we shall use the qcd sum rules in hqet to obtain quantitative estimates of the matrix elements of the kinetic energy and chromomagnetic interaction operators , which are responsible for the kinetic energy and the chromomagnetic mass corrections .
the remainder of this paper is organized as follows . in section [ sec2 ]
we begin with a brief review on the interpolating currents for excited heavy mesons and some important properties of these currents .
some results of two - point functions for the doublets @xmath18 and @xmath19 at the leading order of the @xmath6 expansion are outlined .
section [ sec3 ] is devoted to @xmath0 corrections , the sum rules for the relevant matrix elements of kinetic energy and chromomagnetic interaction operator are derived .
finally , numerical results are presented in section [ sec4 ] .
a basic element in the application of qcd sum rules to excited heavy mesons is to choose a set of appropriate interpolating currents in terms of quark fields which create ( annihilate ) definite states of excited heavy mesons . for excited heavy mesons with arbitrary spin and parity
this problem has been studied in our previous paper @xcite . here
we briefly outline the results of this analysis which will be used in this article . from the general form of the leading order b - s
wave function obtained in @xcite , we can write the general expression of the interpolating current creating a excited heavy meson with arbitrary spin @xmath20 and parity @xmath21 as [ current ] @xmath22 or @xmath23 [ conju ] @xmath34 \label{conju2 } j_{j , p , i}^{'\alpha_1\cdots\alpha_j}(x)&=&\bar q(x)(i\stackrel{\leftarrow } { \not\!{\cal d}}_{x_t})\bar{\gamma}_{j , p , i}^{\{\alpha_1\cdots\alpha_j\}}(\stackrel{\leftarrow}{\cal { d}}_{x_t})h_{v(x)}\;,\end{aligned}\ ] ] where @xmath35 , correspond to the annihilation of the same meson .
the currents defined in ( [ current ] ) and ( [ conju ] ) have nice properties .
we had shown in @xcite that two currents in the sets ( [ current1 ] ) and ( [ current2 ] ) with not identical values of @xmath20 , @xmath21 , @xmath36 never mix in the @xmath37 limit either in the heavy meson or in the quark - gluon level and heavy quark symmetry is explicit with these currents .
this verifies that they are the appropriate interpolating currents for heavy meson states with definite @xmath20 , @xmath21 and the light quark angular momentum @xmath38 which is conserved in qcd in @xmath37 limit .
the properties for currents mentioned above are important for applications to qcd sum rules for excited heavy mesons . if one use other currents or study the sum rules in full qcd , there are in general contributions from two nearby poles corresponding to states of the same @xmath20 , @xmath21 which have different values of the total angular momentum of the light component @xmath39 in the @xmath37 limit .
their contributions may not be separated correctly .
furthermore , the mixing of such two states can be calculated within our formalism by introducing the @xmath0 terms in the lagrangian in hqet .
let @xmath40 be a heavy meson state with the quantum numbers @xmath20 , @xmath21 , @xmath36 in the @xmath37 limit , together with the corresponding interpolating current , we have @xcite @xmath41 where the decay constant @xmath42 has the same value for the two states in the same doublet and @xmath43 is the transverse , symmetric and traceless polarization tensor . in the framework of qcd sum rules
the decay constant can be obtained from the two - point correlator : @xmath44 where @xmath45 is the residual momentum , corresponding to the decomposition @xmath46 of the heavy meson momentum .
this expression has the following form at the leading order @xcite @xmath47 where @xmath48 is twice the external off - shell energy and @xmath49 denotes symmetrization and subtracting the trace terms in the sets @xmath50 and @xmath51 , and @xmath52 is independent of @xmath20 and satisfies dispersion relations of the form : @xmath53 where @xmath54 represents the spectral density .
we shall confine our study to the doublets @xmath18 and @xmath19 here . according to ( [ current1 ] ) ,
( [ current2 ] ) and ( [ gamma2 ] ) , there are two possible choices for currents creating @xmath55 and @xmath56 of the doublet @xmath18 , either @xmath57 or @xmath58 similarly , there are two possible choices for the currents creating @xmath56 and @xmath59 of the doublet @xmath19 .
one is , the coefficient @xmath60 should be @xmath61 . ]
@xmath62 another choice is obtained by adding a factor @xmath63 to ( [ curr5 ] ) and ( [ curr6 ] ) .
note that , without the last term in the bracket in ( [ curr5 ] ) the current would couple also to the @xmath56 state in the doublet @xmath18 even in the limit of infinite @xmath5 .
usually , the currents with the least number of derivatives are used in the qcd sum rule approach .
the sum rules with them have better convergence in the high energy region and often have better stability .
however , there is a motivation for using the currents ( [ curr3 ] ) , ( [ curr4 ] ) for the doublet @xmath18 . in the non - relativistic quark model , which usually gives correct ordering of energy levels of hadron states ,
the doublets @xmath18 and @xmath19 are orbital p - wave states which correspond to one derivative in the space wave functions .
in fact , as proved in @xcite , the coupling constant for the currents ( [ curr3 ] ) , ( [ curr4 ] ) and that for ( [ curr5 ] ) , ( [ curr6 ] ) are proportional to the large components of the b - s wave functions for the doublet @xmath18 and @xmath19 respectively in the non - relativistic approximation . therefore , we shall consider both the currents ( [ curr1 ] ) , ( [ curr2 ] ) and ( [ curr3 ] ) , ( [ curr4 ] ) for the doublet @xmath18 . for the doublet @xmath18 ,
when the currents @xmath64 , @xmath65 in ( [ curr3 ] ) , ( [ curr4 ] ) are used the sum rule ( same for the two states ) after the borel transformation is found to be @xmath66 the corresponding formula when the current @xmath67 and @xmath68 in ( [ curr1 ] ) and ( [ curr2 ] ) are used instead of @xmath64 and @xmath65 is the following @xmath69 when the currents ( [ curr5 ] ) and ( [ curr6 ] ) are used the sum rule for the @xmath19 doublet is found to be @xmath70 here @xmath71 . in the above derivations , we have confined us to terms of the lowest order in perturbation and operators of dimension less than six . above results
shall be used in the next section .
from lorentz covariance , the correlator ( [ sum1 ] ) still has the form ( [ sum2 ] ) beyond the leading order of @xmath6 expansion if we confine @xmath45 to be a longitudinal vector .
but now @xmath72 depends on @xmath20 due to the chromomagnetic interaction operator @xmath2 . inserting the heavy meson eigen - state of the hamiltonian up to the order @xmath0 , the pole term on the hadron side becomes @xmath73 where @xmath74 and @xmath75 are of the order @xmath0 . to extract @xmath74 in ( [ pole ] ) we follow the approach of @xcite and consider the three - point correlation functions @xmath76 where @xmath77 or @xmath2 .
the sum @xmath78 is equal to the correlator ( [ sum1 ] ) in the order @xmath0 when @xmath79 .
the scalar function corresponding to ( [ delta ] ) can be represented as the double dispersion integral @xmath80 the pole parts for them are @xmath81 \label{poles } & & \delta_{\cal s}\pi(\omega,\omega^{'})_{pole}={d_mf^2\sigma\over(2\bar\lambda-\omega)(2\bar\lambda-\omega^{'})}+d_mf\left[{g_{\cal s}(\omega^{'})\over 2\bar\lambda-\omega}+{g_{\cal s}(\omega)\over 2\bar\lambda-\omega^{'}}\right]\;,\end{aligned}\ ] ] where @xmath82 let @xmath79 in ( [ polek ] ) and ( [ poles ] ) , and compare it with ( [ pole ] ) one obtains@xcite @xmath83 the simple pole term in ( [ polek ] ) and ( [ poles ] ) comes from the region in which @xmath84 and @xmath85 is at the pole for a radical excited state or in the continuum .
these terms are suppressed by making double borel transformation for both @xmath86 and @xmath87 .
one obtains thus the sum rules for @xmath88 and @xmath89 as @xmath90 where the spectral densities are obtained from straitforward calculations in hqet . confining us to the leading order of perturbation and the operators with dimension @xmath91 in ope , the relevant feynman diagrams , which contribute to @xmath92 and
@xmath93 are shown in fig . 1 and fig .
2 respectively .
the graphs missing there turn out to be vanishing in the fixed point gauge @xmath94 @xcite .
the spectra functions @xmath95 and @xmath96 can be calculated from these diagrams . for the @xmath97 doublet
, we find [ msum1d ] @xmath98 when the currents ( [ curr3 ] ) and ( [ curr4 ] ) are used and [ msum1 ] @xmath99 when the currents ( [ curr1 ] ) and ( [ curr2 ] ) with no extra derivative are used .
in contrast to ( [ form - s1d ] ) , there is a nonvanishing mixing condensate term in ( [ form - s1 ] ) .
it arises from the diagram depicted in fig .
2(b ) . for @xmath100 doublet , a straitforward calculation yields [ msum2 ] @xmath101 combining ( [ msum1d ] ) , ( [ msum1 ] ) and ( [ msum2 ] ) with ( [ form1 ] ) , ( [ form2 ] ) and ( [ form3 ] ) we can obtain sum rules for @xmath88 and @xmath89 in the three cases .
the above results are not the whole story .
the spin - symmetry violating term @xmath102 not only causes splitting of masses in the same doublet , but also causes mixing of states with the same @xmath20 , @xmath21 but different @xmath38 .
this mixing is characterized by the matrix element @xmath103 this quantity can be extracted from the correlator @xmath104 the double pole term of which is @xmath105 with the methods similar to that used above one can obtain the sum rule of @xmath106 . for two @xmath56 states
we find @xmath107 as an example , here we have used the currents ( [ curr4 ] ) and ( [ curr5 ] ) to do the analysis .
the masses of two @xmath56 states are obtained by diagonalizing the mass matrix @xmath108 \displaystyle{\frac{1}{4m_q}}\;c_{mag}m_{\frac{3}{2},\frac{1}{2}}(1^+ ) & \bar\lambda_{+,\frac{3}{2}}-\displaystyle{{1\over 4m_q}}(k_{+,\frac{3}{2}}+5\;c_{mag}\sigma_{+,\frac{3}{2 } } ) \end{array}\;\right)\;.\end{aligned}\ ] ] the correction to the masses due to the mixing is formally of the order of @xmath109 . however , due to the smallness of the difference between the two diagonal elements there is some possibility that this correction is numerically comparable to the @xmath0 corrections . therefore we shall calculate it also .
we now turn to the numerical evaluation of these sum rules .
we can eliminate the explicit dependence of the sum rules obtained in section [ sec3 ] on @xmath110 and @xmath111 by means of the two - point function sum rules outlined in section [ sec2 ] . dividing the sum rules in ( [ msum1d ] ) , ( [ msum1 ] ) and ( [ msum2 ] ) by the sum rules in ( [ form1 ] ) , ( [ form2 ] ) and ( [ form3 ] ) respectively , we obtain relevant expressions for the @xmath88 and @xmath89 as functions of the borel parameter @xmath112 and the continuum threshold @xmath113 . for the qcd parameters entering the theoretical expressions , we use the following standard values @xmath114
the results for @xmath88 as a function of the borel variable @xmath112 , and for different values of the continuum threshold @xmath113 , are shown in fig .
3 . in particular , in fig .
3a and 3b we depict the results of two kinds of sum rules for doublet @xmath18 , which correspond to using two kinds of currents ( [ curr1 ] ) ( [ curr2 ] ) and ( [ curr3 ] ) ( [ curr4 ] ) respectively . in fig .
3c , we display the result of sum rule associated with doublet @xmath19 . following the qcd sum rule procedure , we should check the existence of stable regions for the results of @xmath88 in the borel variable @xmath112 and the continuum threshold @xmath113 .
as one can see from fig .
3 , stability starts at values of borel parameter @xmath112 slightly larger than @xmath115 , and stretches practically to @xmath116 . the familiar criterion that both the high - order power corrections and the contribution of the continuum model should not be too large , generally speaking , less than @xmath117 ,
rather strongly restricts the ` working ' region .
thus , the stability at large @xmath112 values is not useful , since in this region the sum rules is strongly contaminated by higher resonance states .
this continuum model contamination problem is quite severe also in the case of sum rules analysis of ground state heavy meson to @xmath6 .
it originates from the high power dependence of the spectral densities . according to @xcite ,
it is better to determine the working window by considering the stability of two - point sum rules , and then to use the same values in the evaluation of the three - point sum rules .
the choice of so - called stability window does not necessarily coincide with the stability plateau for the three - point sum rules . in the case of two - point function sum rules
one finds that the working windows are @xmath118 gev , @xmath119 gev for the doublet @xmath18 and @xmath120 gev , @xmath119 gev for the doublet @xmath19 @xcite . from fig .
3 , taking the working regions of the sum rules as that for the two - pont function for the three cases , we obtain the set of values for @xmath88 as following : @xmath121 for the doublet @xmath18 when the currents ( [ curr1 ] ) ( [ curr2 ] ) without the derivative are used , @xmath122 for the same doublet when the currents ( [ curr3 ] ) , ( [ curr4 ] ) with the derivative are used , and @xmath123 for the doublet @xmath19 .
the errors reflect the variations with the borel parameter @xmath112 and the continuum threshold @xmath113 within the working windows . as can be seen when comparing fig . 3a and fig .
3b with each other , @xmath88 value in fig .
3b is more sensitive to the value of @xmath113 than that in fig .
3a owing to the higher dimension of spectral density for the using of current with extra derivative . in fig .
4 we show the numerical results of the sum rules for @xmath89 , corresponding to the three chosen values of the continuum threshold @xmath113 . from these figures , we see that there is no stability plateau for the sum rules of @xmath89 .
this is related to the absence of perturbation terms for @xmath89 in our approximation .
however , due to the strategy mentioned above we can take the working window as that of two - point function sum rules like the case of @xmath88 .
the results for @xmath89 for the doublet @xmath18 in the working regions of the sum rules are @xmath124 when the currents ( [ curr1 ] ) ( [ curr2 ] ) without the derivative are used and @xmath125 when the currents ( [ curr3 ] ) , ( [ curr4 ] ) with the derivative are used . for the doublet @xmath19 , the value of @xmath89 is given by @xmath126 dividing the sum rule in ( [ cross - m ] ) by the sum rules in ( [ form2 ] ) and ( [ form3 ] ) , we also obtain the expression for @xmath127 . in fig .
5 we plot the result of this sum rules as a function of @xmath112 , and for three values of @xmath113 . using the working windows as @xmath128 gev and @xmath119 gev , we obtain @xmath129 after combining with ( [ matrix ] ) , we found that the correction to the masses due to the mixing is negligibly small . in conclusion
, we have calculated the @xmath6 correction to the masses of lowest excited heavy mesons from sum rules within the framework of the hqet .
this study refines the leading order analysis @xcite . from ( 42)-(47 ) , together with ( [ delm ] ) , we can obtain the corrections to the masses of excited @xmath130 mesons . for the doublet @xmath18 , our final results read @xmath131\;,\\ m_{d'_1}-m_{d_0^*}&=&{1\over m_c}\;[(0.023\pm 0.003)~\text{gev}^2]\;,\end{aligned}\ ] ] when the currents ( [ curr1 ] ) and ( [ curr2 ] ) are used and @xmath132\;,\\ m_{d'_1}-m_{d_0^*}&=&{1\over m_c}\;[(0.014\pm 0.003)~\text{gev}^2]\;,\end{aligned}\ ] ] when the currents ( [ curr3 ] ) and ( [ curr4 ] ) are used . as for the doublet @xmath19 , the result is @xmath133\;,\\ m_{d_2^*}-m_{d_1}&=&{1\over m_c}\;[(0.040\pm 0.006)~\text{gev}^2]\;,\end{aligned}\ ] ] where the values of @xmath111 in three cases are given in @xcite .
note that we have reasonably neglected the renormalization coefficient @xmath134 of the chromomagnetic operator for charmed meson in the discussion .
we should also note that the given errors present only the uncertainties of @xmath112 and @xmath113 in the working regions .
the above results for the spin splittings are not theoretically reliable because we have not included the feynman diagrams which are of the one loop order in the perturbation theory and may be of values comparable to that of the gluon condensate term leading in our approximation . nevertheless , they are in approximate agreement with the experimental values for the mass splitting between @xmath135 and @xmath136 @xcite if we take @xmath137 gev .
b. grinstein , nucl .
* b339 * , 253(1990 ) ; e. eichten and b. hill , phys .
* b234 * , 511(1990 ) ; a. f. falk , h. georgi , b. grinstein and m. b. wise , nucl .
b343 * , 1(1990 ) ; f. hussain , j. g. krner , k. schilcher , g. thompson and y. l. wu , phys .
b249 * , 295(1990 ) ; j. g. krner and g. thompson , phys . lett . *
b264 * , 185(1991 ) .
e. bagen , p. ball , v. m. braun and h. g. dosch , phys
b278 * , 457(1992 ) ; m. neubert , phys . rev . * d 45 * , 2451(1992 ) ; d. j. broadhurst and a. g. grozin , phys . lett . * b274 * , 421(1992 ) ; c. a. dominguez and n. paver , phys . lett . *
b276 * , 179(1992 ) ; m. neubert , phys . rep . * 245 * , 259(1994 ) . | the @xmath0 corrections to masses of excited heavy mesons are studied with sum rules in the heavy quark effective theory .
numerical results for the matrix elements of the heavy quark kinetic energy operator @xmath1 and chromomagnetic interaction operator @xmath2 are obtained for the lowest excited doublets @xmath3 and @xmath4 .
= -10 mm |
since the seminal work of thurston @xcite , the concept of orbifold has become a central one in 3-dimensional geometric topology , because it embodies the notion of a 3-manifold and that of a finite group action on that 3-manifold . as a matter of fact
, thurston himself indicated how to prove that , under some natural topological constraints , a 3-orbifold with non - empty singular locus carries one of the 8 geometries , so it is a quotient of a geometric 3-manifold under a finite isometric action ( see @xcite and @xcite for modern detailed accounts of the argument ) . in the case of manifolds ( closed
orientable ones , say ) the two basic obstructions to the existence of a geometric structure are related to the existence of essential spheres and of essential tori .
so one restricts first to irreducible manifolds and next to seifert and atoroidal ones . for this restricted class of manifolds
then one has thurston s _ geometrization conjecture _ ( or _ theorem _ , according to the recent work of perelman @xcite ) .
this result can be regarded as a _
uniformization theorem _ for 3-manifolds in view of the haken - kneser - milnor theorem ( canonical splitting of a 3-manifold along spheres into irreducible 3-manifolds ) and the jaco - shalen - johansson theorem ( canonical splitting of an irreducible 3-manifold along tori into seifert and atoroidal 3-manifolds ) .
the case of 3-orbifolds is similar , where the first basic obstruction to the existence of a geometric structure is removed by restricting to ( suitably defined ) _ irreducible _ 3-orbifolds
. the natural expectation would then be to have an analogue of the haken - kneser - milnor theorem , asserting that each 3-orbifold canonically splits along spherical 2-orbifolds into irreducible 3-orbifolds .
this result has actually been part of the folklore for several years , but it turns out that to state and prove it in detail one has to somewhat refine the ideas underlying the proof in the manifold case .
this paper is devoted to these refinements .
we will state in the present introduction two versions of the splitting theorem that we will establish below , addressing the reader to section [ statements : section ] for the appropriate definitions .
[ basic : splitting : teo ] let @xmath2 be a closed locally orientable @xmath0-orbifold .
suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating .
then @xmath2 contains a finite system @xmath3 of spherical @xmath1-suborbifolds such that : * no component of @xmath4 is punctured discal .
* if @xmath5 is the orbifold obtained from @xmath4 by capping the boundary components with discal @xmath0-orbifolds , then the components of @xmath5 are irreducible . any such system @xmath3 and the corresponding @xmath5 , if viewed as abstract collections of @xmath1- and @xmath0-orbifolds respectively , depend on @xmath2 only .
we inform the reader that this result has a dual interpretation in terms of connected sum , discussed in detail in section [ statements : section ] .
we also note that the existence part of theorem [ basic : splitting : teo ] may seem to have a purely theoretical nature , which is not pleasant in the context of @xmath0-dimensional topology , where a great emphasis is typically given to algorithmic methods @xcite .
the next longer statement emphasizes the constructive aspects of the splitting .
[ splitting : intro : teo ] let @xmath2 be a closed locally orientable @xmath0-orbifold .
suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating . starting with @xmath6 and @xmath7 , consider the ( non - deterministic ) process described by the following steps : * if all ordinary spherical @xmath1-suborbifolds of @xmath5 are inessential , turn to step 2 .
otherwise choose @xmath8 as one such @xmath1-suborbifold , redefine @xmath3 as the given @xmath3 union @xmath8 , redefine @xmath5 as the orbifold obtained from the given @xmath5 by cutting along @xmath8 and capping the result by discal @xmath0-orbifolds , and repeat step 1 .
* if all cyclic spherical @xmath1-suborbifolds of @xmath5 are inessential , turn to step 3 .
otherwise choose @xmath8 as one such @xmath1-suborbifold , redefine @xmath3 and @xmath5 as in step 1 , and repeat step 2 . *
if all vertex spherical @xmath1-suborbifolds of @xmath5 are inessential , output @xmath3 and @xmath5 .
otherwise choose @xmath8 as one such @xmath1-suborbifold , redefine @xmath3 and @xmath5 as in step 1 , and repeat step 3 .
then the process is finite and the final @xmath5 consists of irreducible @xmath0-orbifolds .
moreover , the final @xmath3 and @xmath5 , viewed as abstract collections of @xmath1- and @xmath0-orbifolds respectively , depend on @xmath2 only , not on the specific process leading to them .
in addition , at all steps of the process the essential spherical @xmath1-orbifold @xmath8 can be chosen to be normal with respect to a triangulation of the given orbifold @xmath5 . for the reader familiar with the proof of the splitting theorem for manifolds ( see for instance @xcite )
we will now point out what goes wrong in the orbifold case , and sketch how we have modified the argument . the main steps to show the existence of a splitting of a manifold @xmath9 into prime ones are as follows : * show that if @xmath9 contains a non - separating sphere then it splits as the connected sum of @xmath10 or @xmath11 and some @xmath12 , and the first betti number of @xmath12 is smaller than that of @xmath9 .
deduce that one can restrict to @xmath9 s in which all spheres are separating .
* define a family of spheres @xmath13 to be _ essential _ if no component of @xmath14 is a punctured disc .
show that every essential @xmath3 can be replaced by one having the same number of components and being normal with respect to any given triangulation of @xmath9 .
deduce that all maximal essential systems of spheres are finite . *
prove that if @xmath3 is a maximal essential system then the manifolds obtained by capping off the components of @xmath15 are irreducible .
uniqueness is more elaborate to prove , but its validity only depends on the following fact , that we express using the dual viewpoint of connected sum : * the sphere @xmath16 is the only identity element of the operation of connected sum .
no connected sum with @xmath16 should be employed to realize a manifold . turning to orbifolds
, we can show that none of the points ( a)-(d ) has a straight - forward extension : * the existence of a non - separating spherical @xmath1-suborbifold does not imply the existence of an essential separating one ( see fig.[nonsep : fig ] below ) .
in particular , there are infinitely many 3-orbifolds containing some non - separating but no essential separating spherical @xmath1-orbifold ( see section [ prime : section ] ) .
the difficulty given by non - separating spherical @xmath1-orbifolds is mentioned in @xcite , where the choice is made of cutting along these @xmath1-orbifolds too .
the reasons of our choice of excluding non - separating spherical @xmath1-orbifolds _ tout - court _ will be discussed in section [ existence : section ] .
we only want to mention here that the techniques of the present paper do not seem to adapt easily to include the case of splitting along non - separating spherical @xmath1-orbifolds .
turning to point ( b ) , we follow @xcite and define a family @xmath3 of spherical @xmath1-suborbifolds of a 3-orbifold @xmath2 to be _ essential _ if no component of @xmath17 is a punctured discal @xmath0-orbifold .
we now have : * if @xmath3 is essential and @xmath18 are obtained from @xmath3 by one of the compression moves necessary to normalize @xmath3 , then neither of @xmath19 and @xmath20 need be essential ( see fig .
[ essnocompr : fig ] below ) . this fact could be viewed as a merely technical difficulty , showing that the theory of normal surfaces is inadequate to prove finiteness of maximal essential systems .
the following fact is however more striking : * if @xmath3 is a maximal essential system of spherical @xmath1-suborbifolds of @xmath2 , and @xmath5 is obtained by cutting @xmath2 along @xmath3 and capping the result , the components of @xmath5 need not be irreducible ( see fig . [
essnoirred : fig ] below ) . given this drawback of the notion of essential system ,
we have refined it to a certain notion of _ efficient splitting system _
, where we impose artificially that the capped components of the complement should be irreducible .
the proof of the existence of efficient splitting systems then requires some more efforts than in the manifold case .
in addition , finiteness of the splitting process in theorem [ splitting : intro : teo ] does not follow directly from the existence of finite efficient splitting systems . concerning uniqueness , we have the following : * there are three different types of connected sums of orbifolds , and each has its own identity , different from the other ones . in a sequence of connected sums of irreducible @xmath0-orbifolds ,
the property that a certain sum is `` trivial '' depends on the order in which the sums are performed .
the same notion of efficient system used to prove the existence of a splitting actually allows to deal also with this problem , and hence to prove uniqueness in theorem [ basic : splitting : teo ] .
the same argument employed for uniqueness also proves finiteness of the splitting process in theorem [ splitting : intro : teo ] .
[ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + the author warmly thanks gwnal massuyeau , sergei matveev , luisa paoluzzi , and joan porti for helpful discussions .
in this section we introduce the necessary terminology and we provide alternative statements of theorems [ basic : splitting : teo ] and [ splitting : intro : teo ] .
[ [ local - structure - of - orbifolds ] ] local structure of orbifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + we will not cover here the general theory of orbifolds , referring the reader to the milestone @xcite , to the excellent and very recent @xcite , and to the comprehensive bibliography of the latter .
we just recall that an orbifold of dimension @xmath21 is a topological space with a singular smooth structure , locally modelled on a quotient of @xmath22 under the action of a finite group of diffeomorphisms .
we will only need to refer to the cases @xmath23 and @xmath24 , and we will confine ourselves to orientation - preserving diffeomorphisms .
in addition , all our orbifolds will be compact .
under these assumptions one can see that a 2-orbifold @xmath8 consists of a compact _ support _
surface @xmath25 together with a finite collection @xmath26 of points in the interior of @xmath25 , the _ cone _
points , each carrying a certain _ order _ in @xmath27 .
analogously , a @xmath0-orbifold @xmath2 is given by a compact _ support _
3-manifold @xmath28 together with a _ singular set _ @xmath29 .
here @xmath29 is a finite collection of circles and unitrivalent graphs tamely embedded in @xmath28 , with the univalent vertices given by the intersection with @xmath30 .
moreover each component of @xmath29 minus the vertices carries an _ order _ in @xmath27 , with the restriction that the three germs of edges incident to each vertex should have orders @xmath31 , for arbitrary @xmath32 , or @xmath33 , for @xmath34 .
[ [ bad - spherical - and - discal - orbifolds ] ] bad , spherical , and discal orbifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + an _ orbifold - covering _ is a map between orbifolds locally modelled on a map of the form @xmath35 , naturally defined whenever @xmath36 .
an orbifold is called _ good _ when it is orbifold - covered by a manifold , and _ bad _ when it is not good . in the sequel
we will need the following easy result : [ bad : lem ] the only bad closed @xmath1-orbifolds are @xmath37 , the @xmath1-sphere with one cone point of order @xmath32 , and @xmath38 , the @xmath1-sphere with cone points of orders @xmath39 .
we now introduce some notation and terminology repeatedly used below .
we define @xmath40 to be @xmath41 , the _ ordinary discal _
@xmath0-orbifold , @xmath42 to be @xmath41 with singular set a trivially embedded arc with arbitrary order @xmath32 , and @xmath43 to be @xmath41 with singular set a trivially embedded `` y - graph '' with edges of orders @xmath44 .
we will call @xmath42 and @xmath43 respectively _ cyclic discal _ and _ vertex discal _
@xmath0-orbifolds , and we will employ the shortened notation @xmath45 and @xmath46 to denote cyclic and vertex discal @xmath0-orbifolds with generic orders .
we also define the _ ordinary , cyclic _ , and _ vertex spherical _
@xmath1-orbifolds , denoted respectively by @xmath47 , @xmath48 , and @xmath49 , as the 2-orbifolds bounding the corresponding discal 3-orbifolds @xmath40 , @xmath42 , and @xmath43 .
we also define the _ ordinary , cyclic _ , and _ vertex spherical _ @xmath0-orbifolds , denoted respectively by @xmath50 , @xmath51 , and @xmath52 , as the 3-orbifolds obtained by mirroring the corresponding discal 3-orbifolds @xmath40 , @xmath42 , and @xmath43 in their boundary .
the spherical 2- and 3-orbifolds with generic orders will be denoted by @xmath53 and @xmath54 .
[ [ suborbifolds - and - irreducible-3-orbifolds ] ] 2-suborbifolds and irreducible 3-orbifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we say that a @xmath1-orbifold @xmath8 is a _
suborbifold _ of a @xmath0-orbifold @xmath2 if @xmath25 is embedded in @xmath28 so that @xmath25 meets @xmath29 transversely ( in particular , it does not meet the vertices ) , and @xmath26 is given precisely by @xmath55 , with matching orders .
a spherical @xmath1-suborbifold @xmath8 of a @xmath0-orbifold @xmath2 is called _ essential _ if it does not bound in @xmath2 a discal @xmath0-orbifold .
a @xmath0-orbifold is called _ irreducible _ if it does not contain any bad 2-suborbifold and every spherical 2-suborbifold is inessential ( in particular , it is separating ) .
if a @xmath0-orbifold @xmath2 is bounded by spherical @xmath1-orbifolds , we can canonically associate to @xmath2 a closed orbifold @xmath56 by attaching the appropriate discal @xmath0-orbifold to each component of @xmath57 .
we say that @xmath56 is obtained by _ capping _
@xmath2 .
whenever a @xmath0-orbifold is not irreducible , we can select an essential spherical @xmath1-suborbifold @xmath8 , split @xmath2 along @xmath8 , and cap the result . a naf statement of the theorem we want to prove
would be that _ the successive process of splitting and capping comes to an end in finite time , and the result is a collection of irreducible @xmath0-orbifolds , which depends on @xmath2 only_. with the appropriate assumptions and choices of the essential spherical @xmath1-orbifolds this is eventually true ( theorem [ splitting : intro : teo ] ) but , just as in the case of manifolds , the statement and proof are best understood by performing a simultaneous splitting along a system of spherical @xmath1-orbifolds , rather than a successive splitting along single ones .
[ [ connected - sum ] ] connected sum + + + + + + + + + + + + + to motivate the statement of the main result given below , we reverse our point of view .
the remark here is that , if @xmath2 can be turned into a collection of irreducible @xmath0-orbifolds by the `` split and cap '' strategy mentioned above , then @xmath2 can be reconstructed from the same collection by `` puncture and glue '' operations . as in the case of manifolds , each such operation
will be called a _
connected sum_. but the world of orbifolds is somewhat more complicated than that of manifolds , so we need to give the definition with a little care .
let @xmath58 and @xmath59 be 3-orbifolds , and pick points @xmath60 , such that one of the following holds : 1 . both @xmath61 and @xmath62 are non - singular points .
both @xmath61 and @xmath62 are singular but not vertex points , and they belong to singular edges of the same order @xmath32 .
3 . both @xmath61 and @xmath62 are vertex singular points , and the triple @xmath44 of orders of the incident edges is the same for @xmath61 and @xmath62
. we can then remove from @xmath63 a regular neighbourhood of @xmath64 , and glue together the resulting orbifolds by a homeomorphism which matches the singular loci .
the result is a @xmath0-orbifold @xmath2 that we call a _ connected sum _ of @xmath58 and @xmath59 .
depending on which of the conditions ( i ) , ( ii ) , or ( iii ) is met , we call the operation ( and its result ) , a connected sum of _ ordinary _ type , of _ cyclic _ type _ ( of order @xmath32 ) _ , or of _ vertex _ type _ ( of order @xmath44)_. [ [ trivial - connected - sums ] ] trivial connected sums + + + + + + + + + + + + + + + + + + + + + + in the case of manifolds , there is only one type of operation of connected sum , and its only identity element is the ordinary @xmath0-sphere . for @xmath0-orbifolds we have the easy : _
the unique identity element of the ordinary connected sum ( respectively , of the order-@xmath32 cyclic connected sum , and of the the order-@xmath65 vertex connected sum ) is @xmath50 ( respectively , @xmath51 , and @xmath52 ) .
_ we can then define to be _ trivial _ an operation of connected sum with the corresponding identity element .
however , when one considers a sequence of operations of connected sums , one must be careful , because a sum which appears to be non - trivial in the first place may actually turn out to be trivial later on , as the following example , suggested by joan porti , shows .
[ joan : example ] _ let @xmath66 be any non - trivial knot in @xmath16 , and let @xmath67 be distinct integers . construct a @xmath0-orbifold @xmath2 by first taking the ordinary connected sum of @xmath51 and @xmath68 , and then taking the order-@xmath32 cyclic connected sum of the result with @xmath69 , the orbifold with support @xmath16 and singular set @xmath66 of order @xmath32
. then both sums are non - trivial .
however , if we reverse the order , we can realize the same @xmath2 as the order-@xmath32 cyclic sum of @xmath51 and @xmath69 , followed by the ordinary sum with @xmath68 , and the first sum is now trivial . _ this phenomenon is of course a serious obstacle to proving that each @xmath0-orbifold has a unique expression as a connected sum of irreducible ones . to overcome this obstacle
, we need to redefine the notion of trivial connected sum .
[ [ graphs - and - efficient - connected - sums ] ] graphs and efficient connected sums + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider in this paragraph successive connected sums @xmath70 , stipulating with this notation that the first sum is between @xmath58 and @xmath71 , the second sum is between @xmath59 and @xmath72 , and so on .
note first that to perform the @xmath73-th sum we can always arrange the puncture made to @xmath74 , to be disjoint from the spherical @xmath1-orbifolds along which the punctured @xmath75 have been glued .
this implies that @xmath76 is glued to precisely one of @xmath75 .
we can then construct a graph associated to the connected sum , with nodes the @xmath76 s , and edges labelled by the types of sums performed .
note that this graph is actually a tree .
we will say that a connected sum @xmath70 is _ efficient _ if the @xmath76 s are irreducible and in its graph there is no node @xmath50 with an incident edge of ordinary type , there is no node @xmath77 with an incident edge of cyclic type , and there is no node @xmath78 with an incident edge of vertex type , see fig .
[ inefficientsums : fig ] .
[ inefficientsums : fig ] one should notice that the graph associated to @xmath70 is not quite unique , since to construct it one needs to isotope each puncture away from the previous ones , as described above .
however , one easily sees that two graphs of @xmath70 differ by moves as in fig .
[ manygraphs : fig ] , [ manygraphs : fig ] and one deduces that the notion of efficient connected sum is indeed well - defined .
the following is an alternative statement of our main result : [ connected : sum : teo ] let @xmath2 be a closed locally orientable @xmath0-orbifold .
suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating .
then @xmath2 can be realized as an efficient connected sum of irreducible @xmath0-orbifolds .
any two realizations involve the same irreducible summands and the same types of sums ( including orders ) .
[ [ efficient - splitting - systems ] ] efficient splitting systems + + + + + + + + + + + + + + + + + + + + + + + + + + + as in the case of manifolds , to establish the existence part of theorem [ connected : sum : teo ] one proceeds with the original strategy of `` split and cap '' first mentioned above , but one has to split symultaneously along a ( finite ) system of disjoint spherical @xmath1-orbifolds . to this end
we give the following definitions .
we say that a @xmath0-orbifold is _ punctured discal _ if it is obtained from some @xmath79 by removing a regular neighbourhood of a finite set .
a system @xmath3 of spherical @xmath1-suborbifolds of @xmath2 is called _ essential _ if no component of @xmath4 is punctured discal , and it is called _ coirreducible _ if all the components of @xmath80 are irreducible .
we call _ efficient splitting system _ a finite system of spherical @xmath1-suborbifolds of @xmath2 which is both essential and coirreducible .
_ to each system @xmath3 of separating spherical @xmath1-suborbifolds there corresponds a realization of @xmath2 as a connected sum of the components of @xmath80 .
moreover @xmath3 is efficient if and only if the connected sum is . _
the result we will establish directly below , which is obviously equivalent to theorems [ basic : splitting : teo ] and [ connected : sum : teo ] , and will be used to deduce theorem [ splitting : intro : teo ] , is the following : [ efficient : splitting : teo ] let @xmath2 be a closed locally orientable @xmath0-orbifold .
suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating .
then @xmath2 admits efficient splitting systems .
any two splitting systems coincide as abstract collections of spherical @xmath1-orbifolds , and the capped components of their complements also coincide .
we illustrate in greater detail in this section why it has been necessary to substantially modify the proof of the splitting theorem in passing from the manifold to the orbifold case . concerning the issue of non - separating spherical @xmath1-orbifolds ( point ( a ) of the introduction )
we begin by recalling what is the point for an orientable manifold @xmath9 .
if @xmath81 is a non - separating sphere , one chooses a simple closed curve @xmath82 meeting @xmath8 once and transversely and one takes the boundary @xmath83 of a regular neighbourhood of @xmath84 .
then @xmath83 is again a sphere , it is separating , and one of the capped components of @xmath85 is @xmath10 or @xmath11 , so @xmath9 has one such connected summand .
the very basis of this argument breaks down for orbifolds , because if @xmath8 is a spherical @xmath1-orbifold of cyclic or vertex type , then @xmath83 is not spherical at all ( see fig .
[ nonsep : fig ] ) .
[ nonsep : fig ] we will devote section [ prime : section ] to showing that indeed there is a wealth of @xmath0-orbifolds which contain non - separating spherical @xmath1-orbifolds but can not be expressed as a non - trivial connected sum . turning to points
( b ) and ( c ) of the introduction , recall that a family @xmath3 of spherical @xmath1-suborbifolds of a 3-orbifold @xmath2 is called _ essential _ if no component of @xmath17 is a punctured discal @xmath0-orbifold . to prove finiteness of a maximal essential family using the analogue of orbifolds of haken s normal surface theory
, one should be able to prove that the property of being essential is stable under the `` normalization moves . ''
now one can see ( as we will in section [ uniqueness : section ] ) , that the normalization moves boil down to isotopy and compression along ordinary discs ( _ i.e. _ discs without singular points ) .
and we have the following : [ non : b : exa ] then @xmath2 does not contain any bad @xmath1-suborbifold , every spherical @xmath1-suborbifold of @xmath2 is separating , @xmath3 is essential , @xmath19 and @xmath20 are systems of spherical @xmath1-orbifolds , and they are both inessential in @xmath2
. the failure of the direct orbifold analogue of normal surfaces to establish existence of a maximal essential system may appear to be a technical difficulty , only calling for a different proof of existence .
but the next example , which refers to point ( c ) in the introduction , shows that the notion itself of _ essential _ system does not do the job it was designed for : [ non : c : exa ] @xmath2 does not contain any bad @xmath1-suborbifold , every spherical @xmath1-suborbifold of @xmath2 is separating , @xmath19 is a maximal essential system spherical @xmath1-orbifolds in @xmath2 , but not all the capped components of @xmath86 are irreducible . _ to understand the last two assertions , note that @xmath3 is not essential , because the @xmath0-orbifold in fig .
[ essnoirred : fig ] is indeed punctured discal , but the cyclic component of @xmath3 is essential after cutting along @xmath19 and capping , because it corresponds to a _
cyclic _ connected sum with @xmath52 , which is non - trivial_. it is perhaps worth pointing out the basic fact underlying the previous example , namely that _ a punctured spherical @xmath0-orbifold is a discal one if and only if the puncture has the same type as the original @xmath0-orbifold_. concerning uniqueness ( point ( d ) of the introduction ) , all we had to say about its statement was said in section [ statements : section ] .
about its proof , we only note that milnor s one for manifolds was based on the same method proving existence , _
i.e. _ on the notion of essential spherical system , so of course our proof will be somewhat different .
in this section we prove the existence parts of theorems [ efficient : splitting : teo ] and [ connected : sum : teo ] .
we do so by showing that the process described in theorem [ splitting : intro : teo ] can be carried out with all the components of the splitting system simultaneously normal with respect to a fixed triangulation .
[ [ triangulations - and - normal-2-orbifolds ] ] triangulations and normal @xmath1-orbifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we slightly relax the definition given in @xcite , and define a _
triangulation _ of a @xmath0-orbifold @xmath2 as a triangulation of @xmath28 which contains @xmath29 as a subcomplex of its one - skeleton .
we then define a @xmath1-suborbifold of @xmath2 to be normal with respect to a triangulation if it intersects each tetrahedron in a union of squares and triangles , just as in the case of a surface in a triangulated manifold .
the following fundamental result holds : [ normal : finiteness : prop ] let @xmath2 be a triangulated @xmath0-orbifold .
then there exists a number @xmath87 such that , if @xmath88 is a normal surface with respect to the given triangulation and @xmath88 has more than @xmath87 connected components , then at least one of the components of @xmath89 is a product orbifold .
this result is established exactly as in the case of manifolds , so we omit the proof .
one only needs to notice that if a region of @xmath89 is a product as a manifold then it is also a product as an orbifold . [
[ normalization - moves ] ] normalization moves + + + + + + + + + + + + + + + + + + + we now recall that , in a manifold with a fixed triangulation @xmath90 , there is a general strategy to replace a surface by a normal one , as explained for instance in @xcite . when the surface is a sphere @xmath8 , this strategy consists ( besides isotopy relative to @xmath90 ) of three `` normalization '' moves . _
the first ( respectively , second ) move consists in compressing @xmath8 along a disc @xmath91 contained in the interior of a tetrahedron ( respectively , triangle ) of @xmath90 , thus getting two spheres @xmath83 and @xmath92 , and choosing either @xmath83 or @xmath92 . _
the third move is used to remove the double intersections of the normal discs with the edges , and it is illustrated in fig .
[ secondnorm : fig ] . [ secondnorm : fig ] as a matter of fact
, this move can also be seen as compression along a disc followed by the choice of one of the resulting spheres .
more precisely , one compresses along the disc in fig .
[ secondnorm : fig]-right and one discards the sphere given by the union of the two discs in fig . [ secondnorm : fig ] .
therefore , _
the third move consists of the compression along a disc disjoint from the @xmath93-skeleton of @xmath90 , followed by the elimination of a sphere which is the boundary of a regular neighbourhood of an interior point of an edge of @xmath90 . _
[ [ stepwise - normalization ] ] stepwise normalization + + + + + + + + + + + + + + + + + + + + + + as already mentioned , to prove the existence of an efficient splitting system we will construct a system of spherical @xmath1-orbifolds along the steps of theorem [ splitting : intro : teo ] , making sure each new component of the system is normal with respect to a fixed triangulation of the original orbifold . to show that this is possible
, we start with an easy lemma , that we will use repeatedly below . for its statement
, we stipulate that a spherical 2-orbifold of cyclic type is _ more complicated _ than one of ordinary type , and a that a spherical 2-orbifold of vertex type is _ more complicated _ than one of ordinary or cyclic type .
[ discal : cap : lem ] let @xmath94 be a punctured discal @xmath0-orbifold , and let @xmath8 be a component of @xmath95 .
suppose that no component of @xmath95 is more complicated than @xmath8 . then capping all the components of @xmath95 except @xmath8 we get a discal @xmath0-orbifold .
the next two results are the core of our argument on stepwise normalization
. the first one will be used again in the next section . from now on we fix a @xmath0-orbifold @xmath2 and we suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating . [
ess : ess : prop ] let @xmath3 be an essential system of spherical @xmath1-suborbifolds of @xmath2 .
let @xmath5 be a component of @xmath4 and @xmath8 be a spherical @xmath1-suborbifold of @xmath5 .
suppose that : * @xmath8 is essential in @xmath96 .
* all the components of @xmath3 are at most as complicated as @xmath8 .
then @xmath97 is essential .
suppose that some component @xmath98 of @xmath99 is punctured discal .
the assumptions that @xmath3 is essential and that @xmath8 is separating imply that @xmath98 is contained in @xmath5 and @xmath8 is a boundary component of @xmath98 . by lemma [ discal :
cap : lem ] and the second assumption , capping all the components of @xmath100 but @xmath8 we get a discal @xmath0-orbifold , which implies that @xmath8 bounds a discal @xmath0-orbifold in @xmath101 : a contradiction to the first assumption .
[ step : norm : prop ] fix a triangulation @xmath90 of @xmath2 .
let @xmath3 be an essential system of spherical @xmath1-suborbifolds in normal position with respect to @xmath90 .
let @xmath5 be a component of @xmath4 and @xmath8 be a spherical @xmath1-suborbifold of @xmath5 .
suppose that : * @xmath8 is essential in @xmath96 .
* all the spherical @xmath1-suborbifolds of @xmath96 strictly less complicated than @xmath8 are inessential in @xmath96 .
then @xmath8 can be replaced by a spherical @xmath1-orbifold satisfying the same properties and such that @xmath97 is in normal position with respect to @xmath90 .
we must show that a normalization move applied to @xmath8 can be realized without touching @xmath3 , without changing the type of @xmath8 , and preserving the property that @xmath8 is essential @xmath96 .
the first assertion is a straight - forward consequence of the description of the normalization moves given above . for the second and third assertions ,
we first note that in all cases the compression of a normalization move takes place along a disc disjoint from the @xmath93-skeleton of @xmath90 , whence along an ordinary ( non - singular ) disc @xmath91 which meets @xmath102 in @xmath103 only . to fix the notation ,
let @xmath104 and @xmath105 be the discs bounded by @xmath106 on @xmath8 , and define @xmath107 and @xmath108 . by the assumptions that @xmath2 contains no bad 2-suborbifolds and that @xmath91 is an ordinary disc , we can ( and will ) assume up to permutation that @xmath92 is an ordinary sphere , we now claim that @xmath83 or @xmath92 is essential in @xmath96 . before proving the claim ,
let us see how it implies the conclusion .
since @xmath92 is ordinary , we have @xmath109 .
if @xmath83 is essential , we are done .
otherwise @xmath92 is essential ( and ordinary ) , so @xmath8 is also ordinary by the second assumption , and we can switch @xmath83 and @xmath92 . to prove the claim ,
suppose by contradiction that @xmath83 and @xmath92 are both inessential in @xmath96 . since @xmath92 is ordinary , we deduce that @xmath8 bounds in @xmath101 either a discal @xmath0-orbifold glued along a 2-disc to an ordinary 3-disc , or a 3-orbifold obtained from a discal one by removing an ordinary 3-disc which intersects the boundary in a 2-disc .
the result is in both cases a discal @xmath0-orbifold ( isomorphic to the original one ) , whence the conclusion .
the result just proved easily implies the following : if @xmath2 is triangulated then one can carry out the process of theorem [ splitting : intro : teo ] with the splitting system @xmath3 in normal position at each step .
this corollary , together with propositions [ normal : finiteness : prop ] and [ ess : ess : prop ] immediately yields the result we were heading for : @xmath2 admits finite efficient splitting systems . as promised in the introduction , we explain here why it is crucial for us to assume that all the spherical @xmath1-suborbifolds of @xmath2 are separating .
first of all , the definition itself of essential system would not be appropriate without this assumption , because a non - separating spherical 2-orbifold could constitute an inessential system .
in addition , the assumption was used in a fundamental way in the proof of proposition [ ess : ess : prop ] .
proposition [ step : norm : prop ] also has the next consequence , which proves the last assertion of theorem [ splitting : intro : teo ] : let @xmath98 be a triangulated orbifold and let @xmath8 be an essential spherical @xmath1-suborbifold of @xmath98 .
suppose that all the spherical @xmath1-suborbifolds of @xmath98 strictly less complicated than @xmath8 are inessential .
then there exists an essential spherical @xmath1-suborbifold of @xmath98 of the same type as @xmath8 and in normal position with respect to the fixed triangulation .
in this section we conclude the proofs of theorems [ efficient : splitting : teo ] , [ connected : sum : teo ] , and [ splitting : intro : teo ] . as above
we fix a @xmath0-orbifold @xmath2 and we suppose that @xmath2 does not contain any bad @xmath1-suborbifold , and that every spherical @xmath1-suborbifold of @xmath2 is separating . [
[ gluing - of - discal-3-orbifolds ] ] gluing of discal @xmath0-orbifolds + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our uniqueness result uses the following two easy technical lemmas .
we include a proof of the second one , leaving to the reader the even easier proof of the first one .
[ disc : attach : lem ] let @xmath110 be a @xmath0-orbifold and let @xmath111 be a disc .
let @xmath94 be a punctured discal @xmath0-orbifold .
let @xmath8 be a component of @xmath95 of the same type as @xmath94 .
split @xmath25 along a loop into two discs @xmath104 and @xmath105 , and suppose that @xmath112 are isomorphic to each other , with at most one singular point .
let @xmath113 be obtained from @xmath110 by gluing @xmath94 along a homeomorphism @xmath114 .
then : 1 .
@xmath113 is punctured discal if and only if @xmath110 is .
@xmath115 .
[ sphere : attach : lem ] let @xmath110 be a @xmath0-orbifold with a spherical boundary component @xmath8 .
let @xmath94 be a punctured discal @xmath0-orbifold with at least two boundary components @xmath116 isomorphic to @xmath8 , and all other components not more complicated than @xmath83 and @xmath92 .
let @xmath113 be obtained from @xmath110 by gluing @xmath94 along a homeomorphism @xmath117 .
then : 1 .
@xmath113 is punctured discal if and only if @xmath110 is .
@xmath115 .
we begin by ( 2 ) .
since @xmath83 and @xmath92 are at least as complicated as all the other components of @xmath95 , capping the components different from @xmath83 and @xmath92 we get a product orbifold , which implies the conclusion at once .
let us turn to ( 1 ) .
to be discal for a @xmath0-orbifold @xmath98 means that @xmath118 is some @xmath119 and some component of @xmath100 , or more precisely the most complicated one , is @xmath120 .
now we already know that @xmath115 , and the assumptions imply that the most complicated component of @xmath121 has the same type as the most complicated component of @xmath122 , whence the conclusion .
[ [ finiteness - and - uniqueness - scheme - of - the - proof ] ] finiteness and uniqueness : scheme of the proof + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to conclude the proofs of the results stated so far we must establish the finiteness of the splitting process of theorem [ splitting : intro : teo ] and the uniqueness part of theorems [ splitting : intro : teo ] , [ connected : sum : teo ] , and [ efficient : splitting : teo ] .
the former easily follows from the next result , to be proved below , and proposition [ ess : ess : prop ] .
[ lexi : prop ] let @xmath3 and @xmath19 be finite systems of spherical @xmath1-suborbifolds of @xmath2
. denote by @xmath123 ( respectively , @xmath124 ) the number of components of @xmath3 ( respectively , @xmath19 ) of type @xmath125 .
suppose that @xmath3 is coirreducible and @xmath19 is essential .
then @xmath126 in lexicographical order .
uniqueness easily follows from the next result ( for theorem [ splitting : intro : teo ] one only has to note that the final @xmath3 is obviously coirreducible , and essential by proposition [ ess : ess : prop ] ) : [ uniqueness : prop ] let @xmath3 and @xmath19 be efficient spherical splitting systems for @xmath2
. then @xmath127 and @xmath128 as abstract collections of @xmath1- and @xmath0-orbifolds respectively . [
[ moves - for - spherical - systems ] ] moves for spherical systems + + + + + + + + + + + + + + + + + + + + + + + + + + + we will prove propositions [ lexi : prop ] and [ uniqueness : prop ] in a unified manner .
the underlying idea in both cases is that by certain moves we can modify @xmath3 into a superset of @xmath19 , leaving unaffected the properties of @xmath3 and the topological types of @xmath3 and @xmath80 as abstract orbifolds .
there are two types of move we will need , that we now describe .
these moves apply to any spherical system @xmath129 , under the appropriate assumptions .
_ move @xmath82_.
suppose we have a component @xmath130 of @xmath3 and a disc @xmath131 properly embedded in a component @xmath132 of @xmath4 , with @xmath133 .
let @xmath134 split @xmath130 into discs @xmath135 and @xmath91 , and suppose that @xmath135 and @xmath131 both have at most one singular point .
assume that the component containing @xmath135 of @xmath136 is a punctured disc @xmath94 , and that @xmath95 does not have components more complicated than @xmath137 .
then we replace @xmath138 by @xmath139 .
see fig .
[ alphamove : fig ] .
[ alphamove : fig ] _ move @xmath140_. suppose we have a component @xmath130 of @xmath3 and a spherical @xmath1-orbifold @xmath141 contained in a component @xmath132 of @xmath4 , with @xmath142 .
suppose that the component containing @xmath130 of @xmath143 is a punctured disc @xmath94 , that @xmath130 and @xmath141 have the same type , and that no other component of @xmath95 is more complicated than @xmath130 and @xmath141 .
then we replace @xmath130 by @xmath141 .
see fig .
[ betamove : fig ] .
[ betamove : fig ] [ abmoves : lem ] under any move of type @xmath82 or @xmath140 the following happens : * the type of @xmath3 ( including orders ) does not change .
* the type of @xmath80 does not change ( in particular , coirreducibility is preserved ) .
* essentiality is preserved .
the first assertion is proved by direct examination of figg .
[ alphamove : fig ] and [ betamove : fig ] . to prove the second and third assertions we note the only components of @xmath4 modified by the move are @xmath132 and the other component @xmath144 incident to @xmath130 , and these components
are replaced by the components @xmath145 and @xmath146 incident to @xmath141 .
the relations between them are described as follows : @xmath147 it is now easy to check that in all cases we are under the assumptions of lemmas [ disc : attach : lem ] and [ sphere : attach : lem ] , whence the conclusion at once .
[ [ finiteness - and - uniqueness - conclusion ] ] finiteness and uniqueness : conclusion + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as already mentioned , to prove propositions [ lexi : prop ] and [ uniqueness : prop ] we want to replace @xmath3 by a superset of @xmath19 .
the first step is as follows : [ no : intersection : prop ] suppose that @xmath3 is coirreducible and @xmath19 is essential .
then , up to applying moves @xmath82 and isotopy to @xmath3 , we can suppose that @xmath148 .
let us first isotope @xmath3 so that it is transversal to @xmath19 .
of course it is sufficient to show that if @xmath149 is non - empty then the number of its components can be reduced by a move @xmath82 and isotopy .
suppose some component @xmath83 of @xmath19 meets @xmath3 , and note that the finite set of loops @xmath150 bounds at least two innermost discs in @xmath83 .
since @xmath83 has at most three singular points , we can choose an innermost disc @xmath131 which contains at most one singular point .
let @xmath132 be the component of @xmath4 which contains @xmath131 , let @xmath130 be the component of @xmath3 containing @xmath134 , and denote by @xmath135 and @xmath91 the discs into which @xmath134 cuts @xmath130 , with @xmath135 at most as complicated as @xmath91 .
the assumption that @xmath2 contains no bad 2-suborbifolds easily implies that @xmath137 is a spherical @xmath1-orbifold of ordinary or cyclic type .
coirreducibility of @xmath3 then implies that @xmath137 bounds an ordinary or cyclic discal @xmath0-orbifold @xmath151 in @xmath152 .
now we either have @xmath153 or @xmath154 . in the first case it is easy to see that we are precisely in the situation where we can apply a move @xmath82 , after which a small isotopy allows to reduce the number of components of @xmath149 . in the second case @xmath155 bounds in @xmath152 one of the manifolds obtained from @xmath151 by cutting along @xmath91 .
since @xmath151 is an ordinary or cyclic disc we deduce that @xmath91 is isomorphic to @xmath131 and that @xmath156 bounds a ball in @xmath152 disjoint from @xmath135 .
we can then interchange the roles of @xmath91 and @xmath135 and apply @xmath82 also in this case , whence the conclusion .
propositions [ lexi : prop ] and [ uniqueness : prop ] now easily follow from proposition [ no : intersection : prop ] and the next result : [ ord : fix : prop ] suppose that @xmath3 is coirreducible and @xmath19 is essential .
assume that each component of @xmath19 is either contained in @xmath3 or disjoint from @xmath3 .
then : * up to applying moves @xmath140 and isotopy to @xmath3 , we can suppose that each ordinary component of @xmath19 is contained in @xmath3 .
* if @xmath3 and @xmath19 have the same ordinary components , up to applying moves @xmath140 and isotopy to @xmath3 , we can suppose that each cyclic component of @xmath19 is contained in @xmath3 .
* if @xmath3 and @xmath19 have the same ordinary and cyclic components , up to applying moves @xmath140 and isotopy to @xmath3 , we can suppose that each vertex component of @xmath19 is contained in @xmath3 .
in all the cases , we suppose that a certain component @xmath141 of @xmath19 of the appropriate type is not contained @xmath3 , and we show that a move @xmath140 can be applied to @xmath3 which replaces some component @xmath130 by @xmath141 .
the conclusion then easily follows by iteration . in all cases
we denote by @xmath132 the component of @xmath4 which contains @xmath141 , and we choose @xmath130 among the components of @xmath157 .
we start with the first assertion , so we assume @xmath141 is ordinary .
coirreducibility of @xmath3 implies that @xmath141 bounds an ordinary 3-disc @xmath151 in @xmath152 .
let us choose a component of @xmath19 contained in @xmath151 , not contained in @xmath3 , and innermost in @xmath151 with respect to these properties , and let us rename it @xmath141 .
( note that the new @xmath141 may not be parallel to the old one in @xmath2 , even if it is in @xmath151 ) .
of course @xmath141 is again an ordinary sphere , and it bounds in @xmath5 a punctured ordinary disc @xmath94 such that all the components of @xmath95 except @xmath141 belong to @xmath3 .
essentiality of @xmath19 now implies that there must be a component @xmath130 of @xmath95 which does not belong to @xmath19 .
we have created a situation where a move @xmath140 can be applied , so we are done .
the proof is basically the same for the second and third assertions .
we only need to note that @xmath151 always has the same type as @xmath141 and that the component @xmath130 of @xmath95 not belonging to @xmath19 can not be of type strictly less complicated than @xmath141 , by the assumption that @xmath3 and @xmath19 have the same components of type strictly less complicated than that of @xmath141 .
in this section we treat point ( d ) of the introduction , showing that there are infinitely many @xmath0-orbifolds with non - separating but no essential separating spherical @xmath1-suborbifolds .
our examples arose from discussions with luisa paoluzzi .
as in the case of manifolds , one can define a prime orbifold as one that can not be expressed as a non - trivial connected sum , _
i.e _ an orbifold in which every _ separating _ spherical @xmath1-suborbifold bounds a discal @xmath0-suborbifold .
of course every irreducible orbifold is also prime , but we will show now that there are infinitely many primes which are not irreducible .
this appears as a sharp difference between the orbifold and the manifold case ( in which the only non - irreducible primes are the two @xmath158-bundles over @xmath159 ) .
this difference represents a serious obstacle to promoting theorem [ connected : sum : teo ] to a unique decomposition theorem for orbifolds into prime ones .
[ non : irred : prop ] let @xmath2 be a @xmath0-orbifold bounded by a sphere with @xmath160 cone points of the same order @xmath32 .
assume that : * @xmath2 is irreducible .
* @xmath2 is not the @xmath0-ball with two parallel unknotted singular arcs . * in @xmath2 there is no sphere with three singular points which meets @xmath57 in a disc with two singular points .
consider the orbifold with support @xmath10 and singular set given by two circles of order @xmath32 parallel to the factor @xmath159 .
remove from it a regular neighbourhood of an arc which joins the singular components within a level sphere @xmath161 , and call @xmath5 the result .
let @xmath98 be obtained by attaching @xmath2 and @xmath5 along their boundary spheres , matching the cone points .
then @xmath98 is prime but not irreducible .
a level sphere @xmath161 contained in @xmath5 gives a non - separating spherical @xmath1-suborbifold of @xmath98 , so @xmath98 is not irreducible .
let us consider a separating spherical @xmath1-orbifold @xmath8 in @xmath98 , and the intersection of @xmath8 with the sphere @xmath130 along which @xmath2 and @xmath5 have been glued together .
if @xmath162 is empty then @xmath5 bounds a spherical @xmath0-orbifold , because @xmath2 and @xmath5 are irreducible .
let us then assume @xmath162 to be transverse and minimal up to isotopy of @xmath8 . considering the pattern of circles @xmath162 on @xmath8 we see that @xmath8 contains at least two innermost discs , which are therefore contained either in @xmath2 or in @xmath5 .
since @xmath8 has at most three singular points , we can find one such disc @xmath91 with @xmath163 points .
now consider the loop @xmath106 on @xmath130 , and recall that @xmath130 has @xmath160 singular points
. then @xmath106 encircles @xmath164 singular points .
we will now show that both if @xmath165 and if @xmath166 the conditions @xmath163 and @xmath164 are impossible , whence the conclusion .
suppose first that @xmath165 .
using the fact that @xmath2 is irreducible , the cases @xmath167 and @xmath168 both contradict the minimality of @xmath162 .
the cases @xmath169 and @xmath170 both lead to a bad 2-suborbifold of @xmath2 , which does not exist . in the case @xmath171 , using irreducibility twice , we see that @xmath2 must be the @xmath0-ball with two parallel unknotted singular arcs , but this orbifold was specifically forbidden , so indeed we get a contradiction . the case @xmath172 was also specifically forbidden .
suppose now that @xmath166 .
recalling that @xmath91 is contained in a sphere which is separating in @xmath98 , we see that @xmath91 cobounds a ball with a disc contained in @xmath173 . using
this fact we can prove again that all the possibilities with @xmath164 and @xmath163 are impossible .
as above , @xmath167 and @xmath168 contradict minimality .
the case @xmath171 would imply that one of the singular arcs of @xmath5 can be homotoped to @xmath173 , which is not the case .
the other cases are absurd for similar reasons .
[ [ examples ] ] examples + + + + + + + + whenever @xmath174 and the cone angles are chosen in an admissible fashion , the `` stairway '' of fig . [ stairs : fig ] contained in the 3-ball [ stairs : fig ] provides an example of orbifold @xmath2 satisfying the assumptions of proposition [ non : irred : prop ] . note that for such an @xmath2 the corresponding prime non - irreducible orbifold @xmath98 is always supported by @xmath10 , so @xmath175 is prime as a manifold .
but we can also construct examples of @xmath2 to which proposition [ non : irred : prop ] applies , leading to an orbifold @xmath98 with non - prime @xmath175 .
for instance , one can take an orbifold constructed exactly as @xmath5 was constructed from @xmath10 , starting instead with @xmath176 , where @xmath177 is a closed orientable surface @xmath177 of positive genus .
our interest in the spherical splitting of @xmath0-orbifolds arose from our program of extending matveev s complexity theory @xcite from the manifold to the orbifold context .
we actually succeeded @xcite in generalizing the definition and some of the most significant results , investigating in particular the behaviour under connected sum .
however it turns out that one does not have in general plain additivity as in the manifold case .
we can only prove additivity when there are no vertex connected sums , and we have that complexity is only additive up to a certain correction summand on cyclic connected sums involving at least one purely cyclic singular component .
for this reason in this section we prove that the number of such sums is well - defined : [ cyc:*:prop ] suppose that a @xmath0-orbifold @xmath2 is a connected sum of irreducible orbifolds without vertex connected sums , and fix an efficient realization @xmath178 of @xmath2 as @xmath179 . for all @xmath180 let @xmath181 be the number of @xmath32-cyclic connected sums in @xmath178 involving at least one singular component without vertices .
then @xmath181 is independent of @xmath178 , so it is a function of @xmath2 only . *
reordering of the @xmath182 s , without modification of the efficient system of spherical @xmath1-orbifolds along which @xmath2 is split . *
modification of the splitting system according to one of the moves @xmath82 and @xmath140 described in section [ uniqueness : section ] ( see figg .
[ alphamove : fig ] and [ betamove : fig ] ) .
to deal with the first operation , we define a modified version @xmath183 of the tree associated to the realization @xmath184 . to construct @xmath183
we associate to each @xmath182 one node for each cyclic component of @xmath185 and one for the union of the non - cyclic components of @xmath185 .
we denote by @xmath186 the set of nodes of cyclic type and by @xmath187 the set of nodes of non - cyclic type .
we now define @xmath183 by taking an edge for each @xmath32-cyclic connected sum , with the ends of the edge joining the singular components involved in the sum .
the ideas behind the construction of @xmath183 are as follows .
first , we can not create a new @xmath32-cyclic singular component by performing an ordinary connected sum or a @xmath188-cyclic connected sum for @xmath189 .
second , a cyclic connected sum between two cyclic components gives one cyclic component , while a cyclic connected sum between a non - cyclic component and any other one gives one or two non - cyclic components . with these facts in mind
, it is very easy to see that the invariance of @xmath181 under the first move is implied by the following graph - theoretical statement : _ let @xmath90 be a tree with set of nodes @xmath190 .
let @xmath191 be an ordering of the edges of @xmath90 . for @xmath192 define @xmath193 to be @xmath194 if both the ends of @xmath195 can be joined to nodes in @xmath187 by edges in @xmath196 only , and to be @xmath93 otherwise .
then @xmath197 is independent of @xmath198_. without affecting the sum in question .
this allows us to assume that all the nodes in @xmath187 are `` external , '' _
i.e. _ @xmath93-valent . moreover , using induction
, we can suppose that @xmath187 consists precisely of the external nodes , because if an edge @xmath199 has an end which is external and belongs to @xmath186 then @xmath200 whatever the position of @xmath199 in the ordering , and the value of @xmath82 on the remaining edges does not depend on the position of @xmath199 in the ordering .
[ graph : lem ] let @xmath90 be a tree .
let @xmath203 denote the set of external nodes of @xmath90 .
let @xmath191 be any ordering of the edges of @xmath90 .
define @xmath204 to be @xmath93 if the ends of @xmath195 can be joined to @xmath203 through edges in @xmath196 only , and to be @xmath194 otherwise .
then @xmath205 is independent of @xmath198 , and precisely equal to @xmath206 .
attach a circle to @xmath90 by gluing the nodes in @xmath203 to @xmath207 arbitrarily chosen points on the circle , and denote by @xmath5 the result .
since @xmath90 is a tree , we have @xmath208 .
let @xmath191 be an ordering of the edges of @xmath90 , let @xmath209 be the union of the circle with @xmath210 , and let @xmath211 be the connected component of @xmath209 which contains the circle .
of course @xmath212 is the circle and @xmath213 .
we now have the identity @xmath214 which is easily established by considering separately the cases where @xmath195 has @xmath194 , @xmath93 , or @xmath1 ends on @xmath215 , and using again the fact that @xmath90 is a tree .
the identity implies that @xmath216 to conclude the proof of proposition [ cyc:*:prop ] we must now show that @xmath181 is not affected by the moves @xmath82 and @xmath140 . knowing the independence of the ordering
, this is actually very easy .
for both the moves , we have two realizations @xmath217 and @xmath218 of @xmath2 leading to splitting systems @xmath219 and @xmath220 which differ for one component only , so @xmath221 and @xmath222 .
we then compute @xmath223 by performing first the connected sum which corresponds to @xmath224 . from figg .
[ alphamove : fig ] and [ betamove : fig ] one sees that the connected sums along @xmath130 and @xmath141 give the same contributions to @xmath225 . after performing them ,
all other sums in @xmath217 and @xmath218 are the same , which eventually proves the proposition . | the famous haken - kneser - milnor theorem states that every @xmath0-manifold can be expressed in a unique way as a connected sum of prime @xmath0-manifolds .
the analogous statement for @xmath0-orbifolds has been part of the folklore for several years , and it was commonly believed that slight variations on the argument used for manifolds would be sufficient to establish it .
we demonstrate in this paper that this is not the case , proving that the apparently natural notion of `` essential '' system of spherical @xmath1-orbifolds is not adequate in this context .
we also show that the statement itself of the theorem must be given in a substantially different way .
we then prove the theorem in full detail , using a certain notion of `` efficient splitting system . ''
msc ( 2000 ) : 57m99 ( primary ) , 57m50 , 57m60 ( secondary ) . |
modern wireless communication systems , e.g. , cellular networks and wireless sensor networks ( wsns ) , are featured by larger bandwidth , higher data rate and lower communication delays . the improvement on communication quality and the increased data processing complexity have imposed higher requirement on the quality of power supply to wireless devices ( wds ) .
conventionally , wds are powered by batteries , which have to be replaced / recharged manually once the energy is depleted .
alternatively , the recent advance of radio frequency ( rf ) enabled wireless power transfer ( wpt ) provides an attractive solution to power wds over the air @xcite .
by leveraging the far - field radiative properties of microwave , wds can harvest energy remotely from the rf signals radiated by the dedicated energy nodes ( ens ) @xcite .
compared to the conventional battery - powered methods , wpt can save the cost due to manual battery replacement / recharging in many applications , and also improve the network performance by reducing energy outages of wds .
currently , tens of microwatts ( @xmath0w ) rf power can be effectively transferred to a distance of more than @xmath1 meters .
meters is about @xmath2w . ]
the energy is sufficient to power the activities of many low - power communication devices , such as sensors and rf identification ( rfid ) tags . in the future
, we expect more practical applications of rf - enabled wpt to wireless communications thanks to the rapid developments of many performance enhancing technologies , such as energy beamforming with multiple antennas @xcite and more efficient energy harvesting circuit designs @xcite . in a wireless powered communication network ( wpcn ) , the operations of wds , including data transmissions , are fully / partially powered by means of rf - enabled wpt @xcite . a tdma ( time division multiple access ) based protocol for wpcn is first proposed in @xcite , where the wds harvest rf energy broadcasted from a hybrid access point ( hap ) in the first time slot , and then use the harvested energy to transmit data back to the hap in the second time slot .
later , @xcite extends the single - antenna hap in @xcite to a multi - antenna hap that enables more efficient energy transmission via energy beamforming as well as more spectrally efficient sdma ( space division multiple access ) based information transmission as compared to tdma . to further improve the spectral efficiency
, @xcite considers using full - duplex hap in wpcns , where a hap can transmit energy and receive user data simultaneously via advanced self - interference cancelation techniques . intuitively , using a hap ( or co - located en and information ap ) , instead of two separated en and information access point ( ap ) , to provide information and
energy access is an economic way to save deployment cost , and the energy and information transmissions in the network can also be more efficiently coordinated by the hap .
however , using hap has an inherent drawback that it may lead to a severe doubly - near - far " problem due to distance - dependent power loss @xcite .
that is , the far - away users quickly deplete their batteries because they harvest less energy in the downlink ( dl ) but consume more energy in the uplink ( ul ) for information transmission . to tackle this problem , separately located ens and aps
are considered to more flexibly balance the energy and information transmissions in wpcns @xcite . in this paper , we consider the method using either co - located or separate en and information ap to build a wpcn .
most of the existing studies on wpcns focus on optimizing real - time resource allocation , e.g. , transmit signal power , waveforms and time slot lengths , based on instantaneous channel state information ( csi , e.g. , @xcite ) . in this paper
, we are interested in the long - term network performance optimization based mainly on the average channel gains .
it is worth mentioning that network optimizations in the two different time scales are complementary to each other in practice .
that is , we use long - term performance optimization methods for the initial stage of network planning and deployment , while using short - term optimization methods for real - time network operations after the deployment .
many current works on wpcns use stochastic models to study the long - term performance because of the analytical tractability , especially when the wds are mobile in location .
for instance , @xcite applies a stochastic geometry model in a cellular network to derive the expression of transmission outage probability of wds as a function of the densities of ens and information aps .
similar stochastic geometry technique is also applied to wpt - enabled cognitive radio network in @xcite to optimize the transmit power and node density for maximum primary / secondary network throughput . however , in many application scenarios , the locations of the wds are fixed , e.g. , a sensor network with sensor ( wd ) locations predetermined by the sensed objects , or an iot ( internet - of - things ) network with static wds . in this case , a practical problem that directly relates to the long - term performance of wpcns , e.g. , sensor s operating lifetime , is to determine the optimal locations of the ens and aps .
nonetheless , this important node placement problem in wpcns is still lacking of concrete studies . in conventional battery - powered wireless communication networks ,
node placement problem concerns the optimal locations of information aps only , which has been well investigated especially for wireless sensor networks using various geometric , graphical and optimization algorithms ( see e.g. , @xcite ) .
however , there exist major differences between the node placement problems in battery - powered and wpt - enabled wireless communication networks . on one hand ,
a common design objective in battery - powered wireless networks is to minimize the highest transmit power consumption among the wds to satisfy their individual transmission requirements .
however , such energy - conservation oriented design is not necessarily optimal for wpcns , because high power consumption of any wd can now be replenished by means of wpt via deploying an en close to the wd . on the other hand , unlike information transmission , wpt will not induce harmful co - channel interference to unintended receivers , but instead can boost their energy harvesting performance .
these evident differences indicate that the node placement problem in battery - powered wireless communication networks should be revisited for wpcns , to fully capture the advantages of wpt .
in this paper , we study the node placement optimization problem in wpcns , which aims to minimize the deployment cost on ens and aps given that the energy harvesting and communication performances of all the wds are satisfied .
our contributions are detailed below . 1 .
we formulate the optimal node placement problem in wpcns using either separated or co - located en and ap . to simplify the analysis
, we then transform the minimum - cost deployment problem into its equivalent form that optimizes the locations of fixed number of ens and aps ; 2 .
the node placement optimization using separated en and ap is highly non - convex and hard to solve . to tackle the non - convexity of the problem
, we first propose an efficient cluster - based greedy algorithm to optimize the locations of ens given fixed ap locations .
then , a trial - and - error based algorithm is proposed to optimize the locations of aps given fixed ens locations .
based on the obtained results , we further propose an effective alternating method that jointly optimizes the en and ap placements ; 3 .
for the node placement optimization using co - located en and ap ( or hap ) , we extend the greedy en placement method under fixed aps to solving the hap placement optimization , which is achieved by incorporating additional considerations of dynamic wd - hap associations during hap placement .
specifically , a trial - and - error method is used to solve the wd - hap association problem , which eventually leads to an efficient greedy hap placement algorithm . due to the non - convexity of the node placement problems in wpcns ,
all the proposed algorithms are driven by the consideration of their applicabilities to large - size wpcns , e.g. , consisting of hundreds of wds and en / ap nodes .
specifically , we show that the proposed algorithms for either separated or co - located en and ap placement are convergent and of low computational complexity .
besides , simulations validate our analysis and show that the proposed methods can effectively reduce the network deployment cost to guarantee the given performance requirements .
the proposed algorithms may find their wide application in the future deployment of wpcns , such as wireless sensor networks and iot networks .
the rest of the paper is organized as follows . in section
ii , we first introduce the system models of wpcn where the ens and aps are either separated or co - located .
then , we formulate the optimal node placement problems for the two cases in section iii , and propose efficient algorithms to solve the problems in sections iv and v , respectively . in section vi , simulations are performed to evaluate the performance of the proposed node placement methods .
finally , we conclude the paper and discuss future working directions in section vii .
for the case of separated ens and aps , we consider in fig . [
61 ] a wpcn in @xmath3 consisting of @xmath4 ens , @xmath5 aps and @xmath6 wds , whose locations are denoted by @xmath7 coordinate vectors @xmath8 , @xmath9 , and @xmath10 , respectively .
we assume that the energy and information transmissions are performed on orthogonal frequency bands without interfering with each other .
specifically , the ens are connected to stable power source and broadcast rf energy in the dl for the wds to harvest the energy and store in their rechargeable batteries . at the same time , the wds use the battery power to transmit information to the aps in the ul .
the circuit structure of a wd to perform the above operations is also shown in fig .
[ 61 ] . in a transmission block of length @xmath11 ,
the @xmath4 ens transmit simultaneously on the same bandwidth in the dl , where each en @xmath12 transmits @xmath13,\ i=1,\cdots , m.\ ] ] here , @xmath14 denotes the transmit power , @xmath15 denotes the pseudo - random energy signal used by the @xmath12-th en , which is assumed to be of unit power ( @xmath16=1 $ ] ) and independent among the ens ( @xmath17 = 0 $ ] if @xmath18 ) .
the reason to use random signal instead of a single sinusoid tone is to avoid peak in transmit power spectrum density , for satisfying the equivalent isotropically radiated power ( eirp ) requirement enforced by spectrum regulating authorities @xcite .
notice that the energy beamforming technique proposed in @xcite is not used in our setup , as it requires accurate csi and dl symbol - level synchronization , which may be costly to implement in a highly distributed wpcn network considered in this work .
accordingly , the received energy signal by the @xmath19-th wd is @xmath20 where @xmath21 denotes the equivalent baseband channel coefficient from the @xmath12-th en to the @xmath19-th wd , which is assumed to be constant within a transmission block but may vary independently across different blocks .
let @xmath22 denote the channel power gain , which follows a general distribution with its mean determined by the distance between the en and wd , i.e. , @xmath23 = \beta ||\mathbf{u}_i -\mathbf{w}_k||^{-d_d } , \ \
i=1,\cdots , m , \ k = 1,\cdots k,\ ] ] where @xmath24 denotes the path loss exponent in dl , @xmath25 denotes the @xmath26-norm operator , and @xmath27 with @xmath28 and @xmath29 denoting the downlink antenna power gain and carrier frequency , respectively @xcite . then , each wd @xmath19 can harvest an average amount of energy from the energy transmission within each block given by @xcite @xmath30 = \eta t p_0 \left ( \mathsmaller\sum_{i=1}^m h_{i , k}\right ) , \ k= 1,\cdots , k , \end{aligned}\ ] ] where @xmath31 $ ] denotes the energy harvesting circuit efficiency , and the expectation is taken over the pseudo - random energy signal variations under fixed @xmath32 s within the transmission block .
let @xmath33/t$ ] denote the average energy harvesting rate over the variation of wireless channels ( @xmath32 s ) in different transmission blocks , we have @xmath34 in the ul information transmissions , we assume that each wd transmits data to only one of the aps . to make the placement problem tractable ,
the wd - ap associations are assumed to be fixed , where each wd @xmath19 transmits to its nearest ap @xmath35 regardless of the instantaneous csi , i.e. , @xmath36 here , we assume no co - channel interference for the received user signals from different wds , e.g. , the wds transmit on orthogonal channels .
besides , for the simplicity of analysis , we assume no limit on the maximum number of wds that an ap could receive data from . then , the average power consumption rate for wd @xmath19 is modeled as @xmath37 where @xmath38 denotes the constant circuit power of wd @xmath19 , @xmath39 denotes the average transmit power as a function of the distance between wd @xmath19 and its associated ap @xmath35 , and @xmath40 denotes the ul channel path loss exponent . besides , @xmath41 denotes a parameter related to the transmission strategy used in the ul communication . to achieve a target data rate and maximum allowed outage portability @xmath42 for wd @xmath19
, we have @xmath43 when truncated channel inverse transmission @xcite is used under rayleigh fading channel , where @xmath44 and @xmath45 denote the uplink antenna gain and carrier frequency , respectively , and @xmath46 denotes the exponential integral function . ] in general , the model in ( [ 11 ] ) indicates that the transmit power increases as a polynomial function of the distance between the transmitter and receiver to satisfy certain communication quality requirement , e.g. , minimum data rate or maximum allowed outage probability , which is widely used for wireless network performance analysis @xcite . a special case of the wpcn that we consider in fig . [ 61 ] is when the ens and aps are grouped into pairs and each pair of en and ap are co - located and integrated as a hybrid access point ( hap ) , which corresponds to setting @xmath47 and @xmath48 for @xmath49 . with the network model and hap s circuit structure shown in fig . [ 62 ] , a hap transfers rf power in the dl and receives information in the ul simultaneously on different frequency bands .
although the use of haps is less flexible in placing the ens and aps than with separated ens and aps , the overall deployment cost is reduced , because the production and operation cost of a hap is in general less than the sum - cost of two separate en and ap . for brevity
, we reuse the notation @xmath50 , @xmath49 , to denote the location coordinates of the @xmath4 haps . given other parameters unchanged ,
the expression of the average energy harvesting rate @xmath51 of the @xmath19-th wd is the same as that in ( [ 10 ] ) .
meanwhile , the average power consumption rate @xmath52 can be obtained from ( [ 11 ] ) by replacing @xmath53 with @xmath50 as follows .
@xmath54 where @xmath55 is the index of the hap that wd @xmath19 associates with , i.e. , @xmath56 with the above definitions , the _ net _ energy harvesting rates of the wds in both cases of separate and co - located en and ap are given by @xmath57 in practice , the net energy harvesting rate can directly translate to the performance of device operating lifetime ( see e.g. , @xcite ) .
specifically , given an initial battery level @xmath58 , the average time before the @xmath19-th wd s battery depletes is @xmath59 when @xmath60 , and @xmath61 when @xmath62 .
in other words , given a minimum device operating lifetime requirement @xmath63 , it must satisfy @xmath64 if @xmath65 , and @xmath62 if @xmath66 .
in this paper , we assume that the locations of the wds are known and study the optimal placement of ens and information aps , which are either separated or co - located in their locations .
this may correspond to a sensor network with sensor ( wd ) locations predetermined by the sensed objects , or an iot network with static wds .
in particular , we are interested in minimizing the deployment cost given that the net energy harvesting rates of all the wds are larger than a common prescribed value @xmath67 , i.e. , @xmath68 , where @xmath67 is set to achieve a desired device operating lifetime . when the ens and aps are separated , the total deployment cost is @xmath69 if @xmath4 ens and @xmath5 aps are used , where @xmath70 and @xmath71 are the monetary costs of deploying an en and an ap , respectively . to solve the minimum - cost deployment problem ,
let us first consider the following feasibility problem : [ 8 ] @xmath72,\ \mathbf{v}^n = \left[\mathbf{v}_1,\cdots,\mathbf{v}_n\right]\\ & \text{s . t. } & & \lambda_k \left(\mathbf{u}^m\right ) -\mu_k\left(\mathbf{v}^n\right ) \geq \gamma , \ \
k=1,\cdots , k , \label{56}\\ & & & \mathbf{b}^l\leq \mathbf{u}_i\leq \mathbf{b}^h,\ \ i=1,\cdots , m , \label{54}\\ & & & \mathbf{b}^l\leq \mathbf{v}_j\leq \mathbf{b}^h,\ \ j=1,\cdots , n , \label{55 } \end{aligned}\ ] ] where @xmath51 and @xmath52 are functions of @xmath73 and @xmath74 given in ( [ 10 ] ) and ( [ 11 ] ) , respectively . the inequalities in ( [ 54 ] ) and ( [ 55 ] ) denote element - wise relations . besides , @xmath75 specifies a feasible deployment area for both the ens and aps in @xmath3 , which is large enough to contain all the wds , i.e. , @xmath76 .
evidently , if ( [ 8 ] ) can be efficiently solved for any @xmath4 and @xmath5 , then the optimal node placement solution to the considered minimum - cost deployment problem can be easily obtained through a simple two - dimension search over the values of @xmath4 and @xmath5 , i.e. , finding a pair of feasible @xmath77 that produces the lowest deployment cost @xmath69 . for a pair of fixed @xmath4 and @xmath5 ( @xmath78 and @xmath79 )
, we can see that ( [ 8 ] ) is feasible if and only if the optimal objective of the following problem is no smaller than @xmath67 , i.e. , [ 57 ] @xmath80 we can express ( [ 57 ] ) as its equivalent epigraphic form @xcite , i.e. , @xmath81 given fixed @xmath4 and @xmath5 , ( [ 8 ] ) is feasible if and only if the optimal objective of ( [ 9 ] ) satisfies @xmath82 .
then , the key difficulty of solving the optimal deployment problem is to find efficient solution for problem ( [ 9 ] ) .
when the ens and aps are integrated as haps , the total deployment cost is @xmath83 if @xmath4 haps are used . here
, @xmath84 denotes the cost of deploying a hap , where in general @xmath85 .
similar to the case of separated ens and aps , the minimum - cost placement problem can be equivalently formulated as the following feasibility problem for any fixed number of @xmath78 haps , @xmath86\\ \text{s . t. } & & & \lambda_k \left(\mathbf{u}^m\right ) -\mu_k\left(\mathbf{u}^m\right)\geq \gamma , \ \
k=1,\cdots , k,\\ & & & \mathbf{b}^l\leq \mathbf{u}_i\leq \mathbf{b}^h,\ \ i=1,\cdots , m , \end{aligned}\ ] ] where @xmath87 and @xmath88 are given in ( [ 10 ] ) and ( [ 17 ] ) , respectively .
notice that the study on co - located ens and aps is not a special case of that of separated ens and aps .
in fact , it adds extra constraints ( @xmath89 ) to ( [ 8 ] ) , which leave less flexibility to the nodes placement design and make the problem more challenging to solve .
equivalently , the feasibility of ( [ 18 ] ) can be determined by solving the following optimization problem @xmath90 and then comparing the optimal objective @xmath91 with @xmath67 , to see whether @xmath92 holds . in the following sections iv and v
, we propose efficient algorithms to solve problems ( [ 9 ] ) and ( [ 14 ] ) , respectively .
it is worth mentioning that the placement solution to ( [ 9 ] ) and ( [ 14 ] ) can be at arbitrary locations .
when an en ( or a hap ) is placed at a location very close to an wd , the far - field channel model in ( [ 3 ] ) may be inaccurate .
however , we learn from ( [ 9 ] ) and ( [ 14 ] ) that the optimal value @xmath91 is determined by the performance - bottleneck wd that is far away from the ens and aps ( i.e. , the channel model in ( 3 ) applies practically ) , thus having very low energy harvesting rate and high transmit power consumption . therefore , the potential inaccuracy of ( [ 3 ] ) will not affect the objective values of ( [ 9 ] ) and ( [ 14 ] ) , and the proposed algorithms in this paper are valid in practice .
in this section , we study the node placement optimization for separately located en and ap in problem ( [ 9 ] ) .
specifically , we first study in section iv.a the method to optimize en placement assuming that the locations of aps are fixed in a wpcn . in section iv.b , we further study the method to optimize the placement of aps given fixed en locations . based on the obtained results ,
we then propose in section iv.c an alternating method to jointly optimize the placements of ens and aps .
in addition , an alternative local searching method is considered in section iv.d for performance comparison .
we first consider the optimal en placement problem when the locations of the aps are fixed , i.e. , @xmath53 s are known . in this case , the wd - ap association @xmath35 is known for each wd @xmath19 from ( [ 22 ] ) , and @xmath52 s can be calculated accordingly from ( [ 11 ] ) .
it is worth mentioning that the proposed algorithms under the fixed ap setup can be directly extended to solve en placement problem in other wireless powered networks not necessarily for communication purpose , e.g. , a sensor network whose energy is mainly consumed on sensing and processing data , as long as the energy consumption rates @xmath52 s are known parameters . with @xmath53 s and @xmath52 s being fixed , we can rewrite ( [ 9 ] ) as @xmath93 where @xmath94 .
we can see that ( [ 13 ] ) is a non - convex optimization problem , because @xmath95 is neither a convex nor concave function in @xmath50 .
as it currently lacks of effective method to convert ( [ 13 ] ) into a convex optimization problem , the optimal solution is in general hard to obtain .
however , for a special case with @xmath96 , i.e. , placing only one en , the optimal solution is obtained in the following . by setting @xmath96 , ( [ 13 ] ) can be rewritten as [ 15 ] @xmath97 although ( [ 15 ] ) is still a non - convex optimization problem ( as @xmath98 is not a concave function in @xmath99 ) , it is indeed a convex feasibility problem over @xmath100 when @xmath99 is fixed , which can be efficiently solved using the interior point method @xcite .
therefore , the optimal solution of ( [ 15 ] ) can be obtained using a bi - section search method over @xmath99 , whose pseudo - code is given in algorithm [ 46 ] .
notice that the right hand side ( rhs ) of ( [ 58 ] ) is always positive during the bisection search over @xmath101 .
besides , we can infer that algorithm [ 46 ] converges to the optimal solution @xmath91 , because problem ( [ 15 ] ) is feasible for @xmath102 and infeasible otherwise .
the total number of feasibility tests performed is @xmath103 $ ] , where @xmath104 is a predetermined parameter corresponding to a solution precision requirement .
initialize : @xmath105 , @xmath106 * return *
@xmath107 the last feasible solution @xmath108 since placing one en optimally is solved , we have the potential to decouple the difficult en placement problem ( [ 13 ] ) into @xmath4 relatively easy problems with @xmath109 .
this motivates a greedy algorithm , which places the ens iteratively one - by - one into the network .
intuitively , an optimal deployment solution of ( [ 13 ] ) should spread " the @xmath4 ens among the @xmath6 wds to maximize the minimum energy harvesting rate .
however , the optimal solution obtained from solving ( [ 15 ] ) tends to place the single en around the center of the cluster formed by the @xmath6 wds .
inspired by the optimal solution structures of ( [ 13 ] ) and ( [ 15 ] ) , we propose in the following a cluster - based greedy en placement method , where the newly placed en optimizes the net energy harvesting rates of an expanding cluster of wds , until all the wds are included . in practice , we geographically partition the @xmath6 wds into @xmath4 non - overlapping clusters ( assuming @xmath110 ) , denoted by @xmath111.-means clustering algorithm in this paper . ]
this can be efficiently achieved by , e.g. , the well - known @xmath6-means clustering algorithm @xcite .
then , in the @xmath12-th iteration ( @xmath112 ) , we obtain the optimal location of the @xmath12-th en , denoted by @xmath113 , by maximizing the net energy harvesting rates of the wds in the first @xmath12 clusters as follows .
[ 35 ] @xmath114 where @xmath115 denotes the accumulative rf power harvested at the @xmath19-th wd due to the @xmath116 previously deployed ens , given by @xmath117 in each iteration , @xmath113 can be efficiently obtained using a bi - section search method over @xmath118 similar to that in solving ( [ 15 ] ) .
the pseudo - code of the greedy algorithm is given in algorithm [ 41 ] .
notice in line @xmath119 of algorithm [ 41 ] , the corresponding inequality in ( [ 31 ] ) holds automatically when @xmath120 for some @xmath19 , given a fixed @xmath121 .
therefore , the corresponding constraints can be safely ignored without affecting the feasibility of ( [ 35 ] ) .
ens ( with fixed aps).,scaledwidth=60.0% ] the greedy en placement method is illustrated in fig . [ 63 ] . in this example , we first divide the wds into @xmath122 clusters in fig . [ 63](a ) , then place the @xmath123 ens one - by - one in fig .
[ 63](b)-(d ) .
when placing the @xmath124st en or en@xmath125 , the algorithm only considers the received energy of the wds in the @xmath124st cluster ( shaded wds in fig .
[ 63](b ) ) from the en to be placed ; for the @xmath126nd en or en@xmath127 , it considers the received energy of the wds in the @xmath124st and @xmath126nd clusters from the first @xmath126 ens ; for the last en or en@xmath128 , it considers the received energy of all the wds from the @xmath123 ens .
notice that our greedy algorithm allows multiple ens to be placed in the same cluster , because the placement of the @xmath12-th en considers all the wds in the first @xmath12 clusters .
algorithm [ 41 ] applies algorithm [ 46 ] @xmath4 times , one for placing each en , thus the total number of feasibility tests performed is @xmath129 $ ] , where @xmath130 is a parameter corresponding to a solution precision requirement .
besides , the time complexity of solving each convex feasibility test using the interior point method is @xmath131@xcite .
therefore , the overall time complexity of algorithm [ 41 ] is @xmath132 , which is moderate even for a large - size network consisting of , e.g. , tens of ens and hundreds of wds .
* initialization * : clustering the wds into @xmath133 with @xmath53 s , calculate @xmath52 s using ( [ 22 ] ) and ( [ 11 ] ) * return * @xmath134 we then study in this subsection the method to optimize the placement of aps given fixed en locations , i.e. , @xmath50 s are known . in this case , @xmath51 s are fixed and can be calculated using ( [ 10 ] ) . with @xmath51 s being fixed parameters , we can substitute ( [ 11 ] ) into ( [ 9 ] ) and formulate the optimal ap placement problem under fixed ens as follows @xmath135 where @xmath35 is the index of ap that wd @xmath19 associates with given in ( [ 22 ] ) .
the above problem is non - convex because of the combinatorial nature of wd - ap associations , i.e. , @xmath35 s are discrete indicators .
however , notice that if @xmath35 s are known , ( [ 27 ] ) is a convex problem that is easily solvable .
in practice , however , @xmath35 s are revealed only after ( [ 27 ] ) is solved and the placement of aps is obtained . to resolve this conflict
, we propose in the following a _ trial - and - error _ method to find feasible @xmath35 s and accordingly a feasible ap placement solution to ( [ 27 ] ) .
the pseudo - code of the method to solve ( [ 27 ] ) is presented in algorithm [ 42 ] and explained as follows .
as its name suggests , we first convert ( [ 27 ] ) into a convex problem by assuming a set of wd - ap associations , denoted by @xmath136 , @xmath137 , and then solve ( [ 27 ] ) for the optimal ap placement based on the assumed @xmath136 s .
next , we compare @xmath136 s with the actual wd - ap associations after the optimal ap placement is obtained using ( [ 22 ] ) , denoted by @xmath138 , @xmath137 .
specifically , we check if @xmath139 .
if yes , we have obtained a feasible solution to ( [ 27 ] ) ; otherwise , we update @xmath140 , @xmath137 and repeat the above process until @xmath141 , @xmath142 .
the convergence of algorithm [ 42 ] is proved in the appendix and the convergence rate is evaluated numerically in fig . [ 64 ] of section vi .
intuitively , the trial - and - error method is convergent because the optimal value of ( [ 27 ] ) is bounded , while by updating @xmath140 , we can always improve the optimal objective value of ( [ 27 ] ) in the next round of solving it . as we will show later in fig .
[ 64 ] of section vi , the number of iterations used until convergence is of constant order , i.e. , @xmath143 , regardless of the value of @xmath5 or @xmath6 . there , the time complexity of algorithm [ 42 ] is @xmath144 , as it takes this time complexity for solving ( [ 27 ] ) in each iteration .
* initialization * : + separate the wds into @xmath5 clusters , and place each ap at a cluster center .
use @xmath145 s to denote the initial ap locations with @xmath145 s , calculate @xmath136 s using ( [ 22 ] ) with @xmath50 s , calculate @xmath51 s using ( [ 10 ] ) .
let @xmath146 in this subsection , we further study the problem of joint en and ap placement optimization . in this case , we consider both the locations of ens and aps as variables , such that the joint en - ap placement problem in ( [ 9 ] ) can be expressed as @xmath147 evidently , the optimization problem is highly non - convex because of the non - convex function @xmath95 and the discrete variables @xmath35 s . based on the results in section iv.a and iv.b , we propose an alternating method in algorithm [ 43 ] to solve ( [ 16 ] ) for joint en and ap placement solution . specifically , starting with a feasible ap placement , we alternately apply algorithms [ 41 ] and [ 42 ] to iteratively update the locations of ens and aps , respectively .
a point to notice is that algorithm [ 41 ] ( and algorithm [ 42 ] ) only produces a sub - optimal solution to ( [ 13 ] ) ( and ( [ 27 ] ) ) , thus the objective value of ( [ 16 ] ) may decrease during the alternating iterations . to cope with this problem
, we record the deployment solutions obtained in @xmath148 iterations and select the one with the best performance .
the impact of the parameter @xmath149 to the algorithm performance is evaluated in fig . [ 68 ] of section vi .
given the complexities of algorithms [ 41 ] and [ 42 ] , we can easily infer that the time complexity of algorithm [ 43 ] is @xmath150 .
* initialize * : separate the wds into @xmath5 clusters , and place each ap at a cluster center .
use @xmath53 s to denote the initial ap locations @xmath151 * return : * @xmath152 s and @xmath153 s .
besides the proposed alternating method for solving ( [ 16 ] ) , we also consider an alternative local searching method used as benchmark algorithm for performance comparison .
the local searching algorithm starts with a random deployment of the @xmath4 ens and @xmath5 aps , i.e. , @xmath50 s and @xmath154 s , and checks if the minimum net energy harvesting rate among the wds , i.e. , @xmath155 can be increased by making a random movement to @xmath156 s and @xmath157 s that satisfy @xmath158 where @xmath159 is a fixed positive parameter . if yes , it makes the move and repeats the random movement process . otherwise ,
if @xmath160 can not be increased , the algorithm has reached a local maximum and returns the current placement solution .
several off - the - shelf local searching algorithms are available , where simulated annealing @xcite is used in this paper .
in particular , simulated annealing can improve the searching result by allowing the nodes to be moved to locations with decreased value of @xmath160 to reduce the chance of being trapped at local maximums .
besides , we can improve the quality of deployment solution using different initial node placements , which are obtained either randomly or empirically , and select the resulted solution with the best performance .
in this section , we proceed to study the node placement optimization problem ( [ 14 ] ) for the case of co - located ens and aps .
the problem is still non - convex due to which the optimal solution is hard to be obtained .
inspired by both algorithms [ 41 ] and [ 42 ] , we propose in this section an efficient greedy algorithm for hap placement optimization .
the node placement optimization problem ( [ 14 ] ) is highly non - convex , because the expression of problem ( [ 14 ] ) involves non - convex function @xmath95 in @xmath161 and minimum operator over convex functions in @xmath88 . since
its optimal solution is hard to obtain , a promising alternative is the greedy algorithm , which iteratively places a single hap to the network at one time , similar to algorithm [ 41 ] for solving ( [ 13 ] ) which optimizes the en locations given fixed aps .
however , by comparing problems ( [ 14 ] ) and ( [ 13 ] ) , we can see that the algorithm design for solving ( [ 14 ] ) is more complicated , because each @xmath52 is now a function of @xmath50 s , instead of constant parameter in ( [ 13 ] ) .
similar to the greedy algorithm in section iv.a , we first separate the @xmath6 wds into @xmath4 non - overlapping clusters , denoted by @xmath111 , and add to the network a hap in each iteration .
specifically , in the @xmath12-th iteration , given that the previous @xmath116 haps are fixed , we obtain the optimal location of the @xmath12-th hap , denoted by @xmath113 , by maximizing the net energy harvesting rates of the wds in the first @xmath12 clusters . to simplify the notations , we also use @xmath115 as in section iv.a to denote the accumulative rf harvesting power of the wd @xmath19 from the previously placed @xmath116 haps , which can be calculated using ( [ 53 ] ) .
besides , let @xmath162 denote the energy consumption rate of the @xmath19-th wd after the first @xmath116 haps have been placed , where @xmath163 notice that the only difference between placing the @xmath12-th hap and the @xmath12-th en in section iv.a is that @xmath52 is now a function of @xmath50 instead of a given constant . by substituting ( [ 52 ] ) into ( [ 35 ] ) , the optimal location of the @xmath12-th hap is obtained by solving the following problem [ 49 ] @xmath164 where @xmath165 from ( [ 32 ] ) , we can see that a wd may change its association to the @xmath12-th hap , if the newly placed hap is closer to the wd than all the other @xmath116 haps that have been previously deployed .
this combinatorial nature of wd - ap associations makes problem ( [ 49 ] ) non - convex even if @xmath118 is fixed . in the following ,
we apply the similar trial - and - error technique as that in section iv.b to obtain a feasible solution to problem ( [ 49 ] ) . the basic idea to obtain a feasible solution of ( [ 49 ] ) is to convert it into a convex problem given @xmath118 , and then use simple bi - section search over @xmath118 .
the convexification of ( [ 49 ] ) is achieved by a trial - and - error method similar to that used for finding feasible wd - ap associations proposed in algorithm [ 42 ] .
that is , we iteratively make assumptions on wd - ap associations and update the optimal placement of the @xmath12-th hap obtained from solving ( [ 49 ] ) based on the assumptions in the current iteration . with a bit abuse of notations , here
we reuse @xmath113 in each iteration as the optimal location of the @xmath12-th hap given the current wd - ap association assumptions .
specifically , we assume whether the wds change their associations after the @xmath12-th hap is added , i.e. , assuming either @xmath166 or @xmath167 for each @xmath19 .
then , given a fixed @xmath118 , each constraint on @xmath19 in ( [ 51 ] ) belongs to one of the following four cases : if we assume that wd @xmath19 does not change its wd - hap association after the @xmath12-th hap is placed into the wpcn , or equivalently @xmath166 , we can replace the corresponding constraint in ( [ 51 ] ) with @xmath168 with a fixed @xmath118 , ( [ 47 ] ) is a convex constraint if @xmath169 .
if we still assume @xmath166 , while @xmath170 holds , we can safely drop the constraint in ( [ 51 ] ) without changing the feasible region of @xmath50 . on the other occasion ,
if we assume that wd @xmath19 changes its wd - hap association , or @xmath171 , the corresponding constraint in ( [ 51 ] ) becomes @xmath172 which can be further expressed as latexmath:[\[\label{33 } notice that , given a fixed @xmath118 , ( [ 33 ] ) is a convex constraint if @xmath174 .
otherwise , if we assume @xmath171 and @xmath176 holds , ( [ 33 ] ) is a non - convex constraint , as the left - hand - side ( lhs ) of ( [ 33 ] ) is the difference of two convex functions .
nonetheless , we show that ( [ 33 ] ) can still be converted into a convex constraint in this case .
let us first consider a function @xmath177 where @xmath178 and @xmath179 .
we calculate the first order derivative of @xmath180 and find that @xmath181 increases monotonically when @xmath182^{1/d_u } \triangleq \tau_{i , k},\ ] ] and decreases monotonically if @xmath183 .
notice that @xmath184 and @xmath185 always hold .
therefore , although @xmath180 is not a convex function , @xmath186 can still be equivalently expressed as @xmath187 , with @xmath188 being some positive number satisfying @xmath189 .
the value of @xmath188 can be efficiently obtained using many off - the - shelf numerical methods , such as the classic newton s method or bi - section search method .
a close comparison between the lhs of ( [ 33 ] ) and @xmath180 in ( [ 36 ] ) shows that , by letting @xmath190 , we can equivalently express ( [ 33 ] ) as a convex constraint latexmath:[\[\label{37 } @xmath179 holds . to sum up , given a fixed @xmath118 , we tackle the @xmath19-th constraint in ( [ 51 ] ) using one of the following methods : 1 .
replace by ( [ 47 ] ) if assuming @xmath166 and @xmath169 ; 2 .
drop the constraint if assuming @xmath166 and @xmath170 ; 3 .
replace by ( [ 33 ] ) if assuming @xmath171 and @xmath174 ; 4 . replace by ( [ 37 ] ) if assuming @xmath171 and @xmath192 .
after processing all the @xmath6 constraints in ( [ 51 ] ) , we can convert ( [ 49 ] ) into a convex feasibility problem given a set of wd - hap association assumptions and a fixed @xmath118 .
accordingly , the optimal placement of the @xmath12-th hap ( @xmath113 ) under the assumptions , can be efficiently obtained from solving ( [ 49 ] ) using a bi - section search method over @xmath118 .
similar to the trial - and - error technique used in algorithm [ 42 ] , we check if the obtained @xmath113 satisfies all the assumptions made .
if yes , we have obtained a feasible solution of ( [ 49 ] ) . otherwise , we switch the violating assumptions , then follow the above constraint processing method to resolve ( [ 49 ] ) for a new @xmath113 , and repeat the iterations until all the assumptions are satisfied .
the above trial - and - error method converges .
the proof follows the similar argument as given in the appendix , which proves the convergence of the trial - and - error method used for solving problem ( [ 27 ] ) .
thus , this proof is omitted here . since the placement of a single hap can be obtained via solving ( [ 49 ] ) , we can iteratively place the @xmath4 haps into the wpcn .
the pseudo - code of the revised greedy algorithm is presented in algorithm [ 44 ] .
for example , fig .
[ 66 ] illustrates the detailed steps taken to place the @xmath126nd hap , or hap@xmath127 ( total @xmath123 haps while the first hap , or hap@xmath125 is already placed ) .
specifically , we first assume in fig . [ 66](a ) that all the wds in the @xmath124st cluster associate with hap@xmath125 , and the wds in the @xmath126nd cluster associate with hap@xmath127 , after hap@xmath127 is added into the network .
then , we obtain in fig . [
66](b ) the optimal placement of the hap@xmath127 based on the association assumptions made .
however , the obtained location of hap@xmath127 results in a contradiction with the association assumption made on wd@xmath125 ( assumed to be associated with hap@xmath125 ) .
therefore , we change the association assumption of wd@xmath125 to hap@xmath127 , and recalculate the optimal placement solution for hap@xmath127 ( fig .
[ 66](c ) ) . in fig .
[ 66](d ) , the newly obtained location of hap@xmath127 satisfies all the association assumptions , thus the placement of hap@xmath127 is feasible .
following the similar argument in the appendix , the association assumption update procedure converges , because the optimal objective value of ( [ 49 ] ) is non - decreasing upon each association assumptions update ( lines @xmath193 of algorithm [ 44 ] ) . after obtaining the location of hap@xmath127 , a feasible location of the @xmath123rd hap , hap@xmath128 , can also be obtained using the similar procedures as above . besides
, we can infer that the time complexity of algorithm [ 44 ] is @xmath194 , because it places the @xmath4 haps iteratively , while the trial - and - error method used to place each hap needs @xmath195 complexity .
cluster the wds into @xmath133 * return * the hap locations @xmath134 .
in this section , we use simulations to evaluate the performance of the proposed node placement methods .
all the computations are executed by matlab on a computer with an intel core i5 @xmath196-ghz cpu and @xmath175 gb of memory .
the carrier frequency is @xmath197 mhz for both dl and ul transmissions operating on different bandwidths . in the dl energy transmission
, we consider using powercast tx91501 - 1w power transmitter with @xmath198w ( watt ) transmit power , and p2110 receiver with @xmath199 db antenna gain and @xmath200 energy harvesting efficiency . besides , we assume the path loss exponent @xmath201 , thus @xmath202 . in the ul information transmission , we assume that @xmath203 , @xmath204w and @xmath205 for @xmath137 , where @xmath206 is obtained assuming rayleigh fading and the use of truncated channel inversion transmission @xcite with receiver signal power @xmath207dbm ( equivalent to @xmath208db snr target with -@xmath209dbm noise power ) and outage probability of @xmath210 .
all the wds , ens and aps are placed within a @xmath211 box region specified by @xmath212 and @xmath213 . unless otherwise stated , each point in the following figures is an average performance of @xmath214 random wd placements , each with @xmath215 wds uniformly placed within the box region .
we first evaluate the performance of the proposed alternating optimization method ( algorithm [ 43 ] ) for placing separated ens and aps . without loss of generality , we consider @xmath216 aps and show in fig . [ 68 ] the minimum net energy harvesting rate @xmath160 in ( [ 14 ] ) achieved by algorithm [ 43 ] when the locations of aps are jointly optimized with those of different number of ens ( @xmath4 ) .
evidently , a larger @xmath160 indicates better system performance . for the proposed alternating optimization algorithm ( altopt )
, we show both the performance with @xmath217 and @xmath214 . besides , we also consider the following benchmark placement methods * cluster center method ( cc ) : separate the wds into @xmath4 clusters and place an en at each of the cluster centers .
similarly , separate the wds into @xmath5 clusters and place the @xmath5 aps at the cluster centers ; * optimize only en locations : the aps are placed at the @xmath5 cluster centers ; while the en placement is optimized based on the ap locations using algorithm [ 41 ] . *
local searching algorithm ( ls ) method introduced in section iv , where the initial en and ap locations are set according to the cc method and the best - performing deployment solution obtained during the searching iterations is used . )
.,scaledwidth=60.0% ] evidently , we can see that the proposed alternating optimization has the best performance among the methods considered .
specifically , significant performance gain is observed for altopt over optimizing en placement only .
the ls method has relatively good performance compared to altopt , especially when @xmath4 is small , but the performance gap increases with @xmath4 due to the increasing probability of being trapped at local maximums with a larger @xmath4 .
the cc scheme has the worst performance as it neglects the disparity of energy harvesting / consumption rates among the wds and precludes the case where multiple ap / ens can be placed in a cluster . in practice ,
[ 68 ] can be used to evaluate the deployment cost of each algorithm .
for instance , when @xmath218mw is required , we see that the altopt ( @xmath217 ) on average needs @xmath219 ens , the ls method needs @xmath119 ens , optimizing en placement only requires @xmath220 ens , and the cc method needs @xmath214 ens , with the same number of information aps deployed ( i.e. , @xmath216 ) .
the above results show that , when the ens and aps are separated , significant performance gain can be obtained by jointly optimizing the placements of ens and aps , especially for large - size wpcns that need a large number of ens and aps to be deployed .
in addition , as optimizing only en locations corresponds to a special case of algorithm [ 43 ] with @xmath221 , we can see that the performance gain is significant when @xmath149 increases from @xmath124 to @xmath1 .
however , the performance improvement becomes marginal as we further increase @xmath149 from @xmath1 to @xmath214 . in practice ,
good system performance can be obtained with relatively small number of alternating optimizations , e.g. , @xmath217 in our case .
, as a function of ( a ) @xmath5 under fixed @xmath222 ; and ( b ) as a function of @xmath6 under fixed @xmath223.,scaledwidth=60.0% ] we then show in fig . [
64 ] the convergence rate of algorithm [ 42 ] , for which the convergence is proved in an asymptotic sense in the appendix . in particular
, we plot the average number of iterations ( wd - ap association assumptions ) used until the algorithm converges . here
, we investigate the convergence rate when either the number of aps ( @xmath5 ) or wds ( @xmath6 ) varies . with
fixed @xmath6 in fig .
[ 64](a ) , we see that the number of iterations used till convergence does not vary significantly as @xmath5 increases . similarly in fig .
[ 64](b ) , with a fixed @xmath223 , we do not observe significant increase of iterations when @xmath6 increases from @xmath224 to @xmath225 .
besides , all the simulations performed in fig .
[ 64 ] use at most @xmath226 iterations to converge .
therefore , we can safely estimate that the number of iterations used till convergence is of constant order , i.e. , @xmath143 , which leads to the complexity analysis of algorithm [ 43 ] in section iv.c and algorithm [ 44 ] in section v.c .
next , we evaluate in fig . [ 69 ] the performance of the proposed algorithm [ 44 ] for co - located ens and aps , where the value of @xmath160 achieved by algorithm [ 44 ] is plotted against the number of haps used ( @xmath4 ) . in particular ,
we compare its performance with that of ls ( with @xmath4 cluster centers as the initial searching points ) and the cc placement method , i.e. , the haps are placed at the @xmath4 cluster centers .
we can see that the proposed greedy algorithm in algorithm [ 44 ] has the best performance among the methods considered . nonetheless , the performance gaps over the ls and cc methods are relatively small compared to that in fig .
an intuitive explanation is that the doubly - near - far phenomenon for co - located en and ap renders the optimal haps placement to be around the cluster centers . by comparing fig . [
68 ] and fig .
[ 69 ] , we can see the evident performance advantage of using separated ens and aps over haps .
for instance , the @xmath160 achieved by @xmath219 ens and @xmath219 aps in fig . [ 68 ] is @xmath227mw , while that achieved by @xmath219 haps ( equivalent to @xmath219 ens and @xmath219 aps being co - located ) is only around @xmath228mw .
although the greedy algorithm and the ls method perform closely , they differ significantly in the computational complexity . to better visualize the growth rate of complexity , we plot in fig . [ 70 ] the normalized cpu time of the two methods , where each point on the figure is the normalized against the cpu time of the respective method when @xmath229 .
clearly , we can see that the complexity of the greedy algorithm increases almost linearly with @xmath4 , where the cpu time increases approximately @xmath219 times when @xmath4 increases from @xmath175 to @xmath230 .
this matches our complexity analysis for algorithm [ 44 ] in section v.c that the complexity increases almost linearly in @xmath4 when @xmath6 is much larger .
the ls method , however , has a much faster increase in complexity with @xmath4 , where the cpu time increases by around @xmath231 times when @xmath4 increases from @xmath175 to @xmath230 .
therefore , even in a large - size wpcn with large @xmath4 , the computation time of the proposed greedy algorithm is still moderate , while this may be extremely high for the ls method , e.g. , couple of minutes versus several hours for @xmath232 . finally , we present a case study to compare the cost of node placement achieved by using either separated or co - located ens and aps . here
, we consider a wpcn with @xmath233 wds uniformly placed within the @xmath234 box region , where the detailed placement is omitted due to the page limit .
for the case of separated ens and aps , we use algorithm [ 43 ] to enumerate @xmath77 pairs that can satisfy a given net energy harvesting performance requirement @xmath67 , and select the one with the minimum cost as the solution . for the case of haps , we use algorithm [ 44 ] to find the minimum @xmath4 that can satisfy the performance requirement @xmath67 .
a point to notice is that , the obtained deployment solutions are sub - optimal to the min - cost deployment problems with either separated or co - located ens and aps , i.e. , the cost of the optimal deployment solution can be lower , because algorithms [ 43 ] and [ 44 ] are sub - optimal to solve problems ( [ 9 ] ) and ( [ 14 ] ) , respectively . more effort to further improve the solution performance is needed for future investigations . and in ( b ) @xmath235
, scaledwidth=60.0% ] in fig .
[ 72](a ) , we show the minimum deployment costs achieved by the two methods under different performance requirement @xmath67 .
the number of nodes used by both methods are also marked in the figure . with @xmath236 , we can see that using separated ens and aps can achieve much lower deployment cost than co - located haps . in another occasion in fig .
[ 72](b ) , the two scheme achieve similar deployment cost when the cost of a hap is decreased from @xmath237 to @xmath238 .
we also observe that , by allowing the ens and aps to be separately located , we need much less energy / information access points than they are co - located to achieve the same performance , thanks to the extra freedom in choosing both the numbers and locations of energy / information access points .
for instance , when @xmath239 , we need @xmath230 separated energy / information access points ( @xmath240 and @xmath241 ) , while @xmath242 co - located energy / information access points by the @xmath243 haps
. however , we do not intend to claim that using separated ens and aps is better than the co - located case .
rather , we show that a cost - effective deployment plan can be efficiently obtained using the proposed methods . in practice , the choice of using either separated or co - located ens and aps depends on a joint consideration of the node deployment costs , the network size and the performance requirement .
in this paper , we studied the node placement optimization problem in wpcn , which aims to minimize the node deployment cost while satisfying the energy harvesting and communication performance of wds .
efficient algorithms were proposed to optimize the node placement of wpcns with the ens and information aps being either separated and co - located . in particular , when ens and aps are separately located , simulation results showed that significant deployment cost can be saved by jointly optimizing the locations of ens and aps , especially for large - size wpcns with a large number of ens and aps to be deployed . in the case of co - located ens and aps ( haps ) , however , the performance advantage of node placement optimization is not that significant , where we observed relatively small performance gap between optimized node placement solutions and that achieved by some simple heuristics , i.e. , placing the haps at the cluster centers formed by the wds . in practice , separated ens and aps are more suitable for deploying wpcns than co - located haps , because of the flexibility in choosing both the numbers and locations of ens and aps . nonetheless , because the optimal solution to the node placement problem has not been obtained in this paper , we may expect further improvement upon our proposed methods in the future , especially for the case of hap placement optimization . finally , we conclude with some interesting future work directions for the node placement problem in wpcn .
first , the models considered in this paper can be extended to more general setups . using the uplink information transmission as example
, we assumed in this paper that each wd has fixed association with a single ap . in practice
, dynamic frequency allocation can be applied to enhance the spectral efficiency , where a wd can transmit to different aps , even multiple aps simultaneously , in different transmission blocks .
besides , instead of assuming each ap can serve infinite number of wds , we can allow each ap to serve finite number of wds .
in addition , we may also consider the presence of uplink co - channel interference due to frequency reuse in wpcn .
the extensions require adding corresponding constraints or changing the expression of energy consumption model in the problem formulation of this paper .
second , it is interesting to consider the hybrid node deployment problem that uses both co - located and separated ens / aps .
third , it is practically important to consider the node placement problem with location constraints , e.g. , some areas that may forbid the ens / aps to be placed .
in addition , the density of ens may be constrained to satisfy certain safety consideration on rf power radiation . [ 59 ] let @xmath244 denote the optimal solutions of ( [ 27 ] ) calculated from the @xmath245-th ( @xmath246 ) set of assumptions made on the wd - ap associations , denoted by @xmath247 , @xmath248 .
let @xmath249 denote the set of wds to which the optimal solution @xmath250 s contradict with the wd - ap assumptions ( we consider only @xmath251 , because otherwise the algorithm has reached its optimum ) , i.e. , @xmath252 according to the proposed trial - and - error method , @xmath253 is set for each @xmath137 as @xmath254 let @xmath255 denote the minimum net energy harvesting rate among all the wds given the updated wd - ap association @xmath253 s and the current ap locations @xmath250 s , i.e. , @xmath256 we can see that @xmath257 because the update of wd - ap associations in ( [ 29 ] ) does not increase the energy consumption rate of any wd achived by assuming @xmath247 , @xmath248 .
besides , @xmath258 is a feasible solution of ( [ 27 ] ) under the association assumption @xmath253 s .
therefore , the optimal solution @xmath259 calculated from the association assumption @xmath253 s will lead to @xmath260 .
in other words , the optimal objective of ( [ 27 ] ) is non - decreasing in each trial - and - error update of wd - ap associations .
this , together with the fact that the optimal value of ( [ 27 ] ) is bounded , leads to the conclusion that the proposed trial - and - error method is convergent .
a. georgiadis , g. andia , and a. collado , rectenna design and optimization using reciprocity theory and harmonic balance analysis for electromagnetic ( em ) energy harvesting , " _ ieee antennas wireless propag .
_ , vol . 9 , pp . 444 - 446 , may 2010
a. a. nasir , x. zhou , s. durrani , and r. a. kennedy , wireless - powered relays in cooperative communications : time - switching relaying protocols and throughput analysis , " _ ieee trans .
_ , vol .
63 , no . 5 , pp .
1607 - 1622 , may 2015 .
h. chen , y. li , j. l. rebelatto , b. f. uchoa - filho , and b. vucetic , harvest - then - cooperate : wireless - powered cooperative communications , " _ ieee trans . signal process .
_ , vol . 63 , no . 7 , pp . 1700 - 1711 , apr . 2015 .
y. t. hou , y. shi , h. d. sherali , and s. f. midkiff , on energy provisioning and relay node placement for wireless sensor networks , " _ ieee trans .
wireless commun .
_ , vol . 4 , no . 5 , pp .
2579 - 2590 , sep .
2005 .
n. michelusi , l. badia , r. carli , l. corradini , and m. zorzi , energy management policies for harvesting - based wireless sensor devices with battery degradation , " _ ieee trans .
_ , vol .
61 , no . 12 , pp . 4934 - 4947 , dec . | the applications of wireless power transfer technology to wireless communications can help build a wireless powered communication network ( wpcn ) with more reliable and sustainable power supply compared to the conventional battery - powered network .
however , due to the fundamental differences in wireless information and power transmissions , many important aspects of conventional battery - powered wireless communication networks need to be redesigned for efficient operations of wpcns . in this paper , we study the placement optimization of energy and information access points in wpcns , where the wireless devices ( wds ) harvest the radio frequency energy transferred by dedicated energy nodes ( ens ) in the downlink , and use the harvested energy to transmit data to information access points ( aps ) in the uplink . in particular , we are interested in minimizing the network deployment cost with minimum number of ens and aps by optimizing their locations , while satisfying the energy harvesting and communication performance requirements of the wds . specifically , we first study the minimum - cost placement problem when the ens and aps are separately located , where an alternating optimization method is proposed to jointly optimize the locations of ens and aps . then , we study the placement optimization when each pair of en and ap are co - located and integrated as a hybrid access point , and propose an efficient algorithm to solve this problem .
simulation results show that the proposed methods can effectively reduce the network deployment cost and yet guarantee the given performance requirements , which is a key consideration in the future applications of wpcns . wireless power transfer , wireless powered communication networks , energy harvesting , network planning , node placement optimization . |
a _ hypergraph _ @xmath1 is a pair @xmath5 where @xmath6 is a finite set of _ vertices _ and @xmath7 is a set of subsets of @xmath8 , called _ hyperedges_. a hypergraph is said to be _ sperner _ ( or : a _ clutter _ ) if no hyperedge contains another one , that is , if @xmath9 and @xmath10 implies @xmath11 .
a hypergraph is said to be _ threshold _ if there exist a non - negative integer weight function @xmath12 and a non - negative integer threshold @xmath13 such that for every subset @xmath14 , we have @xmath15 if and only if @xmath16 for some @xmath17 .
threshold hypergraphs were defined in the uniform case by golumbic @xcite and studied further by reiterman et al .
@xcite . in their full generality ( that is , without the restriction that the hypergraph is uniform )
, the concept of threshold hypergraphs is equivalent to that of threshold monotone boolean functions , see , e.g. , @xcite .
a polynomial time recognition algorithm for threshold monotone boolean functions represented by their complete dnf was given by peled and simeone @xcite .
the algorithm is based on linear programming and implies a polynomial time recognition algorithm for threshold hypergraphs .
the mapping that takes every hyperedge @xmath17 to its _ characteristic vector _
@xmath18 , defined by @xmath19 shows that the sets of hyperedges of threshold sperner hypergraphs are in a one - to - one correspondence with the sets of minimal feasible binary solutions to a linear inequality of the form @xmath20 where @xmath21 and @xmath13 .
a set of vertices @xmath14 in a hypergraph is said to be _ independent _ if it does not contain any hyperedge , and _ dependent _ otherwise .
thus , threshold hypergraphs are exactly the hypergraphs admitting a linear function on the vertices separating the characteristic vectors of the independent sets from the characteristic vectors of dependent sets .
sperner hypergraphs can be equivalently defined as the hypergraphs such that every two distinct hyperedges @xmath22 and @xmath23 satisfy @xmath24 this point of view motivated chiarelli and milani to define in @xcite a hypergraph @xmath1 to be _ dually sperner _ if every two distinct hyperedges @xmath22 and @xmath23 satisfy @xmath25 it was shown in @xcite that dually sperner hypergraphs are threshold . in @xcite , these hypergraphs were applied to characterize two classes of graphs defined by the following properties .
every induced subgraph has a non - negative linear vertex weight function separating the characteristic vectors of all total dominating sets @xcite ( resp . , connected dominating sets @xcite ) from the characteristic vectors of all other sets .
the non - negativity of the weight function @xmath26 in the definition of threshold hypergraphs implies that only minimal hyperedges matter for the thresholdness property of a given hypergraph . since dually sperner hypergraphs are threshold , we focus on the family of hypergraphs that are both sperner and dually sperner .
we call such hypergraphs _
@xmath0-sperner_. thus , a hypergraph @xmath1 is @xmath0-sperner if and only if for every pair @xmath2 of distinct hyperedges of @xmath1 both inequalities , hold .
in other words , every two distinct hyperedges @xmath22 and @xmath23 satisfy @xmath27 note that by definition , all hypergraphs with at most one hyperedge ( possibly with no vertices ) are @xmath0-sperner .
the aim of this paper is twofold .
first , we prove a decomposition theorem for @xmath0-sperner hypergraphs and examine several of its consequences , including the fact that the characteristic vectors of the hyperedges of a @xmath0-sperner hypergraph are linearly independent . from this
we derive an upper bound on the size of a @xmath0-sperner hypergraph , as well as a lower bound on the size of a @xmath0-sperner hypergraph without universal , isolated , and twin vertices .
second , we show that within the class of normal sperner hypergraphs ( that is , within the class of maximal clique hypergraphs of graphs ) , the classes of @xmath0-sperner hypergraphs , of threshold hypergraphs , and of @xmath4-asummable hypergraphs coincide .
these classes provide three new characterizations of threshold graphs : a graph is threshold if and only if its clique hypergraph is @xmath0-sperner , or equivalently threshold , or equivalently @xmath4-assumable .
every hypergraph @xmath28 with a fixed pair of orderings of its vertices and edges , say @xmath29 , and @xmath30 , can be represented with its _ incidence matrix _
@xmath31 having rows and columns indexed by edges and vertices of @xmath1 , respectively , and being defined as @xmath32 note the slight abuse of notation above : the incidence matrix does not depend only on the hypergraph but also on the pair of orderings of its vertices and edges
. we will be able to neglect this technical issue often in the paper , but not always .
we will therefore say that two matrices @xmath33 and @xmath34 of the same dimensions are _ permutation equivalent _ , and denote this fact by @xmath35 , if @xmath33 can be obtained from @xmath34 by permuting some of its rows and/or columns . for later use , we state a simple property of incidence matrices of a hypergraph .
[ obs : matrices ] let @xmath28 be a hypergraph with a fixed pair of orderings of its vertices and edges , respectively , and let @xmath36 be the corresponding incidence matrix .
then , any permutation of the vertices and/or edges of @xmath1 results in an incidence matrix that is permutation equivalent to @xmath36 .
moreover , any matrix that is permutation equivalent to @xmath36 is the incidence matrix of @xmath1 with respect to some pair of orderings of its vertices and edges .
we now describe an operation that produces a new @xmath0-sperner hypergraph from a given pair of @xmath0-sperner hypergraphs . given a pair of vertex - disjoint hypergraphs @xmath37 and @xmath38 and a new vertex @xmath39 , the _ gluing of @xmath40 and @xmath41 _ is the hypergraph @xmath42 such that @xmath43 and @xmath44 the operation of gluing can be visualized easily in terms of incidence matrices .
let @xmath45 and @xmath46 for @xmath47 , and let us denote by @xmath48 , resp .
@xmath49 , the @xmath50 matrix of all zeroes , resp . of all ones .
then , the incidence matrix of the gluing of @xmath40 and @xmath41 can be written as @xmath51 see fig .
[ fig:1 ] for an example .
[ prop : gluing ] for every pair @xmath37 and @xmath38 of vertex - disjoint @xmath0-sperner hypergraphs , their gluing @xmath52 is a @xmath0-sperner hypergraph , unless @xmath53 and @xmath54 ( in which case the hypergraph @xmath52 is not sperner ) .
let @xmath22 and @xmath23 be two distinct edges of @xmath52 . if @xmath55 then their differences are the same as the corresponding differences of @xmath56 and @xmath57 , both of which are hyperedges of @xmath40 . if @xmath58 then their differences are the same as the corresponding differences of @xmath59 and @xmath60 , both of which are hyperedges of @xmath41 .
if @xmath61 , then @xmath62 and @xmath63 , unless @xmath64 , and @xmath65 ( which implies @xmath66 and @xmath67 by our assumption that both @xmath40 and @xmath41 are @xmath0-sperner ) .
the case of @xmath68 is symmetric .
the following unary operation also preserves the class of @xmath0-sperner hypergraphs . given a hypergraph @xmath28 , the _ complement _ of @xmath1 is the hypergraph @xmath69 with @xmath70 and @xmath71 , where @xmath72 denotes @xmath73 for any subset @xmath74 . [
prop : complement ] the complement of every @xmath0-sperner hypergraph is @xmath0-sperner .
this follows from the fact that for every two sets @xmath75 , we have @xmath76 and @xmath77 . given a vertex @xmath78 of a hypergraph @xmath1
, we say that a hypergraph @xmath1 is _
@xmath78-decomposable _ if for every two hyperedges @xmath79 such that @xmath61 , we have @xmath80 .
equivalently , if the vertex set of @xmath1 can be partitioned as @xmath81 such that @xmath42 for some hypergraphs @xmath37 and @xmath38 .
we say that a hypergraph @xmath1 is _
@xmath0-decomposable _ if it is @xmath78-decomposable for some @xmath82 . a vertex @xmath83 in a hypergraph @xmath84 _ universal _ , resp .
, _ isolated _ if it is contained in all ( resp .
, in no ) hyperedges .
the following proposition gathers some basic properties of the gluing operation and decomposability .
[ prop : complement-2 ] let @xmath1 be a hypergraph .
then , the following holds : a. if @xmath85 , then @xmath86 ( assuming that in both gluings the same new vertex is used ) . b. if @xmath78 is a vertex of @xmath1 such that @xmath1 is @xmath78-decomposable , then @xmath87 is also @xmath78-decomposable . c. if @xmath83 is an isolated or a universal vertex of @xmath1 , then @xmath1 is @xmath83-decomposable . note that if @xmath88 and @xmath89 are two hypergraphs such that @xmath90 , then their incidence matrices are of the same dimensions , say @xmath91 , and @xmath92 ( in particular , we have @xmath93 ) .
@xmath94 suppose that @xmath85 and let @xmath78 be the new vertex , which is also the new vertex in the gluing of @xmath95 and @xmath96 .
we prove the desired equality @xmath86 by establishing permutation equivalence of the incidence matrices of @xmath1 and of @xmath97 ( cf .
remark [ obs : matrices ] ) . by convention ,
when representing the incidence matrix of the result of a gluing , we use the notation as in fig .
[ fig:1 ] , that is , @xmath78 indexes the first column , vertices of the first factor precede , as column indices , all vertices of the second factor , and hyperedges of the first factor precede , as row indices , all hyperedges of the second factor .
letting @xmath98 , @xmath99 , @xmath100 and @xmath101 for @xmath102 , we thus have @xmath103 @xmath104 let @xmath78 be a vertex of @xmath1 such that @xmath1 is @xmath78-decomposable .
then , there exists a pair of vertex - disjoint hypergraphs @xmath40 and @xmath41 such that @xmath1 is the gluing of @xmath40 and @xmath41 with @xmath105 . by the proof of @xmath94 ( and using the same notation ) , we have @xmath106 , where the first column is indexed by @xmath78 .
thus , @xmath87 is @xmath78-decomposable .
@xmath107 note that @xmath83 is universal in @xmath1 if and only if it is isolated in @xmath87 .
by @xmath104 , it therefore suffices to prove the statement for the case when @xmath83 is an isolated vertex of @xmath1 . in this case
, the column of @xmath36 indexed by @xmath83 is the all zero vector .
it follows that @xmath1 is @xmath83-decomposable , as follows : @xmath108 with @xmath42 , @xmath37 and @xmath38 , where @xmath109 , @xmath110 , and @xmath111 .
let @xmath112 be a hyperedge of maximum possible size in a @xmath0-sperner hypergraph @xmath1 .
then , for every two distinct vertices @xmath113 and every two hyperedges @xmath33 containing @xmath114 and @xmath34 containing @xmath115 , @xmath116 implies @xmath117 .
note that @xmath118 , therefore @xmath119 , since @xmath1 is @xmath0-sperner .
analogously , @xmath120 .
thus , if the sets @xmath121 and @xmath122 were not @xmath123-comparable , the pair @xmath124 would violate the @xmath0-sperner property of @xmath1 .
given a positive integer @xmath125 , a hypergraph @xmath1 is said to be _ @xmath125-uniform _
if @xmath126 for all @xmath127 .
[ lem2 ] let @xmath1 be an @xmath125-uniform @xmath0-sperner hypergraph , where @xmath128 . then
, either there is a subset @xmath129 of vertices of size @xmath130 such that @xmath131 for all @xmath127 or there is a subset @xmath132 of vertices of size @xmath133 such that @xmath134 for all @xmath127 .
the statement of the lemma holds if @xmath1 has at most one hyperedge .
so let us assume that @xmath1 has at least two hyperedges , say @xmath22 and @xmath23 .
let @xmath135 . since @xmath1 is @xmath0-sperner , @xmath136 .
if all hyperedges of @xmath1 contain @xmath129 , then we are done .
if there is a hyperedge @xmath137 such that @xmath138 , say @xmath139 , then @xmath22 and @xmath23 are the only hyperedges containing @xmath129 , since otherwise @xmath137 should contain all vertices of such hyperedges other than @xmath83 , which would imply @xmath140 . consequently , all hyperedges that miss a vertex of @xmath129 are subsets of @xmath141 , and the lemma is proved .
lemma [ lem2 ] suggests the following two families of uniform @xmath0-sperner hypergraphs .
[ ex : stars ] given @xmath128 , an _
@xmath125-star _ is an @xmath125-uniform hypergraph @xmath84 such that there exists sets @xmath142 such that * @xmath143 where @xmath144 , @xmath145 , and @xmath146 , and * @xmath147 .
if this is the case , we say that @xmath1 is the _
( @xmath125-)star generated by @xmath148_. clearly , every @xmath125-star is @xmath0-sperner .
moreover , let us verify that every @xmath125-star is @xmath78-decomposable with respect to every vertex @xmath78 .
let @xmath1 be an @xmath125-star generated by @xmath148 and let @xmath82 . if @xmath149 , then @xmath150 and we have @xmath85 where @xmath40 is the @xmath151-star generated by @xmath152 and @xmath153 and @xmath154 .
if @xmath155 , then we have @xmath85 where @xmath156 and @xmath157 . finally , if @xmath158 , then @xmath78 is isolated and @xmath1 is @xmath78-decomposable by proposition [ prop : complement-2 ] .
[ ex : co - stars ] given @xmath128 , an _
@xmath125-antistar _ is an @xmath125-uniform hypergraph @xmath84 such that there exists sets @xmath142 such that * @xmath143 where @xmath145 and @xmath146 , and * @xmath159 .
if this is the case , we say that @xmath1 is the _
( @xmath125-)antistar generated by @xmath148_. note that every @xmath125-antistar is the complement of an @xmath125-star
. it follows , using propositions [ prop : complement ] and [ prop : complement-2 ] and the properties of stars observed in example [ ex : stars ] , that every antistar is @xmath0-sperner and @xmath78-decomposable with respect to each vertex @xmath78 .
[ thm : decomposition ] every @xmath0-sperner hypergraph @xmath84 with @xmath160 is , that is , it is the gluing of two @xmath0-sperner hypergraphs . by proposition [ prop : complement-2 ]
, we may assume that @xmath1 does not have any isolated vertices .
for every @xmath161 , let @xmath162 and let @xmath163 .
we consider two cases . _ case 1 : not all the @xmath164 values are the same .
_ let @xmath161 be a vertex with the smallest @xmath164 value .
then @xmath165 by the assumption of this case .
suppose first that for every hyperedge @xmath166 such that @xmath167 , we have @xmath168 .
we claim that in this case @xmath1 is @xmath169-decomposable .
this is because for every two hyperedges @xmath9 such that @xmath170 , we have @xmath171 , implying @xmath172 , from what we derive , using the fact that @xmath1 is @xmath0-sperner , that @xmath173 , that is , @xmath174 .
this proves the claim .
assume next that there exists a hyperedge @xmath166 such that @xmath167 and @xmath175 .
let @xmath22 be a hyperedge containing @xmath169 of size @xmath164 , and let @xmath137 be a hyperedge of maximum possible size , that is , @xmath176 .
then @xmath177 , since @xmath178 .
since @xmath1 is sperner , there exists a vertex @xmath179 .
note that @xmath180 , therefore @xmath181 , since @xmath1 is @xmath0-sperner .
moreover , @xmath182 since @xmath183 and @xmath23 does not contain @xmath169 .
we know that @xmath184 , by our choice of @xmath169 . therefore , there exists a hyperedge @xmath185 containing @xmath83 and of size @xmath186 .
since @xmath187 , we have @xmath188 . applying lemma
[ lem1 ] with @xmath189 implies @xmath190 .
applying lemma [ lem1 ] with @xmath191 implies @xmath192 .
consequently , @xmath193 .
on the other hand , @xmath194 .
it follows that @xmath195 , contradicting the sperner property of @xmath1 .
this completes case 1 .
_ case 2 : all the @xmath164 values are the same .
_ let @xmath196 and let @xmath197 .
if @xmath198 , then @xmath1 is @xmath78-decomposable with respect to every vertex @xmath78 .
so suppose that @xmath199 .
consider the subhypergraph @xmath200 of @xmath1 with @xmath201 formed by the hyperedges of @xmath1 of size @xmath202 . by lemma [ lem2 ]
applied to @xmath200 , either there is a subset @xmath129 of vertices of size @xmath203 such that @xmath131 for all @xmath204 or there is a subset @xmath132 of vertices of size @xmath205 such that @xmath134 for all @xmath204 .
suppose first that there is a subset @xmath129 of vertices of size @xmath203 such that @xmath131 for all @xmath204 . if @xmath206 , that is , all hyperedges of @xmath1 are of size @xmath202 , then @xmath1 is @xmath78-decomposable with respect to every vertex @xmath78 ( cf .
example [ ex : stars ] ) .
so we may assume that @xmath207 , that is , that @xmath1 contains a hyperedge @xmath137 of size less than @xmath202 . by the assumption of case @xmath4 ,
we know that @xmath208 . since @xmath1 is sperner , @xmath137 is not contained in any of the hyperedges of @xmath200 ; moreover @xmath137 contains at least two vertices from the set @xmath209 .
if @xmath137 contains at least three vertices from @xmath153 , say @xmath210 , then the sets @xmath211 and @xmath137 would violate the @xmath0-sperner property , since @xmath212 and @xmath213 ( note that @xmath214 ) .
it follows that @xmath215 .
in fact , we have @xmath216 , say @xmath217 , since otherwise , using similar arguments as above , we see that the sets @xmath218 and @xmath137 would violate the @xmath0-sperner property , where @xmath219 .
it follows that @xmath200 has exactly @xmath4 hyperedges , and @xmath220 for every set @xmath221 .
consequently , @xmath1 is @xmath115-decomposable for every @xmath222 : decomposing @xmath1 with respect to @xmath223 , for instance , we have @xmath85 where @xmath37 with @xmath224 , @xmath225 , and @xmath226 .
it remains to consider the case when there is a subset @xmath132 of vertices of size @xmath205 such that @xmath134 for all @xmath204 . since we assume that @xmath227 for all vertices @xmath169 , we have @xmath228 .
let us define @xmath229 and @xmath230 .
then , @xmath145 , and every hyperedge @xmath231 must contain @xmath153 , since @xmath1 is sperner and for every vertex @xmath222 , the set @xmath232 is a hyperedge of @xmath1 .
consequently , @xmath1 is @xmath115-decomposable for every @xmath222 : taking any @xmath222 , we have @xmath85 where @xmath37 with @xmath233 , @xmath234 , and @xmath226 .
let us say that a gluing of two vertex - disjoint hypergraphs @xmath37 , @xmath38 is _ safe _ if it results in a @xmath0-sperner hypergraph . by proposition [ prop : gluing ]
, this is equivalent to requiring that either @xmath235 or @xmath236 .
theorem [ thm : decomposition ] and proposition [ prop : gluing ] imply the following composition result for the class of hypergraphs .
[ cor : composition ] a hypergraph @xmath1 is @xmath0-sperner if and only if it either has no vertices ( that is , @xmath237 ) or it is a safe gluing of two smaller @xmath0-sperner hypergraphs . here , we give a new proof of the thresholdness property of @xmath0-sperner hypergraphs .
recall that it was shown in @xcite that every dually sperner hypergraph is threshold ; in particular , every @xmath0-sperner hypergraph is threshold . that proof is based on the characterization of thresholdness in terms of @xmath202-asummability ( see section [ sec : threshold ] ) .
the proof given here is based on the composition theorem for @xmath0-sperner hypergraphs ( theorem [ cor : composition ] ) and , unlike the proof from @xcite , also constructs the weights and a threshold as in the definition of threshold hypergraphs . [
prop : inclusion ] every @xmath0-sperner hypergraph is threshold
. we will show by induction on @xmath238 that for every @xmath0-sperner hypergraph @xmath84 there exists a _ nice threshold separator _
, that is , a pair @xmath239 such that @xmath240 is a strictly positive integer weight function and @xmath13 is a non - negative integer threshold such that : a. for every subset @xmath14 , we have @xmath241 if and only if @xmath16 for some @xmath17 ; b. if @xmath242 then @xmath243 , and c. if @xmath244 then @xmath245 . in particular , this will establish the thresholdness of @xmath1 . for @xmath246
, we can take the ( empty ) mapping given by @xmath247 for all @xmath248 and the threshold @xmath249 now , let @xmath84 be a @xmath0-sperner hypergraph with @xmath250 . by theorem [ cor : composition ] , @xmath1 is the safe gluing of two @xmath0-sperner hypergraphs , say @xmath42 with @xmath37 and @xmath38 , where @xmath251 , @xmath252 , and @xmath39 . by the inductive hypothesis , @xmath40 and
@xmath41 admit nice threshold separators .
that is , there exist positive integer weight functions @xmath253 and non - negative integer thresholds @xmath254 for @xmath47 such that for every subset @xmath255 , we have @xmath256 if and only if @xmath16 for some @xmath257 .
moreover , if @xmath258 then @xmath259 , and if @xmath260 then @xmath261 .
let us define the threshold @xmath262 , where @xmath263 , and the weight function @xmath240 by the rule @xmath264 let us verify that the so defined weight function is indeed strictly positive .
since @xmath265 for @xmath266 are strictly positive and @xmath267 , we have @xmath268 for all @xmath269 .
moreover , since @xmath270 , @xmath271 , and @xmath272 , we have @xmath273 . if @xmath274 , then @xmath275 and @xmath276 , which implies @xmath53 and @xmath277 , contrary to the fact that the gluing is safe .
it follows that @xmath278 , as claimed .
we claim that @xmath239 is a nice threshold separator of @xmath1 .
we first verify property @xmath104 .
suppose that @xmath242 .
then @xmath279 , which is equivalent to @xmath280 .
this implies @xmath281 .
if @xmath282 , then @xmath283 , a contradiction . therefore , @xmath284 , which by the inductive hypothesis implies @xmath53 . moreover , we have @xmath285 , which implies @xmath286 . consequently , @xmath287 , as claimed .
now , we verify property @xmath107 .
suppose that @xmath244 .
this implies @xmath288 and @xmath289 , and consequently @xmath290 ( since @xmath291 is strictly positive ) and @xmath277 ( by the inductive hypothesis ) .
since the gluing is safe and @xmath277 , it follows that @xmath292 , hence , using @xmath290 , we infer @xmath293 .
we thus have @xmath294 , as claimed .
finally , we verify property @xmath94 , that is , that for every subset @xmath14 , we have @xmath241 if and only if @xmath16 for some @xmath17 .
suppose first that @xmath241 for some @xmath14 .
let @xmath295 for @xmath47 . for later use , we note that @xmath296 suppose first that @xmath149 .
then @xmath297 which implies @xmath298 if @xmath299 then , using , we obtain @xmath300 a contradiction with .
it follows that @xmath301 .
consequently there exists @xmath302 such that @xmath303 , hence the hyperedge @xmath304 satisfies @xmath16 . now , suppose that @xmath305 . in this case ,
@xmath306 which implies @xmath307 we must have @xmath308 since if there exists a vertex @xmath309 , then we would have @xmath310 where the last inequality follows from and @xmath272
. therefore , inequality simplifies to @xmath311 , and consequently there exists a hyperedge @xmath312 such that @xmath313 .
this implies that @xmath1 has a hyperedge @xmath314 such that @xmath315 .
for the converse direction , suppose that @xmath316 is a subset of @xmath8 such that @xmath16 for some @xmath17 .
we need to show that @xmath241 .
we consider two cases depending on whether @xmath317 or not .
suppose first that @xmath317 .
then @xmath318 for some @xmath302 .
due to the property of @xmath291 , we have @xmath319 .
consequently , @xmath320 suppose now that @xmath321 .
then @xmath322 for some @xmath312 .
due to the property of @xmath323 , we have @xmath324 .
consequently , @xmath325 this shows that @xmath241 whenever @xmath316 contains a hyperedge of @xmath1 , and completes the proof .
the same inductive construction of weights and essentially the same proof shows that every @xmath0-sperner hypergraph @xmath84 is also _
equilizable _ , which means that there exist a non - negative integer weight function @xmath12 and a non - negative integer threshold @xmath13 such that for every subset @xmath14 , we have @xmath326 if and only if @xmath327 .
note that equilizable hypergraphs generalize the class of equistable graphs studied , e.g. , in @xcite , in that a graph @xmath328 is equistable if and only if its maximal stable set hypergraph is equilizable .
maximal stable set hypergraph _ of a graph @xmath328 is the hypergraph @xmath329 with vertex set @xmath330 and in which the hyperedges are exactly the maximal stable sets of @xmath328 .
[ prop : equilizable ] every @xmath0-sperner hypergraph is equilizable . for the sake of completeness
, we include the proof of proposition [ prop : equilizable ] in appendix . by @xmath331 , resp .
@xmath332 , we will denote the vector of all zeroes , resp .
ones , of appropriate dimension ( which will be clear from the context ) .
the following lemma can be easily derived from proposition [ prop : equilizable ] .
[ lem : x ] for every @xmath0-sperner hypergraph @xmath84 such that and , there exists a vector @xmath333 such that @xmath334 and @xmath335 .
let @xmath84 be a @xmath0-sperner hypergraph as in the statement of the lemma . by proposition [ prop : equilizable ]
, @xmath1 is equilizable .
let @xmath12 be a non - negative integer weight function and @xmath13 a non - negative integer threshold such that for every subset @xmath14 , we have @xmath326 if and only if @xmath327 .
if @xmath336 , then @xmath337 and consequently @xmath245 , a contradiction .
it follows that @xmath338 , and we can define the vector @xmath333 given by @xmath339 for all @xmath161 .
we claim that vector @xmath114 satisfies the desired properties @xmath334 and @xmath335 . since @xmath326 for all @xmath327
, we have @xmath334 . since @xmath340
, an arbitrary hyperedge @xmath17 shows that @xmath341 .
consequently , we also have [ cor : lambda ] for every @xmath0-sperner hypergraph @xmath84 and every vector @xmath342 we have @xmath343 if or , then the left hand side of the above implication is always false . in all other cases , by lemma [ lem : x ]
, there exists a vector @xmath344 such that @xmath334 and @xmath335 .
therefore , equation @xmath345 implies @xmath346 for every @xmath0-sperner hypergraph @xmath84 such that , the characteristic vectors of its hyperedges are linearly independent in @xmath347 .
we use induction on @xmath348 . if @xmath349 , then the statement holds since @xmath350 .
suppose now that @xmath351 .
then by theorem [ thm : decomposition ] , @xmath1 is the gluing of two @xmath0-sperner hypergraphs , say @xmath42 with @xmath37 and @xmath38 , where @xmath251 , @xmath252 , and @xmath39 .
let @xmath342 be a vector such that @xmath352 .
let @xmath353 and @xmath354 be the restrictions of @xmath355 to the hyperedges corresponding to @xmath356 and @xmath357 , respectively .
the equation @xmath352 implies the system of equations @xmath358 in all cases , the inductive hypothesis implies that @xmath359 and @xmath360 , except in the case when @xmath277 . in this case
, @xmath361 is a single number , say @xmath362 . if @xmath363 , then @xmath364 follows by the induction hypothesis . if @xmath365 , then @xmath366 satisfies @xmath367 and @xmath368 , contradicting corollary [ cor : lambda ] .
[ cor : edges ] for every @xmath0-sperner hypergraph @xmath84 with @xmath160 , we have @xmath369 . recall that a vertex @xmath83 in a hypergraph @xmath84 _ universal _ , resp . , _ isolated _ if it is contained in all ( resp .
, in no ) hyperedges .
moreover , two vertices @xmath370 in a hypergraph @xmath84 are _ twins _ if they are contained in exactly the same hyperedges .
corollary [ cor : edges ] gives an upper bound on the number of hyperedges of a @xmath0-sperner hypergraph in terms of the number of vertices .
can we prove a lower bound of a similar form ?
in general not , since adding universal vertices , isolated vertices , or twin vertices preserves the @xmath0-sperner property and the number of edges , while it increases the number of vertices . however , as we show next , for @xmath0-sperner hypergraphs without universal , isolated vertices , and twin vertices , we prove the following sharp lower bound on the number of hyperedges . [ prop : lower - bound ] for every @xmath0-sperner hypergraph @xmath84 with @xmath371 and without universal , isolated , and twin vertices , we have @xmath372 this bound is sharp .
we use induction on @xmath238 . for @xmath373
, it can be easily verified that the statement holds .
now , let @xmath84 be a @xmath0-sperner hypergraph with @xmath374 and without universal vertices , isolated vertices , and twin vertices . by theorem [ thm : decomposition ] , @xmath1 is the gluing of two @xmath0-sperner hypergraphs , say @xmath42 with @xmath37 and @xmath38 , where @xmath251 , @xmath252 , and @xmath39 .
let @xmath45 and @xmath46 for @xmath47 , and let @xmath375 .
we have @xmath376 , and by the rules of the gluing , @xmath377 . by proposition
[ prop : complement ] , we may assume that @xmath378 ( otherwise , we can consider the complementary hypergraph ) .
in particular , @xmath379 . the fact that @xmath1 does not have a universal vertex implies @xmath40 does not have a universal vertex .
similarly , @xmath41 does not have an isolated vertex .
since @xmath1 does not have any pairs of twin vertices , we have that either @xmath40 does not have a isolated vertex , or @xmath41 does not have a universal vertex .
we may assume that @xmath41 does not have a universal vertex ( otherwise , we consider a different gluing in which we delete the universal vertex from @xmath41 and add an isolated vertex to @xmath40 ) .
first , suppose that @xmath380 , then we apply the inductive hypothesis for @xmath381 and @xmath41 , where @xmath381 is the hypergraph obtained from @xmath40 by deleting from it the isolated vertex ( if it exists ) .
letting @xmath382 and @xmath383 , we thus have @xmath384 and also @xmath385 .
we obtain @xmath386 and @xmath387 consequently , @xmath388 and , since @xmath389 is integer , the desired inequality @xmath390 follows .
suppose now that @xmath391 .
in this case we have @xmath392 , by the sperner property of @xmath41 .
moreover , @xmath40 does not have an isolated vertex , since such a vertex would be a twin in @xmath1 of the unique vertex of @xmath41 .
thus , we can apply the inductive hypothesis on @xmath40 , which yields @xmath393 and consequently @xmath394 as desired .
finally , suppose that @xmath395 . in this case
, since @xmath1 does not have a universal vertex , we must have @xmath277 and @xmath392 .
as above , let @xmath381 be the hypergraph obtained from @xmath40 by deleting from it the isolated vertex ( if it exists ) . letting @xmath382 and @xmath383
, we obtain , by applying the inductive hypothesis to @xmath381 , @xmath396 which implies @xmath397 this completes the proof of the inequality . to see that the inequality is sharp , consider the following family of hypergraphs .
for @xmath199 , let @xmath398 be the hypergraph defined recursively as follows : * @xmath399 .
* for @xmath400 , we set @xmath401 where @xmath402 is the hypergraph obtained from a disjoint copy of @xmath403 by adding to it an isolated vertex .
an inductive argument shows that for every @xmath199 , we have @xmath404 , @xmath405 , and consequently @xmath406
a _ threshold graph _ is a threshold hypergraph in which all hyperedges are of size @xmath4 .
threshold graphs were introduced by chvtal and hammer in 1970s @xcite and were afterwards studied in numerous paper .
many results on threshold graphs are summarized in the monograph by mahadev and peled @xcite .
threshold graphs have many different characterizations .
we now recall two of them and then prove three more . for a set @xmath407 of graphs , we say that a graph @xmath328 is @xmath407-free if no induced subgraph of @xmath328 is isomorphic to a member of @xmath407 . by @xmath408
we denote the @xmath409-vertex path , by @xmath410 the @xmath409-vertex cycle , and by @xmath411 the disjoint union of two copies of @xmath412 ( the complete graph of order two ) .
[ thm : char1 ] a graph @xmath328 is threshold if and only if @xmath328 is @xmath413-free .
[ thm : char2 ] a graph @xmath328 is threshold if and only if @xmath414 where @xmath415 is a clique , @xmath416 is an independent set , @xmath417 , and there exists an ordering @xmath418 of @xmath416 such that @xmath419 for all @xmath420 . given a graph @xmath328 , the _ clique hypergraph of @xmath328 _ is the hypergraph @xmath421 with vertex set @xmath330 in which the hyperedges are exactly the maximal cliques of @xmath328 .
the clique hypergraphs of graphs are exactly those sperner hypergraphs @xmath1 that are also _ normal _ ( or : _ conformal _ ) , that is , for every set @xmath422 such that every pair of elements in @xmath316 is contained in a hyperedge , there exists a hyperedge containing @xmath316 ( see @xcite . )
a necessary condition for thresholdness of a hypergraph @xmath1 is @xmath202-asummability for any @xmath199 .
a hypergraph is _
@xmath202-asummable _ if it has no @xmath202 ( not necessarily distinct ) independent sets @xmath423 and @xmath202 ( not necessarily distinct ) dependent sets @xmath424 such that @xmath425 a hypergraph is known to be threshold if and only if it is @xmath202-asummable for all @xmath202 @xcite .
( more recent and complete information on this can be found in @xcite . )
it is also known that @xmath4-asummability does not imply thresholdness in general @xcite , and thresholdness does not imply @xmath0-spernerness ( as can be seen by considering the edge set of the complete graph @xmath426 ) .
as the next theorem shows , in the class of conformal sperner hypergraphs , the notion of thresholdness coincides with both @xmath4-asummability and @xmath0-spernerness .
moreover , it exactly characterizes threshold graphs .
[ thm : threshold ] for a graph @xmath328 , the following statements are equivalent : 1 .
@xmath328 is threshold .
2 . the clique hypergraph @xmath421 is @xmath0-sperner .
3 . the clique hypergraph @xmath421 is threshold .
the clique hypergraph @xmath421 is @xmath4-asummable .
the implication @xmath427 follows directly from theorem [ thm : char2 ] .
the implication @xmath428 is the statement of proposition [ prop : inclusion ] .
the implication @xmath429 follows from @xcite .
now we prove the implication @xmath430 .
we break the proof into a small series of simple claims .
[ cl-1 ] for any graph @xmath328 , if @xmath431 , @xmath432 , and @xmath433 such that @xmath434 , then the set @xmath435 is an independent set of the hypergraph @xmath421 .
if @xmath436 is a clique of @xmath328 , then @xmath437 is also a clique of @xmath328 , and hence @xmath112 is not maximal .
[ cl-2 ] if @xmath421 is @xmath4-asummable , and @xmath431 , @xmath432 , and @xmath433 then @xmath438 . if @xmath434 then claim [ cl-1 ] and the equality @xmath439 contradicts the @xmath4-asummability of @xmath421 .
[ cl-3 ] if @xmath421 is @xmath4-asummable , and @xmath431 , @xmath432 , and @xmath433 , then the set @xmath440 is an independent set of the hypergraph @xmath421 .
if @xmath441 is a maximal clique of @xmath328 , then * @xmath442 is not a clique , and hence there exists a vertex @xmath443 ( such that @xmath444 ) ; * @xmath445 is not a clique , and hence there exists a vertex @xmath446 ( such that @xmath447 ) .
consequently we have vertices @xmath448 and @xmath449 such that @xmath450 implying @xmath451 .
this contradicts claim [ cl-2 ] , which proves that such a maximal clique @xmath112 can not exist .
[ cl-4 ] if @xmath421 is @xmath4-asummable then @xmath328 is a threshold graph .
let us recall the forbidden subgraph characterization of threshold graphs by theorem [ thm : char1 ] , and assume indirectly that there exists four distinct vertices @xmath452 such that @xmath453 and @xmath454 ( that is , that @xmath455 induces either a @xmath408 , a @xmath410 , or a @xmath411 . )
let us then consider maximal cliques @xmath456 and @xmath457 of @xmath328 . since @xmath453 such maximal cliques
do exist .
our assumptions that @xmath454 imply that @xmath458 and @xmath459
. then we can apply claim [ cl-3 ] and obtain that both @xmath460 and @xmath461 are independent sets of the hypergraph @xmath421 .
since @xmath462 , it also follows that @xmath463 is an independent set of @xmath421 .
consequently , the equality @xmath464 contradicts the @xmath4-asummability of @xmath421 .
this contradiction proves that such a set of four vertices can not exists , from which the claim follows by theorem [ thm : char1 ] .
the last claim proves the implication @xmath430 , completing the proof of the theorem .
finally , we remark that , since the class of threshold graphs is closed under taking complements , one could obtain three further characterizations of threshold graphs by replacing the clique hypergraph @xmath421 with the maximal stable set hypergraph @xmath329 in any of the properties @xmath465@xmath466 in theorem [ thm : threshold ] .
# 10=0=0 0 by1pt#1 10 l. alcn , m. gutierrez , i. kovcs , m. milani , and r. rizzi .
strong cliques and equistability of ept graphs . accepted . c. berge . , volume 45 of _ north - holland mathematical library_. north - holland publishing co. , amsterdam , 1989 .
combinatorics of finite sets , translated from the french .
e. boros , v. gurvich , and m. milani .
on equistable , split , cis , and related classes of graphs . to appear .
n. chiarelli and m. milani .
linear separation of connected dominating sets in graphs .
proceedings of isaim 2014 , international symposium on artificial intelligence and mathematics , fort lauderdale , fl . january 68 , 2014 .
available on the conference web page , http://www.cs.uic.edu / isaim2014/. n. chiarelli and m. milani .
total domishold graphs : a generalization of threshold graphs , with connections to threshold hypergraphs . , 179:112 , 2014 .
c. chow .
boolean functions realizable with single threshold devices . in _ proceedings of the ire ,
39 _ , pages 370371 , 1961 .
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 .
y. crama and p. l. hammer . , volume 142 of _ encyclopedia of mathematics and its applications_. cambridge university press , cambridge , 2011 . c. c. elgot .
truth functions realizable by single threshold organs . in _ proceedings of the second annual symposium on switching circuit theory and logical design , 1961 ( swct 1961 ) .
_ , pages 225245 , oct 1961 .
m. c. golumbic . .
academic press [ harcourt brace jovanovich , publishers ] , new york - london - toronto , ont .
, 1980 .
e. korach , u. n. peled , and u. rotics .
equistable distance - hereditary graphs . , 156(4):462477 , 2008 .
v. e. levit and m. milani .
equistable simplicial , very well - covered , and line graphs .
, 165:205212 , 2014 .
n. mahadev , u. n. peled , and f. sun .
equistable graphs . , 18(3):281299 , 1994 .
n. v. r. mahadev and u. n. peled . , volume 56 of _ annals of discrete mathematics_. north - holland publishing co. , amsterdam , 1995 . .
miklavi and m. milani .
equistable graphs , general partition graphs , triangle graphs , and graph products .
, 159(11):11481159 , 2011 .
m. milani and n. trotignon .
equistarable graphs and counterexamples to three conjectures on equistable graphs . ,
2014 . s. muroga . .
wiley - interscience [ john wiley & sons ] , new york - london - sydney , 1971 .
u. n. peled and u. rotics .
equistable chordal graphs . , 132:203210 , 2003 .
u. n. peled and b. simeone .
polynomial - time algorithms for regular set - covering and threshold synthesis .
, 12(1):5769 , 1985 .
j. reiterman , v. rdl , e. iajov , and m. tma . threshold hypergraphs .
, 54(2):193200 , 1985 .
every @xmath0-sperner hypergraph is equilizable
. we will show by induction on @xmath238 that for every @xmath0-sperner hypergraph @xmath84 there exists a strictly positive integer weight function @xmath240 and a non - negative integer threshold @xmath13 such that for every subset @xmath14 , we have @xmath467 if and only if @xmath327 .
this will establish the equilizability of @xmath1 .
now , let @xmath84 be a @xmath0-sperner hypergraph with @xmath250 . by theorem [ cor : composition ] , @xmath1 is a safe gluing of two @xmath0-sperner hypergraphs , say @xmath42 with @xmath37 and @xmath38 , where @xmath251 , @xmath252 , and @xmath39 . by the inductive hypothesis , @xmath40 and @xmath41
are equilizable , that is , there exist positive integer weight functions @xmath253 and non - negative integer thresholds @xmath254 for @xmath47 such that for every subset @xmath255 , we have @xmath469 if and only if @xmath470 .
let us define the threshold @xmath262 , where @xmath263 , and the weight function @xmath240 by the rule @xmath264 since the weight function defined above coincides with the weight function defined in the proof of proposition [ prop : inclusion ] , the function is indeed strictly positive .
suppose first that @xmath467 for some @xmath14 .
let @xmath295 for @xmath47 . for later use , we note that @xmath471 we consider two cases depending on whether @xmath149 or not .
suppose first that @xmath149 .
then @xmath472 which implies @xmath473 if @xmath299 then , using , we obtain @xmath300 a contradiction with . therefore @xmath301 . moreover , if @xmath474 , then @xmath475 again contradicting .
we infer that @xmath476 and consequently @xmath477 .
equation together with @xmath476 implies that @xmath478 .
since @xmath323 is strictly positive on all @xmath479 , it follows that @xmath480 .
therefore , we have @xmath481 . now
, suppose that @xmath305 . in this case ,
@xmath482 which implies @xmath483 we must have @xmath308 since if there exists a vertex @xmath309 , then we would have @xmath310 where the last inequality follows from and @xmath272
. therefore , equality simplifies to @xmath484 , and consequently there exists a hyperedge @xmath485 .
this implies that @xmath486 . for the converse direction , suppose that @xmath316 is a subset of @xmath8 such that @xmath327 .
we need to show that @xmath467 .
we again consider two cases depending on whether @xmath149 or not .
suppose first that @xmath149 .
then @xmath487 for some @xmath477 .
due to the property of @xmath291 , we have @xmath488 .
consequently , @xmath489 suppose now that @xmath305 .
then @xmath490 for some @xmath485 .
due to the property of @xmath323 , we have @xmath484 .
consequently , @xmath491 this shows that we have @xmath467 whenever @xmath327 , and completes the proof . | we introduce a new class of hypergraphs , the class of @xmath0-sperner hypergraphs .
a hypergraph @xmath1 is said to be _ @xmath0-sperner _ if every two distinct hyperedges @xmath2 of @xmath1 @xmath3 .
we prove a decomposition theorem for @xmath0-sperner hypergraphs and examine several of its consequences , including bounds on the size of @xmath0-sperner hypergraphs and a new , constructive proof of the fact that every @xmath0-sperner hypergraph is threshold .
we also show that within the class of normal sperner hypergraphs , the ( generally properly nested ) classes of @xmath0-sperner hypergraphs , of threshold hypergraphs , and of @xmath4-asummable hypergraphs coincide .
this yields new characterizations of the class of threshold graphs . |
survival of a species of bacteria , translocation of dna through a nano - pore , and diffusion in presence of an absorbing boundary are only few out of many situations , where the central question is the survival , or persistence of the underlying stochastic process . more precisely , persistence , or survival probability @xmath23 of a process is the probability that the process , starting from an initial positive position , stays positive over a time interval @xmath24 $ ] . for many stochastic processes arising in non - equilibrium systems
, persistence decays as a power law @xmath25 , where @xmath26 is called the persistence exponent @xcite . for a simple markov process such as one - dimensional brownian motion , @xmath27 @xcite . on the other hand , the exponent @xmath26 is non - trivial whenever the process is non - markovian , i.e. , has a memory .
in addition to theoretical studies ( for a brief review see @xcite ) , the exponent @xmath26 has been measured in a number of experiments @xcite . even for gaussian non - markovian processes , @xmath26 is non - trivial @xcite . for the latter processes that are
_ close _ to a markov process ( i.e. , whose correlators are close to that of a gaussian markov process ) the exponent @xmath26 was computed perturbatively @xcite .
this perturbation theory has been used for various out - of - equilibrium systems , as the global persistence at the critical point of the ising model in @xmath28 dimensions @xcite , in simple diffusion close to dimension @xmath10 @xcite , and in fluctuating fields such as interfaces @xcite . a quantity that contains more spatial information than persistence @xmath23 is the probability density @xmath7 of the particle at position @xmath8 and at time @xmath9 , given that it has survived ( stayed positive ) up to time @xmath9 .
to investigate @xmath7 , one can equivalently think of a process on the positive semi - infinite line @xmath29 $ ] with _ absorbing _ boundary condition at the origin @xmath30 ( see fig . [ a1 ] ) .
the question is , how does @xmath7 depend on @xmath8 ?
in other words , how does the presence of an absorbing boundary at the origin change the spatial dependence of the probability density of the particle at time @xmath9 ?
in particular , it is clear that @xmath7 must vanish as @xmath31 and @xmath32 .
but how do they vanish there ?
one of the main messages of our paper is that for generic non - markovian processes , @xmath7 vanishes near its boundaries at @xmath30 and @xmath32 in a non - trivial way , characterized by non - trivial exponents . as the persistence @xmath23 , the probability @xmath7 can be computed exactly for a gaussian markov process , as e.g. one - dimensional brownian motion . for non - markovian processes ,
even if they are gaussian , @xmath7 was not known . in this work ,
we consider @xmath7 for a class of one - dimensional gaussian processes known as fractional brownian motion ( fbm ) , which are parametrized by their hurst exponent @xmath2 , with @xmath3 .
the case @xmath4 corresponds to ordinary brownian motion , which is a markov process , while for @xmath6 the process is non - markovian .
the purpose of this paper is to develop a systematic perturbation theory to compute @xmath7 for non - markovian fbm s with @xmath13 , where @xmath14 is the expansion parameter for the perturbation theory .
here we present the result for @xmath7 to @xmath33 .
it can be written as a combination of special functions , i.e. error and hypergeometric functions , see eq .
( [ w ] ) . to our knowledge , this is the first systematic ( exact up to @xmath34 ) calculation of @xmath7 for fractional brownian motion with @xmath6 . denotes the number of the monomer that is crossing the pore at time @xmath9.,title="fig:",width=325 ] before detailing our results , let us position them into a broader context : fractional brownian motion with @xmath6 is relevant for polymer translocation through a nanopore . consider a polymer chain composed of @xmath35 monomers passing through a pore ( translocation ) from left to right , as drawn on fig .
[ translocation ] .
the dynamics of this translocation process has been investigated intensively due to its central role in understanding , e.g. , viral injection of dna into a host , or rna transport through nano - pores , and mastering such applications as fast dna or rna sequencing through engineered channels @xcite .
the translocation coordinate @xmath36 , namely the label of the monomer crossing the pore at time @xmath9 , is key to quantitatively describing the translocation process @xcite , which begins when @xmath37 , and ends when @xmath38 , i.e. , when the first and the last monomer of the chain enter the pore , respectively , see fig .
[ translocation ] . for large @xmath35 ,
when the translocation is not yet complete , one can view @xmath36 as a stochastic process on the semi - infinite line with absorbing boundary conditions at @xmath39 . the absorbing boundary at @xmath39 models that if the chain falls back to the left , i.e. , on the starting side , it will diffuse away and not try again .
the quantity @xmath40 then represents the probability that @xmath8 monomers have translocated to the right at time @xmath9 . to model the process @xmath36
, one observes the following facts : ( i ) scaling arguments and numerical simulations show that @xmath36 is subdiffusive @xcite ; ( ii ) in absence of boundaries , numerical simulations indicate that @xmath36 is a gaussian process @xcite .
based on these observations it was proposed in ref .
@xcite that a good candidate for @xmath36 is a fractional brownian motion with @xmath41 , where the exponent @xmath42 describes the growth of the radius of gyration with the number of monomers ( @xmath43 ) @xcite .
thus for @xmath44 , @xmath45 and hence @xmath36 is generically a non - markovian process , with absorbing boundary conditions at @xmath39 and at @xmath46 .
here we consider the limit of @xmath47 .
thus our results for @xmath7 of a fbm with @xmath6 are directly relevant for polymer translocation .
directions for further applications are numerous : recently a relation was established between the statistics of avalanches associated with the motion of a driven particle in a disordered potential and persistence properties of the latter @xcite .
higher - dimensional generalizations are avalanches of extended elastic objects , for which systematic field - theoretic treatments exist @xcite . in few cases ,
no - hitting probabilities can be calculated for extended ( non - directed ) objects , as self - avoiding random walks avoiding extended objects @xcite . other approaches use real - space renormalization @xcite .
this article is organized as follows : since some of the computations are rather technical , we first provide in section [ summary ] a brief summary of the main definitions and our principal results . in section [ s : prelim ]
, we introduce basic notations and reproduce the known results for @xmath48 .
section [ s : perturbation theory ] explains the basic ideas of our perturbative approach , sketches the calculation , and discusses some of the subtle points .
our predictions are compared to numerical simulations in section [ numerics ] .
conclusions are presented in section [ s : conclusions ] .
more technical points are relegated to two appendices : in appendix [ a : action ] the correction to the action is derived .
appendix [ s : evaluation of z+ ] contains the explicit calculation of the perturbation theory .
finally , appendix [ s : scaling ] reviews the arguments for the scaling law @xmath49 .
consider a particle , located at time @xmath50 at the origin @xmath30 and free to propagate on the real axis .
for gaussian processes , the probability to find the particle inside the interval @xmath51 at time @xmath9 is given by @xmath52 where @xmath53 is the particle s mean square displacement .
a natural scaling variable is @xmath54 and most of the properties of the process are a function of this single variable .
for example , the distribution probability in eq .
( [ propagator1 ] ) becomes @xmath55 in many problems the motion is confined to an interval , finite or semi - infinite . in presence of _ absorbing _ boundaries , the probability distribution of the particle position , subject to the condition that the particle has survived , has no longer a simple gaussian form since it has to vanish at the boundaries .
however , one can still express it as a function of the sole scaling variable @xmath56 defined in eq .
( [ rescaled ] ) , where @xmath53 is the particle s mean square displacement in the _ unconstrained ( without boundaries ) process over the full real line_. in particular , here we discuss the case where the particle can move on the positive semi - axis and is absorbed whenever @xmath57 .
we call @xmath7 and @xmath12 with @xmath56 given in eq . the _ normalized _ probability distribution and the scaling function of the problem in presence of an absorbing boundary at the origin , @xmath58 in contrast to the free case , the functional form of @xmath12 is not the same for all gaussian processes , but depends on the precise nature of the latter .
here we study a particular class of processes , the fractional brownian motion ( fbm ) , for which the autocorrelation function in absence of boundaries is @xmath59 where @xmath2 with @xmath3 is the hurst exponent . for @xmath4 ,
the fbm identifies with brownian motion @xmath60 where @xmath61 is the diffusion constant .
note that only for @xmath4 , the gaussian process @xmath0 is markovian . for other values of @xmath2 ,
the process is non - markovian . for brownian motion ( @xmath4 ) , the form of @xmath12
can be obtained using the method of images ( see section [ s : prelim ] ) , @xmath62 the superscript @xmath63 identifies the case @xmath4 . for other values of @xmath2 , due to the non - markovian nature of the process , the method of images no longer works and the computation of @xmath12 becomes a challenging problem . in this paper
we compute this function , using a perturbative approach for @xmath13 , to first order in @xmath14 .
the final result is [ s : numerics ] defined in eq .
( [ w ] ) .
the solid blue line is the result of the series expansion ( [ wseries ] ) , while the dashed red line is the result of the asymptotic expansion ( [ wasymp ] ) .
we note that @xmath64 .
, width=325 ] @xmath65 \\ \label{w } w ( y)&=&\frac{1}{6 } y^4 \ , _ 2f_2\!\left(1,1;\frac{5}{2},3;\frac{y^2}{2}\right){\nonumber}\\ & & + \pi ( 1-y^2)\ , \text{erfi}\big(\frac{y}{\sqrt{2}}\big)+\sqrt{2 \pi } e^{\frac{y^2}{2 } } y{\nonumber}\\ & & + \left(y^2 - 2\right ) \left[{\ln}\!\big(2 y^2\big)+\gamma_{\mathrm{e } } \right]-3 y^2 \ , \end{aligned}\ ] ] where @xmath66 is euler s constant , @xmath67 a hypergeometric function , and @xmath68 the imaginary error - function .
we can write a _ convergent _ series - expansion @xmath69 - 3 y^2 \end{aligned}\ ] ] where each line is equivalent to the corresponding line in eq .
( [ w ] ) .
both sums converge for all @xmath56 , but problems of numerical precision appear for @xmath70 . in that region , one can use the asymptotic expansion @xmath71 at @xmath72 , the difference between ( [ wseries ] ) and ( [ wasymp ] ) is smaller than @xmath73 .
we obtain , at first order in @xmath14 , the asymptotic expansions of @xmath74 , @xmath75 \nonumber \\ r_+(y ) & \xrightarrow{y \to \infty } & y e^{-y^2/2 } \left [ 1 - 2 \epsilon { \ln}y + \epsilon(1-{\ln}2 -\gamma_{\mathrm e } ) \right ] { \nonumber}\\ & & + \dots \end{aligned}\ ] ] these asymptotics can be recast into @xmath76 where the two exponents @xmath77 and @xmath78 are at first order in @xmath14 given by @xmath79 in a recent publication @xcite ( reviewed in appendix [ s : scaling ] ) , a general scaling relation , valid for arbitrary self - affine processes with stationary increments , was proposed between the exponent @xmath77 , the persistence exponent @xmath26 , and the hurst exponent @xmath2 , @xmath80 for fbm , it is known rigorously that @xmath81 @xcite .
this result predicts that for fbm , @xmath82 one of the objectives of this paper was to verify this scaling relation up to @xmath33 in a perturbation theory around @xmath4 .
using @xmath13 , one expects @xmath83 for fbm .
this is in agreement with our result ( [ final3 ] ) , putting the scaling arguments on a firmer footing
. it is interesting to note that the scaling function @xmath74 given in eq .
( [ propagator2 ] ) has , at least to @xmath33 , the same leading large-@xmath56 behavior @xmath84 as in the unconstrained case ( [ propagator2 ] ) .
this behavior can be understood by a simple heuristic argument : far from the boundary the process is not `` aware '' of the latter .
our calculation reveals that the process nevertheless knows about the boundary , and @xmath12 has a subleading power - law prefactor @xmath85 where @xmath78 is a new ( independent ) exponent , whose result to order @xmath14 is given in eq .
( [ final3 ] ) .
our analytical results are then verified via numerical simulations for @xmath22 .
to simplify notations , we set @xmath86 in the following .
the final result ( [ final1 ] ) , expressed in the variable @xmath56 , is of course independent of this choice .
the spreading of a brownian particle is given by the fokker - planck equation @xmath87 the propagator @xmath88 times @xmath89 gives the probability to find the brownian particle inside the interval @xmath51 at time @xmath9 , knowing that the particle was at @xmath90 at time @xmath50 . with
absorbing boundary conditions at the origin we have , using the method of images @xmath91\ .\ ] ] this propagator is not a probability distribution because it is not normalized .
its normalization , the so - called survival probability , @xmath92 gives the probability that the particle is not yet absorbed by the boundary at @xmath30 .
the survival probability vanishes when @xmath93 ; however , in that limit , the probability distribution for the non - absorbed particles remains well - defined : @xmath94 another quantity with a finite limit for @xmath95 is @xmath96 this allows to write the probability @xmath97 as @xmath98 using in eq .
( [ f10 ] ) the scaling variable defined in ( [ rescaled ] ) , @xmath99 , we recover ( [ bmr+ ] ) .
( [ f10 ] ) is simpler than eq .
( [ 4 ] ) because the @xmath90 dependence is discarded from the beginning
. we will use this definition to compute @xmath100 for @xmath13 .
the process @xmath0 is gaussian for all values of @xmath2 , but it is markovian only for @xmath4 . for all other values of @xmath2 ,
the process is non - markovian and this makes the problem difficult to solve .
our idea is to expand around @xmath4 . in a first step ,
we construct an action , which calculates expectation values of the gaussian process @xmath0 , with bulk expectation values ( [ autocorr1 ] ) . in a second step ,
we obtain the propagator with absorbing boundary conditions at @xmath30 . in a third step
we calculate the probability @xmath7 perturbatively , using the action constructed in step 1 . in the fourth step
, we put together all pieces and interpret our result . for all @xmath2
, @xmath0 is a gaussian process , therefore the statistical weight of a path @xmath103 without any boundary is proportional to @xmath104)$ ] where the action @xmath105 $ ] is quadratic in @xmath8 and given by @xmath106=\int_0^t d t_1 \int_0^t d t_2 \ , \frac12 x(t_1 ) g(t_1,t_2 ) x(t_2).\ ] ] note that we use standard field - theoretic notation , noting @xmath107 a function of the variable @xmath8 , and @xmath105 $ ] a functional , depending on the function @xmath103 , with @xmath108
. the kernel @xmath109 of the action is related to the auto - correlation function of the process via @xmath110 for @xmath4 , the action is simple . in this case , setting @xmath86 , @xmath111^{-1}(t_1,t_2)=\langle x(t_1 ) x(t_2 ) \rangle = 2 \min(t_1 , t_2).\ ] ] using the result ( [ g0 ] ) in eq .
( [ action0 ] ) , we recover the standard brownian action @xmath112=
\frac14 \int_0^t d t ' \left ( \partial_{t ' } x \right)^2 .
\label{s0}\ ] ] for a generic value of @xmath2 the kernel @xmath109 becomes non - local . for @xmath13
one can write @xmath113={\cal s}^{(0)}[x]+\epsilon\ , { \cal s}^{(1)}[x ] + \dots\ ] ] where @xmath114 $ ] is the action ( [ s0 ] ) and @xmath115 $ ] has been computed in appendix [ a : action ] @xmath116&= & - \frac12 \int_0^t d t_1 \int_{t_{1}}^t d t_2 \ , \frac{\partial_{t_1 }
x(t_1 ) \partial_{t_2 } x(t_2)}{|t_1-t_2| } { \nonumber}\\ & & - 2 { \cal s}^{(0)}[x ] ( 1+{\ln}\tau)\ .\end{aligned}\ ] ] note that we have introduced a regularization for coinciding times @xmath117 where @xmath118 is the uv cutoff .
a first - principle definition would necessitate a discretization in time .
it is however sufficient to check that the law ( [ autocorr1 ] ) is correctly reproduced , and that the final result is cutoff independent . for a generic value of @xmath2
, the propagator @xmath120 , denoting the probability that the particle reaches @xmath8 at time @xmath9 , starting from @xmath90 at time @xmath10 , and staying positive over the interval @xmath24 $ ] , can be written using standard path integral notation as @xmath121 \,{e}^{-{\cal s}[x ] } \ , \theta[x]\ .\ ] ] here @xmath122 $ ] is an indicator function that is @xmath102 if the path @xmath103 stays positive over the interval @xmath123 $ ] and @xmath10 otherwise .
the action @xmath105 $ ] is given in ( [ action0 ] ) .
in the limit @xmath93 , we expect , as in the brownian case ( @xmath4 ) , the propagator to vanish as @xmath124 where the yet unknown exponent @xmath125 depends on @xmath2 .
note that for @xmath4 , @xmath126 ( see eq .
( [ 6 ] ) ) . for @xmath13
, we expect that @xmath127 , where @xmath128 is yet unknown .
analogous to eq .
( [ 6 ] ) for @xmath4 we define @xmath129 as @xmath130 \,{e}^{-{\cal s}[x]}\ , \theta[x]\ .\ ] ] using the expansion of the action given in eq .
( [ expans1 ] ) and @xmath131 , we write to leading order in @xmath14 @xmath132 \left(1-\epsilon { \cal s}^{(1)}[x]\right ) { e}^{-{\cal s}^{(0)}[x ] } \ , \theta [ x ] \nonumber \\ & = & \lim_{x_0 \to 0 } \big\ { z_+^{(0)}(x , t)\left[1 -a_1 \epsilon { \ln}(x_0 ) \right ] { \nonumber}\\ & & \ \qquad - \frac \epsilon { x_0 } \int_{x(0)=x_0}^{x(t)=x } { \cal{d}}[x]\ , { \cal s}^{(1)}[x ] \ , { e}^{-{\cal s}^{(0)}[x ] } \ , \theta [ x ] \big\ } \nonumber \\ & = & z_+^{(0)}(x , t ) + \epsilon z_+^{(1)}(x , t)\ , \end{aligned}\ ] ] where @xmath133 is defined in eq .
( [ 6 ] ) and @xmath134 is @xmath135 \ , { \cal s}^{(1)}[x ] \ , { e}^{-{\cal s}^{(0)}[x ] } \,\theta [ x ] { \nonumber}\\ & & -a_1 { \ln}(x_0 ) z_+^{(0)}(x , t ) \big\}\end{aligned}\ ] ] we will see that for @xmath136 , i. e. @xmath137 , @xmath134 is independent of @xmath90 .
the main achievement of this paper is the calculation of @xmath134 defined in eq .
( [ z1def ] ) .
this calculation is rather involved , both conceptually and technically .
therefore , we will relegate several technical calculations to appendix [ s : evaluation of z+ ] .
( [ z1def ] ) can be devided into three pieces : @xmath140 { \nonumber}\\ \label{eqza } z_+^a(x , t)&=&2(1+{\ln}\tau ) \\ & & \times \lim_{x_0 \to 0}\frac{1}{x_0}\int_{x(0)=x_0}^{x(t)=x } { \cal{d}}[x ] { \cal s}^{(0)}[x]\ , { e}^{-{\cal s}^{(0)}[x ] } \,\theta [ x ] { \nonumber}\\ \label{eqzb } z_+^b(x_0,x , t)&= & \frac{1}{4}\int_0^t\int_0^t { \mathrm{d}}t_1 { \mathrm{d}}t_2\\ & & \times \frac{1}{x_0 } \int_{x(0)=x_0}^{x(t)=x } { \cal{d}}[x ] \frac{\dot{x(t_1 ) } \dot{x(t_2 ) } } { |t_1-t_2|}\ , { e}^{-{\cal s}^{(0)}[x ] } \,\theta [ x ] { \nonumber}\end{aligned}\ ] ] the first term , @xmath141 is simple , and is evaluated in appendix [ s : evaluation of z+a ] .
we now come to the evaluation of the contribution @xmath142 , defined in eq .
( [ eqzb ] ) . in fig .
[ f : z+ ] we show a path which contributes to @xmath143 .
the sum of all these paths is a product of transition probabilities .
explicitly , it reads , ordering @xmath144 , which gives an extra factor of 2 compared to ( [ eqzb ] ) : @xmath145\ .\end{aligned}\ ] ] ( 0,0 ) ( 5.5,0 ) ; ( 0,0 ) ( 0,3 ) ; ( 1.5,0 ) ( 1.5,3 ) ; ( 1.8,0 ) ( 1.8,3 ) ; ( 3.5,0 )
( 3.5,3 ) ; ( 3.8,0 )
( 3.8,3 ) ; ( 5,0 )
( 5,3 ) ; ( 0,.2 ) circle ( 1.5pt ) ; ( x ) at ( -.25,.2 ) @xmath146 ; ( x ) at ( 0,-.3 ) @xmath10 ; ( 0,.2 ) ( 1.5,2 . ) ;
( 1.5,2 . )
circle ( 1.5pt ) ; ( tx1 ) at ( 1.5-.2,2.05 ) @xmath147 ; ( tt1 ) at ( 1.5,-.25 ) @xmath148 ; ( 1.8,2.2 ) ( 1.5,2 . ) ; ( 1.8,2.2 ) circle ( 1.5pt ) ; ( tx ) at ( 1.8+.25,2.28 ) @xmath149 ; ( t1 ) at ( 1.85,-.28 ) @xmath150 ; ( 3.5,2 . ) circle ( 1.5pt ) ; ( x2 ) at ( 3.5-.2,2.15 ) @xmath151 ; ( t2 ) at ( 3.5,-.28 ) @xmath152 ; ( 3.8,2.2 ) ( 3.5,2 . ) ; ( 3.8,2.2 ) circle ( 1.5pt ) ; ( tx2 ) at ( 3.8+.25,2.3 ) @xmath153 ; ( tt2 ) at ( 3.85,-.25 ) @xmath154 ; ( 1.8,2.2 )
( 3.5,2 . ) ; ( 5,1.75 ) circle ( 1.5pt ) ; ( tf ) at ( 5,-.3 ) @xmath9 ; ( delta ) at ( 5.2,1.75 ) @xmath8 ; ( 5,1.75 )
( 3.8,2.2 ) ; ( t ) at ( 5.9,0 ) time ; ( space ) at ( 0,3.2 ) space ; @xmath155 , @xmath156 and @xmath157 are defined in ( [ z+0 ] ) . the factors @xmath158 and @xmath159 take into account the terms @xmath160 and @xmath161 in the action @xmath162 $ ] .
@xmath163 finally , we have set @xmath164 , and @xmath165 , since we have taken the limit of their differences to 0 . in order to perform the six integrations in eq .
( [ weight ] ) it turns out to be convenient to evaluate its laplace transform , @xmath166 . from now on , we will always denote with @xmath167 the laplace transform of a function @xmath168 , defined as @xmath169 this laplace transform leads to two important simplifications :
the first simplification is that now the nested time - integrals over @xmath170 and @xmath152 become a product . to see this
, we remind that if @xmath171 and @xmath172 are two functions which depend on @xmath9 , then the laplace transform of their convolution is simply the product of their laplace transforms , @xmath173 { \nonumber}\\ & & = \int_{0}^{\infty}{\mathrm{d}}t\ , \int_{0}^{\infty } { \mathrm{d}}t_{1}\ , \int_{0}^{\infty } { \mathrm{d}}t_{2}\ , \delta ( t - t_{1}-t_{2 } ) { \nonumber}\\ & & \qquad \times f_{1 } ( t_{1}){e}^{-t_{1 } s } f_{2 } ( t_{2}){e}^{-t_{2}s } { \nonumber}\\ & & = \left [ \int_{0}^{\infty } { \mathrm{d}}t_{1 } f_{1 } ( t_{1}){e}^{-t_{1 } s } \right]\left [ \int_{0}^{\infty } { \mathrm{d}}t_{2 } f_2 ( t_{2}){e}^{-t_{2 } s } \right ] { \nonumber}\\ & & = \tilde f_{1 } ( s ) \tilde f_{2 } ( s)\ .
\label{conv - theorem}\end{aligned}\ ] ] this consideration generalizes to 3 and more times .
we obtain for the laplace transform of ( [ weight ] ) @xmath174 \qquad \end{aligned}\ ] ] the second simplification is even more important , and is most easily understood on the example of the bulk propagator @xmath175 its laplace - transform is @xmath176 while integrals over @xmath177 involving ( [ b11 ] ) give error - functions , which are hard to integrate further , the same integrals over ( [ b12 ] ) remain similar exponential functions ; the only complication is that one has to distinguish between @xmath8 smaller or larger than @xmath56 . to evaluate ( [ zblp1 ] ) , we now have to calculate the laplace - transforms of its factors : @xmath178 finally , the term in brackets in eq .
( [ zblp1 ] ) can be rewritten , using a fourier decomposition for @xmath179 , as @xmath180 \nonumber \\ & = & -\int\limits_{-\infty}^{\infty } \frac{{\mathrm{d}}k}{2 \pi } \left
[ e^{i k ( x_1-x_2)}-e^{i k ( x_1+x_2)}\right]\left [ { \ln}([s{+}k^2]\tau ) + \gamma_{\mathrm e } \right ] { \nonumber}\\ \label{48}\end{aligned}\ ] ] note that the time integral in the second line of eq .
( [ 48 ] ) is diverging at small times . since the path integral is defined as discretized in time , a natural approach consist in discretizing this integral , with a step - size @xmath181
. this would indeed be the only possible approach for stronger divergences , like @xmath182 .
however , since our integral is only logarithmically diverging , we can take an easier path , by using a small - time cutoff @xmath181 : @xmath183 \tau ) -\gamma_{\mathrm e } + o(\tau ) \ .{\nonumber}\end{aligned}\ ] ] we note that the regularization by discretization gives the same result apart from the term @xmath184 .
we will check later that it only contributes to the normalization , which will drop from the final result . collecting the results of eqs .
( [ 46 ] ) and ( [ 48 ] ) in eq .
( [ zblp1 ] ) , and doing the remaining space - derivatives we find @xmath185 { \nonumber}\\ & & \times \int\limits_{x_{1}>0}\int\limits_{x_{2}>0 } \left [ e^{i k ( x_1-x_2)}+e^{i k ( x_1+x_2)}\right ] \nonumber \\ & & \times \frac{e^{-\sqrt{s } | x - x_2 |}-e^{-\sqrt{s } ( x_2+x)}}{2 \sqrt{s } } { \nonumber}\\ & & \times \frac{e^{-\sqrt{s } | x_0-x_1 |}-e^{-\sqrt{s } ( x_1+x_0)}}{2 \sqrt{s } } \label{zblp2}\end{aligned}\ ] ] performing the space - integrations , we find @xmath186 \left [ \cos ( k x)- e^{-\sqrt{s } x}\right ] { \nonumber}\\ & & \qquad\qquad \times \frac { k^2 [ { \ln}(\tau(s+k^2))+\gamma_{\mathrm e } ] } { ( s+k^2)^2 } \qquad\end{aligned}\ ] ] note that this is ( rescaling @xmath187 ) @xmath188 { \nonumber}\\ & & \qquad \quad \times \big[\!\cos ( k x_0 \sqrt{s } ) - e^{-x_0 \sqrt{s } } \big ] { \nonumber}\\ & & \qquad \quad \times\frac { k^2 \left [ { \ln}\big(\tau s ( 1+k^{2 } ) \big)+\gamma_{\mathrm e } \right]}{\sqrt{s } ( 1 + k^2)^2 } \qquad \end{aligned}\ ] ] the next step is to invert this laplace transform which is performed in appendix [ ab2 ] .
the final result for @xmath139 is given in eqs .
( [ z1 ] ) and ( [ a+b1+b2 ] ) of appendix [ sub : sum ] , expressed in terms of the scaling variable @xmath190 . note that setting @xmath191 , i. e. @xmath137 , the term @xmath139 does not depend on @xmath90 : @xmath192 - 2 + { \cal i } ( z ) + c ( t ) { \nonumber}\\ c ( t)&= & { \ln}(t)-2 \gamma_{\mathrm{e } } + 2\ \label{z1}\end{aligned}\ ] ] where @xmath193 is defined in eq .
( [ iz ] ) . the first line is arranged as to not contribute to the normalization , whereas @xmath194 is independent of @xmath195 and will not appear in the final conditional probability .
@xmath196 is euler s constant .
the probability distribution , @xmath7 , to find a non - yet - absorbed particle in the interval @xmath197 can be computed following the lines of eq .
( [ f10 ] ) to order @xmath14 as @xmath198 \label{cp0}\end{aligned}\ ] ] note that the term proportional to @xmath199 cancels in normalized objects such as @xmath7 .
therefore , we obtain @xmath200 \big\ } \ , { \nonumber}\end{aligned}\ ] ] where @xmath201 , and @xmath193 is given in eq .
( [ iz ] ) .
the result in eq .
( [ cp1 ] ) still involves both @xmath195 and @xmath9 .
the reason is that for @xmath101 the natural scaling variable is @xmath202 instead of @xmath190 , as can be seen from eq .
( [ rescaled ] ) . to rewrite eq .
( [ cp1 ] ) in terms of @xmath203 , we note that @xmath204 \ln t\right\ } { \nonumber}\\ & = & r_{+}^{(0 ) } ( y)\left\{1- \epsilon \left[y^{2}-2 \right]\ln t \right\}\ .\end{aligned}\ ] ] this gives for eq .
( [ cp1 ] ) up to terms of order @xmath205 @xmath206 \big\ } \ , { \nonumber}\end{aligned}\ ] ] this is the final result announced in equation ( [ final1 ] ) , with @xmath207 calculeted in ( [ iz ] ) and below .
in this section , we compare our analytical results with numerical simulations . more specifically , we consider the super - diffusive process with @xmath208 . .
analytical result for brownian motion @xmath209(solid blue line ) , and simulation data for @xmath210 , and @xmath211 ( red dots ) , as well as for the fbm with @xmath208 ( black diamonds ) .
histograms are performed over @xmath212 paths .
, title="fig:",width=325 ] we aim to sample a fbm processes @xmath0 at discrete times @xmath213 .
the covariance matrix of @xmath214 coincides with the autocorrelation function of the original fbm process in eq .
( [ autocorr1 ] ) , setting @xmath86 , @xmath215 the @xmath216 covariance matrix @xmath217 is symmetric and has positive eigenvalues ; it is thus possible to find a matrix @xmath218 , positive and symmetric , such that @xmath219 .
matrix @xmath218 is called the square root of @xmath217 .
one can simulate paths of a fbm using the standard procedure for gaussian correlated processes : ( i ) determine @xmath218 , the square root of @xmath217 .
( ii ) each path @xmath220 is given by the matrix multiplication @xmath221 .
the vector @xmath222 is a set of @xmath223 independent gaussian numbers with unitary variance and zero mean .
it is easy to check that these paths are characterized by the correct covariance matrix ( [ num1 ] ) .
unfortunately this procedure is time consuming , as for step ( i ) it requires the full diagonalization of c. better results are obtained by making use of the stationarity of the increments @xmath224 ( we set @xmath225 ) . using eq .
( [ num1 ] ) we can compute @xmath226 , the covariance matrix of the increments , @xmath227 , and @xmath228 .
the matrix @xmath226 is symmetric and positive definite like the matrix @xmath217 , but it also is a toeplitz matrix . for toeplitz matrices efficient numerical methods allow to avoid the full diagonalization of @xmath226 . in particular , the levinson algorithm ( for a practical implementation of levinson s algorithm see @xcite and @xcite ) is suitable for first passage problems , as it recursively generates the increment @xmath229 given @xmath230 .
the points of the fbm path are given by @xmath231 .
in our simulation we are interested only in positive paths ( @xmath232 for all @xmath233 ) .
the levinson method allows to discard negative paths whenever a @xmath234 is generated , without building the full path . .
analytical result for brownian motion @xmath209(solid blue line ) , and simulation data for @xmath210 , and @xmath211 ( red dots ) , as well as for the fbm with @xmath208 ( black diamonds ) .
histograms are performed over @xmath235 paths .
, title="fig:",width=325 ] for each positive path we record the final position @xmath236 .
the histogram of the rescaled variable @xmath237 is the scaling function @xmath74 .
the results for @xmath22 and the markovian case @xmath4 are presented on figs .
[ histlin ] and [ histlog ] . for small @xmath56 the scaling function
, @xmath74 behaves as a power - law , with an exponent @xmath77 .
for @xmath4 we expect @xmath238 , for @xmath22 we expect @xmath239 .
inspired by our perturbative calculation we predict that for @xmath18 , @xmath74 behaves like @xmath240 . in order to facilitate the comparison
, we define the scaling function @xmath241 the numerical data for the scaling funtion @xmath242 defined in eq .
( [ r+def ] ) are shown on fig .
[ f : h=2/3 ] for @xmath22 .
they clearly show two distinctive power - law behaviors : for small @xmath56 this power law is @xmath243 with @xmath244 , predicted by the scaling relation @xmath245 . for large @xmath56 a larger exponent @xmath246
is measured .
this is consistent with the perturbative calculation , which suggests @xmath247 for @xmath248 and @xmath249 for @xmath45 . ,
defined in eq .
( [ r+def ] ) for the fbm with @xmath208 ( black diamonds ) , using @xmath210 and @xmath235 paths .
the asymptotic small-@xmath56 behavior is consistent with @xmath250 .
the large-@xmath56 asymptotics ( including the amplitude ) was taken from ( [ r - theory ] ) , with slope @xmath251 .
, width=325 ] a more accurate comparison between the numerical data and the perturbation theory is possible .
our perturbative result given in eq .
( [ final1 ] ) is equivalent to @xmath252 $ ] . in order to compare to numerics , we use @xmath253 while the two expressions are equivalent to order @xmath14 , the latter ( [ r - theory ] ) has the merit to resum the logarithms for small and large @xmath56 into the power - law behavior @xmath254 where the exponents are the order-@xmath14 results @xmath255 for @xmath22 , i.e. @xmath256
, we predict a scaling @xmath257 , @xmath258 , using ( [ final3 + ] ) .
note that the curve drawn is exatly the asymptotic behavior of our analytical result ( [ r - theory ] ) , using ( [ w ] ) , thus also the _ amplitude _ and not only the exponent are estimated .
this can more clearly be seen on fig .
[ f : h=2/3-theory ] , where the solid ( blue ) line represents the theoretical order-@xmath14 prediction , and the dashed line the asymptotic behaviors given in eq .
( [ 44 ] ) .
conversely , relation ( [ r - theory ] ) can be used to extract @xmath259 from @xmath242 , @xmath260 this relation should work the better , the smaller @xmath14 is .
using our numerical results for @xmath261 , we obtain the curve presented on fig .
[ f : w2 ] .
the agreement is quite good for @xmath262 .
it breaks down for larger @xmath56 due to numerial problems . for @xmath263
, the deviations can be attributed to the large value of @xmath14 . ,
defined in eq .
( [ r - theory ] ) for @xmath208 , i.e. @xmath264 .
the dashed lines are the predicted asymptotic behaviors , @xmath243 ( for small @xmath56 ) and @xmath265 ( for large @xmath56 ) . superimposed
are the simulation data shown on fig .
[ f : h=2/3 ] .
note that there is no fitting parameter .
the deviation at small @xmath56 is due to the fact that @xmath266 is rather large , so the order-@xmath14 slope @xmath267 is smaller than the exact result @xmath268 , which evaluates to @xmath269 . for large @xmath56 , but smaller than @xmath138 , the effective cutoff of the simulation , both amplitude and slope are correctly predicted.,width=325 ] , defined in eq .
( [ w ] ) .
black diamonds : estimation of @xmath270 from the numerical data for @xmath208 , using relation ( [ w - extract ] ) .
the agreement is quite good for @xmath271 .
it breaks down for @xmath272 due to numerial problems . for @xmath263
, the deviations can be attributed to the large value of @xmath14.,width=325 ]
in this article , we develop a systematic scheme to calculate the corrections to the universal scaling function @xmath74 for fractional brownian motion , in an @xmath273 expansion .
we compute the full scaling function @xmath74 to first order in @xmath14 .
in particular we find that @xmath12 behaves as @xmath15 as @xmath16 ( near the absorbing boundary ) , while @xmath17 as @xmath18 ( far from the boundary ) , with , at the first order in @xmath14 , @xmath19 and @xmath20 . for small @xmath14
our results confirm the scaling relation found in ref .
@xcite : @xmath15 with @xmath274 . for fractional brownian motion
it is known that @xmath81 , so that @xmath275 .
far from the boundary , i.e. for large @xmath56 , the leading behavior @xmath276 recovers the gaussian propagator in absence of boundaries ; our approach shows that @xmath12 has a subleading power law prefactor @xmath277 , where @xmath78 is a new ( independent ) exponent .
our numerical simulations show that the predictions of the asymptotic behavior of @xmath12 hold at @xmath22 .
in particular the two exponents @xmath78 and @xmath77 have been measured and shown in fig .
[ f : h=2/3 ] .
let us stress that few results are known about non - markovian processes in presence of boundaries .
perturbation theory developed in this paper can provide substantial new insight here .
the method is versatile and can in principle be extended to the calculation of other quantities such as the propagator for a process confined to a finite interval with absorbing boundaries , or alternatively with other , e.g. reflecting boundary conditions . particularly interesting for applications would be the hitting probability @xmath278 , the probability that a generic stochastic process starting at @xmath8 and evolving in a box @xmath279 $ ] hits the upper boundary at @xmath223 before hitting the lower boundary at @xmath10 @xcite . in the context of polymer translocation ,
the hitting probability is the probability that a finite polymer chain will ultimately succeed in translocating through a pore . in the more general framework of anomalous diffusion
, the presence of boundaries has been especially studied for non - gaussian processes .
for instance , lvy flights are markovian superdiffusive processes whose increments obey a lvy stable ( symmetric ) law of index @xmath280 .
the hurst exponent is @xmath281 @xcite . by virtue of the sparre andersen theorem @xcite ,
the persistence exponent is @xmath282 , independent of @xmath283 .
the laplace transform of the scaling function @xmath12 has been computed in @xcite for a generic value of @xmath283 .
a scaling analysis of this laplace transform shows that @xmath12 behaves as @xmath284 as @xmath16 ( this in in agreement with the scaling relation @xmath274 ) , while far from the boundary the lvy - stable behavior is recovered .
an increasing interest is devoted to gaussian processes with self - affine anomalous displacements @xmath285 with @xmath3 @xcite .
our current results apply only to fractional brownian motion , i.e. self - affine gaussian processes defined by the autocorrelation function ( [ autocorr1 ] ) .
in particular for fbm it is known that ( i ) the process has stationary increments , ( ii ) @xmath81 , and ( iii ) @xmath274 . for all other gaussian processes with hurst exponent @xmath2 ,
( i ) the increments are non - stationary , ( ii ) @xmath286 and we particularly emphasize that , ( iii ) no scaling relation is known between @xmath77 and @xmath26 ( unlike in fbm where @xmath274 ) . among such processes it is possible to show that the one , defined by the autocorrelation function @xmath287 describes the subdiffusive behavoir of a tagged monomer in an elastic interface which initially was flat @xcite . for this process
the persistence exponent is known only to first order in @xmath14 @xcite , whereas neither the exponents @xmath77 , nor @xmath288 are known analytically .
it would be interesting to determine the full scaling function @xmath12 for this process within our perturbative framework .
it is a pleasure to thank pierre le doussal , mehran kardar , and andrea zoia for useful discussions .
this work is supported by anr grant 09-blan-0097 - 01/2 .
the aim of this appendix is to determine the action @xmath115 $ ] , the first correction to the brownian action , @xmath114 $ ] , in the expansion of @xmath105 $ ] in eq .
( [ expans1 ] ) . as a first step we expand the autocorrelation function ( [ autocorr1 ] ) around @xmath4 , setting @xmath86 , @xmath289^{-1}(t_1,t_2)+ \epsilon \ ,
k(t_1,t_2 ) + o(\epsilon^2 ) . { \nonumber}\end{aligned}\ ] ]
the first term is the autocorrelation function for @xmath4 , @xmath290^{-1}(t_1,t_2)= 2 \ , \text{min}(t_1 , t_2),\ ] ] the second term gives the correction at first order in @xmath14 , @xmath291.\ ] ] inverting eq .
( [ kdef0 ] ) and expanding up to order @xmath14 one gets @xmath292 where @xmath293 is defined as @xmath294^{-1}(t',t_2)= \delta(t_1-t_2).\ ] ] one can check that the kernel of the brownian action , @xmath114 $ ] , i. e. , @xmath295 satisfies eq .
( [ condition ] ) , namely , @xmath296^{-1}(t',t_2 ) { \nonumber}\\ & = -\int_0^\infty { \mathrm{d}}t'\ , \delta '' ( t_1-t ' ) \text{min}(t',t_2 ) \nonumber \\ & = -\partial_{t_{1}}^{2 } \min ( t_{1},t_{2 } ) = \delta ( t_{1}-t_{2})\ .\end{aligned}\ ] ] it remains to compute the term @xmath297 .
integrating by parts one has @xmath298 using that the first two terms in ( [ a3 ] ) do not contribute since they only depend on one of the times .
the derivative is @xmath299 the second term is not well - defined .
we decide to introduce a regularization for coinciding times @xmath300 where @xmath118 should be thought of as the time - discretization of the path - integral .
let us first give the final result , before commenting on this approximation : @xmath301 .
\end{aligned}\ ] ] this yields for the action @xmath302&=&\int_0^t d t_1 \int_0^t d t_2 \ , \frac12 x(t_1 ) g^{(1)}(t_1,t_2 ) x(t_2 ) { \nonumber}\\ & = & - \frac14 \int_0^t d t_1 \int_0^t d t_2 \ , \frac{\partial_{t_1 } x(t_1 ) \partial_{t_2 } x(t_2)}{|t_1-t_2| } { \nonumber}\\ & & - 2\ , { \cal s}^{(0)}[x ] ( 1+{\ln}\tau).\end{aligned}\ ] ] we see that the only possibly ambiguous term , the term of order @xmath303 , is proportional to the zeroth - order action @xmath304 $ ] , thus equivalent to a change in the diffusion constant @xmath61 . thus its effect is easy to check in the final result , when looking at observables in a domain unaffected by the boundary .
this term is easily evaluated . indeed , eq .
( [ eqza ] ) can be recast in the following form @xmath305 { e}^{- a { \cal s}^{(0)}[x]}\ , \theta [ x ] \nonumber \\ & = & -2(1+{\ln}\tau ) { \nonumber}\\ & & \times \lim_{x_0 \to 0}\frac{1}{x_0 } \left .
\frac{\partial}{\partial a } \right|_{a=1 } \sqrt{\frac{a}{4 \pi t } } \left[e^{-\frac{a}{4 t}(x - x_0)^2}-e^{-\frac{a}{4t}(x+x_0)^2 } \right ] \nonumber \\ & = & ( 1+{\ln}\tau)\frac{x}{\sqrt{4 \pi t^3 } } e^{-\frac{x^2}{4 t}}\left(\frac{x^2}{2 t}-3\right).\end{aligned}\ ] ] in going from the first to the second line we have used the expression of the propagator in the brownian case in eq .
( [ z+0 ] ) , introducing the factor of @xmath306 from the observation that the latter appears together with @xmath307 , and readjusting the normalization . in terms of the variable @xmath190
this gives @xmath308 where @xmath309 is defined in ( [ 6 ] ) and @xmath310 we split @xmath166 into two parts @xmath312{\nonumber}\\ & & \qquad \qquad\times \left[\cos ( k x_0 \sqrt{s } ) - e^{-x_0 \sqrt{s } } \right]{\nonumber}\\ & & \qquad \qquad \times\frac { k^2 { \ln}(1+k^{2 } ) } { \sqrt{s } ( 1 + k^2)^2 } \label{km12b-2 } \\
\tilde i_2(x_0,x , s ) & = & \frac4{x_0 } \int\limits_{-\infty}^{\infty}\frac{{\mathrm{d}}k}{2\pi } \left[\cos ( k x\sqrt{s } ) - e^{-x \sqrt{s } } \right]{\nonumber}\\ & & \qquad \qquad\times \left[\cos ( k x_0 \sqrt{s } ) - e^{-x_0 \sqrt{s } } \right]{\nonumber}\\ & & \qquad \qquad \times\frac { k^2 \left [ { \ln}(\tau s ) + \gamma_{\mathrm e } \right]}{\sqrt{s } ( 1 + k^2)^2 } \label{km12c-2}\end{aligned}\ ] ] the expansion of this term for small @xmath90 must be done with care ; when @xmath90 acts as a regulator , one can not simply expand in it .
we claim , and show below that @xmath314 with @xmath315{\nonumber}\\ & & \qquad \qquad \qquad \times \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } \\ \tilde i_{1}^{b}(x ,
s ) & = & 4\int\limits_{-\infty}^{\infty}\frac{{\mathrm{d}}k}{2\pi } \left [ \cos ( k x\sqrt{s } ) -e^{-x \sqrt{s } } \right ] { \nonumber}\\ & & \qquad \qquad \quad \times \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } \end{aligned}\ ] ] in order to prove this , we group the four terms in ( [ km12b-2 ] ) into two times two terms ; the first combination is @xmath316\frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } { \nonumber}\\ & & = \left[-\frac{4}{x_0 } + 4 \sqrt s + o(x_0)\right ] { \nonumber}\\ & & \qquad \times \int\limits_{-\infty}^{\infty}\frac{{\mathrm{d}}k}{2\pi } \left[\cos ( k x\sqrt{s } ) - e^{-x \sqrt{s } } \right ] \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } { \nonumber}\\ & & = \tilde i_1^{\mathrm{div}}(x_0,x , s)+ \tilde i_1^b(x , s ) + o(x_0)\end{aligned}\ ] ] where the divergent contribution is @xmath317 { \nonumber}\\ & & \qquad \qquad \times \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } \ .\end{aligned}\ ] ] this expansion in @xmath90 is justified since @xmath318 stands outside the integrand , thus does not act as a regulator .
the second contribution to ( [ km12b-2 ] ) is @xmath319 \cos(k x_0\sqrt s)\frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } { \nonumber}\\ & & = \frac{2}{x_0 } \int\limits_{-\infty}^{\infty}\frac{{\mathrm{d}}k}{2\pi } \big[\cos ( k ( x+x_0)\sqrt{s } ) + \cos ( k ( x - x_0)\sqrt{s } ) { \nonumber}\\ & & \qquad\qquad\qquad - 2 e^{-x \sqrt{s } } \cos(k x_0\sqrt s ) \big ] \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } \end{aligned}\ ] ] since @xmath320 , we can taylor - expand @xmath321 and @xmath322 , leading to @xmath323 { \nonumber}\\ & & \qquad \qquad\quad \times \frac { k^2 { \ln}(1+k^{2 } ) } { ( 1 + k^2)^2 } + o(x_0){\nonumber}\\ & & = - \tilde i_1^{\mathrm{div}}(x_0,x , s ) + \tilde i_1^{a}(x_0,x , s)+o(x_0 ) \label{b13}\end{aligned}\ ] ] the contributions proportional to @xmath324 cancel between ( [ b10 ] ) and ( [ b13 ] ) , and we arrive at the decomposition ( [ b7 ] ) .
we now treat the two contributions to ( [ b7 ] ) .
the first contribution @xmath325 can be evaluated analytically .
after integration over @xmath311 we find a bessel function , which can be expanded in @xmath326 as @xmath327+o ( x_0 ) { \nonumber}\end{aligned}\ ] ] the second contribution @xmath328 can be evaluated using the relation @xmath329
\frac{1}{(1+k^2)^{u+1}}.\ ] ] we rewrite @xmath330 as @xmath331 \int\limits_{-\infty}^{\infty } \frac{{\mathrm{d}}k}{2\pi } \frac{e^{i k x \sqrt{s } } - e^{-x \sqrt{s } } } { ( 1+k^2)^{u+1 } } .\ ] ] it can be split into two parts , @xmath332 \int\limits_{-\infty}^{\infty } \frac{{\mathrm{d}}k}{2\pi } \frac{1}{(1+k^2)^{u+1 } } { \nonumber}\\ & ~~= - \left[1+{\ln}(4 ) \right ] e^{-x \sqrt{s } } \\ & \tilde{i}_{1b}(x , s ) { \nonumber}\\ & ~~= \left[\frac{{\mathrm{d}}}{{\mathrm{d}}u}\bigg|_{u=1}- \frac{{\mathrm{d}}}{{\mathrm{d}}u}\bigg|_{u=0 } \right ] 4 \int\limits_{-\infty}^{\infty } \frac{{\mathrm{d}}k}{2\pi } e^{i k x \sqrt{s } } ( 1+k^2)^{-(u+1)}\label{b11}\end{aligned}\ ] ] to do the @xmath311-integral in @xmath333 , it is useful to introduce the integral representation @xmath334 this gives @xmath335 \bigg [ \frac{1}{\gamma(1+u ) } { \nonumber}\\ & & \times \int_{-\infty}^{\infty}\frac{{\mathrm{d}}k}{2\pi } e^{i k x\sqrt{s } } \int_0^\infty { \mathrm{d}}z\ , z^u e^{-(1+k^2)z } \bigg]\ , \ \
\qquad \end{aligned}\ ] ] and performing the gaussian integral over @xmath311 yields @xmath336 \bigg[\frac{1}{\gamma(1+u ) } { \nonumber}\\ & & \qquad \times \int_0^\infty \frac{{\mathrm{d}}z}{2 \sqrt{\pi } } z^{u-1/2 } e^{-\frac{s x^2}{4 z } -z } \bigg]\ .\quad \end{aligned}\ ] ] @xmath338 can be calculated using residue calculus .
we use @xmath339 to expand the expression , choosing every pole in the half - plane in which the corresponding exponential factor converges .
the result is @xmath340\qquad \
\\end{aligned}\ ] ] expanding for small @xmath90 yields @xmath341+o\left(x_0 \right)\ ] ] it is useful to re - organize @xmath343 as the sum of three contributions : @xmath344 the first term depends on @xmath90 , @xmath345\ , \ ] ] while the other two terms are @xmath346 \bigg[\frac{1}{\gamma(1+u ) } { \nonumber}\\ & & \qquad \times \int_0^\infty \frac{{\mathrm{d}}z}{2 \sqrt{\pi } } z^{u-1/2 } e^{-\frac{s x^2}{4 z } -z } \bigg ] \\
\tilde j_2(x , s)&= & - x \sqrt{s } e^{-x \sqrt{s } } \left [ \gamma_{\mathrm{e } } + { \ln } ( \tau s ) \right]\qquad \qquad \end{aligned}\ ] ] the inversion of @xmath349 is done by observing that @xmath350 this yields @xmath351 where @xmath352 , and @xmath353 the inverse laplace transform of the second term can be done directly , @xmath354 { \nonumber}\\ & & = 2 \left[\frac{{\mathrm{d}}}{{\mathrm{d}}u}\bigg|_{u=1}- \frac{{\mathrm{d}}}{{\mathrm{d}}u}\bigg|_{u=0 } \right ] \bigg [ \frac{1}{\gamma(1+u ) } { \nonumber}\\ & & \qquad \times \int_0^\infty \frac{{\mathrm{d}}z}{\sqrt{\pi } } z^{u-1/2 } e^{-z } \delta\left(\frac{x^2}{4 z } -t\right ) \bigg]\ .\qquad \end{aligned}\ ] ] we observe that @xmath355 and obtain @xmath356 \frac{\big(\frac{x^2}{4t}\big)^{u+1/2}}{\gamma(1+u)}\ .\ ] ] finally @xmath357\ .\qquad ~~\end{aligned}\ ] ] introducing the variable @xmath190 we have : @xmath358 where @xmath359 , see eq .
( [ 6 ] ) , and @xmath360@xmath361 this completes the laplace inversion of @xmath362 .
the laplace inversion of the second term @xmath363 is more complicated , and we split it as @xmath364 x \sqrt{s } e^{-x \sqrt{s } } \nonumber \\ & = & x ( { \ln}\tau + \gamma_{\mathrm e } ) \frac{{\mathrm{d}}}{{\mathrm{d}}x } e^{-x \sqrt{s } } + x \frac{{\mathrm{d}}}{{\mathrm{d}}x } \left({\ln}s \ , e^{-x \sqrt{s } } \right ) \nonumber \\ & = & \tilde{j}_{2a } ( x , s)+\tilde{j}_{2b } ( x , s).\end{aligned}\ ] ] it is easy to perform the laplace inversion of the first term : @xmath365 the inverse laplace transform of the second term can be written as @xmath366 where @xmath367 the idea is to inverse - laplace transform @xmath368 and @xmath369 , and then to calculate @xmath370 as convolution of @xmath371 and @xmath372 , using ( [ conv - theorem ] ) .
these inverses are @xmath373 the convolution is @xmath374)+\gamma_{\mathrm e}}{\sqrt{t - t'}}\ .{\nonumber}\end{aligned}\ ] ] using ( [ uno ] ) we have @xmath375 e^{-\frac{x^2}{4 t ' } } \frac{{\ln}(4[t - t'])+\gamma_{\mathrm e}}{\sqrt{t - t ' } } \label{finalea}\ ] ] making a change of variables @xmath376 , and using @xmath377 , this gives @xmath378{\nonumber}\\ & & \qquad \times e^{-\frac{z^2}{2 u } } \left[{\ln}(4 t ) + \gamma_{\mathrm e } + { \ln}(1-u ) \right]\qquad \label{finaleaa}\end{aligned}\ ] ] the integral contains two pieces , which we note @xmath379 the first piece is @xmath380 the second integral @xmath381 is more difficult , but can be performed using mathematica .
a convenient substitution @xmath382 allows to write @xmath383where @xmath384 @xmath385 is the imaginary error - function , @xmath386 the hypergeometric function @xmath387 can be defined by its series expansion @xmath388 the error - function and the exponential function can be combined in another converging series , @xmath389 while problems of numerical precision appear for @xmath70 , we can use the asymptotic expansion @xmath390 at @xmath391 , the relative numerical agreement of ( [ iasymp ] ) and ( [ iz ] ) is about @xmath73 .
note that @xmath392 , thus @xmath393 does not contribute to the normalization .
the sum @xmath394 can be expressed using the variable @xmath190 as @xmath395 where @xmath396 and @xmath397 in summary , @xmath398\end{aligned}\ ] ] where @xmath309 is defined in eq.([6 ] ) .
the terms in question are given in eqs .
( [ az ] ) , ( [ b0 ] ) , ( [ b1 ] ) and ( [ b2 ] ) , and repeated here : @xmath399 their sum is @xmath400 - 2 \right\ } + { \cal i } ( z){\nonumber}\\ & & -(4+a_1){\ln}x_0+c ( t ) { \nonumber}\\ c ( t)&= & { \ln } ( t)+2 - 2 \gamma_{\mathrm{e}}\ .\qquad \end{aligned}\ ] ] the result is arranged such that the term in the curly brackets , when multiplied by @xmath401 , integrates to zero , as does @xmath402 . the propagator @xmath403 becomes independent of @xmath90 if @xmath137 , equivalent to @xmath404 .
as expected , @xmath405 , see eq .
( [ final3 ] ) .
since @xmath194 only contributes to the ( time - dependent ) normalization , it does not enter the scaling function @xmath12 .
on the other hand , the only contribution to the normalization of the propagator @xmath129 comes from @xmath194 .
since @xmath406 integrated over @xmath8 equals 1 , we conclude that the survival - probability is @xmath407{\nonumber}\\ & \sim & t^{-\theta}\ , \qquad \theta = \frac{1}{2}- \epsilon + o(\epsilon^2)\end{aligned}\ ] ] in agreement with @xmath408 .
this is a non - trivial check of our calculations .
consider a process @xmath409 , starting at @xmath410 , and arriving at @xmath8 at time @xmath9 , without having crossed zero , i.e. @xmath411 for all @xmath412 .
denote @xmath413 its arrival probability density at @xmath8 .
further denote @xmath414 the survival probability or the persistence up to time @xmath9 . at late times and fixed @xmath90 , for many processes , this survival probability decays algebraically @xmath415 where @xmath26 is the persistence exponent @xcite .
let us now assume that the process @xmath0 is self - affine .
this simply means that the process is characterized by a single growing length scale @xmath416 where @xmath2 is the hurst exponent of the process .
for example , ordinary brownian motion is a self - affne process with @xmath4 .
since the only length scale is @xmath417 , the survival probability @xmath418 is a function of only the scaled variable @xmath419 , i.e , @xmath420 . in order
that @xmath421 for large @xmath9 and fixed @xmath90 , the scaling function @xmath422 , for small @xmath56 , must behave as @xmath423 we next define @xmath424 as the conditional probability density of finding the walker , given that it has not been absorbed at any previous time : @xmath425 note that following eq .
( [ 4 ] ) , the probability distribution of a non - adsorbed particle is for @xmath426 @xmath427 we anticipate the following scaling form for @xmath428 @xmath429 in terms of the scale variables @xmath430 and @xmath419 we get from ( [ conditional ] ) and ( [ scaling_1 ] ) @xmath431 where @xmath432 is the conditional probability density ( [ conditional ] ) expressed in terms of the rescaled variables . in the long - time limit
, @xmath433 and @xmath434 can be factorized as @xmath435 let us now consider the limit @xmath436 and suppose that @xmath437 .
the process is time - reversible invariant , since its increments are stationary , i.e. , a path from @xmath90 to @xmath8 forward in time plays the same role as a path from @xmath8 to @xmath90 backward in time . as a consequence ,
@xmath438 is a symmetric function , @xmath439 .
factorization of probabilities for @xmath8 and @xmath146 to zero and symmetry thus implies @xmath440 and it follows the proposed scaling relation @xmath274 .
b. derrida , a.j .
bray , and c. godrche , j. phys .
a 27 , l357 ( 1994 ) ; a.j .
bray , b. derrida , and c. godrche , europhys .
27 , 175 ( 1994 ) ; b. derrida , v. hakim , and v. pasquier , phys .
75 , 751 ( 1995 ) .
s. redner , _ a guide to first - passage processes _ , cambridge university press , new york , ( 2001 ) .
majumdar , _ persistence in nonequilibrium systems _ , curr .
sci . * 77 * , 370 ( 1999 ) .
m. marcos - martin , d. beysens , j .-
bouchaud , c. godrche , and i. yekutieli , physica d * 214 * , 396 ( 1995 ) .
b. yurke , a.n .
pargellis , s.n .
majumdar , and c. sire , phys .
e * 56 * , r40 ( 1997 ) .
tam , r. zeitak , k.y .
szeto , and j. stavans , phys .
* 78 * , 1588 ( 1997 ) .
g. p. wong , r. w. mair , r. l. walsworth , and d. g. cory phys .
lett . * 86 * , 4156 ( 2001 ) .
d. b. dougherty et al . , phys .
lett . * 89 * , 136102 ( 2002 ) . j.mj .
leeuwen , v.w.a .
de villeneuve , h.n.w .
lekkerkerker j. stat .
mech . : theory and experiment , p09003 ( 2009 ) .
j. soriano , i. braslavsky , d. xu , o. krichevsky , and j. stavans , phys .
lett . * 103 * , 226101 ( 2009 ) .
majumdar , c. sire , a.j .
bray , and s.j .
cornell , phys .
lett . * 77 * , 2867 ( 1996 ) ; b. derrida , v. hakim , and r. zeitak , ibid . | fractional brownian motion is a gaussian process @xmath0 with zero mean and two - time correlations @xmath1 , where @xmath2 , with @xmath3 is called the hurst exponent . for @xmath4 , @xmath5 is a brownian motion , while for @xmath6 , @xmath0 is a non - markovian process . here
we study @xmath5 in presence of an absorbing boundary at the origin and focus on the probability density @xmath7 for the process to arrive at @xmath8 at time @xmath9 , starting near the origin at time @xmath10 , given that it has never crossed the origin .
it has a scaling form @xmath11 .
our objective is to compute the scaling function @xmath12 , which up to now was only known for the markov case @xmath4 .
we develop a systematic perturbation theory around this limit , setting @xmath13 , to calculate the scaling function @xmath12 to first order in @xmath14 .
we find that @xmath12 behaves as @xmath15 as @xmath16 ( near the absorbing boundary ) , while @xmath17 as @xmath18 , with @xmath19 and @xmath20 .
our @xmath14-expansion result confirms the scaling relation @xmath21 proposed in ref .
@xcite .
we verify our findings via numerical simulations for @xmath22 .
the tools developed here are versatile , powerful , and adaptable to different situations . |
for metallic transport to occur in arrays of spherical ncs that touch each other at transport limiting facets , the ncs themselves need to be metallic , i.e. the number of electrons @xmath21 in a nc is large .
hence , the electron gas can be described with the fermi wave vector : @xmath22 here @xmath23 is the density of electrons in a nc .
below , @xmath24 serves as a measure of the concentration @xmath19 . in ref .
@xcite we show that if @xmath25 the nc has random energy spectrum filled upto @xmath26 due to random potential of donors . in the opposite quantum - confined case , @xmath27 , for spherically symmetric ncs
, electrons occupy states with different radial and angular momentum quantum numbers @xmath28-shells , each of them being degenerate with respect of azimuthal quantum number @xmath29 . if the total number of electrons in the nc is @xmath30 , several @xmath28-shells are occupied .
still , when quantum numbers are large , bohr s correspondence principle allows us to consider the average density of states of electrons quasiclassically and introduce the fermi wave vector @xmath24 and the fermi energy @xmath31 .
this description is correct if the fermi energy @xmath31 is a good estimate for the energy of the top shell . at the critical concentration @xmath32 for cdse ncs with diameter @xmath33 ,
each nc has 13 electrons and the top shell is the half - filled 1d - shell .
the fermi energy @xmath31 at the concentration @xmath0 is 50% smaller than the shell energy @xmath34 .
hence , our degenerate gas description is accurate within 50% , which is a measure for the accuracy of our @xmath0 predictions . to derive the mit condition
, we consider the conductance of two metallic nc connected by a small facet contact .
when @xmath35 , the conductance of such a `` point '' contact was previously studied quasiclassically @xcite : @xmath36 where @xmath37 is the contact area .
this conductance can be easily understood with the help of the landauer formula @xcite .
the number of conducting channels in the contact area is @xmath38 and each of them additively contributes @xmath39 to @xmath40 .
it was proven that the mit occurs if the average conductance between two neighboring ncs @xmath40 in an array of ncs is equal to the minimal conductance @xmath41 @xcite : @xmath42 substituting @xmath40 from eq .
( [ eq : sharvin_2 ] ) into ( [ eq : minimal_g ] ) yields the general criterion for the mit @xmath43 which can easily be rewritten in terms of @xmath0 with help of eq .
( [ eq : fermi_wave ] ) to yield eq .
( [ eq : result ] ) .
the origin of eq .
( [ eq : mit_final ] ) is illustrated in fig .
[ fig : sheme ] : the condition @xmath44 describes electron wave packets with a size small enough to pass through the contact facet .
we now discuss the effect of the contact facet size , which for metal chalcogenide ncs can be large @xcite .
considering as an example an octahedron - shaped particle that is circumscribed by a sphere of diameter @xmath12 , the area of each facet is @xmath45 . equating this to @xmath46 , we find an effective @xmath47 at @xmath48 , we get @xmath49 , which is not far from @xmath50 for the spherical case . this @xmath51 results in a 3 times smaller @xmath0 than @xmath52 .
the nc contact through facets is , of course , the best case scenario that defines the lower bound for @xmath0 for spherical ncs . for ncs that do not touch through facets , a finite tunneling distance @xmath53 in the medium between ncs
should be taken into account . here
@xmath54 is the work function .
an electron can move between neighboring ncs only in a disc that we call @xmath55-contact , in which the distance between ncs is smaller than @xmath55 .
the radius of such a @xmath55-contact is @xmath56 ( see si1 [ sec : b_contact ] )
. the small ratio of the effective electron mass in the semiconductor nc to the free electron mass makes the @xmath55-contact transparent ( see similar effect in ref . ) .
usually , for ncs in vacuum ( air ) , @xmath57 which is much smaller than the lattice constant @xmath11 ; relying on only @xmath55-contacts increases @xmath0 upto 10 times .
so far we have considered ncs with bare surfaces .
if ncs are covered by a thin shell of ligands or oxide leading to a nc separation @xmath58 , the conductance @xmath40 acquires an additional factor @xmath59 and eqs .
( [ eq : minimal_g ] ) , ( [ eq : mit_final ] ) yield @xmath60 in this case , the mit may become unreachable .
we also can calculate the low temperature mobility @xmath61 in the vicinity of the mit . substituting the conductivity from eq .
( [ eq : sharvin_2 ] ) into the expression @xmath62 ( where the factor @xmath63 accounts for the difference between the concentration of electrons inside ncs and the average concentration in the film ) we find the low - temperature metallic mobility @xmath64 for cdse ncs with @xmath65 , @xmath61 is on the order of @xmath66 at @xmath67 and is close to the experimentally observed room - temperature mobility of @xmath68 for cdse @xcite .
note that this mobility mostly is due to the contact resistance , while the in the case of bulk semiconductors the low temperature mobility is due to the scattering by donors .
we now discuss the role of disorder .
the number of donors @xmath69 in a nc randomly fluctuates between ncs with a gaussian distribution .
if each nc were neutral ( @xmath70 ) , gaussian fluctuations of donor number , @xmath71 , would lead to substantial fluctuations @xmath72 of @xmath31 from one nc to another . to establish a unique chemical potential of electrons ( the fermi level ) , electrons move from ncs with larger than average @xmath19 to ones with smaller than average @xmath19 . accordingly , most ncs attain net charges @xmath73 .
this leads to large fluctuations of the coulomb potential and moves the nc electron energy levels with respect to the fermi level .
in an insulating nc array ( @xmath74 ) , this replaces the global charging energy gap of the density of states by the coulomb gap which leads to the efros - shklovskii variable range hopping @xcite ( see below ) .
this theory is based on the criterion ( [ eq : minimal_g ] ) , which guarantees that energy levels of ncs have a width comparable to the average energy difference between the adjacent levels @xmath75 .
this also eliminates the anderson localization . at the same time , at @xmath76 , there is at least one electron channel in the contact disk with almost perfect transparency .
this guarantees @xcite that the charging energy of every single nc , @xmath77 , is reduced to a value much smaller than @xmath75 ( @xmath78 is effective dielectric constant of the nc film ) . accordingly , the mott - hubbard localization is eliminated at the same time as the anderson localization .
we emphasize that criterion ( [ eq : minimal_g ] ) is universal and holds for heavily doped nc films regardless of whether ncs are quantum - confined or not .
a detailed discussion of this universality is presented in si2 [ sec : anderson ] .
another generic disorder effect is the variation of nc sizes @xcite .
remarkably , in heavily doped ncs , this variation does not lead to anderson localization of electrons , because their spectrum is already random .
thus , as for contacting metallic ncs , small variations of the diameter are inconsequential . on the other hand ,
the @xmath79 15% size dispersion in our experiments below may complicate the matching of nc facets and therefore increase @xmath0 .
to test the predictions of our theory , we studied the electron transport in films of heavily phosphorous ( p)-doped si ncs .
freestanding si ncs were synthesized in a nonthermal radio - frequency plasma reactor as reported previously @xcite .
we investigated six si nc films with different p concentrations .
we refer to them using their nominal doping concentration @xmath80 , the fractional flow rate , defined as @xmath81/([ph_{3}]+[sih_{4}])}\times 100\%$ ] , where [ ph@xmath82 and [ sih@xmath83 are the flow rates of phosphine and silane .
[ tab1 ] .*parameters of p - doped si ncs . *
@xmath80 is the nominal doping , @xmath84 is the atomic fraction of p in si ncs measured from icp - oes , @xmath12 is the average diameter of ncs , @xmath85 is the plasmonic peak in wavenumber for si nc films , @xmath19 is the electron concentration estimated from the plasmonic peak , @xmath86 is the localization length calculated from the electrical transport data . [ cols="^,^,^,^,^,^",options="header " , ] to quantify p incorporation in si ncs , we utilized inductively coupled plasma optical emission spectroscopy ( icp - oes ) . table i shows the p atomic fraction in si ncs for each nominal doping concentration .
we observe a monotonic increase in the incorporated p fraction with increasing nominal doping concentration , and the incorporation efficiency is about 50% when @xmath80 is below 5% .
however , this technique only measures the elemental composition but not the concentration of active dopants . for samples at sufficiently high doping concentration
, we can determine the free electron concentration through the localized surface plasmon resonance ( lspr ) .
the position of the plasmonic peak depends on the free electron concentration @xmath19 as described by the equation @xcite @xmath87 where @xmath88 is the localized surface plasmonic resonance frequency , @xmath5 is the dielectric constant for bulk si ( 11.7 ) and @xmath89 is the dielectric constant for the surrounding medium , taken as @xmath90 1 for nitrogen atmosphere in this study .
as shown in figure [ fig : concentration ] , the plasmonic peaks are at 1110 , 1260 and 1360 @xmath91 in the infrared absorption spectra and the free electron concentrations are @xmath92 , and @xmath93 @xmath94 for 5% , 10% and 20% p - doped si ncs , respectively .
no plasmonic peaks were observed for doping concentrations lower than nominal 5% .
all known parameters of si ncs are summarized in table i. si nc sizes were determined by x - ray diffraction and transmission electron microscopy , as presented in the si3 [ sec : characterization ] .
next , we examined the electrical transport in the p - doped si nc films .
figure [ fig : g_t ] depicts the temperature dependence of the ohmic conductance @xmath40 for p - doped si nc films at nominal doping concentrations from 1% to 20% .
as shown in figure [ fig : g_t]a , the conductance of p - doped si nc films monotonically increases with the nominal doping concentration
. however , si nc film at _ x_@xmath95 = 20% shows lower conductance than the film at _
x_@xmath95 = 10% .
the reason is not clear at this time . over the entire range of doping concentrations under investigation ,
the film conductance @xmath96 follows efros - shklovskii ( es ) law : @xmath97,\ ] ] where @xmath98 zabrodskii analysis @xcite ( not shown here ) confirms this result .
the fact that es conductivity is seen even at the smallest studied donor concentration implies that even in this case the average number @xmath69 of donors per nc is large . as explained above , fluctuations of this number lead to charging of majority of ncs , the es coulomb gap and es conductivity .
we extract the characteristic temperature @xmath99 from the slope of linear fits for @xmath100 vs @xmath101 using eq . .
we estimate the effective dielectric constant @xmath78 of the nc film from the canonical maxwell - garnett formula @xcite(the film density is assumed to be @xmath90 50% ) and find @xmath102 .
knowing @xmath99 , the dielectric constant and using eq .
we compute the localization length @xmath86 . as the doping concentration increases
, @xmath86 grows from 1.4 nm at @xmath103 to 26.8 nm at @xmath104 , as displayed in figure [ fig : g_t]b and shown in table i. for larger nominal doping @xmath105 , @xmath86 exceeds the nc diameter and reaches three nc diameters for the highest doping level .
this indicates the approach to the mit with growing @xmath19 .
these data are consistent with the predicted @xmath106 by eq .
( [ eq : result ] ) . similar growth of @xmath86 was observed in ref .
@xcite for cdse ncs .
we also studied the effect of the nc separation to verify the theory prediction , eq .
( [ eq : n_exp ] ) .
the p - doped si ncs are prone to oxidation if exposed to air .
an oxide shell starts to grow from the outer surface towards the core by consuming the original si lattice .
the neighboring ncs are now separated by two oxide shells , whose combined thickness @xmath58 grows with time . according to eq .
the critical electron concentration @xmath0 increases with increasing @xmath58 and therefore , @xmath86 decreases . in si4
[ sec : oxidation ] we report our study of @xmath107 and find a qualitative agreement with eq .
( [ eq : n_exp ] ) .
we derived the mit criterion given by eq . for films of semiconductor ncs analogous to the mott criterion for bulk semiconductors .
according to this criterion , mit occurs in si nc films at a critical concentration @xmath108 .
we investigated the electron transport in p - doped si ncs to test this theory .
es variable range hopping conduction was found for all doping concentrations @xmath19 up to @xmath109 .
the localization length increases with increasing doping concentration and exceeds the diameter of a nc at @xmath110 , which indicates the approach to the mit in p - doped si nc films , in agreement with our theory . recently in the ref .
@xcite we focused on the variable - range hopping of electrons in semiconductor nanocrystal ( nc ) films below the critical doping concentration @xmath0 at which it becomes metallic .
we studied how the localization length grows with the doping concentration @xmath19 in the film of touching ncs . for
that we calculated the electron transfer matrix element @xmath111 between neighboring ncs .
we used the ratio of @xmath111 to the disorder - induced nc level dispersion to find the localization length of electrons due to the multistep elastic co - tunneling process and showed that the localization length diverges at concentration @xmath19 equal to @xmath0 given by eq . .
the authors would like to thank k.a .
matveev , c. leighton , b. skinner and a. kamenev for helpful discussions , c. d. frisbie for the use of his equipment and r. knurr for assistance with the icp - oes analysis .
ting chen ( electrical transport studies ) and k.v .
reich(theory ) were supported primarily by the national science foundation through the university of minnesota mrsec under award number dmr-1420013 .
nicolaas kramer(materials synthesis ) was supported by the doe center for advanced solar photophysics .
part of this work was carried out in the college of science and engineering characterization facility , university of minnesota , which has received capital equipment funding from the nsf through the umn mrsec program .
part of this work also used the college of science and engineering nanofabrication center , university of minnesota , which receives partial support from nsf through the nnin program .
reich , han fu . and b.i .
shklovskii created the theory . ting
chen performed the structural and electrical characterization , nicolaas kramer synthesized materials , uwe r. kortshagen discussed and supervised the work .
all authors participated in the discussion and interpretation of the results and co - wrote the manuscript .
the authors declare no financial interests .
freestanding p - doped si ncs were synthesized in a nonthermal radio frequency plasma with a frequency of 13.56 mhz .
the detailed description of synthesis can be found elsewhere @xcite .
the doping concentration is controlled by changing the flow rate of phosphine ( ph@xmath112 ) while maintaining constant flow rates for ar and sih@xmath113 .
typical flow rates used in this work are 0.4 standard cubic centimeters per minute ( sccm ) of sih@xmath113 , 55 sccm of ar , and 0.028 - 0.66 sccm of ph@xmath112 diluted to 15% in hydrogen
. the plasma is operated at a pressure of 0.9 torr with a nominal power of 110 w. the crystallinity and the particle size of si ncs were characterized by xrd using a bruker - axs microdiffractometer with a 2.2 kw sealed cu x - ray source at 40 kv and 40 ma ( wavelength 0.154 nm ) .
the xrd pattern was recorded for dry powders of si ncs deposited on a glass substrate . the high resolution bright field tem employed fei tecnai g2 f-30 tem with a schottky field - emission electron gun operated at 100 kv accelerating voltage .
the tem sample was prepared by collecting si ncs directly onto a copper lacey carbon grid in the plasma reactor .
ftir measurements were performed using a bruker alpha ir spectrometer equipped with a diffuse reflectance ( drifts ) accessory with a deuterated triglycine sulfate ( dtgs ) detector .
all spectra were recorded from 375 to 7000 @xmath91 at 2 @xmath91 resolution , and averaged over 20 scans . the p incorporation in si ncs
was quantified by icp - oes .
si ncs were digested in a mixture of hydrochloric acid ( hcl ) , nitric acid ( hno@xmath112 ) and hydrofluoric acid ( hf ) .
the elemental analysis was calibrated by the standards of si and p samples .
lateral two - terminal devices were fabricated on sio@xmath114 substrates with prepatterned au interdigitated electrodes inside a nitrogen - filled glovebox .
the spacing of electrodes is 30 @xmath61 m , and the aspect ratio is 5317 .
the substrates were precleaned by sequential ultrasonication for 10 min each in acetone , methanol and isopropyl alcohol , and were treated in uv / ozone for 20 min . as - produced si nc powders were dissolved in anhydrous 1,2-dichlorobenzene ( dcb ) , and cloudy stable suspensions were formed by ultrasonication .
si nc films were spin - coated from dispersions of 10 mg ml@xmath115 .
previous work has shown that low temperature annealing leads to an increase in the free electron concentration of the p - doped si ncs and this is primarily attributed to the reduction of dangling bond defects during the annealing process @xcite .
we notice that the annealed si nc films exhibit higher conductance and improved stability compared with fresh - made films . for this study , all p - doped si nc films were annealed at 125 c overnight inside the glovebox before measurement .
the o@xmath114 level was controlled less than 0.1 ppm to minimize the oxidation of ncs during annealing .
the devices were then transferred into another nitrogen - filled glovebox for subsequent electrical measurements .
all handling and testing of devices was performed without air exposure .
the current - voltage ( _ i - v _ ) characteristics of the nc films were recorded in a desert cryogenics ( lakeshore ) probe station in a nitrogen - filled glovebox with keithley 236 and 237 source measuring units and homemade labview programs .
low temperature measurements employed a lakeshore 331 temperature controller with a fixed ramp rate of 4 k / min .
all electrical measurements were carried out in the dark and under vacuum at the pressure of @xmath90 10@xmath116 torr .
44ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1126/science.1117908 [ * * , ( ) ] link:\doibase 10.3402/nano.v1i0.5202 [ * * , ( ) ] link:\doibase 10.1021/nl5029655 [ * * , ( ) ] link:\doibase 10.1021/nn500182b [ * * , ( ) ] link:\doibase 10.1126/science.271.5251.933 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2009.89 [ * * , ( ) ] link:\doibase 10.1021/ja00072a025 [ * * , ( ) ] link:\doibase 10.1126/science.1116703 [ * * , ( ) ] link:\doibase 10.1126/science.1084424 [ * * , ( ) ] link:\doibase 10.1126/science.291.5512.2390 [ * * , ( ) ] link:\doibase 10.1021/nn3057356 [ * * , ( ) ] link:\doibase 10.1126/science.1143802 [ * * , ( ) ] link:\doibase 10.1126/science.1196321 [ * * , ( ) ] link:\doibase 10.1021/nl300880 g
[ * * , ( ) ] link:\doibase 10.1021/jz300048y [ * * , ( ) ] link:\doibase 10.1021/nl403033f [ * * , ( ) ] link:\doibase 10.1515/zpch-2014 - 0587 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.40.677 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.125430 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.60.3963 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevb.51.1743 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.79.469 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.096803 [ * * , ( ) ] link:\doibase 10.1038/nnano.2011.46 [ * * , ( ) ] link:\doibase 10.1021/nl301104z [ * * , ( ) ] link:\doibase 10.1021/ja308200f [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.205316 [ * * , ( ) ] link:\doibase 10.1021/nl050066y [ * * , ( ) ] link:\doibase 10.1116/1.581520 [ * * , ( ) ] link:\doibase 10.1063/1.362670 [ * * , ( ) ] link:\doibase 10.1021/nl4001184 [ * * , ( ) ] link:\doibase 10.1080/13642810108205796 [ * * , ( ) ] http://books.google.com/books?hl=en&lr=&id=wtexaqaamaaj&oi=fnd&pg=pr1&dq=j.+c.+maxwell,+a+treatise+on+electricity+and+magnetism&ots=jd5zjrowzk&sig=zxqac0n6fqoit7lxtwx49dmvvke[__ ] , vol .
( , ) @noop * * , ( ) link:\doibase 10.1016/s0924 - 8099(05)80050 - 2 [ _ _ ] , edited by , , , , and , , vol .
( , ) pp .
http://pubs.acs.org/doi/abs/10.1021/la0477183 [ * * , ( ) ] link:\doibase 10.1016/0039 - 6028(95)01014 - 9 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.155327 [ * * , ( ) ]
link:\doibase 10.1103/physrevb.70.115317 [ * * , ( ) ] in http://www.sciencedirect.com/science/article/pii/s0360056408605635[__ ] , vol . , ( , ) pp .
link:\doibase 10.1063/1.2245191 [ * * , ( ) ] link:\doibase 10.1063/1.2897291 [ * * , ( ) ]
it specifies the conductance between ncs at which the anderson localization is eliminated .
one can wonder why large energy gaps @xmath118 between consecutive shells are not making the anderson delocalization more difficult in the degenerate case than in the non - degenerate one . indeed , due to random charges of surrounding ncs the whole spectrum of a nc is shifted up and down with respect to the spectrum of nearest - neighbor ncs and the fermi level of the system ( see fig .
[ fig : tunneling ] ) .
the reason for the universality in this case is that a large number of degenerate levels in shells closest to the fermi level in both contacting ncs compensates the large value of @xmath118 .
below we verify this compensation generalizing arguments of ref .
@xcite to the degenerate case .
the overlap integral between levels of two nearest neighbor ncs @xmath119 where @xmath120 are wave functions on two contacting ncs . following ref .
@xcite we calculate the amplitude to find an electron on the nc @xmath121 when it is localized on the nc @xmath122 . the electron tunnels from @xmath122 to @xmath121 following a particular path with @xmath123 intermediate states in sequential ncs ( see fig .
[ fig : tunneling ] and ref .
the corresponding amplitude is @xmath124 where @xmath118 stands for the estimate of the absolute value of differences between the energy of the tunneling electron and the ( degenerate ) energy levels of an intermediate nc . due to the @xmath125 degeneracy of each energy shell
the total number of such paths is @xmath126 .
since the intermediate energy is the same , these tunneling amplitudes have the same sign . adding all of them we get @xmath127 this leads to the tunneling probability @xmath128 when @xmath129 the tunneling probability exponentially diverges if @xmath130 .
this means the delocalization of the electron . we can define a quantity @xmath131 as the density of states participating in the tunneling process within each dot .
then the universal criterion for the mit is we used x - ray diffraction ( xrd ) to estimate the diameters of ncs .
figure [ fig : size]a shows well - defined xrd peaks corresponding to diamond cubic structure in p - doped si ncs with nominal doping concentration _ x_@xmath95 from 1% to 20% .
the diameters of ncs shown in table i are calculated from the peak broadening in xrd spectra with spherical correction @xcite . as _
x_@xmath95 increases from 1% to 20% , the nc diameter decreases from 8.1 nm to 7.1 nm .
this size reduction is likely caused by h@xmath114 etching @xcite , since ph@xmath112 is diluted in h@xmath114 with a volume fraction 15% .
a typical bright field transmission electron microscopy ( tem ) image for nominal 10% p - doped si ncs is shown in figure [ fig : size]b .
the p doping does not alter the spherical shape of si ncs and the nc diameter dispersion is @xmath90 15% for all ncs used in this study . with the diameter of ncs and the concentration of free electrons @xmath19
, we can calculate the average number of electrons per nc @xmath21 . for @xmath135 % ,
we get @xmath136 electrons per nc . all known parameters of si ncs are summarized in table i. in previous work , we found that oxidation follows the cabrera - mott mechanism with a characteristic time @xmath137 min @xcite .
based on this oxidation mechanism , we can investigate the dependence of the localization length on the separation @xmath58 between ncs .
the 10% p - doped si nc film was exposed to the air for different periods of time .
temperature dependence of the ohmic conductance for films exposed to air from 1 min to 4 hrs is plotted against _
t_@xmath138 in figure [ fig : oxidation]a .
the film conductance decreases with decreasing temperature for all measurements , and es variable range hopping was observed for the majority of temperature range .
@xmath99 is getting larger with increasing air exposure time , which indicates the decrease of the localization length with oxidation .
this effect can be understood as follows @xcite . in es variable range hopping ,
when an electron tunnels to a distant , non - neighboring nc at a distance _ x _ , its tunneling trajectory involves passing through a chain of intermediate ncs , and the decay of the electron wave function is dominated by passage through gaps between neighboring ncs along the chain .
as a consequence , the wave function is suppressed by a factor of @xmath139 $ ] @xcite , so that the localization length is given by @xmath140 .
the oxide growth on the si nc surface increases @xmath58 and reduces @xmath86 .
the dynamics of the oxide growth on nc surface can be characterized by the elovich equation @xcite : @xmath141 , where @xmath142 and @xmath143 are reaction rate and characteristic time , respectively .
now we have , the inverse localization length for oxidized si nc films is plotted against air exposure time in a linear - log scale as shown in figure [ fig : oxidation]b , and the red solid line is the fit with eq . .
the experimental data are in good agreement with cabrera - mott oxidation mechanism .
the characteristic time @xmath145 for p - doped si ncs is found to be 28 min , twice larger than the intrinsic h - terminated si ncs @xcite .
this means faster oxidation of p - doped si ncs . in cabrera - mott mechanism ,
an electron from the cleaved si - si bond is transferred to an adsorbed o@xmath114 molecule and the resulting o@xmath146 ion drifts toward the cleaved si - si bond with assistance of the electrostatic potential @xcite .
since oxidation of si ncs requires electron tunneling , faster oxidation is expected in @xmath19-type doped si ncs @xcite , as we see in our system . | to fully deploy the potential of semiconductor nanocrystal films as low - cost electronic materials , a better understanding of the amount of dopants required to make their conductivity metallic is needed . in bulk semiconductors ,
the critical concentration of electrons at the metal - insulator transition is described by the mott criterion . here
, we theoretically derive the critical concentration @xmath0 for films of heavily doped nanocrystals devoid of ligands at their surface and in direct contact with each other . in the accompanying experiments ,
we investigate the conduction mechanism in films of phosphorus - doped , ligand - free silicon nanocrystals . at the largest electron concentration achieved in our samples , which is half the predicted @xmath0
, we find that the localization length of hopping electrons is close to three times the nanocrystals diameter , indicating that the film approaches the metal - insulator transition .
semiconductor nanocrystals ( ncs ) have shown great potential in optoelectronics applications such as solar cells @xcite , light emitting diodes @xcite , and field - effect transistors @xcite by virtue of their size - tunable optical and electrical properties @xcite and low - cost solution - based processing techniques @xcite .
these applications require conducting nc films and the introduction of extra carriers through doping can enhance the electrical conduction .
several strategies for nc doping have been developed .
remote doping , the use of suitable ligands as donors in the vicinity of nc surface , increased the conductivity of pbse nc films by 12 orders of magnitude@xcite .
electrochemical doping , which tunes the carrier concentration accurately and reversibly , resulted in conducting nc films @xcite . lately
, stoichiometric control has emerged as a strategy to dope lead chalcogenide ncs @xcite .
finally , electronic impurity doping of ncs , originally impeded by synthetic challenges @xcite , was recently achieved in inas @xcite and cdse @xcite ncs . while many experimental studies have been directed towards increasing the conductivity of nc films ,
there is still no clear consensus on the fundamental question : what is the condition for the metal - insulator transition ( mit ) in nc films @xcite ? in a bulk semiconductor ,
the critical electron concentration @xmath1 for the mit depends on the bohr radius @xmath2 according to the well - known mott criterion @xcite @xmath3 where @xmath4 is the effective bohr radius ( in gaussian units ) , @xmath5 is the dielectric constant of the semiconductor , and @xmath6 is the effective electron mass .
it is obvious that a dense film of undoped semiconductor ncs is an insulator , while a film of touching metallic ncs with the same geometry is a conductor .
therefore , the mit has to occur in semiconductor nc films at some critical concentration of electrons @xmath0 , i.e. there should be an analog to the mott criterion in a dense film of touching semiconductor ncs . here , we focus on ncs that touch each other through small facets of radius @xmath7 without any ligands that impede conduction ( fig .
[ fig : sheme ] ) .
we derive below that for such touching ncs the mit criterion is @xmath8 where @xmath9 is the number of equivalent minima in the conduction band of the semiconductor . as to be expected , eq .
( [ eq : result ] ) predicts @xmath0 for nc films that is much larger than @xmath1 for the bulk .
for instance , for close to spherical particles , the facet radius imposed by the discretness of the crystal lattice can be approximated as @xmath10 , where @xmath11 is the lattice constant and @xmath12 is the nc diameter .
for such facets and cdse ncs with @xmath13 nm , eq .
( [ eq : result ] ) gives @xmath14 , while mott s criterion equation ( [ mott_eq ] ) yields @xmath15 for bulk cdse .
for an array of si ncs with @xmath16 we find @xmath17 , compared to @xmath18 from the mott criterion .
below , we derive eq .
( [ eq : result ] ) and discuss its applicability and limitations . to test the predictions of our theory
, we investigate the electron transport in dense films of phosphorus - doped , ligand - free si ncs over a wide range of doping concentration .
we find that the electron localization length grows with @xmath19 and becomes 3 times larger than @xmath12 at @xmath20 , where @xmath0 is predicted by eq .
( [ eq : result ] ) .
this signals that the mit is indeed occurring close to predicted @xmath0 . |
this study shoots out from two inspiring sources : progress in genetic reconstructions of our species history and experience in accumulating and analysis rich dental data on living and fossil human populations .
the innovation is in combining both advantages to provide new knowledge on eurasian ancestry .
principles of anthropophenetics determine the basic approach to the research and allow using methods of population genetics .
phenes ( discrete irreducible morphological traits ) yield to genes in number and in marking precision but win in extent of genome covering .
dental traits provide the best possibility to examine directly time records in populations .
mapping applications can detect many different patterns hidden in numerous tabled data ; each pattern seems to have a certain historical content .
computer maps provide both analysis and visualization of the enormous volume of data accumulated in dental anthropology .
the study is the first experience of this sort .
the study involves data from 498 samples , 50257 individuals in total , drawn from living populations in eurasia and africa .
the material was taken from a great number of publications , the major its part is presented in two generalizing books by @xcite ( @xcite , @xcite ) ; the other part is our own recent data on populations of the caucasus ( 86 samples ) , the far east ( 3 samples ) and on the russians ( 27 samples ) @xcite . to process this rich information the universal system of analysis , visualization and mapping of dental data _ eurasia _
has been developed by the authors .
all dental data on living and fossil eurasian populations available to this moment are managed by the mysql relational database .
data refer to 830 populations , 32 dental traits , no less than 120 dental phenes as several grades or discrete variations of a trait , and 12 historical periods from the palaeolithic to the present .
all the data operations including statistics and visualization are implemented as routines written in python . in several cases c modules are used to improve the performance .
the basic statistics are executed using the principal component analysis ( pca ) method with the help of pca module algorithms ( http://folk.uio.no/henninri/pca_module/ ) .
mapping of separate dental markers frequencies and pc scores is accomplished via the matplotlib basemap toolkit ( http://matplotlib.sourceforge.net/basemap/doc/html/ ) .
the whole system is managed through web interface built within django framework ( http://www.djangoproject.com/ ) allowing to handle the database , to generate dynamic graphics and to save it in vector or bitmap formats .
the study program is common in the russian federation and includes 32 non - metric dental traits ( @xcite , @xcite ; @xcite ) .
but the real situation is that only few markers are usually presented in published tables of frequencies , so we had to find reasonable balance between number of populations and number of markers involved into pca .
thus the numbers are 498 and 8 respectively . in total ,
143 phenogeographical maps have been created , but in the present short paper only 4 of them , i.e. mapping scores of the four pcs , are overviewed and discussed .
the maps were constructed by interpolating the pc score distribution with the gaussian as a weight function .
we have adopted the following parameters for constructing maps in the case of living populations : the averaging window @xmath0 , the weight function range @xmath1 and the total number of grid knots 50,400 .
small black points indicate the location of the populations under investigation . only aboriginal groups are investigated .
conclusions derived from maps interpretation are often ultimately compiled on the basis of how authors envisage their data fit with established genetic , archaeological or linguistic theories .
we try to make such conclusion in the most independent way and on the basis of experience in dental anthropology and anthropophenetics .
only this approach can provide really new knowledge .
the pca results are presented in table [ table - pca ] .
= 0.8ex .the loadings of 8 dental traits in the first 4 pc scores . [ cols="<,^,^,^,^",options="header " , ] the weight of a certain trait is defined as loadings for a corresponding normalized @xmath2-transformed ( arcsine - transformed ) frequency in the linear combination specifying the component .
pca was applied to the among - group dental variability .
the longitudinal variability of phene pool in eurasian populations seems to be the most important regularity revealed by mapping and pca ( fig.[1pc ] ) .
the geographical factor provides the main contribution to the revealed diversity .
the 1@xmath3 pc explains 53% of the total phenetic variation .
all populations under investigation are divided into two main provinces : the western area with high pc1 scores and the eastern one with low scores .
several scenarios of different time series could determine this pattern .
the larger is the space embracing populations that share similar frequencies , the deeper is the time in their divergence .
so , we can suggest the most ancient scenario in the history of eurasian populations was developing from two perspective different groups .
africa presented by populations from ethiopia @xcite and the republic of mali @xcite joins the western province .
the map shows clines of evident phene flows from near east north - eastward to siberia .
it could be backgrounded to intensive post - neolithic expansion or to any earlier events .
it s the matter for further research .
another flow can be easily traced from east to west along the steppe belt of the continent .
it is explained by the latest ( early medieval ) expansion from inner asia evoking oscillatory migratory waves in population settled along the steppe belt , thus comprising a complicated system in populational interaction .
the kalmyks in the south of east europe is the western final point in this expansion .
the contact zone between the provinces occupies the urals , west siberia , middle asia and india . regarding the split in two main provinces
it should be noted that this phenomenon in eurasia can be traced since homo erectus .
indeed , archaic western forms show a low grade of shoveling and poor differentiation in odontoglyphical patterns on molars versus extremely developed shoveling and richness in odontoglyphics in the eastern province ( @xcite , pp.196197 ) . chronology and dynamics in different morphological systems evolution
seem to be rather independent , while dental characteristics demonstrate much antiquity in phylogeny and provide a direct bridge from the present to the far past inaccessible for cranial traits , as our recent research on ancient and living caucasian populations showed @xcite .
it is worth to mention that both western and eastern forms of homo erectus had five - cusped and six - cusped lower molars @xcite , their gracilization is a peculiar characteristic of homo sapiens , still eastern living populations keep higher frequencies in five - six - cusped lower molars .
it is difficult to ignore these most important data provoking an assumption of the replacement in hominines in the west of the continent and of the hetero - level assimilation in the eastern province .
the 2@xmath4 pc explains 13% of the total phenetic variation ( fig.[2pc ] ) .
the map shows more the latitudinal variability of phene pool in eurasian populations .
the scores of the pc are high in dravidian and munda groups of india , in other indian and some far east populations as well as in many populations in the south of west asia and in the north of europe and siberia .
in fact , the 2@xmath4 pc presents the paradox combination of eastern ( the distal trigonid crest on the first lower molar ) and western markers ( four - cusped lower molars , precisely the first one ) . in our previous study on the caucasian populations we suggested both southern and northern gracile subsets in west eurasia
had developed from one ancestral eastern group @xcite .
the pattern on the map supports this assumption . for the first time we find the traces of the initial group in the east province .
we can envisage the movement of this ancient group from south asia to the east and to the west , subsequent splitting the west flow into northern and southern subsets , probably as a result of populating postglacial continental space .
the 3@xmath5 pc explains 9% of the total phenetic variation ( fig.[3pc ] ) .
high scores of the 3@xmath5 pc are determined by loadings mainly in western traits ( carabelli cusp and four - cusped lower first molars ) .
we ve revealed this combination for west living and fossil populations of the continent in our previous investigation of the caucasus in the anthropohistorical space of eurasia @xcite .
the wide spread of this component through the whole continent seen in the map [ 3pc ] was an absolutely unexpected and amazing phenomenon .
it provides evidence for wide scale ancient populating the continent by a group or close groups mainly from west to east .
maybe this event could happen at different times and maybe repeatedly .
the 4@xmath6 pc describes 7% of the total phenetic variation .
this pc presents again a combination of eastern and western traits ( distal trigonid crest and carabelli cusp ) , and it covers south regions of eurasia .
the revealed landscape more obviously divides the continent into southern and northern halves .
the northern zone is occupied by another mixed combination ( deflecting wrinkle and four - cusped first lower molars ) common to finn - ugric populations ( @xcite ; @xcite ) .
the map ( fig.[4pc ] ) again detects hidden patterns and more wide traces of this combination in southern , central and eastern areas of the continent .
the highest scores of the last combination are found in some marginal coastline and in central mountain populations of eurasia .
the map traces the dispersal of a branch of this combination to the west via the middle urals and its subsequent irradiation in the european territory .
it must be emphasized that two of the four pcs are composed by combinations of western and eastern markers .
diminishing eastern traits frequencies at different grades in different groups of west continental populations @xcite and worldwide in shoveling @xcite is a discovered phenomenon .
so , we can suggest that ancestral polymorphic or assimilated populations should have even more expressed eastern component .
evolutionary factors , including genetic drift , selection and gene flows , may have altered the patterns of phenetic frequency and distribution in existing populations . however , the time depths of the revealed landscapes are still not known exactly , so associating them with particular historic and demographic events seems to be speculative at the moment . to provide clear dating additional studies in integrating with established genetic , archaeological and linguistic evidence should be launched . since dental markers provide the best possibility to examine directly time records in populations ,
an alternative perspective is in phenetic investigations of fossil eurasian groups .
east europe and adjacent areas rich in fossil data seem to be the region to start with .
in spite of the enormous territory and the revealed divergence the populations of the continent have undergone wide scale and intensive time - space interaction .
the maximal phenetic diversity was detected in india , respectively lesser in north europe , west siberia and near east .
many details in the revealed landscape could be backgrounded to different historical events .
the maps visualize the most important results in analysis : the wide spread of the western combination through the whole continent till the pacific coastline and the envision of the dispersal of the paradox combinations of eastern and western markers from south or central asia to the east and to the west . taking into account that no additional eastern combinations in the total variation in asian groups have been found but mixed or western markers sets and that eastern dental characteristics are traced in asia since homo erectus , the choice between the ancestral polymorphism and the hetero - level assimilation in the eastern province is made in favour of the latter .
the study was supported by the russian foundation of basic research , grant 08 - 06 - 00124 .
the authors are grateful to prof .
elena balanovska , research centre for medical genetics , russian academy of medical sciences , for the initial support of the idea of this research ; and to petr voitsik , astro space center , the lebedev physical institute of the russian academy of sciences , for his valuable aid in software development ; and , last but not least , to the organizers and the participants of the 2010 meeting inqua - seqs for the inspiring communication .
kashibadze v.f . , 2006 .
the caucasus in the anthropohistorical space of eurasia .
southern scientific centre , russian academy of sciences publishing house , rostov - on - don , russia ( in russian , with english summary ) .
keita b. , 1977 .
the anthropology of the republic of mali population . ph.d .
thesis , institute of ethnology and anthropology , russian academy of sciences , moscow , russia ( in russian ) .
mizoguchi y. , 1985 . shovelling : a statistical analysis of its morphology .
university of tokyo press , japan .
shinkarenko v.s . ,
naumkin v.v . ,
, zoubov a.a . , 1984 .
the anthropological investigations on the sokotra island .
sovetskaya etnografia 4 , 53 - 62 ( in russian ) .
zoubov a.a . , 1968
odontology : methods in anthropological research .
nauka , moscow ( in russian ) .
zoubov a.a . ,
ethnical odontology .
nauka , moscow ( in russian ) .
zoubov a.a .
, khaldeeva n.i . , 1989 .
odontology in current anthropology .
nauka , moscow ( in russian ) . | on the base of advantages in gene geography and anthropophenetics the phenogeographical method for anthropological research is initiated and experienced using dental data .
statistical and cartographical analyses are provided for 498 living eurasian populations .
mapping principal components supplied evidence for the phene pool structure in eurasian populations and for reconstructions of our species history on the continent . the longitudinal variability seems to be the most important regularity revealed by principal components analysis ( pca ) and mapping proving the division of the whole area into western and eastern main provinces .
so , the most ancient scenario in the history of eurasian populations was developing from two perspective different groups : western group related to ancient populations of west asia and the eastern one rooted by ancestry in south and/or east asia . in spite of the enormous territory and the revealed divergence
the populations of the continent have undergone wide scale and intensive time - space interaction .
many details in the revealed landscapes could be backgrounded to different historical events .
the most amazing results are obtained for proving migrations and assimilation as two essential phenomena in eurasian history : the wide spread of the western combination through the whole continent till the pacific coastline and the envision of the movement of the paradox combinations of eastern and western markers from south or central asia to the east and to the west .
taking into account that no additional eastern combinations in the total variation in asian groups have been found but mixed or western markers sets and that eastern dental characteristics are traced in asia since homo erectus , the assumption is made in favour of the hetero - level assimilation in the eastern province and of net - like evolution of our species . dental markers , pca , mapping , eurasia |
the group - theoretic methods are a modern mathematical apparatus widely used to investigation of different mathematical models based both on the ordinary and partial differential equations .
the sources of the group analysis of differential equations are found in the fundamental papers of sophus lie @xcite .
however , only after the classical works of g. birkhoff @xcite and l.v .
ovsiannikov @xcite have been published , the group - theoretical methods would have been widely used in applications . to the present day , the symmetry properties of many well - known equations from mechanics , hydrodynamics , quantum physics etc .
have been studied in detail ( see the monographs @xcite and the papers sited therein ) .
this notwithstanding , it should be stressed that only a few papers consider differential equations with initial and boundary conditions .
to the best of our knowledge , first works in this direction were published only at the end of 60 s years of the last century @xcite .
nevertheless , interest to the group - theoretic analysis of bvps have been growing only in the 90 s of the twentieth century ( see the monographs @xcite and the papers sited therein ) . at present
, the generally - accepted view is that a symmetry ( in lie s sense ) of a boundary - value problem must satisfy such three criteria @xcite : * a symmetry of the governing differential equation ; * a smooth bijective mapping of the domain to itself ; * a mapping of the set of boundary data to itself . if a differential equation contains as coefficients arbitrary functions ( formally speaking , they can be constants ) then the group classification problem springs up .
such kind of problems was formulated and solved for a class of non - linear heat equations in the pioneering work @xcite ( see also @xcite ) . at the present time
, there are algorithms for rigorous solving group classification problems ( see , e.g. , @xcite and the papers cited therein ) , which were successfully applied to different classes of partial differential equations .
thus , if a governing equation and/or auxiliary ( initial or boundary ) conditions contain as coefficients arbitrary functions then one should formulate and solve the group classification problem for such class of bvps . in this paper
we consider from group - theoretic point of view a class of ( 1 + 1)dimensional nonlinear bvps modeling the processes of heat transfer in the semi - infinity conductive domain occupying the region @xmath0 : @xmath1 where @xmath2 is the unknown temperature field , @xmath3 is the nonlinear coefficient of thermal conductivity , @xmath4 is the some known function describing the heat flux of energy absorbing ( or radiating ) at the surface @xmath5 , @xmath6 is the some known temperature at infinity .
additionally , we will assume that all functions arising in problem ( [ 0.1])([0.4 ] ) are sufficiently smooth .
it should be noted that bvps with governing heat conduction equation have a wide field of applications in various branches of science and technology . that is why the problems of this type were a subject of especial attention of researchers throughout the history of development of the heat conduction theory . in the case of linear heat equation ( [ 0.1 ] ) , the problems of the type ( [ 0.1])([0.4 ] ) can be easy solved @xcite . if the equation is nonlinear than an analysis of the problems like ( [ 0.1])([0.4 ] ) is very complicated , and each such a problem requires special investigation .
starting from the classical works of ya.b . zeldovich and g.i .
barenblatt ( see the monograph @xcite ) , the bvps for nonlinear heat conduction have been intensively studied by different both exact and approximate methods ( see @xcite and the paper sited therein ) . however , to the beast of our knowledge , there are a few papers dealing with group - theoretic investigation of problems for nonlinear heat equation and its generalizations @xcite , but class ( [ 0.1])([0.4 ] ) has not yet been the subject of study by researchers working in the field of symmetry analysis of differential equations .
the paper is organized as follows . in section 2 , we formulate a definition of invariance in lie s sense for a wide class of bvps . the problem of group classification in classes of bvps parameterized by arbitrary elements is studied and an algorithm for its solution is also formulated . in section 3 ,
we perform the complete group classification of the class of bvps ( [ 0.1])([0.4 ] ) . in section 4 ,
we carry out the reduction of the bvp in question in the special case when @xmath7 , @xmath8 , and @xmath9 to a bvp of lower dimensionality , and an example of its exact solution is constructed . finally , we present some conclusions in the last section .
let us consider a bvp of @xmath10th order ( @xmath11 ) for a system of @xmath12 evolution equations with @xmath13 independent @xmath14 ( hereafter @xmath15 ) and @xmath12 dependent @xmath16 variables .
we assume that the basic equations @xmath17 are defined on a domain @xmath18 with smooth boundaries .
hereafter the subscripts @xmath19 and @xmath20 denote differentiation w.r.t .
these variables , the symbol @xmath21 denotes the set of all partial derivatives of @xmath22 w.r.t .
@xmath20 of order @xmath23 .
we , also , will use the symbol @xmath24 to designate the set of all partial derivatives of @xmath22 w.r.t .
all independent variables @xmath14 of order @xmath23 . along with system ( [ 1 ] ) , we also consider two types of boundary conditions that often appear in applications @xmath25 and @xmath26 here @xmath27 and @xmath28 are the given numbers , @xmath29 and @xmath30 are some known functions .
we assume that all functions arising in ( [ 1])([3 ] ) are sufficiently smooth .
consider an @xmath31parameter ( local ) lie group @xmath32 of point transformations of the variables @xmath33 in the euclidean space @xmath34 ( open subset of @xmath34 , respectively ) given by equations @xmath35 where @xmath36 , @xmath37 ; @xmath38 are the group parameters . according to the general lie group theory , one may construct the corresponding @xmath31dimensional lie algebra @xmath39 with the basic generators @xmath40 where @xmath41 .
[ def1 ] the boundary value problem of the form ( [ 1])([3 ] ) is called to be invariant w.r.t .
the lie group @xmath32 ( [ 4 ] ) if : * the manifold , determined by the system of equations ( [ 1 ] ) in the space of variables @xmath42 , is invariant w.r.t .
the @xmath10th order prolongation of the group @xmath32 ; * each manifold , determined by conditions ( [ 2 ] ) with any fixed number @xmath43 in the space of variables @xmath44 , where @xmath45 , is invariant w.r.t .
the @xmath46th order prolongation of the group @xmath32 ; * each manifold , determined by conditions ( [ 4 ] ) with any fixed number @xmath47 in the space of variables @xmath48 , where @xmath49 , is invariant w.r.t .
the @xmath50th order prolongation of the group @xmath32 .
[ def2 ] a group of invariance of bvp ( [ 1])([3 ] ) is called the maximal group of invariance ( mgi ) if any group of invariance of the bvp in question is contained in the mgi as a subgroup .
it is to be noted that the notion of maximal algebra of invariance ( mai ) of bvp ( [ 1])([3 ] ) can be formulated in a quite similar way .
it immediately follows from definitions [ def1 ] and [ def2 ] the algorithm to seek for the mgi of some bvp of the form ( [ 1])([3 ] ) : * using the classical lie algorithm , to find the mgi of the system of differential equations ( [ 1 ] ) ; * to find the maximal subgroup of the lie group obtained in item @xmath51 under which the boundary conditions ( [ 2 ] ) and ( [ 3 ] ) are simultaneously invariant . in section 4 we will use the notion of invariant solution of bvp ( [ 1])([3 ] ) .
so , the relevant definition must be done .
[ def3 ] @xmath52 , with components @xmath53 , is an invariant solution of bvp ( [ 1])([3 ] ) corresponding to the lie group ( [ 4 ] ) if the following conditions take place : * @xmath52 satisfies equation ( [ 1 ] ) and the boundary conditions ( [ 2 ] ) and ( [ 3 ] ) ; * the manifold @xmath54 is an invariant manifold of the lie group ( [ 4 ] ) .
let us now consider a class of bvps of the form ( [ 1])([3 ] ) : @xmath55 where @xmath56 @xmath57 @xmath58 are arbitrary elements satisfying several auxiliary equations and ( or ) inequalities ; the other designations are similar those used in bvp ( [ 1])([3 ] ) .
[ def4 ] an equivalence transformation of the class of bvps ( [ 2.11])([2.13 ] ) is a non - degenerate local change of variables @xmath59 carrying every bvp of the form ( [ 2.11])([2.13 ] ) into a bvp of the same form @xmath60 the functions @xmath61 , @xmath62 and @xmath63 representing the new arbitrary elements , may be , in general , different from the original functions @xmath64 , @xmath65 and @xmath66 , respectively .
the set of all such equivalence transformations of the class of bvps ( [ 2.11])([2.13 ] ) forms a group called the equivalence group .
hereafter we denote one by @xmath67 .
let now the functions @xmath68 , @xmath69 and @xmath70 are fixed ; another words , formulae ( [ 2.11])([2.13 ] ) prescribe some bvp .
let the mgi of such bvp is denoted by @xmath71 , and @xmath72 is the respective mai .
[ def5 ] the common part of @xmath71 , i.e. @xmath73 where @xmath68 , @xmath69 , and @xmath70 run all admissible values , is called the kernel of mgis ( principal group ) of bvps from the class under study .
the respective lie algebra @xmath74 is called the algebra of kernel of mgis .
the group classification problem for the class under study is to find the kernel @xmath75 of mgis of bvps from the class ( [ 2.11])([2.13 ] ) and to describe all such cases of the arbitrary elements @xmath68 , @xmath69 , and @xmath70 admitting nonequivalent expansions of the group @xmath75 .
we propose the following algorithm of the group classification for the class of bvps ( [ 1])([3 ] ) : * to construct the equivalence group @xmath76 of the class of partial differential equations systems ( [ 2.11 ] ) ; * to find the equivalence group @xmath67 of the class of bvps ( [ 2.11])([2.13 ] ) by extending the space of @xmath76action on the prolonged space , where all arbitrary elements @xmath69 and @xmath70 arising in the boundary conditions ( [ 2.12 ] ) and ( [ 2.13 ] ) are treated as new dependent variables ; * to perform the group classification of the class of systems ( [ 2.11 ] ) up to local transformations from the group @xmath67 ; * to find the kernel @xmath75 of mgis of bvps from the class ( [ 2.11])([2.13 ] ) ; * to describe all possible @xmath67inequivalent bvps of the form ( [ 2.11])([2.13 ] ) admitting mgis of higher dimensionality than @xmath75 .
the algorithm proposed leads to a list of nonequivalent ( up to the local transformations from the equivalence group @xmath67 ) bvps of the form ( [ 2.11])([2.13 ] ) with the relevant mgis ( or mais ) .
the problem of group classification is considered to solve in full if it was proved that : * the groups ( algebras ) constructed are the mgis ( mais , respectively ) of the bvps obtained ; * all bvps from the list are nonequivalent with respect to the transformations from the group @xmath67 . it should be noted that the class ( [ 0.1])([0.4 ] ) is parameterized by two arbitrary functions and a constant , so , to investigate symmetry properties of its , we will exploit the algorithm proposed .
in this section we will perform the complete group classification of the class of bvps ( [ 0.1])([0.4 ] ) . in conformity with
the algorithm proposed , at the first step we should seek for the equivalence group @xmath76 of the class of nonlinear heat equations ( [ 0.1 ] ) . however , this group is well - known @xcite @xmath77 where @xmath78 .
the class of bvps ( [ 0.1])([0.4 ] ) admits the equivalence group of point transformations @xmath67 @xmath79 @xmath80 where @xmath81
. * proof . * according to definition [ def4 ] , the group of equivalent transformations @xmath67 is generated by all such point transformations of the independent and dependent variables mapping the class of bvps ( [ 0.1])([0.4 ] ) into itself , i.e. @xmath82 substituting transformations ( [ 3.1 ] ) into the class of boundary conditions ( [ 3.7 ] ) one obtains @xmath83 taking into account ( [ 0.3 ] ) , from ( [ 3.6 ] ) we immediately receive @xmath84 if in a quite similar way we consider the class of boundary conditions ( [ 3.8 ] ) , one can arrive at transformations ( [ 3.3 ] ) and ( [ 3.4 ] ) .
the proof is now complete .
@xmath85 [ t1 ] all possible mais ( up to the equivalent transformations from the group @xmath67 ) of equation ( [ 0.1 ] ) for any
fixed strictly positive function @xmath3 are presented in table 1 .
any other equation of the form ( [ 0.1 ] ) is reduced by an equivalence transformation from the group @xmath67 to one of those given in table 1 . *
the result of group classification of the class of nonlinear heat equations ( [ 0.1 ] ) * * up to the equivalent transformations from the group @xmath67 * * proof . * it follows directly from the result of group classification of the nonlinear heat equation carried out up to local transformations from the group @xmath76 @xcite .
the proof is now complete .
@xmath85 [ t2 ] all possible mgis ( up to equivalent transformations from the group @xmath67 ) of the nonlinear bvp ( [ 0.1])([0.4 ] ) for any fixed set of arbitrary elements @xmath86 , where @xmath3 are presented in table 2 .
any other bvp of the form ( [ 0.1])([0.4 ] ) is reduced by an equivalence transformation from the group @xmath67 to one of those with the arbitrary elements given in table 2 .
[ rem1 ] in table 2 , we used the following designations for the one - parameter lie groups @xmath87 where @xmath88 with indexes are the relevant group parameters .
[ rem2 ] up to local transformations from the group @xmath67 , the parameter @xmath89 may be set equal to @xmath90 . *
the result of group classification of the class of bvps ( [ 0.1])([0.4 ] ) * * up to the equivalent transformations from the group @xmath67 * * proof .
* let us consider the case of the arbitrary function @xmath3 . in accordance with theorem [ t1 ] ,
the mgi of the governing equation ( [ 0.1 ] ) is the three - parameter lie group generated by the one - parameter groups @xmath91 , and the group @xmath92 first , we will study the invariance of the bvps from the class in question w.r.t .
the lie group @xmath93 . according to item 2 ) of definition [ def1 ] , the boundary conditions ( [ 0.3 ] ) is invariant under @xmath93 if the conditions @xmath94 are satisfied on the manifold @xmath95 = @xmath96 . the first equation from ( [ 16 ] )
is identically fulfilled , while the second one leads to the requirement @xmath97 from which we immediately obtain @xmath98 , where @xmath89 is an arbitrary real number , or @xmath99 .
the invariance of the boundary condition ( [ 0.4 ] ) w.r.t .
the lie group under study is evident .
thus , one can conclude that bvp ( [ 0.1])([0.4 ] ) with the arbitrary function d(u ) is invariant w.r.t .
@xmath93 if and only if @xmath98 ( case 1 in table 2 ) , or @xmath99 .
it is easy to obtain that the bvp in question is not invariant w.r.t .
the lie group @xmath100 , since the boundary curve @xmath101 is not invariant under the point transformations from @xmath100 . at the same time , the invariance of bvp ( [ 0.1])([0.4 ] ) w.r.t .
@xmath102 leads to the requirement @xmath103 ( case 2 in table 2 ) . in order to complete examination of the case of the arbitrary function @xmath104 ,
let us study the invariance of the bvps from class ( [ 0.1])([0.4 ] ) w.r.t .
the one - parameter lie groups corresponding to the linear combination of the basic generators of the algebra @xmath105 of equation ( [ 0.1 ] ) @xmath106 where @xmath107 are arbitrary real parameters .
let @xmath108 , then the lie group @xmath109 @xmath110 corresponds to the generator @xmath111 .
for the boundary curve @xmath5 to be invariant w.r.t .
@xmath109 , the condition @xmath112 must hold .
therefore , the group @xmath109 is reduced to @xmath113 , which we completely considered above .
now let @xmath114 . in this case , the operator @xmath111 generates the one - parameter lie group @xmath115 the invariance of the boundary condition ( [ 0.4 ] ) w.r.t .
@xmath116 is evident , while the invariance of the boundary condition ( [ 0.1 ] ) leads to the restrictions @xmath112 and @xmath117 .
up to local transformations from @xmath67 , one can set @xmath118 and reduce the group @xmath116 to @xmath119 , with the function @xmath4 taking the form @xmath98 .
thus , any linear combination of the basic generators of the nonlinear heat equation ( [ 0.1 ] ) with the arbitrary function @xmath104 does not lead to any new one - parameter group of invariance .
taking into account the invariance conditions of the bvps from the class under study w.r.t .
the groups @xmath102 and @xmath93 , one obtains that bvp ( [ 0.1])([0.4 ] ) is invariant under the two - parameter lie group @xmath120 if and only if @xmath99 ( case 3 in table 2 ) .
it means that the case of the arbitrary function @xmath104 is completely investigated .
let us now consider case 2 from table 1 , i.e. when @xmath121 . here
the nonlinear heat equation ( [ 0.1 ] ) is invariant w.r.t .
the four - parameter lie group generated by the one - parameter groups @xmath102 , @xmath100 , @xmath93 , and the group @xmath122 obviously , the bvps from class ( [ 0.1])([0.4 ] ) are not invariant w.r.t . the lie group @xmath123 . indeed , for the boundary condition ( [ 0.4 ] ) to be invariant under @xmath123 , the following conditions must satisfy @xmath124 for an arbitrary value @xmath125 , where @xmath126 = @xmath127 . taking into account ( [ a1 ] ) , from the second equality one immediately obtains @xmath128 .
thus , the contradiction is obtained and we can conclude that class ( [ 0.1])([0.4 ] ) is not invariant w.r.t . @xmath123 . in a quite similar way one can show that the bvps from the class in question is not invariant w.r.t . the one - parameter lie groups corresponding to any linear combination of the generator @xmath129 with the generators of the algebra @xmath105 .
thereby , we conclude that the case @xmath121 does not lead to any new group of invariance .
let us consider the most interesting case 3 from table 1 , i.e. @xmath130 .
in this case , equation ( [ 0.1 ] ) is invariant w.r.t .
the four - parameter lie group , generated by the one - parameter groups @xmath102 ,
@xmath100 , @xmath93 , and the group @xmath131 first , we will investigate the invariance of the bvps from class ( [ 0.1])([0.4 ] ) w.r.t .
the group @xmath132 . according to item 2 ) of definition [ def1 ] , the boundary condition ( [ 0.3 ] ) is invariant under @xmath132 if , on the manifold
@xmath133 = @xmath134 , the following equalities @xmath135 are true . from ( [ dod1 ] )
one obtains the relation @xmath136 which immediately leads to the conditions : @xmath137 , @xmath4 is arbitrary ( case 8 in table 2 ) and @xmath138 , @xmath139 .
the boundary condition ( [ 0.4 ] ) is invariant w.r.t .
@xmath132 if and only if @xmath9 .
taking into account the invariance of bvps ( [ 0.1])([0.4 ] ) w.r.t .
the one - parameter groups @xmath140 , and @xmath132 , one can draw a conclusion that it are invariant w.r.t .
the two - parameter lie group @xmath141 ( case 9 in table 2 ) and the three - parameter lie group @xmath142 ( case 7 in table 2 ) if and only if the following conditions @xmath143 and @xmath144 hold , respectively .
let us now consider the invariance of the bvps from the class under study w.r.t .
the one - parameter lie groups corresponding to any linear combination of the generator @xmath145 with the generators from the algebra @xmath105 @xmath146 where @xmath107 are arbitrary real parameters .
the operator @xmath147 generates depending from the values of the parameters @xmath148 and @xmath10 such one - parameter lie groups * @xmath149 @xmath150 * @xmath151 @xmath152 * @xmath108 @xmath153 first , we will study the case of the group @xmath154 .
the invariance of the boundary conditions ( [ 0.3 ] ) w.r.t .
@xmath154 ( see equalities ( [ dod1 ] ) ) gives @xmath155 the second equality of ( [ 3.14 ] ) leads to the condition at the function @xmath4 @xmath156 where @xmath89 is an arbitrary real constant .
making in ( [ 3.15 ] ) the change of the variable @xmath118 , one obtains @xmath157 while the transformations from the group @xmath154 get such form @xmath158 let us now reduce the form of function @xmath4 using the substitution @xmath159 , from which we can easy obtain @xmath160 . plugging this value of @xmath161 into @xmath162 , one arrives at the group @xmath163 ( see remark [ rem1 ] ) .
the invariance of the boundary condition ( [ 0.4 ] ) w.r.t .
@xmath163 gives directly @xmath9 .
thus , we proved that the bvp from class ( [ 0.1])([0.4 ] ) is invariant w.r.t .
the one - parameter lie group @xmath163 if and only if the arbitrary elements @xmath164 , and @xmath6 have such values @xmath165 it is exactly case 4 in table 2 . the case of the group @xmath166 is considered in a quite similar way , and as a result one obtains the conditions @xmath167 while the group @xmath166 reduces to the form @xmath168 it is easy to see if we set @xmath169 than the group @xmath170 is a particular case of the group @xmath163 . hence ,
the case of the group @xmath166 does not lead to any new group of invariance of the bvps from the class under study .
consider the case of the group @xmath171 .
the invariance of the boundary condition ( [ 0.4 ] ) w.r.t .
@xmath171 gives again the condition @xmath9 , while , from the invariance of the boundary conditions ( [ 0.3 ] ) , we arrive at the restrictions @xmath172 it follows from the second equality in ( [ 3.16 ] ) that @xmath173 where @xmath89 is an arbitrary real constant .
making the change of the variable @xmath174 , one obtains that the group @xmath171 is reduced to the group @xmath175 .
hence , the bvp from class ( [ 0.1])([0.4 ] ) is invariant w.r.t .
the group @xmath175 if and only if the arbitrary elements have such form @xmath176 thus , case 5 in table 2 is obtained .
taking into account the invariance conditions of the bvp from class ( [ 0.1])([0.4 ] ) w.r.t .
the one - parameter lie groups @xmath177 , and @xmath175 , we can conclude that such bvp is invariant w.r.t .
the two - parameter lie group @xmath178 ( case 6 in table 2 ) if and only if @xmath179 thus , the case of the function @xmath180 is completely examined .
let us now consider the case of the conformal power @xmath181 , i.e. @xmath182 . according to theorem [ t1 ] , in this case equation ( [ 0.1 ] ) is invariant w.r.t .
the five - parameter lie group generated by the one - parameter groups @xmath183 , and the group @xmath184 first , we stress that the one - parameter groups ( with @xmath185 ) listed in cases 47 of table 2 are the groups of invariance of the bvps in question ( with @xmath185 ) under the same restrictions on the function @xmath4 and constant @xmath6 .
now , our purpose is to show that the group @xmath186 does not lead to any case of invariance of bvps from class ( [ 0.1])([0.4 ] ) . indeed , for the boundary condition ( [ 3.3 ] ) to be invariant w.r.t .
@xmath186 , the following conditions must hold @xmath124 for arbitrary value of @xmath187 , where @xmath126 = @xmath127 .
however @xmath188 therefore , the condition @xmath189 is satisfied if and only if the group parameter @xmath190 .
thus , the contradiction is obtained , and we can draw a conclusion that the bvps from class ( [ 0.1])([0.4 ] ) are not invariant w.r.t . @xmath186 .
let us now examine the invariance the bvps from the class under study w.r.t .
the one - parameter lie groups corresponding to any linear combination of the conformal operator @xmath191 with the operators of the algebra @xmath192 @xmath193 where @xmath194 are arbitrary real constants .
first , we note that the invariance of the boundary curve @xmath5 w.r.t . the generator @xmath195 immediately yields @xmath112 .
let now the following conditions @xmath196 hold . in this case
, the transformation of the variable @xmath20 from the corresponding group has such form @xmath197 taking into account ( [ 3.17 ] ) , the equality @xmath189 gives @xmath198 the cases when at least one of the parameters @xmath161 , or @xmath199 are not zero can be examined in a quite similar way , whereas the cases @xmath200 , or @xmath201 lead to the group @xmath186 considered above .
thus , we showed that any bvp from class ( [ 0.1])([0.4 ] ) is not invariant w.r.t .
each one - parameter lie group corresponding to any linear combination of the conformal operator @xmath191 with the operators from the mai of equation ( [ 0.1 ] ) with @xmath182 .
the proof is now complete .
in this section we will use the results of the group classification of the class of bvps ( [ 0.1])([0.4 ] ) obtained in the previous section for reduction one bvp from the class and construction an exact invariant solution for its .
let us restrict ourselves to the case 6 from table 2 .
another words , we consider a bvp of the form @xmath202 according to theorem 2 , bvp ( [ 3.18])([3.20 ] ) admits the two - parameter lie group @xmath203 induced by the following one - parameter groups @xmath204 @xmath205 it is easy to obtain that the two - dimensional lie algebra @xmath206 corresponds to the group @xmath203 .
consider the reduction of the bvp in question by the generator @xmath207 .
using the standard procedure , one can easy obtain the ansatz @xmath208 substituting ( [ 4.2 ] ) in bvp ( [ 3.18])([3.20 ] ) , and making relevant calculations , we arrive at the reduced bvp @xmath209 now one sees that the problem of construction of exact solutions of bvp ( [ 3.18])([3.20 ] ) is reduced to solving the non - linear ordinary differential equation ( [ 4.3 ] ) . if the parameter @xmath10 is arbitrary , than a general solution of equation ( [ 4.3 ] ) is unknown , but several particular solutions of ones , for some values of the parameter @xmath10 ,
are presented in the handbook @xcite ( see , also , @xcite ) . here
we will consider only one instance when @xmath210 . in this case ,
equation ( [ 4.3 ] ) has the general solution written in the parametric form @xmath211 where @xmath212 .
+ substituting ( [ 4.10 ] ) into the boundary conditions ( [ 4.4 ] ) and ( [ 4.5 ] ) , and making relevant calculations , we arrive at the solution of bvp ( [ 4.3])([4.5 ] ) with @xmath210 @xmath213 now , in accordance with ( [ 4.2 ] ) , formulae ( [ 4.11 ] ) lead to the solution in implicit form of bvp ( [ 3.18])([3.20 ] ) @xmath214 where @xmath215 , @xmath216 . )
([3.20 ] ) with @xmath210 , @xmath217 at the moments of time : 1 ) @xmath218 and 2 ) @xmath219.,title="fig:",width=302 ] ) ([3.20 ] ) with @xmath210 , @xmath217 at the moments of time : 1 ) @xmath218 and 2 ) @xmath219.,title="fig:",width=302 ]
in this paper we proposed a new definition of invariance in lie s sense of a bvp for systems of evolution differential equations .
the definition formulated is essential extension of the known definitions of bvp invariance available in the literature .
it is applicable to a wide class of bvps and can be easily generalized to bvps with hyperbolic and elliptic basic equations
. the problem of group classification in a class of bvps parameterized by arbitrary elements is formulated , and an algorithm for its solution is also proposed . the main result of the paper is the complete group classification ( carried out in section 3 ) of the class of bvps ( [ 0.1])([0.4 ] ) with ( 1 + 1)dimensional governing nonlinear heat equation . it should be stressed that the nonlinear heat equation ( [ 0.1 ] ) with the thermal conductivity coefficient depending on temperature as a power law , i.e. @xmath220 , has the mai for the conformal power @xmath185 . however , bvp ( [ 0.1])([0.4 ] ) has the particular power @xmath137 ( see cases 89 in table 2 ) , which is not a special one for equation ( [ 0.1 ] ) .
this fact emphasizes nontrivial character of the problem of investigation of symmetry properties for partial differential equations with both boundary and initial conditions .
cherniha , nonlinear evolution equations : galilei invariance , exact solutions and their applications .
thesis for a full doctor s degree by specialization 01.01.03 mathematical physics , institute of mathematics , national academy of sciences of ukraine , kyiv , 2003 . | a definition of invariance in lie s sense for a boundary value problem ( bvp ) with the basic evolution differential equations is proposed .
a problem of group classification at a wide class of bvps parameterized by arbitrary elements is formulated and an algorithm for its solution is also proposed .
the group classification of a class of bvps for an ( 1 + 1)dimensional nonlinear heat equation arising from mathematical modeling of heat conduction in semi - infinite body is carried out .
an example of invariant solution of a bvp from the class under study is presented .
* group theoretic analysis of a class of boundary value problems for a nonlinear heat equation * * sergii kovalenko * _ institute of mathematics , ukrainian national academy of sciences , + tereshchenkivska street 3 , kyiv 01601 , ukraine _
e - mail : [email protected] 2010 mathematics subject classification : 35k61 , 22e70 .
keywords : lie point symmetry , heat conduction , boundary - value problem , exact solution . |
a chemical reaction system that is self - sustaining and collectively autocatalytic is believed to represent an important step in the emergence of early life @xcite .
these systems are defined by two properties : ( i ) each molecule can be built up from a small subset of pre - existing ` food ' molecules by some reaction in the system , and ( ii ) each reaction is catalysed by some product of another reaction ( or an element of the food set ) .
moreover , recent experimental work has demonstrated at least the possibility ( and viability ) of such sets @xcite .
it is also of interest to develop a mathematical framework that allows us to study the entire universe of possible self - sustaining autocatalytic sets , so that general results can be established , and predictions made . here
, we further explore one approach ( ` raf theory ' ) which has provided a tractable and incisive tool for addressing computational and stochastic questions .
raf theory grew out of two strands : stuart kauffman s pioneering work on random autocatalytic networks from the 1970s and 1980s @xcite , and analysis of the first emergence of cycles in random directed graphs by bollobas and rasmussen @xcite .
both of these earlier studies were explicitly motivated by origin - of - life considerations .
the approach is related to , but different from chemical organisation theory ( cot ) @xcite and other formal approaches of a similar flavour , which include petri nets @xcite , rosen s ( m ; r ) systems @xcite , and eigen and schuster s hypercycle theory @xcite . in earlier work @xcite
@xcite , @xcite we have established a series of results concerning the structure , discovery and probability of the formation of raf sets in a variety of catalytic reaction systems .
when such a system contains a self - sustaining autocatalytic set ( an ` raf ' , defined below ) , this set can often be broken down into smaller rafs until we arrive at the smallest ` building block ' rafs that can not be broken down any further ( _ c.f . _ @xcite ) . in this paper , we investigate the structure of these irreducible rafs , and bounds on the size of the smallest rafs within a catalytic reaction system . along the way , we derive some new facets of raf theory , exploring further its relationship to cot , and the related weaker notions of pseudo - rafs and co - rafs , which can be co - opted by a raf to form a larger raf system . while it is easy to determine whether a chemical reaction system contains an raf ( in which case
there is a unique largest one @xcite ) , we prove that finding a smallest raf is an np - hard problem .
nevertheless , the structure of the smallest ( ` irreducible ' ) rafs allows us to present efficient algorithms to find lower bounds on their size , and to determine whether a given collection contains the smallest raf in the system .
we begin by recalling some definitions before proceeding to the combinatorial and algorithmic aspects of rafs .
we then apply mathematical arguments and simulations to study the size and distribution of irreducible rafs in kauffman s random binary polymer model @xcite , and show that at a level of catalysis at which rafs first form , small rafs are highly unlikely .
we end with a short discussion .
to formalize the notion of a chemical reaction system ( crs ) , the following basic notation and definitions are useful : * let @xmath0 be a set of _ molecule types _ : each element @xmath1 represents a different type of molecule .
* let @xmath2 be a _ food set _
, containing molecule types that are assumed to be freely available in the environment .
* let @xmath3 be a chemical _ reaction _ , transforming a set of @xmath4 _ reactants _ ( molecule types @xmath5 ) into a set of @xmath6 _ products _ ( molecule types @xmath7 ) . in principle there is no restriction on the number of reactants or products ,
although in the specific model we use ( see below ) @xmath4 and @xmath6 are at most two .
* let @xmath8 be a set of ( chemically possible ) reactions .
* let @xmath9 and @xmath10 denote , respectively , the set of all reactants of @xmath11 and the set of all products of @xmath11 , and for any subset @xmath12 of @xmath13 , let @xmath14 and @xmath15 .
* let @xmath16 be a _
catalysis set _
, i.e. , if the molecule - reaction pair @xmath17 then molecule type @xmath18 catalyses reaction @xmath11 . a _ chemical reaction system _ ( or , equivalently , a _ catalytic reaction system
_ ; crs ) is now defined as a tuple @xmath19 consisting of a set of molecule types , a set of ( possible , or allowed ) reactions , and a catalysis set .
based on @xcite , we can visualise a crs as a _ reaction graph _ with two types of vertices ( molecules and reactions ) and two types of directed edges ( from molecules to reactions and vice versa , and from catalysts to the reactions they catalyse ) .
informally , a subset of reactions @xmath12 is an raf ( reflexively - autocatalytic and @xmath20-generated ) set if it satisfies the following property : 1 cm every reactant of every reaction in @xmath12 can be built up by starting from @xmath20 and using just reactions in @xmath12 , and so that all reactions are eventually catalysed by at least one molecule that is either a product of some reaction in @xmath12 or is an element of @xmath20 . to define an autocatalytic set more formally , we first need to define the notion of `` closure ''
informally , the closure of a set of molecule types relative to a set of reactions , is the initial set of molecule types together with all the molecule types that can be created from it by repeated application of reactions from the given set of reactions .
more formally , given a crs @xmath19 , the _ closure _ @xmath21 of @xmath22 relative to @xmath23 is the ( unique ) minimal set @xmath24 that contains @xmath25 and satisfies the condition that , for each reaction @xmath26 ( with @xmath27 being a set of reactants and @xmath28 a set of products ) , @xmath29 notice that when @xmath30 the set @xmath21 is still defined , and it equals @xmath25 .
our mathematical definition of raf sets is now as follows ( note that this is the definition from @xcite , which is slightly modified from the original definition in @xcite ) . given a crs @xmath19 and a food set @xmath2 , a non - empty subset @xmath23 is said to be : * _ reflexively autocatalytic _ if , for all reactions @xmath31 , there is at least one molecule type @xmath32 such that @xmath17 ; * _ @xmath20-generated _ if @xmath33 ; * _ reflexively autocatalytic and @xmath20-generated _ ( raf ) for @xmath34 if @xmath12 is both reflexively autocatalytic and @xmath20-generated .
because the union of rafs for @xmath34 is also an raf for @xmath34 it follows that any crs that contains an raf has a unique maximal raf called the ` maxraf ' ; any other raf is called a ` subraf ' of this maximal raf .
we say that an raf is an _ irreducible raf _ ( or , more briefly , an ` irrraf ' ) if no proper subset is also an raf .
in contrast to the uniqueness of the maximal raf , there may be many ( indeed exponentially many ) irrrafs @xcite .
we have already defined the concept of being@xmath20-generated , however , it will be useful to explore this further for the following reasons : * to better understand the distinction between rafs and ` pseudo - rafs ' ( defined shortly ) ; * to explain the link between @xmath20-generated sets and ` organisations ' in chemical organisation theory ; * to provide a characterisation that we will require later in the proof of our main stochastic theorem ( theorem [ nosmall ] ) . given a crs @xmath35 and a food set @xmath20 , the closure set @xmath36 has two further equivalent descriptions .
firstly , it is the intersection of all subsets of @xmath37 that contain @xmath20 and that are closed relative to @xmath12 .
it also has an explicit constructive definition as follows : @xmath36 is the final set @xmath38 in the sequence of nested sets @xmath39 where @xmath40 is equal to the union of @xmath41 and the set of products of reactions in @xmath12 whose reactants lie in @xmath41 , and where @xmath42 is the first value of @xmath43 for which @xmath44 . with this in hand , we now examine the definition of @xmath20-generated sets of reactions more closely .
recall from the earlier definitions that a subset of reactions @xmath12 is @xmath20-generated provided that every reactant of every reaction in @xmath12 lies in @xmath36 .
note that saying @xmath12 is @xmath20-generated implies but is * strictly stronger * than the condition that the reactant of each reaction in @xmath12 is either a molecule in @xmath20 or it is a product of another reaction in @xmath12 .
@xmath20-generated is also strictly stronger than requiring that the molecules of @xmath37 that are ` used up ' in maintaining the reactions in @xmath12 is precisely @xmath20 .
an example that demonstrates both these strict containments is provided in fig .
[ figure_x ] for the set @xmath45 , which is not @xmath20-generated ( since @xmath46 .
we now provide precise characterizations of when a set of reactions is @xmath20-generated .
[ lemequiv2 ] given a crs @xmath35 , a food set @xmath20 and a non - empty subset @xmath12 of @xmath13 , the following are equivalent : * @xmath12 is @xmath20-generated .
* @xmath47 .
* @xmath12 has a linear ordering @xmath48 so that the reactants of @xmath49 are molecules in @xmath20 , and for each @xmath50 the reactants of @xmath51 are contained in @xmath52 * @xmath12 has a linear ordering @xmath48 so that for each @xmath53 each reactant of @xmath51 is either an element of @xmath20 or is a product of some reaction @xmath54 where @xmath55 . _
proof : _ the equivalence @xmath56 is from @xcite ( lemma 4.3 ) and the equivalence @xmath57 is easily verified , as the ordering of @xmath13 that applies for either part , also works for the other ( from the definitions ) .
thus , to establish this four - way equivalence , it suffices to show that @xmath58 , and @xmath59 . to establish @xmath60 ,
suppose that @xmath12 is @xmath20-generated .
we construct an ordering satisfying @xmath61 as follows : let @xmath62 denote the reactions in @xmath12 that have their reactants in @xmath20 , and for @xmath63 let @xmath64 denote the reactions in @xmath12 that have their reactants in the set @xmath65 , where @xmath66 is the sequence of nested sets described in the preamble to this lemma . then take any ordering on @xmath12 for which the reactions in @xmath64 all come before @xmath67 for @xmath68 .
this ordering satisfies the property described in part @xmath61 . to establish @xmath59
, we only need to observe that @xmath69 for all @xmath70 , so if @xmath71 is a subset of the first set , it is necessarily a subset of the second set .
this completes the proof of lemma [ lemequiv2 ] .
we now point out a consequence of this lemma that sheds some light on why the subset @xmath12 in fig .
[ figure_x ] fails to be @xmath20-generated . given a crs @xmath35 , a food set @xmath20 and a subset @xmath12 of @xmath13 , consider the directed graph @xmath72 that has vertex set @xmath12 and an arc from @xmath11 to @xmath73 precisely if there is a reactant @xmath18 of @xmath73 that is a product of @xmath11 and , in addition , if @xmath74 .
this last condition states that molecule @xmath18 can not be built up from @xmath20 using only the reactions in @xmath12 that do not include @xmath11 .
note that a vertex of @xmath72 is permitted to have a loop ( i.e. an arc from a reaction to itself ) . as an example of this graph , for the reactions shown in fig .
[ figure_x](a ) , @xmath72 is a directed three - cycle , while in part ( b ) of that figure , @xmath72 has no directed cycle .
[ corol ] given a crs @xmath35 , a food set @xmath20 , a non - empty subset @xmath12 of @xmath13 is @xmath20-generated if and only if the following two conditions hold : * every reactant of a reaction in @xmath12 is either an element of @xmath20 or is a product of some reaction in @xmath12 ; and * the graph @xmath72 has no directed cycle ( including loops ) . _
proof : _ suppose that @xmath12 is @xmath20-generated .
then condition ( a ) in the theorem follows by part ( ii ) of lemma [ lemequiv2 ] ; moreover , there exists an ordering @xmath75 of @xmath12 that satisfies the condition described in part ( iii ) of that lemma .
now , if @xmath76 is an arc in @xmath72 , we must have @xmath77 , since otherwise , if @xmath78 , part ( iii ) of lemma [ lemequiv2 ] gives : @xmath79 and the containment @xmath80 would preclude the arc @xmath76 from @xmath72 .
so , if @xmath72 had a directed cycle @xmath81 , we would have : @xmath82 , a contradiction .
thus if @xmath12 is @xmath20-generated , condition ( b ) in the theorem also holds .
conversely , suppose that @xmath12 satisfies conditions ( a ) and ( b ) .
we first show that there exists a reaction @xmath83 that has all its reactants in @xmath20 , i.e. @xmath84 .
suppose to the contrary that this were not the case ( we will show this contradicts condition ( b ) ) .
then for every reaction @xmath11 in @xmath12 , we can select a molecule @xmath85 that is a reactant of @xmath11 .
moreover , by property ( a ) and the condition that @xmath85 it follows that @xmath86 is the product of some other reaction , which we will write as @xmath87 .
thus , starting with any given reaction , @xmath88 , consider the alternating sequence of molecules and reactions @xmath89 that we generate from @xmath88 by setting @xmath90 and @xmath91 .
since @xmath12 is finite , this sequence must have @xmath92 for some @xmath93 .
moreover , we can not have @xmath94 for all @xmath95 $ ] ; otherwise , in the graph @xmath72 , there would be an arc from @xmath96 to @xmath97 for all @xmath98 $ ] and so we would obtain a directed cycle in @xmath72 , and by part ( b ) , no such cycle exists .
this contradiction ensures there exists some molecule @xmath99 for @xmath98 $ ] for which @xmath100 .
however , if the closure of @xmath20 under _ any _ set of reactions contains a molecule outside of @xmath20 , then some reaction in the collection must have all its reactants in @xmath20 ( by lemma [ lemequiv2 ] ) .
this justifies our claim that there is a reaction @xmath83 with @xmath84 .
we now use induction on @xmath101 to establish that conditions ( a ) and ( b ) imply that @xmath12 is @xmath20-generated . for @xmath102 and the non - existence of a loop from this reaction to itself ( by ( b ) ) , we see that @xmath12 is @xmath20-generated .
therefore suppose that the implication holds for any @xmath103 satisfying ( a ) and ( b ) , and that we have @xmath104 .
now , consider @xmath105 and @xmath106 , where @xmath107 is the reaction in @xmath12 with @xmath84 .
notice that @xmath108 satisfies property ( a ) .
moreover , we claim that property ( b ) also holds for @xmath108 since if @xmath109 is an arc of @xmath110 then it is also an arc of @xmath72 . to verify this ,
observe that if @xmath109 is an arc of @xmath110 then there exists a reactant @xmath18 of @xmath73 that is a product of @xmath11 and for which @xmath111 .
however : @xmath112 and so @xmath113 , which implies that @xmath114 is indeed an arc of @xmath72 .
consequently , the arcs of @xmath110 are a subset of the set of arcs of @xmath72 that do not contain @xmath107 and so @xmath110 can not contain a directed cycle ( or else @xmath72 would ) .
thus , since @xmath108 satisfies properties ( a ) and ( b ) , it follows ( by the induction hypothesis ) that @xmath108 is @xmath115-generated , and this implies that @xmath12 is @xmath20-generated .
this completes the proof of the converse result .
chemical organisation theory ( cot ) @xcite provides another way to study chemical reaction systems , and the concept of a ( chemical ) _ organisation _ shares two key properties with rafs : closure and self - maintenance ( for precise definitions , see @xcite , and for recent relevant results , see @xcite and @xcite ) . however , the latter concept ( ` self maintenance ' ) is defined somewhat differently : while rafs require the property of being @xmath20-generated , an organisation is defined as self - sustaining under chemical dynamics , as encoded by the stoichiometric matrix . more precisely ,
if @xmath116 is the stoichiometric matrix for the system , with its rows indexed by molecules and its columns by reactions , then self - maintenance requires a column vector @xmath117 with strictly positive coordinates for which : @xmath118 in words , this is the condition that the reactions can proceed at positive rates , so that the net rate of production of each molecule in the system is not less than the rate at which it is used up ( otherwise such a molecule would disappear from the system ) .
this is a weaker requirement than being @xmath20-generated , since self - maintenance requires only that the system be self - sustaining once it exists , but does not address the question of whether the system could form in the first place from a set of molecules in @xmath20 ; we describe an example to illustrate this shortly .
a second difference is that organisations allow but do not explicitly require reactions to be catalysed , though an extension to allow this has been discussed recently in @xcite .
note that rafs easily extend to allow some reactions not to be catalysed by introducing a putative new element of @xmath20 to act as a catalyst for any reactions that otherwise do not require catalysis .
a third important difference is algorithmic and we will discuss this shortly ( a further minor difference is that organisations are subsets of molecules , while an raf is a subset of reactions and molecules ) .
the following lemma shows that there is a close but not identical relationship between @xmath20-generated sets and organisations ; part ( i ) was discovered by @xcite .
given a crs @xmath119 and food set @xmath20 , consider the set @xmath120 of reactions that formally generate @xmath20 without using other molecules in @xmath37 .
* if @xmath12 is @xmath20-generated then the set of molecules @xmath36 forms an organisation , for the reactions @xmath121 .
* it is possible for a set @xmath122 of molecules to form an organisation for a set of reactions @xmath121 but for @xmath12 to fail to be @xmath20-generated .
part ( i ) of the lemma was established in @xcite ( corollary 1 ) . here , we show how it also follows as a consequence of lemma [ lemequiv2 ] .
firstly , if @xmath12 is @xmath20-generated , then it is closed by the implication ( i ) @xmath123 ( ii ) in lemma [ lemequiv2 ] .
moreover , we may order the reactions in @xmath121 so that the reactions in @xmath124 come first ( in any order ) and so that the order of the subsequent reactions from @xmath12 is such that the reactants of each reaction are either elements of @xmath20 or products of earlier reactions the existence of such an ordering for @xmath12 is provided by the implication ( i ) @xmath123 ( iv ) in lemma [ lemequiv2 ] . consider the corresponding stoichiometric matrix @xmath116 .
then the first non - zero element in each row of @xmath116 is + 1 .
now for _ any _ real matrix with this last property , there is a strictly positive column vector @xmath117 for which @xmath125 , since if @xmath116 has @xmath126 columns , and if the largest absolute value of any negative entry of @xmath116 is @xmath127 then we can take @xmath117 to be the strictly positive vector that has its @xmath43-th coordinate given by : @xmath128 for @xmath129 part ( ii ) is established by considering the example shown in fig .
[ figure_x](a ) with @xmath130 and @xmath45 . ordering @xmath122 as @xmath131 and @xmath121 as @xmath132
, we obtain the following @xmath133 stoichiometric matrix ( rows are indexed by molecules ; columns , by reactions ) : @xmath134\ ] ] it is now clear that @xmath135^t$ ] for the strictly positive vector @xmath136^t$ ] and therefore the self - maintenance inequality ( [ sm ] ) holds .
since @xmath122 is closed relative to the six reactions , it follows that @xmath122 forms an organisation , but @xmath12 fails to be @xmath20-generated , since @xmath137 .
this completes the proof .
a further difference between cot and raf theory is that determining whether or not a crs contains a non - empty organisation is an np - complete problem ( _ c.f . _
@xcite , section 6.2 ) , while determining whether there exists an raf ( necessarily non - empty ) within any crs can be decided by a polynomial time algorithm .
we describe this now .
the usual raf algorithm ( @xcite ) starts with the full set of reactions and iteratively prunes out reactions until the set stabilises . for completeness
, we describe this explicitly now . given a crs @xmath35 and a food set @xmath20 ,
define the following nested ( decreasing ) sequence of subsets of reactions @xmath139 as follows : * @xmath140 ; and for @xmath141 , * @xmath142 ; * @xmath42 is the first value @xmath43 for which @xmath143 .
it can be shown that if @xmath144 then @xmath13 contains no raf ; otherwise , @xmath145 is the unique maximal raf contained in @xmath13 ( for further details , see @xcite ) . throughout this paper
, we will let @xmath146 denote the terminal set ( @xmath145 ) obtained by applying this process to an arbitrary subset @xmath12 of @xmath13 .
note that an raf @xmath12 for @xmath147 satisfies the following two properties : * every reaction in @xmath12 is catalysed by the product of another reaction from @xmath12 or by an element of @xmath20 ; and * each reactant of every reaction in @xmath12 is either an element in @xmath20 or a product of another reaction in @xmath12 .
we will call any subset @xmath12 of @xmath13 that is non - empty and that satisfies properties ( i ) and ( ii ) a _ pseudo - raf _ for @xmath147 .
not every pseudo - raf is an raf , as the example in fig [ figure_y ] ) shows .
however , pseudo - rafs satisfy some of the properties of rafs ; in particular , the union of two or more pseudo - rafs for @xmath34 is a pseudo - raf for @xmath34 .
it follows that any pair @xmath147 either contains no pseudo - raf ( in which case @xmath147 contains no raf either ) or @xmath147 has a unique maximal pseudo - raf that contains all other pseudo - rafs of @xmath34 as well as the unique maximal raf for @xmath34 .
an analogous algorithm to the raf algorithm applies for constructing the maximal pseudo - raf ( when it exists ) , the only change being that @xmath148 is replaced by @xmath149 in the construction of @xmath67 from @xmath64 ( where @xmath150 is the set of products of reactions in @xmath64 ) .
the raf and pseudo - raf algorithms have a similar flavour to the ` unit propagation ' method of solving the propositional logic problem ` horn - sat ' , and in @xcite we showed that horn - sat can be solved by an extension of the raf algorithm .
although a pseudo - raf can not become established by itself ( since it is not @xmath20-generated ) , it can nevertheless become established in the presence of another raf .
this property is not unique to pseudo - rafs , and we formalise and investigate this notion as follows . given a crs @xmath119 and a food set @xmath20 , we will say that a subset @xmath12 of @xmath13 is a _ co - raf _ for @xmath34 if @xmath12 is a non - empty set for which there exists some raf @xmath151 for @xmath152 , which is disjoint from @xmath12 and whose union with @xmath12 , @xmath153 , forms an raf for @xmath152 .
a simple example of a co - raf is the set @xmath154 in fig .
[ figure - u ] .
informally , a co - raf is a system that may not have enough structure to form an raf by itself , but which another ( disjoint ) raf can co - opt to form a larger raf .
note that a co - raf may fail to be an raf because either a reactant or a catalyst ( or both ) can fail to be in the closure of @xmath20 ; in either case , @xmath151 can provide the missing @xmath20-generated reactant or catalyst .
the relationship between an raf @xmath151 and an associated co - raf @xmath12 is similar to the relationship between a ` viable core ' and an associated ` periphery ' in @xcite .
the requirement that @xmath12 and @xmath151 are disjoint in the definition of a co - raf is not a serious restriction , since if @xmath155 is an raf for @xmath34 , where @xmath12 overlaps ( but is not strictly contained within ) an raf @xmath151 , then @xmath156 is a co - raf for @xmath147 .
determining whether a given subset @xmath12 of @xmath13 is a co - raf for @xmath147 can be solved in polynomial time by virtue of the following result ( the equivalence of parts ( i ) and ( ii ) ) .
we also give two other alternative descriptions of co - rafs .
the proofs of these results are presented in the appendix .
[ proprop ] given a crs @xmath157 and a food set @xmath20 , let @xmath12 be a non - empty subset of @xmath13 .
the following are equivalent : * @xmath12 is a co - raf for @xmath147 ; * @xmath158 and @xmath159 is an raf for @xmath147 ; * @xmath160 * @xmath12 is an raf for @xmath161 where @xmath162 , for some raf @xmath151 for @xmath147 that is disjoint from @xmath12 .
note that the equivalence @xmath163 provides a simple way to generate co - rafs : any non - maximal raf @xmath164 for @xmath152 has a co - raf ; simply let @xmath165 be the maximal raf , and take @xmath166 .
given a crs @xmath35 and a food set @xmath20 , we can find an irrraf efficiently ( i.e. in polynomial time ) , but finding a minimal - sized raf is much harder we will show it is an np - hard problem even to determine this minimal size .
nevertheless , it is possible to test whether a given irrraf for @xmath147 is the only irrraf for @xmath147 and if it is , then it is necessarily a minimal sized raf . more generally , if we generate irrrafs for @xmath147 ( each in polynomial time ) , and have found only a relatively small number of them ( e.g. @xmath167 or so ) then it is possible to test whether these are the only irrrafs for @xmath34 and , if so , the one(s ) of smallest size are the minimal - sized rafs for @xmath147 .
we now show how this can be solved efficiently ( in polynomial time ) , provided that we bound the number of irrrafs .
suppose that a crs @xmath168 with a food set @xmath169 , has an raf .
let @xmath170 be a collection of distinct irrrafs that have been constructed from this raf ( e.g. by our search algorithm ) .
we would like to be able to determine whether these are _ all _ the irrrafs for @xmath171 .
the following result provides a way to do this for moderate values of @xmath172 .
recall that for a subset @xmath12 of @xmath13 , @xmath146 is the result of applying the raf algorithm to @xmath12 .
suppose that a crs @xmath168 with a food set @xmath20 , has an raf .
then a collection @xmath173 of distinct irrrafs constitutes the set of all the irrrafs for @xmath171 if and only if the following condition holds : for all @xmath174 , we have @xmath175 _ proof : _ suppose first that for some @xmath174 we have @xmath176 .
then @xmath177 is an raf and so it contains at least one irrraf , say @xmath12 . since @xmath178 @xmath12 can not equal @xmath64 for any @xmath43 , since @xmath12 does not contain @xmath97 ,
but @xmath64 does .
thus , @xmath173 does not constitute the set of all irrrafs of @xmath171 .
conversely , suppose that @xmath173 is not the set of all irrrafs .
let @xmath12 be any other irrraf .
then @xmath64 is not strictly contained within @xmath12 for any @xmath43 because otherwise @xmath12 would not be an irrraf .
thus for each @xmath43 , there exists some reaction @xmath179 and thus a sequence @xmath174 .
now consider @xmath177 . since @xmath12 is a subset of @xmath180
, it follows that @xmath181 and so @xmath182 .
this completes the proof .
* remark : * for any given value of @xmath172 , determining whether or not we have all the irrrafs can be solved in polynomial time ( in the size of the crs ) .
of course , the exponent in the polynomial involves @xmath172 , so it would also be interesting to see if this exponential dependency on @xmath172 can be removed ( and , if not , whether the problem is fixed parameter tractable in @xmath172 ) . given a crs and a food set @xmath147 , finding a largest raf can be solved by a polynomial time algorithm .
this raises an obvious question : is there an efficient way to find the _
raf for @xmath34 , or at least to calculate its size ?
a related question replaces ` smallest raf ' with ` smallest irrraf ' , but it is clear that any smallest raf must also be irreducible so the two questions are equivalent .
consider then the decision problem : [ npthm ] * the decision problem min - raf is np - complete .
* counting the number of sub - rafs ( or smallest sub - rafs ) of an arbitrary raf is # p - complete .
the proof of this theorem involves a reduction of min - raf to the graph theory problem vertex cover , by associating with each crs a graph that has its vertex covers of size @xmath42 in one - to - one correspondence with the sub - rafs of the crs of size @xmath42+constant .
the details of the construction and the full proof of theorem [ npthm ] are provided in the appendix . in the light of theorem [ npthm ]
, an interesting question is whether we can efficiently compute lower bounds on the size of an raf .
the first lower bound is easily computed .
consider a catalytic reaction system @xmath152 and a food set @xmath20 .
let @xmath183 then every raf for @xmath147 has size at least @xmath184 .
_ proof : _ let @xmath12 be an raf for @xmath147 .
suppose that @xmath185 .
if @xmath186 then @xmath187 , which is not possible , since an raf is non - empty , by definition .
thus , @xmath188 .
since this holds for all @xmath31 it follows that @xmath189 , and so @xmath190 this completes the proof . part ( ii ) of the following lemma provides a further computable lower bound on the smallest raf , if we require the raf to have the additional property that none of its reactions are catalysed by a food molecule . given a crs @xmath152 ,
let @xmath191 be the graph with the vertex set @xmath12 and with an arc from reaction @xmath11 to reaction @xmath73 precisely if some product of @xmath11 is a catalyst of @xmath73 . part ( i ) of the lemma is essentially the ` loop theorem ' of @xcite ( theorem 2 ) .
[ nof ] consider a catalytic reaction system @xmath119 and a food set @xmath20 .
* if @xmath12 is an raf for @xmath147 and no reaction in @xmath12 is catalysed by any food molecule then @xmath191 contains a directed cycle . *
provided that @xmath192 ( i.e. @xmath193 has an raf ) , the smallest raf for @xmath147 for which no reaction is catalysed by a food molecule is at least as large as the length of the shortest directed cycle in @xmath194 , and this can be computed in polynomial time in the size of @xmath152 .
_ proof : _ _ part ( i ) : _ a classic , elementary result ( _ c.f .
_ @xcite proposition 1.4.2 ) states that any digraph that has no vertex of in - degree 0 must have a directed cycle
. now if @xmath31 then @xmath11 has a catalyst in @xmath36 and so this catalyst is either the product of some reaction in @xmath12 or it is in @xmath20 .
however , the latter possibility is ruled out by the stated assumption concerning @xmath12 .
thus each vertex of @xmath191 has positive in - degree and so this digraph has a directed cycle . _
part ( ii ) : _ suppose @xmath12 is the smallest raf for @xmath34 . from part ( i ) @xmath12 contains a directed cycle of some length @xmath172 , so @xmath195 .
moreover , since @xmath196 , @xmath172 is at least the size of the smallest directed cycle in @xmath194 , as claimed .
moreover , since @xmath197 can be computed in polynomial time ( by the raf algorithm from section [ rafalg ] ) , and thus @xmath194 can be also , one can find the shortest directed cycle in this graph by an application of the floyd warshall algorithm , or via dijkstra s algorithm ( see , for example , @xcite ) .
to investigate the issue of the smallest rafs empirically , we used the _ binary polymer model _ to collect statistics on the sizes of raf and irrraf sets .
this model has all binary sequences of length at most @xmath4 as its molecules , and the reactions consist of ligation reactions ( joining two sequences to form a longer sequence ) , together with the reversal of this operation ( cleavage reactions , in which a sequence is split into two subsequences )
. examples of ligation and cleavage reactions are @xmath198 and @xmath199 , respectively . in this model , a ligation reaction and its associated cleavage reaction are often regarded as the same ( reversible ) ` cleavage - ligation ' reaction .
we let @xmath200 denote this set of cleavage - ligation reactions , and for a subset @xmath12 of @xmath201 , the set @xmath202 will be taken to be the set of of products of the cleavage and ligation reactions associated with @xmath12 ( and so the closure of @xmath20 relative to @xmath12 is the closure of @xmath20 relative to the union of the associated cleavage and ligation reactions ) . in the simplest form of this model ,
each molecule @xmath18 catalyses any given cleavage - ligation reaction @xmath11 independently with probability @xmath203 , which depends on @xmath4 . the food set @xmath20 is usually chosen to be all binary sequences of length at most @xmath204 for a small value of @xmath204 ( typically , @xmath205 , in which case @xmath206 ) . in previous work
, we already studied how the probability of raf sets existing in this model scales with the value of @xmath4 ( the maximum length of molecules ) . here
, we simply chose one value ( @xmath207 ) and computed the sizes of raf sets for various values of @xmath208 ( the probability that a given molecule catalyses a given cleavage - ligation reaction ) or , equivalently , the level of catalysis @xmath209 ( the average number of reactions catalysed per molecule ) . fig .
[ fig : rafsize ] shows the average sizes of raf sets ( black squares ) and irrraf sets ( crosses ) for increasing levels of catalysis .
these data points are averages over 1000 instances of the model for each value of @xmath208 . when the level of catalysis is too low ( @xmath210 ) , no raf sets are found at all , i.e. , their sizes are equal to zero . however , at a level of catalysis just above @xmath211 , the first raf sets are starting to show up .
initially , they are found in only 6 out of 1000 model instances , but with increasing levels of catalysis @xmath212 , they become more and more frequent , and their sizes seem to increase linearly with @xmath212 .
in contrast , the average size of irrrafs remains constant ( for each non - empty raf set , one ( arbitrary ) irrraf set was generated ) as the rate of catalysis increases across this narrow interval . in the binary polymer model . ]
an interesting feature of fig .
[ fig : rafsize ] is that the sizes of the raf sets when they first start appearing ( around @xmath211 ) are already quite large : 1222 reactions on average in the six raf sets and 624 reactions in the corresponding six irrraf sets ( with @xmath213 for the full reaction set ) .
so , it seems there are no `` small '' rafs when they are only just starting to appear .
this observation is formalised in the following theorem , which shows that at the catalysis levels at which rafs have a moderate probability of occurring , the smallest rafs have a size that grows exponentially with @xmath4 .
[ threshold catalysis rafs have exponential size in @xmath4 ] [ nosmall ] consider the binary polymer model @xmath215 for sequences up to length @xmath4 .
select any fixed value @xmath216 and then select the catalysis probability @xmath203 so that @xmath217 then , for any constant @xmath218 : @xmath219 as @xmath220 . _
proof : _ for any subset @xmath12 of @xmath201 with @xmath221 we have @xmath222 , and so the probability that an arbitrary reaction @xmath31 is catalysed by at least one element of @xmath36 is at most @xmath223 .
consequently , if , in addition , @xmath12 has size @xmath172 , the probability that @xmath12 is reflexively autocatalytic is at most @xmath224 now , we can provide a further upper bound on this last probability by an expression that involves just @xmath172 ( and not @xmath225 ) by observing that : @xmath226^k,\ ] ] by noting that @xmath227 , since each reaction in @xmath12 is associated with at most three distinct molecules . in summary , the probability that any subset @xmath12 of @xmath201 of size @xmath172 is reflexively autocatalytic
is , at most : @xmath228^k.\ ] ] let @xmath229 be the number of subsets of @xmath201 of size @xmath172 that are @xmath20-generated .
boole s inequality , combined with the upper bound ( [ upper_bound ] ) , implies that the probability that @xmath215 has an raf of size @xmath172 is bounded above by : @xmath230^k$ ] .
thus : @xmath231^k.\ ] ] now , the value of @xmath232 for which @xmath233 is bounded above by @xmath234 for some value @xmath235 dependent only on @xmath236 ( by @xcite [ theorem 4.1 ] , and @xcite [ proposition 8.1 ] ) .
thus : @xmath237^k.\ ] ] now , by lemma [ lemequiv2 ] , any set of reactions is @xmath20-generated if and only if the reactions can be linearly ordered so that every reaction in the sequence has its reactants provided either from @xmath20 or from the products of earlier reactions in the sequence ( or both ) .
therefore , @xmath229 is bounded above by the collection of ordered sequences @xmath238 where , for all @xmath239 : * @xmath240 is a cleavage or ligation reaction involving one or two ( respectively ) molecules of @xmath241 ( taking @xmath242 ) .
now , each reaction in the sequence @xmath238 creates , at most , two new molecules , and so @xmath243 for all @xmath244 .
since @xmath245 , we have for all @xmath246 : @xmath247 ) , the number of possible choices for @xmath240 to satisfy condition ( * ) above is , at most : @xmath248 since the first term in this sum is an upper bound on the number of possible ligation reactions , while the second term is an upper bound on the number of cleavage reactions .
combining this with ( [ ups ] ) gives the following upper bound on the number of sequences @xmath238 satisfying ( * ) .
@xmath249 \leq \left[(|f|+2k)(n+|f|+2k)\right]^k \leq ( n+|f|+2k)^{2k},\ ] ] and so @xmath250 applying this inequality to ( [ peq ] ) , with the asymptotic equivalence @xmath251 , gives : @xmath252^k.\ ] ] notice that we can provide an upper bound for the term on the right by the expression : @xmath253^k = \theta/(1-\theta),\ ] ] where @xmath254 $ ] .
it follows that if @xmath255 for @xmath256 , then @xmath257 ( and thereby @xmath258 ) converges to zero as @xmath220 , and therefore so too does the expression for the probability in ( [ sumeq ] ) .
this completes the proof .
* comments * * this result is interesting in the light of theorem 11 of @xcite , as the probability that the length of a first cycle is @xmath172 when a first cycle appears in a random digraph is @xmath259 , and so short cycles have considerable probability in that model .
+ by contrast , when the first rafs appear , there are no small ones , since any raf requires the simultaneous satisfying of two properties : it must be reflexively autocatalytic and also @xmath20-generated ; the former property is equivalent to the existence of a directed cycle in the catalysis graph ( at least in the case @xmath260 for @xmath261 ) ; while there might be a small cycle , it is unlikely to be @xmath20-generated .
* theorem [ nosmall ] provides an interesting complement to the earlier theorem [ npthm ] , which showed that there is , in general , no efficient way to determine the size of the smallest raf in a crs .
thus , it could be difficult to exclude the possibility a small raf in the binary polymer model for large values of @xmath4 , by searching for the smallest irrrafs .
however , theorem [ nosmall ] provides a theoretical guarantee that , with high probability , there will be no small rafs when they first appear within this model . *
the final inequality in the proof of theorem [ nosmall ] allows us to place explicit bounds on the likely minimal size of rafs for finite values of @xmath4 .
for example , for @xmath262 , the probability that there exists an raf of size 1000 when the existence of an rafs has a probability of 0.5 is less that 0.01 ( taking @xmath206 and the conservative value for @xmath235 of @xmath263 from theorem 4.1(ii ) of @xcite ) . *
it is easy to show that when the rate of catalysis becomes sufficiently large , we will expect to find small rafs in the binary polymer model .
thus the initially largely flat line for irrraf sizes in fig .
[ fig : rafsize ] must eventually decrease to small values ( in the limit of size 1 ) as the rate of catalysis continues to increase .
moreover , small catalytic reaction systems ( of size 16 ) that form rafs ( and which contain even smaller rafs ) have recently been discovered in real rna replicator systems @xcite . that such small sets form rafs can be partly explained by the high catalysis rate @xcite . with theorem [ npthm ] above , we proved that finding the smallest ( irr)raf set is a hard problem , so we can not hope to have a polynomial time algorithm to do this .
however , it is still possible to get an idea of the distribution of the sizes of the irrraf sets that exist inside an raf set .
this can be done as follows . in @xcite
, we described a polynomial time algorithm for finding one possible irrraf in a given raf @xmath12 by removing one reaction @xmath97 from @xmath12 and applying the raf algorithm to the set @xmath264 .
if this results in an empty set ( @xmath265 ) , then reaction @xmath97 is essential and needs to remain in @xmath12 . otherwise ,
replace @xmath12 by the non - empty subraf @xmath266 .
now repeat this procedure with every next reaction @xmath97 in @xmath12 until all reactions have been considered .
the result of this is an irrraf of @xmath12 .
this algorithm was used to generate the data on irrraf sizes in fig . [
fig : rafsize ] .
note that the particular irrraf in @xmath12 that is found by this algorithm depends on the order in which the reactions @xmath267 are considered for possible removal .
so , by repeating the above algorithm a number of times and randomly re - ordering the reactions in @xmath12 each time , we can generate a sample of irrrafs of @xmath12 .
[ fig : irrraf ] shows two histograms of the sizes of 1000 irrrafs generated this way from two of the raf sets that were found at a level of catalysis of about @xmath211 , i.e. , when raf sets are just starting to show up . in both cases ,
the sample is dominated by one particular irrraf size , with the rest being relatively close in size , although the histogram on the right shows a case where the smallest irrraf is about 100 reactions smaller than the dominant one . since this is only a random sample ,
there is no guarantee that this is indeed the _ smallest _ irrraf .
however , the fact that even the smallest irrraf in these samples is still rather large ( close to 600 reactions ) is probably a good indication that , indeed , there are no small rafs when they just start appearing .
raf theory provides a way to address one aspect of the complex question , how did life arise ?
the existence of rafs does not represent a sufficient condition , but it would seem to be a necessary one . moreover , the approach is sufficiently general that it can be applied to other emergence phenomena both inside chemistry and in quite disparate fields ( for an application to a ` toy ' problem in economics , see @xcite ) .
rafs are based on two key ideas every molecule must be able to be built up from the available set of ` food ' molecules by reactions from the set , and each reaction must ` eventually ' be catalysed . here ` eventually ' refers to the fact that some reactions may need to proceed uncatalysed ( at a lower rate ) in order to get the system going , but eventually , all reactions are catalysed .
a stronger requirement would be that all reactions must be catalysed by the available molecules as the system develops ( from the food molecules or products of reactions that have already occurred ) .
this notion of a ` constructively autocatalytic @xmath20-generated ' ( caf ) set from @xcite seems an unnecessarily strong condition ( since reactions can generally proceed , at a lower rate , without catalysis ) and the mathematical properties of cafs ( and the probability that they form ) are quite different from rafs @xcite .
a weaker requirement is that only some reactions need to be catalysed this fits perfectly easily within the current raf framework , as we may simply formally allow a food molecule to act as a putative catalyst for those reactions .
another weakening of the raf concept is to consider a closed chemical reaction system , which , once established , will continue to be self - maintaining . this underlies the notion of an ` organisation ' in chemical organisation theory .
the property of rafs of being @xmath20-generated was shown in @xcite to imply the property of being an organisation ; we have shown here that the converse need not hold in other words , an organisation may not be able to be built starting just with the food set , without the presence of some other reactant to get it started .
this property of an organisation has a superficial similarity to the property that a raf can allow one or more some reactions to proceed uncatalysed until the catalyst is formed .
however , there is an important difference , since an uncatalysed reaction can proceed ( at a lower rate ) , while this a reaction that lacks one of its reactants can not take place .
the focus of this paper has been on small rafs , as these are , in some sense , the ` simplest ' systems that could be of interest in origin - of - life studies .
it is of interest to know whether within some crs that harbours an raf , there is a very small one present , or instead whether all subrafs are quite large .
the smallest rafs are irreducible , though not all irreducible rafs are of the smallest size .
in contrast to the maximal rafs , where there is a unique object ( maxraf ) that can be constructed in polynomial time ( by the raf algorithm ) , there may be exponentially many irreducible rafs , and finding a smallest raf is , in general , np - hard .
nevertheless , we can find irrrafs in polynomial time , and we can describe computable lower bounds on the size of irrrafs and also determine if a given ( small ) collection comprises all the irrrafs .
it is also of interest to consider the size and distribution of rafs in simple settings such at the binary polymer model , where simulations suggest that when rafs first appear , small irrrafs are unlikely , a result that has been verified formally in theorem [ nosmall ] .
however , as the level of catalysis increases , one is guaranteed to eventually find small irrrafs .
an interesting problem for future work would be to develop better bounds and approximations for the minimal size of a raf within a catalytic reaction system . for example , is it possible to obtain a bound for the size the smallest raf that is within some constant factor of optimal ?
it would also be of interest to investigate an extension of rafs that allow some molecules to not only catalyse some reactions , but also to inhibit other reactions ; in this case determining whether an analogue of an raf exists within an arbitrary crs has been shown to np - hard @xcite , but in certain cases the raf algorithm can be adapted to solve this problem @xcite .
we thank the _ allan wilson centre for molecular ecology and evolution _ for helping fund this work .
99 ashkenasy , g. , jegasia , r. , yadav , m. and ghadiri , m.r .
design of a directed molecular network . pnas .
101(30 ) : 1087210877 .
bang - jensen , j. and gutin , g. ( 2001 ) .
digraphs : theory , algorithms and applications .
springer - verlag .
bollobas , b. and rasmussen , s. ( 1989 ) .
first cycles in random directed graph processes .
discrete mathematics 75 : 5568 .
bonchev , d. and mekenyan , o. ( 1994 ) .
graph theoretical approaches to chemical reactivity .
centler , f. , kaleta , c. speroni di fenizio , p. and dittrich , p. ( 2008 ) . computing chemical organisations in biological networks .
bioinformatics 24 ( 14 ) : 16111618 .
contreras , d.a .
, pereira , u. , hernndez , v. , reynaert , b. and letelier , j.c .
( 2011 ) . a loop conjecture for metabolic closure . in : advances in artificial life ,
ecal 2011 : proceedings of the eleventh european conference on the synthesis and simulation of living systems .
mit press ; pp . 176183 .
dittrich , p. and speroni di fenizio , p. ( 2007 ) .
chemical organisation theory .
bulletin of mathematical biology .
69 : 11991231 .
dowling , w and gallier , j. ( 1984 ) .
linear - time algorithms for testing the satisfiability of propositional horn formulae .
journal of logic programming , 1(3):267284 .
dyson , f.j .
( 1982 ) . a model for the origin of life .
journal of molecular evolution . 18 : 344360 .
eigen , m. , schuster , p. ( 1977 ) .
the hypercycle : a principle of natural self - organisation .
part a : emergence of the hypercycle .
naturwissenschaften 64 , 541565 .
garey , m. r. and johnson , d. s. ( 1979 ) .
computers and intractability : a guide to the theory of np - completeness .
hayden , e.j . , von kiedrowski , g. and lehman , n. ( 2008 ) .
systems chemistry on ribozyme self - construction : evidence for anabolic autocatalysis in a recombination network .
angewandte chemie international edition , 120 : 8552 - 8556 .
hordijk , w. and steel , m. ( 2004 ) .
detecting autocatalyctic , self - sustaining sets in chemical reaction systems .
journal of theoretical biology 227(4 ) : 451 - 461 .
hordijk , w. , kauffman , s. , and steel , m. ( 2011 ) . required levels of catalysis for the emergence of autocatalytic sets in models of chemical reaction systems .
international journal of molecular sciences ( special issue : origin of life 2011 ) 12 : 30853101 .
hordijk , w. , steel , m. and kauffman , s. ( 2012 ) .
the structure of autocatalytic sets : evolvability , enablement , and emergence .
acta biotheoretica 60 : 379392 .
hordijk , w. and steel , m. ( 2012 ) .
autocatalytic sets extended : dynamics , inhibition , and a generalization .
journal of systems chemistry , 3:5 ( in press ) .
hordijk , w.and steel , m. ( 2012 ) . a formal model of autocatalytic sets emerging in an rna replicator system .
submitted to journal of systems chemistry ( arxiv:1211.3473 ) .
jaramillo , s. , honorato - zimmer , r. ; pereira , u. , contreras , d. , reynaert , b. , hernndez , v. , soto - andrade , j. , crdenas , m. , cornish - bowden , a. ; letelier , j. ( 2010 ) .
( m , r ) systems and raf sets : common ideas , tools and projections . in proceedings of the alife xii conference , odense , denmark ,
1923 august , 2010 ; pp . 94100 .
kauffman , s.a .
cellular homeostasis , epigenesis and replication in randomly aggregated macromolecular systems .
journal of cybernetics 1(1 ) , 7196 .
kauffman , s.a .
autocatalytic sets of proteins .
journal of theoretical biology 119 , 124 kauffman , s.a .
the origins of order .
oxford university press .
kreyssig , p. , escuela , g. , reynaert , b. , veloz , t. ibrahim , b. , dittrich , p. ( 2012 ) .
cycles and the qualitative evolution of chemical systems . plos 7(1 ) : e45772 .
lee , d.h . ,
severin , k. , ghadiri , m.r .
autocatalytic networks : the transition from molecular self - replication to molecular ecosystems .
current opinion in chemical biology , 1 : 491496 .
letelier , j.c .
, soto - andrade , j. , abarzua , f.g .
, cornish - bowden , a. , crdenas , m.l .
organizational invariance and metabolic closure : analysis in terms of ( m;r ) systems .
journal of theoretical biology 238 , 949961 .
mossel , e. and steel , m. ( 2005 ) .
random biochemical networks and the probability of self - sustaining autocatalysis .
journal of theoretical biology 233(3 ) , 327 - 336 .
oparin , a.i .
the pathways of the primary development of metabolism and artificial modeling of this development in coacrvate drops . in fox ,
the origins of pre - biological systems and of their molecular matrices .
sharov , a. ( 1991 ) .
self - reproducing systems : structure , niche relations and evolution .
biosystems , 25 : 237249 .
sievers , d. and von kiedrowksi , g. ( 1994 ) .
self - replication of complementary nucleotide - based oligomers .
369 : 221224 .
taran , o. , thoennessen , o. , achilles , k. and von kierdoswki . g. ( 2010 ) .
synthesis of information - carrying polymers of mixed sequences from double stranded short deoxynunleotides .
journal of systems chemistry , 1(9 ) .
steel , m. ( 2000 ) .
the emergence of a self - catalysing structure in abstract origin - of - life models . applied mathematics letters 13 ( 3 ) : 91 - 95 .
vadhan , s.p .
the complexity of counting in sparse , regular , and planar graphs .
siam journal on computing 31 : 398 427 .
vaidya , n. , manapat , l. , chen , i.a . and xulvi - brunet , r. ( 2012).spontaneous network formation among cooperative rna replicators . nature , 491 : 72 - 77 .
vasas v. , fernando c. , santos m. , kauffman s. , szathmry e. ( 2012 ) .
evolution before genes .
biology direct , 7:1 .
wchterhuser , g. ( 1990 ) .
evolution of the first metabolic cycles .
proceedings of the national academy of sciences usa 87 : 200 - 204 .
zhang , h. and stickel , m. ( 1996 ) .
an efficient algorithm for unit - propagation . in proceedings of the fourth international symposium on artificial intelligence and mathematics ( ai - math96 ) , fort lauderdale ( florida usa ) .
to establish @xmath268 , suppose that @xmath269 and that @xmath270 is an raf for @xmath34 .
@xmath271 is an raf for @xmath272 , where @xmath273 with @xmath274 , and so is certainly an raf for @xmath275 .
furthermore @xmath276 , since @xmath277 .
hence @xmath12 is a co - raf for @xmath34 . to establish @xmath278
suppose that @xmath12 is a co - raf for @xmath34 .
then there exists an raf @xmath151 for @xmath34 , such that @xmath279 and @xmath155 is an raf for @xmath34 .
consider @xmath271 . since @xmath151 is an raf for @xmath34 and
is a subset of @xmath271 , we must have @xmath158 .
it remains to show that @xmath159 is an raf for @xmath34 .
suppose that @xmath280 .
then either @xmath188 , in which case all the reactants of @xmath11 and at least one catalyst are contained in @xmath281 ( since @xmath155 is an raf for @xmath34 ) , while if @xmath282 then all the reactants of @xmath11 and at least one catalyst is contained in @xmath283 ( since @xmath271 is an raf for @xmath34 ) .
now , @xmath155 and @xmath271 are both subsets of @xmath159 , and so @xmath281 and @xmath283 are both subsets of @xmath284 .
consequently , every reaction in @xmath270 has all its reactants and at least one catalyst in @xmath284 , which implies that @xmath270 is an raf for @xmath34 . to establish @xmath60 ,
suppose that @xmath12 is a co - raf for @xmath34 .
then there exists an raf @xmath151 for @xmath34 , such that @xmath287 and @xmath155 is an raf for @xmath34 .
trivially , @xmath288 and clearly @xmath289 , since @xmath12 is non - empty by the definition of a co - raf , so take @xmath290 and @xmath291 . to establish @xmath292 ,
suppose that @xmath12 is a co - raf for @xmath34 .
then there exists an raf @xmath151 for @xmath34 such that @xmath293 and @xmath294 is an raf for @xmath34 .
it suffices to show that @xmath12 is an raf for @xmath295 , where @xmath296 .
first we prove that @xmath12 is generated from @xmath115 : i.e. @xmath297 . let @xmath298 .
@xmath294 is @xmath20-generated , so there exists an ordering @xmath299 of its reactions @xmath300 satisfying part ( iv ) of lemma [ lemequiv2 ] for the food set @xmath20 .
we herein refer to an ordering satisfying part ( iv ) of lemma [ lemequiv2 ] for some food set @xmath20 as a _ proper ordering relative to f_. @xmath151 is @xmath20-generated so there exists a proper ordering relative to @xmath20 , @xmath301 , of its reactions @xmath97 . define @xmath302 to be the ordering of the reactions of @xmath12 obtained by deleting from @xmath303 every reaction that also appears in @xmath304 , preserving the order of the remaining reactions .
we claim that the concatenation @xmath305 , a reordering of @xmath303 , is a proper ordering relative to @xmath20 .
consider any reaction @xmath306 and a reactant @xmath307 .
@xmath294 is @xmath20-generated and @xmath308 , so by part ( ii ) of lemma [ lemequiv2 ] at least one of the following holds : ( i ) @xmath309 , ( ii ) @xmath310 for some @xmath311 , or ( iii ) @xmath312 for some @xmath313 .
if ( i ) alone is true , @xmath73 trivially does not prevent the reordering from being a proper ordering relative to @xmath20 .
if ( ii ) alone is true , every @xmath311 precedes @xmath73 in @xmath305 , so @xmath73 certainly does not prevent the reordering from being a proper ordering relative to @xmath20 .
if ( iii ) alone is true , @xmath314 must precede @xmath73 in @xmath303 and the order of the reactions of @xmath12 in @xmath303 is preserved in @xmath305 , so @xmath314 precedes @xmath73 in @xmath305 .
if more than one of ( i)-(iii ) are true , then since @xmath303 is a proper ordering , at least one of the conclusions will hold , which is sufficient .
therefore our claim that @xmath305 is a proper ordering relative to @xmath20 is justified .
it follows that @xmath315 , and for each @xmath316 , @xmath317 , so moreover @xmath318 alone is a proper ordering relative to @xmath115 .
then , by the implication @xmath319 in lemma 3.1 , @xmath12 is generated from @xmath115 .
it remains to show that @xmath12 is reflexively autocatalytic .
since @xmath12 and @xmath294 are @xmath20-generated , we can apply part ( ii ) of lemma [ lemequiv2 ] to @xmath320 and @xmath321 to deduce that they are equal .
now , since @xmath294 is reflexively autocatalytic then certainly @xmath12 is reflexively autocatalytic ( by definition ) . to establish @xmath319 , it suffices to show that @xmath294 is an raf for @xmath34 , since we already have that @xmath151 is an raf for @xmath34 and @xmath293 .
first we prove that @xmath294 is @xmath20-generated .
@xmath151 is @xmath20-generated , so there exists a proper ordering relative to @xmath20 of its reactions @xmath322 .
similarly for @xmath12 there exists a proper ordering relative to @xmath323 of its reactions @xmath324 .
hence the concatenation @xmath325 is a proper ordering relative to @xmath20 of the reactions in @xmath294 , so @xmath294 is @xmath20-generated .
it remains to show that @xmath294 is reflexively autocatalytic .
since @xmath326 and @xmath294 are each @xmath20-generated , we can apply part ( ii ) of lemma [ lemequiv2 ] to each of @xmath327 , @xmath320 and @xmath321 to deduce that @xmath328 .
now since @xmath151 and @xmath12 are reflexively autocatalytic then certainly @xmath294 is reflexively autocatalytic .
this completes the proof . _
proof : _ min - raf is clearly in the complexity class np , since one can verify in polynomial time if a given subset of @xmath13 has size , at most , @xmath172 and forms an raf .
we will reduce the graph theory problem vertex cover to min - raf . recall that for a graph @xmath329 , a _
vertex cover _ of @xmath330 is a subset @xmath331 of @xmath332 with the property that each edge of @xmath330 is incident with at least one vertex in @xmath331 ; vertex cover has as its instance a graph @xmath329 and an integer @xmath42 and we ask whether or not @xmath330 has a vertex cover of size , at most , @xmath42 .
this is a well - known np - complete problem @xcite ( indeed , one of karp s original 21 np - complete problems ) .
given an instance @xmath333 of vertex cover , we show how to construct an instance @xmath334 , of min - raf for which the answers to the two decision problems are identical .
we first construct @xmath335 and @xmath336 .
for each @xmath337 , let @xmath338 be two distinct elements of @xmath335 and let @xmath339 be an element of @xmath340 .
order @xmath341 as @xmath342 and for each @xmath343 , let @xmath344 be a distinct element of @xmath20 and @xmath345 an element of @xmath340
. let @xmath346 be another distinct element of @xmath335 .
thus @xmath335 consists of the @xmath347 elements : @xmath348 and @xmath340 consists of @xmath349 elements : @xmath350 for each @xmath351 , define a reaction : @xmath352 for each @xmath353 , define the reaction : @xmath354 and for @xmath355 , let : @xmath356 for any subset @xmath357 of @xmath332 let : @xmath358 @xmath359 and set @xmath360 * if @xmath364 ( where @xmath365 ) then @xmath366 is catalysed by both @xmath367 and @xmath368 ( but by no other molecules ) . * in addition , each reaction @xmath369 is catalysed by @xmath370 and by no other molecule - we call the molecule @xmath370 the _ super - catalyst_. * @xmath362 is an raf for @xmath371 . *
a subset @xmath12 of @xmath362 is an raf for @xmath371 if and only if @xmath373 for a vertex cover @xmath331 of @xmath330 .
* the vertex covers of @xmath330 of size @xmath42 are in one - to - one correspondence with the sub - rafs of @xmath362 of size @xmath374 . to establish the second claim ,
suppose that @xmath331 is a vertex cover of @xmath330 .
then every reaction in @xmath375 is catalysed by the product of least one reaction in @xmath376 .
moreover , the product of @xmath377 catalyses all the remaining reactions .
thus , @xmath12 is reflexively autocatalytic , and it is also clear that @xmath12 is @xmath20-generated ; thus @xmath12 is an raf and it has @xmath374 reactions .
conversely , suppose that @xmath108 is an raf for @xmath371 of size at most @xmath374 .
if @xmath378 is not in @xmath108 then the super - catalyst is not produced by any reaction in @xmath108 so none of the reactions in @xmath379 is catalysed ; moreover , because the products from these last reactions provide the only catalysts for @xmath375 it follows that @xmath380 .
thus , since @xmath108 is non - empty ( being an raf ) , @xmath377 must be an element of @xmath108 , and in order to construct the reactants of @xmath381 , all the reactions @xmath375 must form a subset of @xmath108 . in order for all these reactions to be catalysed , at least one of the reactions @xmath382 and @xmath383 must lie in @xmath108 for each @xmath384 .
thus @xmath385 is a vertex cover of @xmath330 and it has size , at most , @xmath386 as claimed .
this establishes the required reduction , and thereby completes the proof of the second claim .
the third claim follows by the noting that the association @xmath387 maps vertex covers of @xmath330 of size @xmath42 onto sub - rafs of @xmath362 of size @xmath374 ( by the previous claim ) and two different vertex covers are mapped to distinct sub - rafs .
this completes the proof .
part ( i ) of theorem [ npthm ] now follows from the first two claims , while part ( ii ) of theorem [ npthm ] follows from the third claim , combined with the # p - completeness of counting vertex covers of a graph and minimum vertex covers of a graph ( see @xcite ) .
* remark : * we have ensured in the proof above that each reaction has just two reactants , in line with the binary polymer model . however , the attentive reader will notice that @xmath20 may have to be quite large .
nevertheless , it is quite straightforward to modify this example so that @xmath20 is kept small ( e.g. of size 6 ) , and to implement the construction within the constraints of the binary polymer cleavage ligation model . | self - sustaining autocatalytic chemical networks represent a necessary , though not sufficient condition for the emergence of early living systems .
these networks have been formalised and investigated within the framework of raf theory , which has led to a number of insights and results concerning the likelihood of such networks forming . in this paper
, we extend this analysis by focussing on how _ small _ autocatalytic networks are likely to be when they first emerge .
first we show that simulations are unlikely to settle this question , by establishing that the problem of finding a smallest raf within a catalytic reaction system is np - hard .
however , irreducible rafs ( irrrafs ) can be constructed in polynomial time , and we show it is possible to determine in polynomial time whether a bounded size set of these irrrafs contain the smallest rafs within a system . moreover , we derive rigorous bounds on the sizes of small rafs and use simulations to sample irrrafs under the binary polymer model .
we then apply mathematical arguments to prove a new result suggested by those simulations : at the transition catalysis level at which rafs first form in this model , small rafs are unlikely to be present .
we also investigate further the relationship between rafs and another formal approach to self - sustaining and closed chemical networks , namely chemical organisation theory ( cot ) . catalytic reaction system , random autocatalytic network , origin of life 1 cm `` individual chemical reactions in living beings are strictly coordinated and proceed in a certain sequence , which as a whole forms a network of biological metabolism directed toward the perpetual self - preservation , growth , and self - reproduction of the entire system under the given environmental conditions '' oparin ( 1965 ) @xcite |
a _ graph property _ @xmath11 of @xmath2-vertex graphs is a collection of graphs on the vertex set @xmath12=\{1,\dots , n\}$ ] that is invariant under relabeling of the vertices .
a property @xmath11 is called _ monotone _ ( decreasing ) if it is preserved under the deletion of edges .
the trivial graph properties are the empty set and the set of all graphs .
a class of examples are the _ forbidden subgraph _ properties : for a fixed graph @xmath3 , let @xmath13 denote the class of @xmath2-vertex graphs that do not contain a ( not necessarily induced ) subgraph isomorphic to @xmath3 .
we view a set of labeled graphs on @xmath2 vertices as a boolean function on the @xmath14 variables describing adjacency .
a boolean function on @xmath0 variables is _ evasive _ if its deterministic query ( decision - tree ) complexity is @xmath0 .
the long - standing aanderaa - rosenberg - karp conjecture asserts that _ every nontrivial monotone graph property is evasive .
_ the problem remains open even for important special classes of monotone properties , such as the forbidden subgraph properties . in this note , @xmath2 always denotes the number of vertices of the graphs under consideration .
aanderaa and rosenberg ( 1973 ) @xcite conjectured a lower bound of @xmath15 on the query complexity of monotone graph properties .
rivest and vuillemin ( 1976 ) @xcite verified this conjecture , proving an @xmath16 lower bound .
kleitman and kwiatkowski ( 1980 ) @xcite improved this to @xmath17 karp conjectured that nontrivial monotone graph properties were in fact evasive .
we refer to this statement as the aanderaa - rosenberg - karp ( ark ) conjecture . in their seminal paper , kahn , saks , and sturtevant @xcite
observe that non - evasiveness of monotone boolean functions has strong topological consequences ( contracibility of the associated simplicial complex ) .
they then use results of r. oliver about fixed points of group actions on such complexes to verify the ark conjecture when @xmath2 is a prime - power . as a by - product
, they improve the lower bound for general @xmath2 to @xmath18 since then , the topological approach of @xcite has been influential in solving various interesting special cases of the ark conjecture .
yao ( 1988 ) @xcite proves that non - trivial monotone properties of bipartite graphs with a given partition @xmath19 are evasive ( require @xmath20 queries ) .
triesch ( 1996 ) @xcite shows ( in the original model ) that any monotone property of bipartite graphs ( all the graphs satisfying the property are bipartite ) is evasive .
chakrabarti , khot , and shi ( 2002 ) @xcite introduce important new techniques which we use ; we improve over several of their results ( see section [ sec : main ] ) .
dirichlet s theorem ( 1837 ) ( cf .
@xcite ) asserts that if @xmath21 then there exist infinitely many primes @xmath22 .
let @xmath23 denote the smallest such prime @xmath24 .
let @xmath25 .
linnik s celebrated theorem ( 1947 ) asserts that @xmath26 for some absolute constant @xmath5 ( cf .
heath - brown @xcite shows that @xmath27 .
chowla @xcite observes that under the extended riemann hypothesis ( erh ) we have @xmath28 for all @xmath29 and conjectures that @xmath30 suffices : for every @xmath31 and every @xmath32 we have @xmath33 .
this conjecture is widely believed ; in fact , number theorists suggest as plausible the stronger form @xmath34 @xcite .
turn @xcite proves the tantalizing result that for almost all @xmath35 we have @xmath36 .
let us call a prime @xmath24 an _ @xmath37-near fermat prime _ if there exists an @xmath38 such that @xmath39 and @xmath40 .
we need the following weak form of chowla s conjecture : for every @xmath31 there exist infinitely many @xmath37-near fermat primes . in other words
, the weak conjecture says that for every @xmath37 , for infinitely many values of @xmath41 we have @xmath42 . for a graph property @xmath43
we use @xmath11 to denote the set of graphs on vertex set @xmath12 $ ] with property @xmath43 .
we say that @xmath43 is _ eventually evasive _ if @xmath11 is evasive for all sufficiently large @xmath2 .
our first set of results states that the `` forbidden subgraph '' property is `` almost evasive '' under three different interpretations of this phrase .
[ thm : forbidden ] for all graphs @xmath3 , the forbidden subgraph property @xmath13 ( a ) is eventually evasive , assuming the weak chowla conjecture ; ( b ) is evasive for almost all @xmath2 ( unconditionally ) ; and ( c ) has query complexity @xmath8 for all @xmath2 ( unconditionally ) .
part ( b ) says the asymptotic density of values of @xmath2 for which the problem is not evasive is zero .
part ( c ) improves the bound @xmath44 given in @xcite .
parts ( a ) and ( c ) will be proved in section [ sec : forbidden ] .
we defer the proof of part ( b ) to the journal version .
the term `` monotone property of graphs with @xmath45 edges '' describes a monotone property that fails for all graphs with more than @xmath32 edges .
[ thm : density ] all nontrivial monotone properties of graphs with at most @xmath46 edges are eventually evasive , where ( a ) under chowla s conjecture , @xmath47 for any @xmath31 ; ( b ) under erh , @xmath48 ; and ( c ) unconditionally , @xmath49 for some constant @xmath9 .
( d ) unconditionally , all nontrivial monotone properties of graphs with no cycle of length greater than @xmath50 are eventually evasive ( for all @xmath31 ) . part ( c ) of theorem [ thm : density ] will be proved in section [ sec : sparse - uncond ] .
parts ( a ) and ( b ) follow in section [ sec : sparse - cond ] .
the proof of part ( d ) follows along the lines of part ( c ) ; we defer the details to the journal version of this paper .
we note that the proofs of the unconditional results ( c ) and ( d ) in theorem [ thm : density ] rely on haselgrove s version @xcite of vinogradov s theorem on goldbach s conjecture ( cf .
[ sec : vinogradov ] ) .
recall that a _
topological subgraph _ of a graph @xmath51 is obtained by taking a subgraph and replacing any induced path @xmath52 in the subgraph by an edge @xmath53 ( repeatedly ) and deleting parallel edges .
a _ minor _ of a graph is obtained by taking a subgraph and contracting edges ( repeatedly ) .
if a class of graphs is closed under taking minors then it is also closed under taking topological subgraphs but not conversely ; for instance , graphs with maximum degree @xmath54 are closed under taking toopological subgraphs but every graph is a minor of a regular graph of degree 3 .
[ cor : top - sub ] let @xmath43 be a nontrivial class of graphs closed under taking topological subgraphs .
then @xmath43 is eventually evasive .
this unconditional result extends one of the results of chakrabarti et al .
@xcite , namely , that nontrival classes of graphs closed under taking minors is eventually evasive .
corollary [ cor : top - sub ] follows from part ( c ) of theorem [ thm : density ] in the light of mader s theorem which states that if the average degree of a graph @xmath51 is greater than @xmath55 then it contains a topological @xmath56 @xcite .
theorem [ thm : density ] suggests a new stratification of the ark conjecture . for a monotone ( decreasing ) graph property @xmath11 , let @xmath57 we can now restate the ark conjecture : if @xmath11 is a non - evasive , non - empty , monotone decreasing graph property then @xmath58
for the basics of group theory we refer to @xcite . all groups in this paper
are finite . for groups
@xmath59 we use @xmath60 to denote that @xmath61 is a subgroup ; and @xmath62 to denote that @xmath61 is a ( not necessarily proper ) normal subgroup .
we say that @xmath63 is a @xmath24-group if @xmath64 is a power of the prime @xmath24 . for a set @xmath65 called the `` permutation domain , '' let @xmath66 denote the _ symmetric group _ on @xmath65 , consisting of the @xmath67 permutations of @xmath65 . for @xmath68=\{1,\dots , n\}$ ] , we set @xmath69)$ ] . for a group @xmath63 , a homomorphism @xmath70
is called a _ @xmath63-action _ on @xmath65 .
the action is _ faithful _ if @xmath71 . for @xmath72 and @xmath73
we denote by @xmath74 the image of @xmath75 under @xmath76 .
for @xmath72 we write @xmath77 and call it the _ orbit _ of @xmath75 under the @xmath63-action . the orbits partition @xmath65 .
let @xmath78 denote the set of @xmath79-subsets of @xmath65 .
there is a natural induced action @xmath80 which also defines a natural @xmath63-action on @xmath78 .
we denote this action by @xmath81 .
similarly , there is a natural induced @xmath63-action on @xmath82 .
the orbits of this action are called the _ orbitals _ of @xmath63
. we shall need the undirected version of this concept ; we shall call the orbits of the @xmath63-action on @xmath83 the _ u - orbitals _ ( undirected orbitals ) of the @xmath63-action . by an action of the group @xmath63 on a structure @xmath84 such as a group or a graph or a simplicial complex we mean a homomorphism @xmath85 where @xmath86 denotes the automorphism group of @xmath84 .
let @xmath63 and @xmath87 be groups and let @xmath88 be a @xmath87-action on @xmath63 .
these data uniquely define a group @xmath89 , the _ semidirect product _ of @xmath63 and @xmath87 with respect to @xmath90 .
this group has order @xmath91 and has the following properites : @xmath92 has two subgroups @xmath93 and @xmath94 such that @xmath95;@xmath96 ; and @xmath97 . moreover , identifying @xmath63 with @xmath98 and @xmath87 with @xmath99 , for all @xmath100 and @xmath101 we have @xmath102 .
@xmath92 can be defined as the set @xmath103 under the group operation @xmath104 for more on semidirect products , which we use extensively , see ( * ? ? ?
* chap . 7 ) .
the group @xmath105 of affine transformations @xmath106 of @xmath107 ( @xmath108 , @xmath109 ) acts on @xmath107 . for each @xmath110
, @xmath105 has a unique subgroup of order @xmath111 ; we call this subgroup @xmath112 .
we note that @xmath113 and @xmath114 is cyclic of order @xmath115 and is isomorphic to a subgroup @xmath87 of @xmath105 ; @xmath112 can be described as a _
semidirect product _ @xmath116 .
an _ abstract simplicial complex _
@xmath117 on the set @xmath65 is a subset of the power - set of @xmath65 , closed under subsets : if @xmath118 then @xmath119 . the elements of @xmath117 are called its _ faces_. the _ dimension _ of @xmath120 is @xmath121 ; the dimension of @xmath117 is @xmath122 .
the _ euler characteristic _ of @xmath123 is defined as @xmath124 let @xmath12 : = \{1,2 , \ldots , n\}$ ] and @xmath125}{2}$ ] .
let @xmath11 be a subset of the power - set of @xmath65 , i.e. , a set of graphs on the vertex set @xmath12 $ ] .
we call @xmath11 a _ graph property _ if it is invariant under the induced action @xmath126 .
we call this graph property _ monotone decreasing _ if it is closed under subgraphs , i.e. , it is a simplicial complex
. we shall omit the adjective `` decreasing . ''
let @xmath127 be an abstract simplicial complex with a @xmath63-action .
fixed point complex _ @xmath128 action is defined as follows .
let @xmath129 be the @xmath63-orbits on @xmath65 . set @xmath130
\mid \bigcup_{i \in s } \omega_i \in { { \mathcal k}}\}.\ ] ] we say that a group @xmath63 satisfies * oliver s condition * if there exist ( not necessarily distinct ) primes @xmath131 such that @xmath63 has a ( not necessarily proper ) chain of subgroups @xmath132 such that @xmath133 is a @xmath24-group , @xmath134 is cyclic , and @xmath135 is a @xmath136-group .
[ thm : oliver ] assume the group @xmath63 satisfies oliver s condition . if @xmath63 acts on a nonempty contractible simplicial complex @xmath117 then @xmath137 in particular , such an action must always have a nonempty invariant face .
the topological approach to evasiveness , initiated by kahn , saks , and sturtevant , is based on the following key observation .
[ lem : kss ] if @xmath11 is a non - evasive graph property then @xmath11 is contractible .
kahn , saks , and sturtevant recognized that lemma [ lem : kss ] brought oliver s theorem to bear on evasiveness .
the combination of lemma [ lem : kss ] and theorem [ thm : oliver ] suggests the following general strategy , used by all authors in the area who have employed the topological method , including this paper : we find primes @xmath131 , a group @xmath63 satisfying oliver s condition with these primes , and a @xmath63-action on @xmath11 , such that @xmath138 . by oliver s theorem and the kss lemma
this implies that @xmath11 is evasive .
the novelty is in finding the right @xmath63 .
kss @xcite made the assumption that @xmath2 is a prime power and used as @xmath139 , the group of affine transformations @xmath106 over the field of order @xmath2 . while we use subgroups of such groups as our building blocks , the attempt to combine these leads to hard problems on the distribution of prime numbers .
regarding the `` forbidden subgraph '' property , chakrabarti , khot , and shi @xcite built considerable machinery which we use .
our conclusions are considerably stronger than theirs ; the additional techniques involved include a study of the orbitals of certain metacyclic groups , a universality property of cyclotomic graphs derivable using weil s character sum estimates , plus the number theoretic reductions indicated . for the `` sparse graphs '' result ( theorem [ thm : density ] ) we need @xmath63 such that all u - orbitals of @xmath63 are large and therefore @xmath140 . in both cases ,
we are forced to use rather large building blocks of size @xmath136 , say , where @xmath136 is a prime such that @xmath141 has a large divisor which is a prime for theorem [ thm : density ] and a power of 2 for theorem [ thm : forbidden ] .
in this section we prove parts ( a ) and ( c ) of theorem [ thm : forbidden ] .
a _ homomorphism _ of a graph @xmath3 to a graph @xmath142 is a map @xmath143 such that @xmath144 .
( in particular , @xmath145 is an independent set in @xmath3 for all @xmath146 . )
let @xmath147}$ ] be the set of those @xmath142 with @xmath148 $ ] that do not admit an @xmath149 homomorphism .
let further @xmath150 where @xmath151 denotes the number of vertices of @xmath3 .
the following is the main lemma of chakrabarti , khot , and shi @xcite .
[ lem : even_char ] if @xmath152 then @xmath153 } ) \equiv 0 \pmod 2 $ ] .
let @xmath136 be an odd prime power and @xmath115 an even divisor of @xmath154 consider the graph @xmath155 whose vertex set is @xmath156 and the adjacency between the vertices is defined as follows : @xmath157 @xmath155 is called a _
generalized paley graph_. [ lem : paley_clique ] if @xmath158 then @xmath159 contains a clique on @xmath151 vertices .
this follows from the following lemma which in turn can be proved by a routine application of weil s character sum estimates ( cf .
@xcite ) .
[ extension_lemma ] let @xmath160 be distinct elements of the finite field @xmath161 assume @xmath162 .
then the number of solutions @xmath163 to the system of equations @xmath164 is @xmath165 let @xmath112 be the subgroup of order @xmath111 of @xmath105 defined in section [ sec : groups ] .
each u - orbital of @xmath112 is isomorphic to @xmath155 .
if @xmath166 then each u - orbital of @xmath112 contains a clique of size @xmath167 the numbers in the title were defined in section [ sec : primes ] . in this section
we prove theorem [ thm : forbidden ] , part ( a ) .
let @xmath3 be a graph on @xmath151 vertices .
if there are infinitely many @xmath168-near - fermat primes then @xmath13 is eventually evasive .
[ near - fermat ] _ proof .
_ fix an odd prime @xmath169 such that @xmath170 if there are infinitely many @xmath168-near - fermat primes then infinitely many of them belong to the same residue class mod @xmath24 , say @xmath171 .
let @xmath172 be the @xmath173-th @xmath168-near - fermat prime such that @xmath174 and @xmath175 let @xmath176 and @xmath177 then @xmath178 and therefore @xmath179 for some @xmath180 .
now in order to use lemma [ lem : even_char ] , we need to write @xmath2 as a sum of @xmath181 terms where @xmath182 .
we already have @xmath183 of these terms ; we shall choose each of the remaining @xmath184 terms to be @xmath24 or @xmath185 .
if there are @xmath79 terms equal to @xmath185 then this gives us a total of @xmath186 terms , so we need @xmath187 . by assumption , @xmath188 ; therefore such a @xmath79 exists ; for large enough @xmath2 , it will also satisfy the constraints @xmath189 , let now @xmath190 acting on @xmath191 $ ] with @xmath79 orbits of size @xmath185 and @xmath192 orbits of size @xmath24 as follows : on an orbit of size @xmath193 ( @xmath194 ) the action is @xmath195 .
the additive groups act independently , with a single multiplicative action on top . @xmath196
acts on @xmath197 through the group homomorphism @xmath198 defined by the map @xmath199 .
let @xmath200 denote an orbit of @xmath201 on @xmath202 $ ] .
now the orbit of any pair @xmath203 is a clique of size @xmath204 , therefore a @xmath201-invariant graph can not contain an intra - cluster edge .
let @xmath205 be the largest power of 2 that divides @xmath206 let @xmath207 be the subgroup of @xmath208 of order @xmath209 let @xmath210 acting on @xmath211 $ ] with @xmath183 orbits of sizes @xmath212 in the obvious manner . from lemma [ lem : paley_clique ]
we know that the orbit of any @xmath213 \choose 2}$ ] must contain a clique of size @xmath167 hence , an invariant graph can not contain any intra - cluster edge .
overall , let @xmath214 , acting on @xmath12.$ ] since @xmath215 we have @xmath216 thus , @xmath63 is a `` @xmath217-group extension of a cyclic extension of a @xmath24-group '' and therefore satisfies oliver s condition ( stated before theorem [ thm : oliver ] ) .
hence , assuming @xmath13 is non - evasive , lemma [ lem : kss ] and theorem [ thm : oliver ] imply @xmath218 on the other hand , we claim that the fixed - point complex @xmath219 is isomorphic to @xmath220}$ ] .
the ( simple ) proof goes along the lines of lemma 4.2 of @xcite .
thus , by lemma [ lem : even_char ] we have @xmath153 } ) \equiv 0 \pmod 2,$ ] a contradiction . in this section
, we prove part ( c ) of theorem [ thm : forbidden ] .
[ near_eva ] for every graph @xmath3 there exists a number @xmath222 such that the query complexity of @xmath13 is @xmath223 _ proof . _
let @xmath151 be the number of vertices of @xmath3 .
let @xmath24 be the smallest prime such that @xmath224 and @xmath169 .
so @xmath225 for some function @xmath226 by dirichlet s theorem ( we do nt need any specific estimates here ) . since @xmath227 we have @xmath228 and therefore @xmath229 .
now , by the chinese remainder theorem , select the smallest positive integer @xmath230 satisfying @xmath231 and @xmath232 . note that @xmath233 .
let @xmath234 ; so we have @xmath235 .
let @xmath236 .
consider the following boolean function @xmath237 on @xmath238 variables .
consider graphs @xmath239 on the vertex set @xmath12 $ ] with the property that they have no edges among their last @xmath230 vertices .
these graphs can be viewed as boolean functions of the remaining @xmath238 variables .
now we say that such a graph has property @xmath237 if it does not contain @xmath3 as a subgraph
. * claim . *
the function @xmath237 is evasive .
the claim immediately implies that the query complexity of @xmath13 is at least @xmath238 , proving the theorem with @xmath240 . to prove the claim , consider the groups @xmath241 and @xmath242 . here
@xmath243 acts on @xmath244 $ ] in the obvious way : we divide @xmath244 $ ] into @xmath245 blocks of size @xmath24 ; @xmath246 acts on each block independently and @xmath247 acts on the blocks simultaneously ( diagonal action ) so on each block they combine to an @xmath248-action .
@xmath249 acts as a @xmath230-cycle on the remaining @xmath230 vertices .
so @xmath63 is a cyclic extension of a @xmath24-group ( because @xmath250 ) .
if @xmath237 is not evasive then from theorem [ thm : oliver ] and lemma [ lem : kss ] , we have @xmath251 .
on the other hand we claim that , @xmath252},$ ] where @xmath253 the proof of this claim is exactly the same as the proof of lemma 4.2 of @xcite .
thus , from lemma [ lem : even_char ] , we conclude that @xmath153})$ ] is even .
this contradicts the previous conclusion that @xmath153 } ) = 1.$ ] specific estimates on the smallest dirichlet prime can be used to estimate @xmath222 .
linnik s theorem implies @xmath254 , extending theorem [ near_eva ] to strong lower bounds for variable @xmath3 up to @xmath255 for some positive constant @xmath256 .
we prove part ( c ) of theorem [ thm : density ] .
if the non - empty monotone graph property @xmath11 is not evasive then @xmath257 [ superlin ] assume in this section that @xmath258 where @xmath24 is prime .
let @xmath259 .
we construct the group @xmath260 acting on @xmath12.$ ] let @xmath261 .
let @xmath260 be the semidirect product @xmath262 with respect to the @xmath87-action on @xmath263 defined by @xmath264 we describe the action of @xmath260 on @xmath12 $ ] .
partition @xmath12 $ ] into @xmath180 clusters of size @xmath265 each .
identify each cluster with the field of order @xmath266 i.e. , as a set , @xmath12 = [ k ] \times \mathbb{f}_{p^{\alpha}}.$ ] the action of @xmath267 is described by @xmath268 an unordered pair @xmath269 $ ] is termed an _ intra - cluster edge _ if both @xmath173 and @xmath270 are in the same cluster , otherwise it is termed an _ inter - cluster edge . _ note that every u - orbital under @xmath63 has only intra - cluster edges or only inter - cluster edges .
denote by @xmath271 and @xmath272 the minimum sizes of u - orbitals of intra - cluster and inter - cluster edges respectively .
we denote by @xmath273 the minimum size of an orbit in @xmath274 $ ] under @xmath275 and by @xmath276 the minimum size of a u - orbital in @xmath274.$ ] we then have : @xmath277 let @xmath278 and define @xmath279 as the minimum size of a u - orbital in @xmath12.$ ] then @xmath280 the goldbach conjecture asserts that every even integer can be written as the sum of two primes . vinogradov s
theorem @xcite says that every sufficiently large odd integer @xmath180 is the sum of three primes @xmath281 .
we use here haselgrove s version @xcite of vinogradov s theorem which states that we can require the primes to be roughly equal : @xmath282 .
this can be combined with the prime number theorem to conclude that every sufficiently large even integer @xmath180 is a sum of four roughly equal primes .
let @xmath283 where @xmath24 is prime .
assume @xmath180 is not bounded .
write @xmath180 as a sum of @xmath284 roughly equal primes @xmath285 .
let @xmath286 where @xmath287 denotes the cyclic group of order @xmath285 and the direct product is taken over the _
@xmath285 .
@xmath275 acts on @xmath274 $ ] as follows : partition @xmath180 into parts of sizes @xmath288 and call these parts @xmath289.$ ] the group @xmath287 acts as a cyclic group on the part @xmath289.$ ] in case of repetitions , the same factor @xmath287 acts on all the parts of size @xmath290 we follow the notation of section [ wreath - with - autos ] and consider the group @xmath260 with our specific @xmath275 .
we have @xmath291 and hence we get , from equation : [ additive partition minsize lemma ] let @xmath283 where @xmath24 is a prime . for the group @xmath292 , we have @xmath293 where @xmath279 denotes the minimum size of a u - orbital .
let @xmath258 where @xmath265 is the largest prime power dividing @xmath2 ; so @xmath294 ; this will be a lower bound on the size of u - orbitals .
our group @xmath63 will be of the general form discussed in section [ wreath - with - autos ] .
@xmath295 + let @xmath296 . following the notation of section [ wreath - with - autos ] , we get @xmath297 and this yields that @xmath298 oliver s condition is easily verified for @xmath63 . case 2 .
@xmath299 + consider the @xmath300 acting on @xmath12 $ ] where @xmath275 is as described in section [ semidirect product with additive partition ] .
the minimum possible size @xmath279 of a u - orbital is @xmath301 by lemma [ additive partition minsize lemma ] . finally , since @xmath302
, we obtain @xmath303 if all @xmath285
are co - prime to @xmath304 then @xmath305 becomes a cyclic group and @xmath63 becomes a cyclic extension of a @xmath24-group . since @xmath306 for all @xmath173 and @xmath307
, size considerations yield that at most one @xmath285 divides @xmath308 and @xmath309 does not .
suppose , without loss of generality , @xmath310 divides @xmath311 let @xmath312 then @xmath115 must be co - prime to each @xmath290 thus , @xmath313 thus , @xmath87 is a @xmath310-group extension of a cyclic group .
hence , @xmath63 satisfies oliver s condition ( cf .
theorem [ thm : oliver ] ) . for almost all
@xmath314 our proof gives a better dimension lower bound of @xmath315
in this section we prove parts ( a ) and ( b ) of theorem [ thm : density ] .
let @xmath316 where @xmath24 and @xmath181 are prime numbers .
let @xmath136 be a prime divisor of @xmath317 we partition @xmath12 $ ] into two parts of size @xmath318 and @xmath181 , denoted by @xmath191 $ ] and @xmath319 $ ] respectively .
we now construct a group @xmath320 acting on @xmath12 $ ] as a direct product of a group acting on @xmath191 $ ] and a group acting on @xmath319,$ ] as follows : @xmath321 here , @xmath322 acts on @xmath191 $ ] and is as defined in section [ semidirect product with additive partition ] , and involves choosing a partition of @xmath180 into upto four primes that are all @xmath323 @xmath324 is defined as the semidirect product @xmath325 with @xmath326 viewed as a subgroup of the group @xmath327 it acts on @xmath319 $ ] as follows : we identify @xmath319 $ ] with the field of size @xmath328 let @xmath329 be a typical element of @xmath330 where @xmath331 and @xmath332 then , @xmath333 thus , @xmath334 acts on @xmath12.$ ] let @xmath279 be the minimum size of the orbit of any edge @xmath335 \choose 2}$ ] under the action of @xmath336 one can show that @xmath337 we shall choose @xmath338 carefully such that ( a ) the value of @xmath279 is large , and ( b ) oliver s condition holds for @xmath320 . the extended riemann hypothesis ( erh )
implies the following strong version of the prime number theorem for arithmetic progressions .
let @xmath339 denote the numer of primes @xmath340 , @xmath341 .
then for @xmath342 we have @xmath343 where @xmath344 and the constant implied by the big - oh
notation is absolute ( cf .
* ch.7 , eqn.(5.12 ) ) or ( * ? ? ?
* thm.8.4.5 ) ) .
[ dirichlet - bertrand ] assume erh .
suppose the sequence @xmath345 satisfies @xmath346 .
then for all sufficiently large @xmath2 and for any @xmath347 relatively prime to @xmath345 there exists a prime @xmath348 such that @xmath349 we want to write @xmath316 where @xmath24 and @xmath181 are primes , and with @xmath136 a prime divisor of @xmath350 as described in section [ sparse - conditional setup ] .
specifically , we try for : @xmath351 we claim that under erh , such a partition of @xmath2 is possible .
to see this , fix some @xmath352 such that @xmath353 fix some @xmath354 now , @xmath355 and @xmath356 solves to @xmath357 for some @xmath35 such that @xmath358 since @xmath359 we can conclude under erh ( using lemma [ dirichlet - bertrand ] ) that there exists a prime @xmath357 such that @xmath360 this gives us the desired partition .
one can verify that our @xmath63 satisfies oliver s condition .
equation gives @xmath361 this completes the proof of part ( b ) of theorem [ thm : density ] .
we can use chowla s conjecture , along with the general setup of section [ sparse - conditional setup ] , to obtain a stronger lower bound on @xmath367 the new bounds we hope to achieve are : @xmath368 such a partition is always possible assuming chowla s conjecture . to see this ,
first fix @xmath369 then fix @xmath370 and find the least solution for @xmath355 and @xmath371 which is equivalent to solving for @xmath372 for some @xmath373 the least solution will be greater than @xmath374 unless @xmath35 happens to be a prime . in this case , we add another constraint , say @xmath375 and resolve to get the least solution greater than @xmath376 note that @xmath377 now , from equation ( [ pqr_bound ] ) , we get the lower bound of @xmath378 this completes the proof of part ( a ) of theorem [ thm : density ] . | a boolean function on @xmath0 variables is called _ evasive _ if its decision - tree complexity is @xmath0 .
a sequence @xmath1 of boolean functions is _ eventually evasive _ if @xmath1 is evasive for all sufficiently large @xmath2 .
we confirm the eventual evasiveness of several classes of monotone graph properties under widely accepted number theoretic hypotheses .
in particular we show that chowla s conjecture on dirichlet primes implies that ( a ) for any graph @xmath3 , `` forbidden subgraph @xmath3 '' is eventually evasive and ( b ) all nontrivial monotone properties of graphs with @xmath4 edges are eventually evasive .
( @xmath2 is the number of vertices . )
while chowla s conjecture is not known to follow from the extended riemann hypothesis ( erh , the riemann hypothesis for dirichlet s @xmath5 functions ) , we show ( b ) with the bound @xmath6 under erh .
we also prove unconditional results : ( a@xmath7 ) for any graph @xmath3 , the query complexity of `` forbidden subgraph @xmath3 '' is @xmath8 ; ( b@xmath7 ) for some constant @xmath9 , all nontrivial monotone properties of graphs with @xmath10 edges are eventually evasive . even these weaker , unconditional results rely on deep results from number theory such as vinogradov s theorem on the goldbach conjecture .
our technical contribution consists in connecting the topological framework of kahn , saks , and sturtevant ( 1984 ) , as further developed by chakrabarti , khot , and shi ( 2002 ) , with a deeper analysis of the orbital structure of permutation groups and their connection to the distribution of prime numbers .
our unconditional results include stronger versions and generalizations of some result of chakrabarti et al .
lszl babai anandam banerjee raghav kulkarni vipul naik |
how do real - world networks evolve with time ?
while empirical studies provide many intuitions and expectations , many questions remain open . in particular , we lack tools to characterize and quantitatively compare temporal graph dynamics . in turn ,
such tools require good observables to quantify the ( temporal ) relationships between networks .
in particular , the few network dynamics models that currently exist are often _ oblivious of the network type_. this is problematic , as complex networks come in many different flavors , including social networks , biological networks , or physical networks .
it seems highly unlikely that these very different graphs evolve in a similar manner .
a natural prerequisite to measure evolutionary distances are good metrics to _ compare _ graphs .
the classic similarity measure for graphs is the _ graph edit distance ( @xmath0 ) _
@xcite : the graph edit distance @xmath1 between two graphs @xmath2 and @xmath3 is defined as the minimal number of _ graph edit operations _ that are needed to transform @xmath2 into @xmath3 .
the specific set of allowed graph edit operations depends on the context , but typically includes node and link insertions and deletions .
while graph edit distance metrics play an important role in computer graphics and are widely applied to pattern analysis and recognition , @xmath4 is not well - suited for measuring similarities of networks in other contexts @xcite : the set of graphs at a certain graph edit distance @xmath5 from a given graph @xmath6 exhibit very diverse characteristics and seem unrelated ; being oblivious to semantics , the @xmath4 does not capture any intrinsic structure typically found in real - world networks . a similarity measure that takes into account the inherent structure of a graph
may however have many important applications .
a large body of work on graph similarities focusing on a variety of use cases have been developed in the past ( see our discussion in section [ sec : relwork ] ) .
depending on the context in which they are to be used , one or another is more suitable . in particular
, we argue that graph similarities and graph distance measures are also an excellent tool for the analysis , comparison and prediction of temporal network traces , allowing us to answer questions such as : _ do these two networks have a common ancestor ? _ _ are two evolution patterns similar ? _ or _ what is a likely successor network for a given network ?
_ however , we argue that in terms of graph similarity measures , there is no panacea : rather , graphs and their temporal patterns , come with many faces . accordingly , we in this paper , propose to use a parametric , _ centrality - based approach _ to measure graph similarities and distances , which in turn can be used to study the evoluation of networks .
more than one century ago , camille jordan introduced the first graph _ centrality _ measure in his attempt to capture `` the center of a graph '' . since then the family of centrality measures has grown larger and is commonly employed in many graph - related studies .
all major graph - processing libraries commonly export functionality for degree , closeness , betweenness , clustering , pagerank and eigenvector centralities . in the context of static graphs ,
_ centralities _ have proven to be a powerful tool to extract meaningful information on the structure of the networks , and more precisely on the _ role _ every participant ( node ) has in the network . in social network analysis ,
centralities are widely used to measure the importance of nodes , e.g. , to determine key players in social networks , or main actors in the propagation of diseases , etc . today
, there is no consensus on good " and bad " centralities : each centrality captures a particular angle of a node s topological role , some of which can be either crucial or insignificant , depending on the application .
_ am i important because i have many friends , because i have important friends , or because without me , my friends could not communicate together ? _
the answer to this question is clearly context - dependent . in this paper
, we argue that the perceived quality of network similarities or distances measuring the difference between two networks depends on the focus and application just as much . instead of debating the advantages and disadvantages of a set of similarities and distances , we provide a framework to apply them to characterize network evolution from different perspectives .
in particular , we leverage centralities to provide a powerful tool to quantify network changes .
the intuition is simple : to measure how a network evolves , we measure the change of the nodes roles and importance in the network , by leaving the responsibility to quantify node importance to centralities .
[ [ our - contributions ] ] our contributions + + + + + + + + + + + + + + + + + this paper is motivated by the observation that centralities can be useful to study the dynamics of networks over time , taking into account the individual roles of nodes ( in contrast to , e.g. , isomorphism - based measures , as they are used in the context of anonymous graphs ) , as well as the context and semantics ( in contrast to , e.g. , graph edit distances ) .
in particular , we introduce the notion of _ centrality distance _
@xmath7 for two graphs @xmath8 , a graph similarity measure based on a _ node centrality _ @xmath9 .
we demonstrate the usefulness of our approach to identify and characterize the different faces of graph dynamics . to this end , we study five generative graph models and seven dynamic real world networks in more details .
our evaluation methodology comparing the quality of different similarity measures to a random baseline using data from actual graph evolutions , may be of independent interest .
in particular , we demonstrate how centrality distances provide interesting insights into the structural evolution of these networks and show that actual evolutionary paths are far from being random .
moreover , we build upon the centrality distance concept to construct dynamic graph signatures .
the intuition is simple : we measure the probability of an update to be considered as an outlier compared to a uniformly random evolution .
this allows us to quantify the deviation of a given dynamic network from a purely random evolution ( our null - model ) of the same structure for a set of centrality distances .
the signature consisting of the resulting deviation values enables the comparison of different dynamisms on a fair basis , independently from scale and sampling considerations .
[ [ examples ] ] examples + + + + + + + + to motivate the need for tools to analyse network evolution , we consider two simple examples .
* example 1 . * [ local / global scenario ] consider three graphs @xmath2 , @xmath3 , @xmath10 over five nodes @xmath11 : @xmath2 is a line , where @xmath12 and @xmath13 are connected ; @xmath3 is a cycle , i.e. , @xmath2 with an additional link @xmath14 ; and @xmath10 is @xmath2 with an additional link @xmath15 . in this example , we first observe that @xmath3 and @xmath10 have the same graph edit distance to @xmath2 : @xmath16 , as they contain one additional edge .
however , in a social network context , one would intuitively expect @xmath10 to be closer to @xmath2 than @xmath3 .
for example , in a friendship network a short - range `` _ _ triadic closure _ _ '' @xcite link may be more likely to emerge than a long - range link : friends of friends may be more likely to become friends themselves in the future .
moreover , more local changes are also expected in mobile environments ( e.g. , under bounded human mobility and speed ) .
as we will see , the centrality distance concept of this paper can capture such differences .
* example 2 . * [ evolution scenario ] as a second synthetic example , consider two graphs @xmath17 and @xmath18 , where @xmath17 is a line topology and @xmath18 is a `` shell network '' ( see also figure [ fig : methodology ] ) .
how can we characterize evolutionary paths leading from the @xmath17 topology to @xmath18 ?
note that the graph edit distance does not provide us with any information about the likelihood or the role changes of _ evolutionary paths _ from @xmath17 to @xmath18 , i.e. , on the order of edge insertions : there are many possible orders in which the missing links can be added to @xmath17 , and these orders do not differ in any way when comparing them with the graph edit distance .
in reality , however , we often have some expectations on how a graph may have evolved between two given snapshots @xmath17 and @xmath18 . for example , applying the triadic closure principle to our example , we would expect that the missing links are introduced one - by - one , from left to right . to a shell graph @xmath18 . ]
the situation may look different in technological , man - made networks .
adding links from left to right only slowly improves the `` routing efficiency '' of the network : after the addition of @xmath19 edges from left to right , the longest shortest path is @xmath20 hops , for @xmath21 .
a more efficient evolution of the network is obtained by connecting @xmath22 to the furthest node , adding links to the middle of the network , resulting in a faster distance reduction : after @xmath19 edge insertions , the distance is roughly reduced by a factor @xmath19 .
thus , different network evolution patterns can be observed in real networks . instead of defining application - dependent similarities with design choices focusing on which evolution patterns are more expected from a certain network , we provide a framework that allows the joint characterization of graph dynamics along different axes .
[ [ organization ] ] organization + + + + + + + + + + + + the remainder of this paper is organized as follows .
section [ sec : background ] provides the reader with the necessary background .
section [ sec : graph - dist ] introduces our centrality distance framework and section [ sec : methodology ] our methodology to study the different graph dynamics empirically . section [ sec : experiments ] reports on results from analyzing real and generated networks . after reviewing related work in section [ sec : relwork ] ,
we conclude our contribution in section [ sec : conclusion ] .
this paper considers _ labeled _ graphs @xmath23 , where vertices @xmath24 have unique identifiers and are connected via undirected edges @xmath25 . in the following ,
we denote as @xmath26 the set of neighbors of node @xmath27 : @xmath28 .
a temporal network trace is a sequence @xmath29 $ ] , where @xmath30 represents the network at the @xmath31 snapshot .
we focus on _ node centralities _ , a centrality being a real - valued function assigning `` importance values '' to nodes . obviously , the notion of importance is context - dependent , which has led to many different definitions of centralities .
we refer to @xcite for a thorough and formal discussion on centralities . a _ centrality _
@xmath9 is a function @xmath32 that , given a graph @xmath23 and a vertex @xmath33 , returns a non - negative value @xmath34 .
the centrality function is defined over all vertices @xmath35 of a given graph @xmath6 . by convention
, we define the centrality of a node without edges to be 0 .
we write @xmath36 to refer to the vector in @xmath37 where the @xmath31 element is @xmath38 for a given order of the identifiers .
centralities are a common way to characterize networks and their vertices . frequently studied centralities include the _ degree centrality _
( @xmath39 ) , the _ betweenness centrality _ ( @xmath40 ) , the _ closeness centrality _ ( @xmath41 ) , and the _ pagerank centrality _ ( @xmath42 ) among many more . a node is @xmath43-central if it has many edges : the degree centrality is simply the node degree ; a node is @xmath44-central if it is on many shortest paths : the betweenness centrality is the number of shortest paths going through the node ; a node is @xmath45-central if it is close to many other nodes : the closeness centrality measures the inverse of the distances to all other nodes ; and a node is @xmath46-central if the probability that a random walk on @xmath6 visits this node is high .
we use the classical definitions for centralities , and the exact formulas are presented in [ appendix : centrality ] for the sake of completeness .
finally , throughout this paper , we will define the graph edit distance @xmath0 between two graphs @xmath2 and @xmath3 as the minimum number of operations to transform @xmath2 into @xmath3 ( or vice versa ) , where an _ operation _ is one of the following : link insertion and link removal .
the canonical distance measure is the graph edit distance , @xmath0 .
however , @xmath0 often provides limited insights into the graph dynamics in practice .
figure [ fig : methodology ] shows an example with two evolutionary paths : an incremental ( _ left _ ) and a binary ( _ right _ ) path that go from @xmath17 to @xmath18 . with respect to @xmath0 , there are many equivalent shortest paths for moving from @xmath17 to @xmath18 .
however , intuitively , not all traces are equally likely for dynamic networks , as the structural roles that nodes in networks have are often preserved and do not change arbitrarily . clearly ,
studying graph evolution with ged thus can not help us to understand how structural properties of graphs evolve .
the graph edit distance @xmath0 does not provide much insights into graph evolution .
we in this paper aim to enrich the graph similarity measure with semantics . at the heart of our approach lies the concept of _ centrality distance _ : a simple and flexible tool to study the similarity of graphs .
essentially , the centrality distance measures the similarity between two centrality vectors .
it can be used to measure the distance between two arbitrary graphs , not only between graphs with graph edit distance 1 . given a centrality @xmath9
, we define the centrality distance @xmath47 between any two graphs as the sum of the node - wise difference of the centrality values : @xmath48 thus , the centrality distance intuitively measures the magnitude by which the roles of different nodes change .
while we focus on the 1-norm in this paper , the concept of centrality distance can be useful also for other norms . both the importance of node roles as well as the importance of node role changes is application - dependent . due to the large variety of processes
dynamic graphs can capture , there is no one - size - fits - it - all measure of importance . to illustrate this point ,
let us consider the `` intuitive '' similarity properties proposed by faloutsos et al .
for instance , the proposed edge importance property should penalize changes that create disconnected components more than changes that maintain the connectivity properties of the graphs .
now imagine a cycle graph of 100 nodes @xmath49 , and a single additional node @xmath27 connected to @xmath50 . according to the proposed edge importance property the most important link is @xmath51 .
indeed , it is the only link whose removal would create a disconnected component ( containing @xmath27 alone ) .
yet the removal of any other link would double the diameter of the structure . or in an information dissemination network
all nodes would have to update half of their routing tables . so which link is more important ?
the answer clearly depends on the context .
similar examples can be found for other properties proposed in @xcite , e.g. , regarding submodularity and focus - awareness .
not only are these properties hard to formalize , their utility varies from application to application .
we conclude by noting that given two centralities @xmath52 and @xmath53 and two arbitrary graphs @xmath2 and @xmath3 with @xmath54 nodes , the respective distances are typically different , i.e. , @xmath55 .
hence , using a set of different centrality distances , we can explore the variation of the graph dynamics in more than one `` dimension '' .
in order to characterize the different faces of graph dynamics and to study the benefits of centrality - based measures , we propose a simple methodology . intuitively , given a centrality capturing well the roles of different nodes in a real - world setting , we expect the centrality distance between two consecutive graph snapshots @xmath56 and @xmath57 to be smaller than the typical distance from @xmath56 to other graphs that have the same ged . to verify this intuition , we define a _ null model _ for evolution .
a null model generates networks using patterns and randomization , i.e. , certain elements are held constant and others are allowed to vary stochastically .
ideally , the randomization is designed to mimic the outcome of a random process that would be expected in the absence of a particular mechanism @xcite . applied to our case
, this means that starting from a given snapshot @xmath56 that represents the fixed part of the null model , if the evolution follows a null model , then any graph randomly generated from @xmath56 at the given @xmath0 is evenly likely to appear . concretely ,
for all consecutive graph pairs @xmath56 and @xmath57 of a network trace , we determine the graph edit distance ( or `` radius '' ) @xmath58 .
then , we generate a set @xmath59 of @xmath60 sample graphs @xmath61 at the same @xmath4 @xmath62 from @xmath56 _ uniformly at random_. that is , to create @xmath63 , we first start from a copy of @xmath56 and select @xmath62 node pairs , @xmath64 , uniformly at random . for each of these pairs @xmath65
we add the edge @xmath65 to @xmath63 if it does not exist in @xmath56 or we remove it if it was in @xmath56 originally . such randomly built sample graphs at the same graph edit distance allow us to assess the impact of a uniformly random evolution of the same magnitude from the same starting graph @xmath56 : @xmath66 in other words , @xmath56 is the pattern and the evolution to @xmath63 at graph edit distance @xmath62 is the randomized part of the null model describes results obtained for a null model that guarantees the average degree of @xmath57 in the sample graphs . ] . as a next step ,
given a centrality @xmath9 , we compare @xmath57 with the set @xmath59 that samples the evolution following the null model .
we consider that @xmath57 does not follow the null model if it is an _ outlier _ in the set @xmath59 for the centrality @xmath9 .
practically , @xmath57 is considered an outlier if the absolute value of its distance from @xmath56 minus the mean distance of @xmath59 to @xmath56 is at least twice the standard deviation , i.e. , if @xmath67 given a temporal trace @xmath68 , we define @xmath69 as the fraction of outliers in the trace for centrality @xmath9 .
an ensemble of such values @xmath70 for a set of centralities @xmath71 is called a _
dynamic signature _ of @xmath68 .
based on our centrality framework and methodology , we can now shed some light on the different faces of graph dynamics , using real world data sets . * * caida ( as ) : * this data captures the autonomous systems relationships as captured by the caida project .
each of the @xmath72 snapshots represents the daily interactions of the @xmath73 first as identifiers from august 1997 until december 1998 @xcite .
* * icdcs ( icdcs ) : * we extracted the most prolific authors in the icdcs conference ( ieee international conference on distributed computing systems ) and the co - author graph they form from the dblp publication database ( http://dblp.uni-trier.de ) .
this trace contains 33 snapshots of 691 nodes and 1076 collaboration edges .
the timestamp assigned to an edge corresponds to the first icdcs paper the authors wrote together . clearly ,
the co - authorship graph is characterized by a strictly monotonic densification over time . * * uci social network ( uci ) : * the third case study is based on a publicly available dataset @xcite , capturing all the messages exchanges realized on an online facebook - like social network between 1882 students at university of california , irvine over 7 months .
we discretized the data into a dynamic graph of 187 time steps representing the daily message exchanges among users .
* * hypertext ( ht ) : * face - to - face interactions of the acm hypertext 2009 conference attendees .
113 participants were equipped with rfid tags .
each snapshot represents one hour of interactions @xcite . * * infectious ( in ) : * face - to - face interactions of the `` infectious : stay away '' exhibition held in 2009 .
410 participants were equipped with rfid tags .
each snapshot represents 5 minutes of the busiest exhibition day @xcite . *
* manufacture ( ma ) : * daily internal email exchange network of a medium - size manufacturing company ( 167 nodes ) over 9 months of 2010 @xcite . * * souk ( sk ) : * this dataset captures the social interactions of 45 individuals during a cocktail , see @xcite for more details .
the dataset consists of @xmath74 snapshots , describing the dynamic interaction graph between the participants , one time step every @xmath75 seconds @xcite . in the network traces.,scaledwidth=70.0% ] figure [ fig : datasets ] provides a temporal overview on the evolution of the number of edges in the network and the @xmath4 between consecutive snapshots .
some of the seven datasets exhibit very different dynamics : one can observe the time - of - day effect of attendees interactions on _ hypertext _ , and the day - of - week effect on _
manufacture_. _ uci _ , _ hypertext _ , _ infectious _ and _ manufacture _ all exhibit a high level of dynamics with respect to their number of links .
this is expected for _ infectious _ , as visitors come and leave regularly and rarely stay for long , but rather surprising for _
the density of _ caida _ slowly increases , and with a steady @xmath4 .
similarly , the number of co - author edges of _ icdcs _ steadily increases over the years , while the number of new edges per year is relatively stable .
the number of days of the conference _
hypertext _ and the fact that conference participants sleep during the night and do not engage in social activity is evident in the second trace . the dynamic pattern of the online social network _
uci _ has two regimes : it has a high dynamics for the first 50 timestamps , and is then relatively stable , whereas _ souk _ exhibits a more regular dynamics .
generally , note that @xmath4 can be at most twice as high as the maximal edge count of two consecutive snapshots . and @xmath57 in dashed red lines and between @xmath56 and 100 graphs with the same ged as @xmath56 and @xmath57 in solid blue lines representing the median , @xmath76 bars in grey .
ec : ego centrality , bc : betweenness centrality , cc : closeness centrality , kc : cluster centrality , pc : pagerank centrality.,title="fig : " ] and @xmath57 in dashed red lines and between @xmath56 and 100 graphs with the same ged as @xmath56 and @xmath57 in solid blue lines representing the median , @xmath76 bars in grey .
ec : ego centrality , bc : betweenness centrality , cc : closeness centrality , kc : cluster centrality , pc : pagerank centrality.,title="fig : " ] and @xmath57 in dashed red lines and between @xmath56 and 100 graphs with the same ged as @xmath56 and @xmath57 in solid blue lines representing the median , @xmath76 bars in grey .
ec : ego centrality , bc : betweenness centrality , cc : closeness centrality , kc : cluster centrality , pc : pagerank centrality.,title="fig : " ] + hypertext infectious icdcs figure [ fig : chronogram ] presents examples of the results of our comparison of random graphs with the same graph edit distance ged as real - world network traces .
the red dashed lines represent the centrality distances of @xmath56 and @xmath57 .
the distribution of @xmath77 values from @xmath56 to the @xmath78 randomly sampled graphs of @xmath59 is represented as follows : the blue line is the median , while the gray lines represent the @xmath76 outlier detection window . for most graphs under investigation and for most centralities
it holds that the induced centrality distance between @xmath56 and @xmath57 is often lower than between @xmath56 and an arbitrary other graph with the same distance .
there are however a few noteworthy details . _
hypertext _ and _ infectious _ exhibit very similar dynamics compared from a ged perspective as shown in figure [ fig : datasets ] . yet from the other centralities perspective , their dynamism is very different .
consider for instance _ infectious _ for @xmath42 , where the measured distance is consistently an order of magnitude less than the sampled one .
this can be understood from the link creation mechanics : in _ infectious _ , visitors at different time periods never meet . by connecting these in principle very remote visitors ,
the null model dynamics creates highly important links .
this does not happen in _ hypertext _ where the same group of researchers meet repeatedly .
in the monotonically growing co - authorship network of icdcs , we can observe that closeness and ( ego ) betweenness distances grow over time , which is not the case for the other networks in figure [ fig : chronogram ] .
when looking at other centrality distances , we observe that even though the local structure changes , a different set of properties remains mostly unaltered across different networks .
moreover , for some _ ( graph , distance ) _ pairs , like @xmath79 on _ icdcs _ , @xmath45 on _ hypertext _ , or @xmath46 on _ infectious _ , the measured distance is orders of magnitude lower than the median of the sampled ones .
this underlines a clear difference between random evolution and the observed one from this centrality perspective : the link update dynamics is biased .
figure [ fig : spider ] summarizes the @xmath80 signatures for @xmath81 applied to 7 real and 5 synthetic graphs in the form of a histogram chart for synthetic graphs , each point is the average of 50 independent realizations of the model , and @xmath82 .
that is , each chart represents the probability of having graph evolutions being outliers with respect to the null model for the corresponding centralities .
interestingly , this `` distinction ratio '' is not uniform among datasets .
on _ caida _ , _ infectious _ and _ uci _ , the ratio is high for local centralities such as _ pagerank _ and _ clustering _ , and low for global centralities such as _ closeness _ or _ betweenness_. on the contrary , _ hypertext _ and _ manufacture _ exhibit large ratios for global centralities and small ratios for local centralities .
both local and global centralities perform well on souk .
the difference of these behaviors show that these graphs adhere to different types of dynamics . , the probability that @xmath57 is an outlier w.r.t . the null model for the corresponding centrality .
synthetic scenarios are depicted in _
blue _ , real scenarios in _ red_. the black line at @xmath83 represents the null model , i.e. , the fraction of graphs that are at distance at least @xmath76 from the mean in a normal distribution .
synthetic datasets : ba : barabasi - albert , cmhalf : preferential attachment equiprobable nodes and edge events cmlog : preferential attachement node events decay in log , er : erds - rnyi , rr : random regular .
real - life datasets : ca : caida , icdcs : icdcs co - authors , uci : online social network of uci , ht : hypertext conference , in : infectious ma : manufacture mails , sk : souk cocktail . ]
to complement our observations on real networks with graph snapshots produced according to a model , we investigated graph traces generated by some of the most well - known models : _ erds - rnyi _
er @xcite , _ random regular _ rr @xcite , _ barabasi - albert _ ba @xcite and _ preferential attachment _ @xcite graphs with an equal number of node and edge events ( cmhalf ) and with the number of node events depending logarithmically on the time ( cmlog ) .
perhaps the most striking observation is that all tested dynamic network models have low @xmath69 values for all @xmath9 .
this is partly due to the fact that the graph edit distance between two subsequent snapshots is one and thus the centrality vectors do not vary as much as between the snapshots and the sampled graphs of the same graph edit distance for the real networks .
moreover , these randomized synthetic models are closer to the null model , and lack some of the characteristics ( like link locality ) of real world networks . furthermore , we observe that each random network model exhibits distinct dynamics signatures , with er being closest to the null model .
to the best of our knowledge , our paper is the first to combine the concepts of centralities and graph distances . in the following , we review related work in the two fields in turn , and subsequently discuss additional literature on dynamic graphs . * graph characterizations and centralities .
* graph structures are often characterized by the frequency of small patterns called _ motifs _
@xcite , also known as _ graphlets _
@xcite , or _ structural signatures _ @xcite
. another important graph characterization , which is studied in this paper , are _ centralities _ @xcite .
dozens of different centrality indices have been defined over the last years , and their study is still ongoing , with no unified theory yet .
we believe that our centrality distance framework can provide new inputs for this discussion . * graph similarities and distances .
* graph edit distances have been used extensively in the context of inexact graph matchings in the field of pattern analysis .
we refer the reader to the good survey by gao et al .
soundarajan et al @xcite compare twenty network similarities for anonymous networks .
they distinguish between comparison levels ( node , community , network level ) and identify vector - based , classifier - based , and matching - based methods .
surprisingly they are able to show that the results of many methods are highly correlated .
netsimile @xcite allows to assess the similarity between @xmath84 networks , possibly with different sizes and no overlaps in nodes or links .
netsimile uses different social theories to compute similarity scores that are size - invariant , enabling mining tasks such as clustering , visualization , discontinuity detection , network transfer learning , and re - identification across networks .
the deltacon method @xcite is based on the normed difference of node - to - node affinity according to a belief propagation method . more precisely , the similarity between two graphs is the root euclidean distance of their two affinity matrices or an approximation thereof .
the authors provide three axioms that similarities should satisfy and demonstrate using examples and simulations that their similarity features the desired properties of graph similarity functions .
our work can be understood as an attempt to generalize the interesting approach by faloutsos et al . in @xcite , which derives a distance from a normed matrix difference , where each element depends on the relationships among the nodes .
in particular , are argue that there is no one - size - fits - it - all measure , and propose an approach parametrized by centralities .
interestingly , we also prove that distances derived in our framework satisfy the axioms postulated in @xcite
. * dynamic graphs . * among the most well - known evolutionary patterns are the shrinking diameter and densification @xcite .
a lot of recent work studies link prediction algorithms @xcite .
others focus on methods for finding frequent , coherent or dense temporal structures @xcite , or the evolution of communities and user behavior @xcite .
another line of research attempts to extend the concept of centralities to dynamic graphs @xcite .
some researchers study how the importance of nodes changes over time in dynamic networks @xcite .
others define temporal centralities which to rank nodes in dynamic networks and study their distribution over time @xcite .
time centralities which describe the relative importance of time instants in dynamic networks are proposed in @xcite .
in contrast to this existing body of work , our goal is to facilitate the direct comparison of entire networks and their dynamics , not only parts thereof .
a closely related work but using a different approach is by kunegis @xcite .
kunegis studies the evolution of networks from a spectral graph theory perspective .
he argues that the graph spectrum describes a network on the global level , whereas eigenvectors describe a network at the local level , and uses these results to devise link prediction algorithms
. * bibliographic note . * an early version of this work appeared at the acm fomc 2013 workshop @xcite .
this paper was motivated by the observation that in terms of graph similarity measures , there is no `` one size fits it all '' .
in particular , we have proposed a centrality - based distance measure , and introduced a simple methodology to study the different faces of graph dynamics . indeed
, our experiments confirm that the evolution patterns of dynamic networks are not universal , and different networks need different centrality distances to describe their behavior .
we observe that the edges in networks represent structural characteristics that are inherently connected to the roles of the nodes in these networks .
these structures are maintained under changes , which explains the inertia of centrality distance which capture these properties .
this behavior can be used to distinguish between natural and random network evolution . after analyzing a temporal network trace with a set of distance centralities
, one can guess with confidence for future snapshots if they belong to the trace .
we believe that our work opens a rich field for future research . in this paper , we focused on five well - known centralities and their induced distances , and showed that they feature interesting properties when applied to the use case of dynamic social networks .
however , we regard our approach as a _ similarity framework _ , which can be configured with various additional centralities and metrics , which may not even be restricted by distance metrics , but can be based on the angles between centrality vectors or use existing correlation metrics ( e.g. , pearson correlation , tanimoto coefficient , log likelihood ) .
finally , exploiting the properties of centrality distances , especially their ability to distinguish and quantify between similar evolutionary traces , also opens the door to new applications , such as graph interpolation ( what is a likely graph sequence between two given snapshots of a trace ) and extrapolation , i.e. , for link prediction algorithms based on centralities .
the authors would like to thank clment sire for insightful remarks on a previous version of this document .
* degree centrality @xmath43 : * recall that @xmath26 is the set of neighbors of a node @xmath27 . the _ degree centrality _
is defined as : @xmath85 * betweenness centrality @xmath44 : * given a pair @xmath86 , let @xmath87 be the number of shortest paths between @xmath27 and @xmath88 , and @xmath89 be the number of shortest paths between @xmath27 and @xmath88 that pass through @xmath90 .
the _ betweenness centrality _ is : @xmath91 for consistency reasons , we consider that a node is on its own shortest path , i.e. , @xmath92 , and , by convention , @xmath93 . if @xmath6 is not connected , each connected component is treated independently ( @xmath94 ) .
* ego centrality @xmath95 : * let @xmath96 be the subgraph of @xmath6 induced by @xmath97 .
the _ ego centrality _ is : @xmath98 * closeness centrality @xmath45 : * let @xmath99 be the length of a shortest path between vertices @xmath100 and @xmath101 in @xmath6 .
the _ closeness centrality _ is defined as : @xmath102 * pagerank centrality @xmath46 : * let @xmath103 be a damping factor ( e.g. , the probability that a random person clicks on a link @xcite ) .
the _ pagerank centrality _ of @xmath6 is defined as : @xmath104 * cluster centrality @xmath79 * : the _ cluster centrality _ of a node @xmath27 is the cluster coefficient of @xmath27 , i.e. , the number of triangles in which @xmath27 is involved divided by all possible triangles in @xmath27 s neighborhood . by convention ,
@xmath105 for @xmath106 , and @xmath107 for @xmath108 . for higher degrees:@xmath109
we present here some additional results related to an alternative choice of the null model . as described in the article , we base our methodology on a uniformly random evolutionary null model that is based on the graph edit distance and
hence may not preserve some of intrinsic characteristics of networks under study , such as their density . to complete our study , figure [ figure : altnull ]
provides the results of applying the methodology described in the article using such an alternative null model .
more precisely , we ran the same experiments where the null model is a random process that ensures that the average degree of all sample graphs @xmath63 is the same as for @xmath57 .
figure [ figure : orignull ] recalls the results we obtained for the uniformly random null model for comparison . for 4 out of 5 datasets , namely ht ( hypertext conference ) , in ( infectious )
, ma ( manufacture mails ) and sk ( souk cocktail ) , results obtained in both cases are very similar . for all networks the dynamic signatures are strong , in the sense that the networks are outliers for many of the studied centralities and the signatures of different networks vary , illustrating their unique evolution paths . as expected , the ability of the presented method to distinguish the real network evolution compared to the networks generated according to the more refined null model decreases for most network traces and centralities .
yet , results are strikingly different from the more general null model in the main part of the paper for the case of ca , the caida dataset .
caida differs from the other datasets in the sense that it does not directly derive from human activity ( caida captures autonomous systems relationships ) , and the density in this dataset is much higher than in other considered datasets , while the graph edit distance between different snapshots does not vary much . | the topological structure of complex networks has fascinated researchers for several decades , resulting in the discovery of many universal properties and reoccurring characteristics of different kinds of networks .
however , much less is known today about the _ network dynamics _ : indeed , complex networks in reality are not static , but rather _ dynamically evolve over time_. our paper is motivated by the empirical observation that network evolution patterns seem far from random , but _
exhibit structure_. moreover , the specific patterns appear to depend on the network type , contradicting the existence of a `` one fits it all '' model .
however , we still lack observables to quantify these intuitions , as well as metrics to compare graph evolutions . such observables and metrics
are needed for extrapolating or predicting evolutions , as well as for interpolating graph evolutions . to explore the many faces of graph dynamics and to quantify temporal changes , this paper suggests to build upon the concept of centrality , a measure of node importance in a network .
in particular , we introduce the notion of _ centrality distance _ , a natural similarity measure for two graphs which depends on a given centrality , characterizing the graph type . intuitively , centrality distances reflect the extent to which ( non - anonymous ) node roles are different or , in case of dynamic graphs , have changed over time , between two graphs .
we evaluate the centrality distance approach for five evolutionary models and seven real - world social and physical networks .
our results empirically show the usefulness of centrality distances for characterizing graph dynamics compared to a null - model of random evolution , and highlight the differences between the considered scenarios .
interestingly , our approach allows us to compare the dynamics of very different networks , in terms of scale and evolution speed .
+ * keywords : * network dynamics , graph evolution .
[ theorem]corollary [ theorem]lemma [ theorem]claim [ theorem]definition [ theorem]example [ theorem]fact |
the nuclear medium effects on meson properties are interesting and have been investigated extensively .
the data from relativistic heavy ion collision at cern @xcite and at lower energies at bevalac @xcite , may indicate either a lowering of the @xmath5 meson mass or a large broadening of its width . in the near future , experiments at gsi(hades collaboration @xcite ) could clarify the situation by providing better statistics and mass resolution .
the recent discovery of the deeply bound state of @xmath6 in pb @xcite reveals an upward shift of the pion mass of about 20 mev , and confirms the repulsive s - wave interaction of the @xmath6 in nuclear matter already established from studies of the bulk pionic atom data @xcite .
the low energy magnitudes of the @xmath0-@xmath1 interaction and the properties of the @xmath0 meson in the medium , are still open questions .
the @xmath0 mesic nuclei are expected to provide such informations and are searched at several facilities .
for example , the @xmath7 , @xmath8 and @xmath9 reactions are investigated at gsi @xcite . in this paper
, we study the properties of the @xmath0 in the medium from a theoretical point of view .
we evaluate the effective s - wave @xmath0-@xmath1 interaction in the medium in a chiral unitary approach .
the self - energy of @xmath0 is obtained by summing the @xmath10 interaction in the medium over the nucleons in the fermi sea .
the present work follows closely the method used in @xcite to evaluate the @xmath11 nucleus optical potential in a self - consistent way . in the next section
, we explain our approach .
we show our input related with the nuclear medium in section 3 .
the results are given in section 4 .
section 5 is devoted to the conclusions .
we want to obtain the self - energy of the @xmath0 meson in nuclear matter at various densities @xmath5 , as a function of the @xmath0 energy @xmath12 and the momentum @xmath13 in the nuclear matter frame . in this paper , we calculate it by means of @xmath14 where @xmath15 and @xmath16 are the momentum of the neutron and the fermi momentum at density @xmath5 respectively , and @xmath17 is the @xmath0-neutron in - medium s - wave interaction , with the total 4-momentum of the system @xmath18 in the nuclear matter frame , namely @xmath19 and @xmath20 . here , the isospin symmetry , @xmath21 , is assumed and the amplitude is summed over nucleons in the fermi sea as shown in fig.[fig : etasw ] .
we shall be concerned about the s - wave @xmath0 self - energy . at low energies of the @xmath0
this part of the potential is largely dominant .
= 35 mm we evaluate the in - medium amplitude @xmath22 in a chiral unitary approach . for this purpose
we follow the model for the free space @xmath2-@xmath1 and coupled channels scattering of ref.@xcite , which reproduces the experimental data of @xmath23 scattering up to energies above the @xmath3 region . a similar chiral approach which covers a wider energy range in isospin 1/2 , although with more free parameters , is also done in ref.@xcite . in ref.@xcite the bethe - salpeter equation
is considered with eight coupled channels including two @xmath24 states , namely \{@xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32}. the kernels for the meson - baryon two - body sector are taken from the lowest order chiral lagrangians and improved by applying a form factor corresponding to a vector meson exchange in the t - channel .
the kernels for @xmath33 transitions are determined so that they account for both the @xmath23 elastic and @xmath34 processes .
that model reproduces well the the @xmath23 scattering amplitudes , especially in isospin 1/2 , for the center of mass energy energies from threshold to 1600 mev .
it reproduce also the @xmath35 cross section at the region where the p - wave contribution is negligible . in this coupled channels approach
, the model also provides the @xmath0-@xmath36 interaction in free space and generates dynamically the @xmath3 resonance providing the width and branching ratios for its decay in good agreement with experiment , among them the @xmath10 branching ratio which is quite large for that resonance .
the agreement of the model with the different available data around @xmath3 resonance region and the adequate description of the properties of the resonance , in particular the strong coupling to the @xmath10 state , give us confidence that the model is rather accurate to make predictions on the @xmath37 interaction .
therefore we use this model as a base to take nuclear matter effects into account .
= 30 mm analogously to the free space case , @xmath17 is given by one matrix element of the matrix equation in the space of coupled channels @xmath38^{-1 } v ( { \sqrt s } ) \label{eqn : bseq}\ ] ] which is the solution of bethe - salpeter equation(fig.[fig : bseq ] ) , where the kernel matrix @xmath39 have been factorized on shell as shown in ref.@xcite and ref.@xcite , and @xmath40 is a diagonal matrix of loop functions .
the kernel has nothing to do with nuclear matter and does not depend on density and hence can be parameterized in terms of only the invariant energy @xmath41 or @xmath42 .
one can find the explicit form of it in ref.@xcite .
the effects of the nuclear medium are taken into account through the loop functions @xmath43 , which describe the propagation of intermediate states in the medium .
the meson baryon loop functions for the free space are given by the integral @xmath44 which is the zero density limit of the loop function in medium .
the generalization to the case where the initial meson baryon system is not in the cm frame ( @xmath45 ) , is a necessity in order to evaluate the meson baryon amplitude for arbitrary meson and baryon momenta in the frame of nuclear matter at rest , where the @xmath0 self - energy is evaluated ( see eq.([eqn : selfint ] ) ) .
we regularize the integral by means of a cut off rather than by dimensional regularization as done in ref.@xcite .
the choice of a cut off is preferable when one performs the calculation in nuclear matter since lorentz covariance is manifestly broken given the fact that one has a privileged frame of reference , the one where nuclear matter is at rest .
the integration variable @xmath46 is the momentum of the meson in the loop , but in order to obtain a lorentz invariant quantity when @xmath47 and @xmath48 , the limits in eq.([eqn : labint ] ) must be implemented in such a way that the meson momentum in the rest frame of the original meson baryon state , @xmath49 , should be smaller than the cut off taken .
another possibility is of course to make a boost and work in the meson baryon cm frame which is the option followed in ref.@xcite .
appropriate choices of the cut off and the subtraction constants(@xmath50 ) give the equivalent loop functions as in dimensional regularization ( see also ref.@xcite )
. evaluating the integral in cm frame of the meson baryon system , we find that the choice @xmath51 gives meson baryon loop functions equivalent to those in our previous paper . in order to find @xmath52 which appears in eq.([eqn : bseq ] ) we shall use the formulation of eq.([eqn : labint ] ) and eq.([eqn : subconst ] ) in the free case but will take into account pauli blocking in the intermediate nucleon states , plus the meson and baryon self - energy in the intermediate states .
the pauli blocking is one of the important nuclear matter effects .
we incorporate this by replacing the free nucleon propagator as @xmath53 with the unit step function @xmath54 , which prevents the scattering to intermediate nucleon states below the fermi momentum .
another important nuclear matter effect is the dressing of hadrons .
all the baryons and mesons in the intermediate loops interact with nucleons of the fermi sea and their dispersion relations are changed .
for the baryons , we incorporate this effect , within a mean - field approach , as a momentum - independent binding potential . due to baryon number conservation ,
only the difference between the nucleon and hyperon potentials is relevant for our purpose and is introduced in the hyperon propagator in the kaon - hyperon loops .
for the mesons , we incorporate the dressing effect by introducing the self - energy function @xmath55 .
the meson propagators are replaced , using the lehmann representation , as @xmath56 where @xmath57 is meson spectral function given by @xmath58 } { | ( q^0)^2 - { \vec q\,}^2 - m_l^2 -\pi_{l } ( q^0,\vec q~;\rho)|^2 } \ ] ] as a function of energy and momentum in the nuclear matter frame .
summarizing , the in - medium @xmath23 loop function , for example , is calculated by @xmath59 with @xmath60 \vec p + \vec q\ ] ] from the lorentz boost .
the 2-loop function needed for the @xmath61 channels is also modified in nuclear matter .
we still take a zero constant for the real part according to our previous study in the free space @xcite . on the other hand ,
the imaginary part is modified along the lines of @xmath62 and we obtain @xmath63 \!\!\ ! & = & \!\!\ ! - \int \!\ ! \frac{d^3 \vec
q_1}{(2 \pi)^3 } \ ! \ ! \int \!\ ! \frac{d^3 \vec q_2}{(2 \pi)^3 } ( \vec q_1 - \vec q_2)^2 \frac{m_n}{e_n } \theta\left ( |\vec p-\vec q_1-\vec q_2|-k_f \right ) \\ & & \quad \times \theta\left ( p^0-e_n \right ) \pi \ !
\int_0^{\infty } \ ! \ ! \ ! \ ! d \omega
s_{\pi}(\omega , \vec q_1 ~ ; \rho ) s_{\pi}(p^0-e_n-\omega ,
\vec q_2 ~ ; \rho ) \nonumber\end{aligned}\ ] ] with @xmath64 when both the pauli blocking and the dressed pion are considered .
we use commonly accepted values of baryon binding potentials which are @xmath65 mev for the nucleon and @xmath66 mev for the hyperons .
hence , the difference @xmath67 mev is added to the hyperon energy @xmath68 in the hyperon propagator .
the effective @xmath69-@xmath1 interaction in the medium is studied in ref.@xcite and the resulting kaon self - energy is @xmath70 approximately .
we use this real constant self - energy for kaons in the kaon - hyperon loops .
= 40 mm = 40 mm the pion self - energy which we use in the @xmath23 and @xmath24 loops , consists of a p - wave term and an s - wave term as shown in fig.[fig : piself ] .
the p - wave term includes n - h , @xmath71-h and @xmath72-h excitations , and is basically given by @xmath73 where @xmath74 , and @xmath75 , @xmath76 and @xmath77 are the lindhard functions corresponding to the above excitations respectively .
for example , @xmath77 is given by @xmath78 \label{eqn : unst}\ ] ] @xmath79 where we use @xmath80 .
this term is small compared to the @xmath81-h excitation term whose coupling is @xmath82 , but has a sizable contribution for energetic pions .
the expression for @xmath83 is analogous to that in eq.([eqn : unst]-[eqn : defzdash ] ) and is given in ref.@xcite , the factor @xmath84 in eq.([eqn : unst ] ) is replaced by @xmath85 and the coupling , mass and width of the @xmath86 are replaced by the corresponding magnitudes for the @xmath81 . in the actual calculation we use the p - wave contribution eq.([eqn : pipw ] ) improving with a form factor , a short range correlation and a recoil factor as @xmath87 respectively , where @xmath88 is a monopole form factor with cutoff of 1.2 gev and @xmath89 .
the s - wave term of the pion self - energy is taken in the so called `` @xmath90 '' approximation , @xmath91 using the free space s - wave @xmath2-@xmath1 scattering amplitudes calculated in the present model .
this term includes the contribution of @xmath3-h excitation through the isospin 1/2 amplitude @xmath92 , and is small compared to the p - wave one and becoming only sizable for very energetic pions with about 700 mev momentum .
= 58 mm the total self - energy of pion is given by @xmath93 and in fig.[fig : piprop ] we show the self - energy(left ) , propagator(center ) and spectral function(right ) of the pion for 500 mev momentum for normal nuclear matter density .
in the self - energy(left ) graph , the accompanying dotted and dashed dotted lines correspond to the case when we take only the @xmath1-h and @xmath81-h excitations .
they show that the contribution of @xmath72 is relatively small .
one of the needed input in our evaluation is the self - energy of the @xmath0 which appears in the @xmath27 loop .
however , this is not known beforehand since this is what we want to calculate here .
we perform for this purpose a self - consistent calculation . in a first step ,
the @xmath90 approximation is used since we have previously calculated the free @xmath27 t matrix . with this
we obtain a new self - energy of the @xmath0 which is introduced in the second step , and so on until convergence is reached and one obtains an output @xmath0 self - energy equal to the input one .
= 55 mm fig.[fig : t66free ] shows the free space @xmath0-neutron scattering amplitude @xmath22 obtained in the present model .
it is plotted as a function of the invariant energy @xmath41 , which is 1487 mev at the threshold .
one sees that the amplitude changes drastically , even from attractive to repulsive , above the threshold .
this is due to the coupling to the @xmath3 resonance which is generated dynamically in the present model .
note that the threshold is about 50 mev below the center of the resonance .
we can obtain the @xmath0 self - energy as an explicit function of @xmath12 and @xmath13 as independent variables .
one of the interesting applications is the study of @xmath0 bound states in the nucleus , where the @xmath0 will have small energy and momenta . for this purpose
we can evaluate the @xmath0 self - energy for @xmath94 and @xmath95 as a good approximation .
non - local effects from the consideration of the energy and momentum dependence of the self - energy give very small corrections for weakly bound states as demonstrated in ref.@xcite for @xmath96 atoms . in order to evaluate the @xmath0 self - energy in a first step
, we substitute the free @xmath27 amplitude in eq.([eqn : selfint ] ) , setting @xmath97 , and obtain the @xmath0 self - energy for zero momentum shown in fig.[fig : iiabc ] top .
this simplest calculation is nearly equal to the @xmath90 approximation .
the magnitude of the self - energy increases almost linearly in @xmath5 and its shape reflects that of @xmath22 .
= 55 mm = 55 mm = 55 mm then we turn on the pauli blocking in the way explained in section 2 .
the resulting @xmath0 self - energy for zero momentum is shown in fig.[fig : iiabc ] middle . comparing to the top one , we see the interesting effects of the pauli blocking .
first of all , the resonant shape is strongly enhanced .
this can be understood as the reduction of the decay width of the resonance , since the pauli blocking forbids the decay to nucleons with momentum smaller than @xmath16 .
second , the peak of the resonance , seen in the imaginary part of the amplitudes , or equivalently in the zero of the real part , is only shifted moderately to higher energies .
this is in contrast with the @xmath11 case , where the @xmath98 resonance in the @xmath99 interaction , is about 80 mev shifted upward @xcite . as pointed out in ref.@xcite , this is natural because a large part of the @xmath3 resonance generated in the present model , comes from the @xmath100 , @xmath30 and @xmath28 components @xcite , which have nothing to do with pauli blocking . in ref.@xcite
it is argued that the fermi motion compensates the decay width reduction due to the pauli blocking .
there also is some effect of this sort here , but we still find a net reduction of the @xmath3 width due to the consideration of both effects , resulting in an enhanced strength of both the real and imaginary parts of the @xmath0 self - energy . a major difference in our model from ref.@xcite , at the present stage of the calculation , is the presence of the 3-body @xmath24 channels . also , our free space @xmath101 shown in fig.[fig : t66free ] , is already different from that in ref.@xcite , where the peak of the resonance is seen at around 1590 mev . finally , we turn on the dressing of the hadrons in addition to pauli blocking .
this study is the main novelty of the present work .
we use the input explained in the previous section . besides this , we need the spectral function of the @xmath0 in order to calculate the @xmath27 loop function .
as stated in the previous section , we determine it in a self - consistent way by iterating the bethe - salpeter equation .
this means that we need to study the finite momentum case even if we are interested only in the @xmath0 with zero momentum .
we , hence , evaluate the @xmath0 self - energy also for finite momentum and , for example , fig.[fig : sp2d ] left shows a rough sketch of the @xmath0 spectral function obtained at normal nuclear matter density as a function of @xmath12 and @xmath13 .
the results in what follows include the dressed @xmath0 accounting for the @xmath0 self - energy as a function @xmath12 and @xmath13 .
= 60 mm = 55 mm the @xmath0 self - energy for zero momentum obtained in this self - consistent approach is shown in fig.[fig : iiabc ] bottom .
one can see the effects of the hadron dressing in comparison to the middle figure of the panel .
apparently , the strength is spread wider .
this is expected because the @xmath0 spectrum is distributed as stated above .
even then , the resonant shape is still clearly visible .
the center of the resonance is shifted upward , for example , about 25 mev for normal nuclear matter density .
this means that the sum of the repulsion on @xmath69 and so on , surpasses the attraction on the pions .
the difference of the hyperon binding energy and the nucleon one , works as a repulsion for this calculation .
the dressed @xmath0 also shifts it upward slightly because the @xmath0 with large momentum(larger than about 250 mev ) feels a weak repulsion , as shown in fig.[fig : sp2d ] right .
recall that the @xmath102 , @xmath103 and @xmath27 components are more important than the @xmath23 and @xmath61 components for the resonance in the present model .
the 25 mev shift for normal nuclear matter density is small(less than 2% of the @xmath3 mass ) and consistent with @xmath0 photo - production data @xcite , which together with the theoretical analysis of ref.@xcite assuming no shift of the resonance , suggest that the in - medium mass of the resonance is almost the same as in free space .
= 55 mm = 55 mm fig.[fig : spfull ] left shows the spectral function of a zero momentum @xmath0 corresponding to the final self - energy .
we see still a narrow peak even for normal nuclear matter density and , hence , it makes sense to regard the @xmath0 in the medium as a quasi - particle with a modified mass .
it is interesting to note a second peak in the spectral function at higher energies , which , as noted in ref.@xcite , comes from the coupling of the @xmath0 to the @xmath3-h .
we can observe in the figure that as the density increases the two peaks move away from each other .
we define the effective mass @xmath104 as the energy @xmath12 which satisfies @xmath105=0 ~ , \label{eqn : effmass}\ ] ] which is almost the position of the peak in the spectral function .
the result is plotted in fig.[fig : spfull ] right , which indicates that approximately one has @xmath106 in short .
this moderate downward shift agree with the result of @xcite . while , a much stronger , about 11% , downward shift for normal nuclear matter density , is reported in @xcite where the quark meson coupling model is used .
it should be interesting to test these predictions from experimental data of @xmath0 mesic nuclei .
the optical potential @xmath107 and the effective @xmath0-neutron scattering length @xmath108 defined as @xmath109 are useful when we study the bound state of an @xmath0 in nuclei using the local density approximation .
we plot them in fig.[fig : optfull ] .
one can see that the real and imaginary parts of the optical potential deviate as the density increases .
the real part becomes larger than the linear @xmath110 approximation while the imaginary part is nearly @xmath110 .
this behavior is better seen in the effective scattering length , where the real part increases and the imaginary part almost stays the same .
the depth of the optical potential at normal nuclear matter density is @xmath111 mev .
this value is not in contradiction with the 5% ( 28 mev ) mass shift of the @xmath0 , because the self - energy is evaluated at an @xmath0 energy equal to the @xmath0 mass , while the effective @xmath0 mass , @xmath12 in eq.([eqn : effmass ] ) , is obtained using the @xmath0 self - energy calculated at same @xmath12 energy .
it is interesting to compare these result with the ones obtained in the literature . in ref.@xcite
the potential obtained was @xmath112 which provides an imaginary part similar to ours but the real part is about one half of the one obtained here . in ref.@xcite
the potential obtained assuming that the mass of the @xmath3 did not change in the medium , as we showed it was approximately the case here , was @xmath113 at @xmath114 , somewhere in between the two former results .
= 55 mm = 55 mm in order to facilitate the task of making an accurate as possible prediction for the @xmath0 bound states in nuclei , we parameterize our results for the @xmath0 self - energy as a function of energy and density . given the strong energy dependence of the @xmath115 seen in fig.[fig : iiabc ] for @xmath116 , the consideration of this explicit energy dependence in the klein - gordon equation should be important .
we can parameterize our results in the region @xmath117 , as @xmath118 & = & a(\rho ) + b(\rho)(k^0-m_{\eta } ) + c(\rho)(k^0-m_{\eta})^2 + d(\rho)(k^0-m_{\eta})^3 \\
\mbox{im } [ \pi_{\eta}(k^0 , \vec 0~ ; \rho ) ] & = & e(\rho ) + f(\rho)(k^0-m_{\eta } ) + g(\rho)(k^0-m_{\eta})^2 + h(\rho)(k^0-m_{\eta})^3\end{aligned}\ ] ] with @xmath119 by using the klein - gordon equation , and substituting @xmath120 in the spirit of the local density approximation , one can then obtain binding energies and widths of the @xmath0 state in different nuclei .
in the present paper we have used a chiral unitary approach successfully applied to the study of the @xmath23 interaction and its coupled channels , in particular the @xmath10 channel , in order to evaluate the @xmath0 self - energy in a nuclear medium .
we have taken into account the standard many body effects like fermi motion and the pauli blocking in the intermediate @xmath1 states .
in addition we have also included the self - energy of the baryons and mesons in the intermediate states , including the @xmath0 self - energy in a self - consistent way .
the results obtained are interesting . while qualitatively similar to other ones obtained before , we obtain however a deeper potential with a real part about twice as big as in former , more simplified , studies .
in addition we show that the energy dependence of the @xmath0 self - energy is very pronounced below the @xmath0 threshold and it would be interesting to consider it in studies of @xmath0 bound states in nuclei . for this purpose
we have parameterized our results in an easy analytical form which can be used to solve the klein - gordon equation for @xmath0 bound states or to interpret some physical processes where @xmath0 states close to threshold play some role @xcite .
the stronger real part of our potential compared to previous ones and the fact that the imaginary part of the potential decreases rapidly as the @xmath0 energy decreases , open new hopes that @xmath0 mesic states , relatively wide , but narrow enough compared to the binding energy , could exist and be identified in actual experiments .
we would like to thank prof .
vicente vacas for useful discussions .
this work has been partly supported by the spanish ministry of education in the program `` estancias de doctores y tecnlogos extranjeros en espaa '' , by the dgicyt contract number bfm2000 - 1326 and by the eu tmr network eurodaphne , contact no .
erbfmrx - ct98 - 0169 . | the self - energy of an @xmath0 meson in the nuclear medium is calculated in a chiral unitary approach .
a coupled channel bethe - salpeter equation is solved to obtain the effective @xmath0-@xmath1 interaction in the medium .
the base model reproduces well the free space @xmath2-@xmath1 elastic and inelastic scattering at the @xmath0-@xmath1 threshold or @xmath3 region .
the pauli blocking on the nucleons , binding potentials for the baryons and self - energies of the mesons are incorporated , including the @xmath0 self - energy in a self - consistent way .
our calculation predicts about @xmath4 mev for the optical potential at normal nuclear matter for an @xmath0 at threshold but also shows a strong energy dependence of the potential . |
a relevant quantity to understand the breaking of the @xmath2 symmetry in qcd is the topological susceptibility @xmath3 in pure yang - mills theory @xcite @xmath4 where @xmath5 is the topological charge density .
the prediction is @xcite @xmath6 where @xmath7 is the relevant number of flavours .
( [ 180 ] ) implies a well defined prescription @xcite to deal with the @xmath8 singularity in eq .
( [ qchi1 ] ) . in @xcite @xmath3 was evaluated at zero and finite temperature for @xmath1 yang - mills theory .
the value obtained at zero temperature was in agreement with the prediction of eq .
( [ 180 ] ) .
the value of @xmath3 at finite temperature displayed a sharp drop beyond the deconfinement transition .
here we give a short review of a similar calculation for the @xmath0 gauge group @xcite .
in addition we discuss the comparison with the geometric method , and we show that , after a proper renormalization of the latter , the two procedures give consistent results @xcite .
let @xmath9 be any definition of the topological charge density on the lattice .
the lattice topological susceptibility @xmath10 is defined as @xmath11 the lattice regulated @xmath9 is related to the continuum @xmath12 @xmath13 by a finite renormalization @xcite @xmath14 where @xmath15 in the usual notation . since @xmath10 does not obey in general the prescription @xcite leading to eq .
( [ 180 ] ) , besides the multiplicative renormalization of eq .
( [ z ] ) there is also an additive renormalization @xmath16 where @xmath17 contains mixings with operators of dimension @xmath18 . in order to extract the physical signal @xmath3 from eq .
( [ m ] ) , we need a determination of the renormalization constants @xmath19 and @xmath20 .
we will determine them using the non perturbative method of ref .
we have used wilson action and the usual heat - bath updating algorithm .
the scale @xmath21 was fixed by using the results of ref .
@xcite .
we have used various definitions for @xmath23 .
the @xmath24-smeared field theoretical @xmath25 is defined as @xmath26 @xmath27 is the plaquette in the @xmath28 plane constructed with @xmath24-times smeared links @xmath29 @xcite .
we call @xmath30 and @xmath31 the additive and multiplicative renormalization constants for the @xmath24-smeared operators .
of course @xmath3 must be independent of the choice of the operator @xmath23 .
this is visible in figure 1 , for @xmath3 at zero temperature . up and down - triangles
indicate the value of @xmath3 as obtained from the 0-smear and 2-smear data respectively .
there is good scaling and @xmath32 mev , the first error is statistical and the second comes from the error in @xmath33 @xcite . in fig .
1 we also report the geometric susceptibility @xmath34 @xcite .
the usual determination is done by identifying @xmath35 , and claiming that the geometrical susceptibility has no additive renormalization .
@xmath34 is shown in figure 1 by the stars : it is one order of magnitude bigger than the field theoretical determinations and no scaling is observed . by using the same method as for the field theoretical determination , we have measured the multiplicative renormalization @xmath36 , finding @xmath37 within errors . at the same time
we have determined and subtracted the additive renormalization @xmath38 .
this brings down the resulting @xmath3 by a factor of 10 .
the results are shown by the circles in figure 1 and are in agreement with the field theoretical results . by renormalizing
we have eliminated the so called dislocations . to support the necessity of the subtraction of @xmath38 ,
we have also computed the correlation function @xmath39 by reflection positivity we expect that @xmath40 at @xmath41
. since @xmath42 , the susceptibility is mainly determined by the singularity at @xmath43 .
this correlator , for @xmath44 lying along a coordinate axis , is shown in figures 2 for the smeared charges and in figure 3 for the geometrical charge .
the peak at @xmath43 for the geometric charge is @xmath45 orders of magnitude larger than for the @xmath24-smeared charges , indicating that @xmath38 is much bigger than @xmath46 , and is @xmath47 of the observed @xmath10 .
the data show again a drop at the transition .
however this is less sharp than for the @xmath1 case @xcite . in figure 4
we show the behaviours for @xmath0 and @xmath1 of the ratio @xmath52 , where @xmath53 indicates the physical susceptibility at temperature @xmath54 .
the slope for the @xmath1 data is steeper . in both cases the data at @xmath55 show a constant value consistent with the value at @xmath56 .
9 e. witten , nucl .
b156 ( 1979 ) 269 .
g. veneziano , nucl .
b159 ( 1979 ) 213 .
b. alls , m. delia , a. di giacomo , nucl .
b494 ( 1997 ) 281 .
b. alls , m. delia , a. di giacomo , hep - lat/9706016 , to appear in phys .
b. b. alls , m. delia , a. di giacomo , r. kirchner , in preparation .
m. campostrini , a. di giacomo , h. pana - gopoulos , phys .
b212 ( 1988 ) 206 .
a. di giacomo , e. vicari , phys .
b275 ( 1992 ) 429 .
b. alls , m. campostrini , a. di giacomo , y. gnd , e. vicari , phys .
d48 ( 1993 ) 2284 .
j. engels , f. karsch , k. redlich , nucl .
b435 ( 1995 ) 295 . j. fingberg , u. heller , f. karsch , nucl .
b392 ( 1993 ) 493 . c. christou , a. di giacomo , h. panagopoulos , e. vicari , phys .
d53 ( 1996 ) 2619 .
m. lscher , commun .
85 ( 1982 ) 39 .
a. phillips , d. stone , commun . math .
103 ( 1986 ) 599 . | new results on the topology of the @xmath0 yang - mills theory are presented . at zero temperature
we obtain the value of the topological susceptibility by using the recently introduced smeared operators as well as a properly renormalized geometric definition .
both determinations are in agreement . at non - zero temperature
we study the behaviour of the topological susceptibility across the confinement transition pointing out some qualitative differences with respect to the analogous result for the @xmath1 gauge theory . |
extreme conditions in plasma such as high temperatures and pressures can lead to different features of nonlinear wave dynamics , including propagation and collision , from those encountered in ordinary ionized environments @xcite .
it is well known that dispersive dynamics arise due to quantum diffraction phenomenon in many of degenerate plasma kinds @xcite .
it has also been confirmed that in a degenerate plasma there may be critical values which govern the nonlinear dynamics @xcite .
the degeneracy feature which is a fundamental aspect of ordinary solids arise due to exclusion mechanism when the de broglie thermal wavelength @xmath0 is comparable or higher than inter - particle distances @xcite and leads to a much higher pressure on degenerated species compared to the non - degenerated ones .
it has been shown @xcite that in a completely degenerate astrophysical object such as a white dwarf the electrons , which follow fermi - dirac statistics rather than that of boltzmann , become relativistically degenerate due to gigantic inward gravitational pressure and this can lead to softening of the degeneracy pressure giving rise to the ultimate collapse of the star .
the effects of the relativistic degeneracy has recently been considered in electron - positron - ion plasmas on large - amplitude nonlinear wave dynamics by number of authors in ultra - relativistic and non - relativistic degeneracy limits @xcite . on the other hand ,
more recently , the dynamics of nonlinear ion waves has been investigated in electron - ion plasma considering a wide range of relativistic degeneracy and it has been confirmed that the relativity parameter plays a crucial role in propagation of such waves in dense fermi - dirac plasmas @xcite .
* however , to our current knowledge , effect of the relativistic degeneracy parameter , @xmath1 ( defined in next section ) , which is directly related to quantities such as the electron number - density , @xmath2 , and mass - density of plasma , @xmath3 , has not been considered in collision dynamics of solitary structure .
* historically , discovery of the remarkable shape - preservation feature of these waves during interactions by zabusky et.al .
@xcite in 1965 , made their first important applications in communication technology .
later , washimi et.al @xcite showed that such waves , in a weakly nonlinear regime , can be mathematically modeled by the well known korteweg - de vries ( kdv ) equations and oikawa et.al . @xcite
have used an extended approach to consider the interaction of such waves as the superposition of two single kdv - type solitons .
electron - positron - ion plasmas are important because they appear in many astrophysical environment such as , active - galactic - nuclei , pulsar magnetospheres , neutron stars and supernovas etc .
they may even play important role in * evolution of the early universe * @xcite .
electron - positron - ion plasmas can also be produced in intense - laser matter interaction experiments @xcite .
it has been noted @xcite that the characteristic nonlinear wave frequencies are comparably higher compared to that of pair annihilation in such dense plasmas so that these waves * survive in these extreme environments*. in the current study we consider in a magnetized electron - positron - ion plasma the propagation and head - on collisions of ion - acoustic solitary waves in a wide range of relativity parameter , i.e. plasma mass - density , thus the findings may well be applicable to inertial confinement fusion electron - positron - ion plasmas .
the article is organized in the following way .
the basic normalized hydrodynamics equations are introduced in section [ basic ] .
evolution equations along with collision parameters are derived in section [ shift ] . numerical analysis and
* discussion are * presented in section [ discussion ] and final remarks are drawn in section [ conclusion ] .
considering a quantum fully degenerate plasma with * relativistically degenerate * electrons / positrons , we will use the conventional quantum hydrodynamic ( qhd ) fluid equations to describe the dynamics of nonlinear excitations .
the pair - annihilation rate is ignored compared to characteristic plasma frequencies @xcite and the plasma is considered as collision - less due to pauli - blocking mechanism . therefore , the closed set of qhd equations may be written in the following dimensional form @xmath4 where , @xmath5 , @xmath6 and @xmath7 , for electrons and positrons , respectively , and @xmath8 is the normalized plank constant .
it is noted that , in fully - degenerate configuration from fermi - dirac statistical model , the relativistic degeneracy pressure is expressed in the following general form , which is valid to the electrons / positrons with arbitrary degree of relativistic degeneracy @xcite @xmath9 } \right\},\ ] ] * in which the relativity parameter @xmath10 @xcite ( @xmath11 ) is the ratio of electron / positron fermi relativistic momentum @xmath12 to @xmath13 . * thus , with the new relativistic degeneracy pressure term the basic equations may be rewritten as @xmath14^{1/2 } + { \frac{{{m_j}}}{{{m_i}}}}\left ( \frac{{{h_r^{2}}}}{2}\right)\nabla \left ( { \frac{{{\nabla ^2}\sqrt { { n_j } } } } { { \sqrt { { n_j } } } } } \right ) , \\ { \nabla ^2}\phi = - 4\pi e\left ( { \sum\limits_j { { s_j}{n_j } } + { n_i } } \right ) , \\ \end{array}\ ] ] where , @xmath15 ( @xmath16 for electrons and positrons , respectively ) is a measure of the relativistic effects ( called the relativity parameter ) and @xmath17 , is the quantum diffraction coefficient which is related to the relativity parameter via @xmath18 . on the other hand , the parameter @xmath19 is related also to the mass - density ( of white dwarf , for instance ) through the relation @xmath20 or @xmath21 with @xmath22 , where , @xmath23 is the proton mass . note that , the density @xmath24 is exactly within the range of mass - density of a typical white dwarf and a contribution of @xmath25 is included in mass - density definition for the electron - positron pair production / anihilation effect
. the density of typical white dwarfs can be in the range @xmath26 , which , neglecting electron - positron pair production / anihilation , results in values of @xmath27 for the relativity parameter .
* before proceeding with calculations , a clear definition of the relativistic degeneracy and distinction between a low - pressure relativistic plasma from the relativistically degenerate quantum fermi - gas is in order .
the relativistic degeneracy is a completely quantum phenomenon ruled by the uncertainty principle and is increased due to the decrease in inter - fermion distances in degenerated plasmas . although the relativistic effects arise due to increase in fermion number - density in a dense degenerate plasma , however , unlike for the low - pressure relativistic plasmas the degeneracy pressure in the fermion momentum fluid equation usually dominates the relativistic dynamic effects in super - dense plasma state .
chandrasekhar @xcite , combining the relativity and the quantum statistics , showed that for dense degenerate fermi - gas such as a white - dwarf with a mass - density , @xmath3 , the degeneracy pressure turns from @xmath28 ( with polytropic index @xmath29 ) dependence for normal degeneracy for the limit @xmath30 to @xmath31 ( with polytropic index @xmath32 ) dependence for ultra - relativistic degeneracy case in the limit @xmath33 .
the relativistic degeneracy starts at mass density of about @xmath34 of the order in the core a @xmath35 white dwarf , which corresponds to a fermi - momentum @xmath36 ( corresponding to the relativistic degeneracy parameter value of @xmath37 ) or the threshold velocity of @xmath38 ( the fermi relativistic factor @xmath39 ) . now comparing the terms in the momentum equation eq .
( [ normal ] ) reveals that the term containing mass ratio is still negligible compared to the degeneracy pressure term .
the similar treatment of ultrarelativistic quantum plasma as in our case has been considered in refs .
. however , this assumption breaks down for a relativistic low - pressure plasma and one has to follow a coherent relativistic formalism given in refs . @xcite .
thus , for the range of the relativistic degeneracy parameter @xmath40 , which is used in our analysis ( see discussion ) , it is safe to neglect the relativistic dynamic effects compared to that of the dominant relativistic degeneracy pressure . on the other hand
, it should be noted that the current analysis includes the whole range of degeneracy parameter , @xmath19 , and is not limited only to the relativistic degeneracy range .
* the normalized set of qhd equations can be obtained from the following scaling @xmath41 where , @xmath42 , @xmath43 and @xmath44 are the vacuum light speed and the characteristic plasma frequency and relativistic sound - speed , respectively , and the parameter @xmath45 denotes the electron equilibrium density .
now , neglecting the terms containing the small mass - ratio , @xmath46 and assuming that @xmath47 , we arrive at the following simplified set of basic normalized equations @xmath48^{1/2},\\ { \nabla ^2}\phi = - { \sum\limits_j { { s_j}{n_j } } - { n_i } } , \\
\end{array}\ ] ] where , @xmath49 with the ion - cyclotron frequency of @xmath50 .
solving eqs .
( [ comp ] ) for the electron and positron number - densities in terms of electrostatic potential with the appropriate boundary requirements ( @xmath51 and @xmath52 ) , results in the following energy relations @xmath53^{3/2 } } \\ { n_p } = { \alpha ^ { - 1}}r_0^ { - 3}{\left [ { r_0 ^ 2{\alpha ^{4/3 } } - \phi ( 2\sqrt { 1 + r_0 ^ 2{\alpha ^{4/3 } } } - \phi ) } \right]^{3/2 } } , \\ \end{array}\ ] ] where , the equilibrium charge neutrality condition is given by poisson s relation as @xmath54 transforming the normalized plasma equations ( eqs .
( [ normal ] ) ) to the appropriate strained coordinate defined below admits successful separation of variables and allows elimination of secular terms , which consequently , leads to the desired evolution equations and the corresponding collision parameters @xcite @xmath55 where , the interrelated functions @xmath56 and
@xmath57 ( @xmath58 ) denote the phase of the traveling solitary waves to be determined together with the evolution equations in next section .
the solitons are assumed to travel at directions described by cosine indices @xmath59 and the direction of magnetic field with respect to the collision line is presented by @xmath60 angle which is determined in the following relation @xmath61 now , expanding the dependent plasma variables around equilibrium state through the smallness , @xmath62 parameter which is of the order of perturbation amplitude being a measure of nonlinearity strength @xcite , we get @xmath63 = \left [ { \begin{array}{*{20}{c } } 1-\alpha \\ { \begin{array}{*{20}{c } } 0 \\ 0 \\ 0 \\ \end{array } } \\ 0 \\ \end{array } } \right ] + { \varepsilon ^2}\left [ { \begin{array}{*{20}{c } } { n_i^{(1 ) } } \\ { \begin{array}{*{20}{c } } 0 \\ 0 \\ { w_i^{(1 ) } } \\ \end{array } } \\ { { \varphi ^{(1 ) } } } \\ \end{array } } \right ] + { \varepsilon ^3}\left [ { \begin{array}{*{20}{c } } { n_i^{(2 ) } } \\ { \begin{array}{*{20}{c } } { u_i^{(1 ) } } \\ { v_i^{(1 ) } } \\ { w_i^{(2 ) } } \\ \end{array } } \\ { { \varphi ^{(2 ) } } } \\ \end{array } } \right ] + { \varepsilon ^4}\left [ { \begin{array}{*{20}{c } } { n_i^{(3 ) } } \\ { \begin{array}{*{20}{c } } { u_i^{(2 ) } } \\ { v_i^{(2 ) } } \\ { w_i^{(3 ) } } \\ \end{array } } \\ { { \varphi ^{(3 ) } } } \\ \end{array } } \right ] + \ldots\ ] ] the reduced set of plasma equations in new strained coordinate is presented in appendix a. from the lowest - orders in @xmath62 we obtain the following relations [ leading ] @xmath64 @xmath65 @xmath66 @xmath67 @xmath68{\varphi ^{(1)}},\ ] ] which leads to the following first - order approximations of plasma variables @xmath69\left [ { \varphi ^{(1)}(\xi , \tau ) + \varphi ^{(1)}(\eta , \tau ) } \right ] , \\
u_i^{(1 ) } = - \frac{{l}}{{\bar \omega}}\left [ { { \partial _ \xi}\varphi ^{(1)}(\xi , \tau ) + { \partial _ \eta } \varphi ^{(1)}(\eta , \tau ) } \right],\hspace{3 mm } v_i^{(1 ) } = \frac{{k}}{{\bar \omega}}\left [ { { \partial _ \xi } \varphi ^{(1)}(\xi , \tau ) + { \partial _ \xi } \varphi ^{(1)}(\eta , \tau ) } \right ] , \\
w_i^{(1 ) } = \frac{{c}}{{m(1-\alpha ) } } 3r_0^ { - 2}\left [ { \sqrt { 1 + r_0 ^ 2 } + { \alpha ^ { - 1/3}}\sqrt { 1 + r_0 ^ 2{\alpha ^{4/3 } } } } \right]\left [ { \varphi ^{(1)}(\xi , \tau ) -
\varphi ^{(1)}(\eta , \tau ) } \right],\\ \end{array}\ ] ] on the other hand , the compatibility relation can be written as @xmath70,\ ] ] with the normalized phase - speed , @xmath42 , given as @xmath71 from the next higher - order in @xmath62 , we get the second - order approximations for variables which are of the following shapes * * [ second ] @xmath72 @xmath73 @xmath74 @xmath75 @xmath76{\varphi ^{(2)}},\ ] ] yielding the following second - order approximates @xmath77\left [ { \varphi ^{(2)}(\xi , \tau ) + \varphi ^{(2)}(\eta , \tau ) } \right ] , \\
u_i^{(2 ) } = \frac{{ck}}{{{{\bar \omega } ^2}}}\left [ { { \partial _ \xi } \varphi ^{(2)}(\xi , \tau ) - { \partial _ \eta } \varphi ^{(2)}(\eta , \tau ) } \right ] - \frac{{l}}{{\bar \omega}}\left [ { { \partial _ { \xi \xi } } \varphi ^{(1)}(\xi , \tau ) + { \partial _ { \eta \eta } } \varphi ^{(1)}(\eta , \tau ) } \right ] , \\
v_i^{(2 ) } = \frac{{cl}}{{{{\bar \omega } ^2}}}\left [ { { \partial _ \xi } \varphi ^{(2)}(\xi , \tau ) - { \partial _ \eta } \varphi ^{(2)}(\eta , \tau ) } \right ] + \frac{{k}}{{\bar \omega}}\left [ { { \partial _ { \xi \xi } } \varphi ^{(1)}(\xi , \tau ) + { \partial _ { \eta \eta } } \varphi ^{(1)}(\eta , \tau ) } \right ] , \\
w_i^{(2 ) } = \frac{{c}}{{m(1-\alpha ) } } 3r_0^ { - 2}\left [ { \sqrt { 1 + r_0 ^ 2 } + { \alpha ^ { - 1/3}}\sqrt { 1 + r_0 ^ 2{\alpha ^{4/3 } } } } \right]\left [ { \varphi ^{(2)}(\xi , \tau ) -
\varphi ^{(2)}(\eta , \tau ) } \right ] .
\\ \end{array}\ ] ] where , @xmath78 and @xmath79 describe the first - order amplitude evolution and @xmath80 and @xmath81 describe the second - order amplitude evolution components of two distinct solitary excitations in the oblique directions @xmath82 and @xmath83 ( @xmath84 ) , respectively .
in what follows , we will always employ the notations @xmath85 and @xmath86 instead of @xmath78 and @xmath79 for clarity .
proceeding to the next higher - order approximation , again by solving the coupled differential equations in this approximation level and by making use of compatibility relation ( eq .
[ dispers ] ) and the previous plasma approximations , we obtain @xmath87\xi - \frac{n}{4}\left [ { \frac{{\partial \varphi _ \xi^{(1)}}}{{\partial \tau } } - a\varphi _ \xi^{(1)}\frac{{\partial \varphi
_ \xi^{(1)}}}{{\partial \xi } } + b\frac{{{\partial ^3}\varphi _
\xi^{(1)}}}{{\partial { \xi^3 } } } } \right]\eta + \\ { \frac{e_2}{4}}\left [ { { p_0}(\eta,\tau ) + \frac{e_1}{e_2}\int { \varphi _ \eta^{(1)}d\eta } } \right]\frac{{\partial \varphi _ \xi^{(1)}}}{{\partial \xi } } -{\frac{e_2}{4}}\left [ { { q_0}(\xi,\tau ) - \frac{e_1}{e_2}\int { \varphi _ \xi^{(1)}d\xi } } \right]\frac{{\partial \varphi _ \eta^{(1)}}}{{\partial \eta } } + \\ \frac{n}{4}\left [ { \int { \frac{{\partial \varphi _ \xi^{(1)}}}{{\partial \tau}}d\xi } - \int { \frac{{\partial \varphi _ \eta^{(1)}}}{{\partial \tau}}d\eta}}\right ] - \frac{e_1}{8}\left [ { { { ( \varphi _ \xi^{(1)})}^2 } - { { ( \varphi _ \eta^{(1)})}^2 } } \right ] - \frac{e_1}{4}\left [ { \frac{{{\partial ^2}\varphi
_ \xi^{(1)}}}{{\partial { \xi^2 } } } - \frac{{{\partial ^2}\varphi _
\eta^{(1)}}}{{\partial { \eta^2 } } } } \right ] + \\
f(\xi,\tau ) + g(\eta,\tau ) , \\ \end{array}\ ] ] where , the entities @xmath88 and @xmath89 are assumed to be the homogenous solutions of differential equations . the coefficients in eq . ( [ n3 ] )
are as follows [ coeffs ] @xmath90^{3/2}}\cos \gamma,\ ] ] @xmath91^{3/2}}\cos \gamma,\ ] ] @xmath92^ { - 3/2}}\sec \gamma,\ ] ] @xmath93 @xmath94 as it is realized already , the two first terms in eqs .
( [ n3 ] ) are secular and they diverge at @xmath95 and @xmath96 , and therefore must vanish .
thus , we obtain two distinct kdv evolution equations one for each traveling solitary structure . furthermore , the next two terms in eq .
( [ n3 ] ) may become secular @xcite at the next higher - order and they must also vanish .
the later conditions determine the collision parameters introduced in eqs .
( [ stretch ] ) .
however , the full determination soliton dynamics is given by the following coupled differential equations @xmath97
@xmath98 @xmath99 @xmath100 in order to get single - soliton solutions for eqs .
( [ kdv1 ] ) and ( [ kdv2 ] ) , with multi - soliton solutions , we require that the perturbed potential components and their derivatives vanish at infinity , i.e. @xmath101 then , we obtain @xmath102 @xmath103 where , @xmath104 and @xmath105 represent the soliton amplitude and width , respectively , and @xmath106 is the relativistic mach - speed .
close inspection of eqs .
( [ coeffs ] ) , reveals that the kdv coefficients @xmath107 and @xmath108 in eqs .
( [ kdv1 ] ) and ( [ kdv2 ] ) change sign at @xmath109 , a critical value which determines weather the solitons are compressive or rarefactive .
moreover , it is remarked that the soliton amplitude is independent of the strength of magnetic field , while it strictly depends on applied field direction , @xmath110 .
however , the soliton width depends on both magnitude and direction of the external magnetic field .
this is in agreement with previous findings @xcite .
the collision phase - shifts of solitary excitations are obtained using eqs .
( [ p0 ] ) and ( [ q0 ] ) together with the kdv solutions ( eqs .
( [ phi - x ] ) and ( [ phi - y ] ) ) as @xmath111 @xmath112 we can write , up to order @xmath113 @xmath114 - \varepsilon^2\frac{{{{e'}_1}}}{{{{e'}_2}}}{\varphi _ { \xi0}}{\delta _ \xi}\tanh ( \frac{{\xi + { u_{\xi0}}\tau}}{{{\delta _ \xi}}})+o(\varepsilon^3 ) , \\ \end{array}\ ] ] we can also calculate the overall phase - shifts by comparing the phases of each wave long before and after the collision event in the following form @xmath115- \mathop { \lim } \limits_{\xi=0,\eta \to - \infty } [ \varepsilon ( { k}x + { l}y + { m}z + { c}t ) ] ,
\\ \delta { q_0 } = q_{post - collision}-q_{past - collision}=\\ \mathop { \lim } \limits_{\eta=0,\xi \to + \infty } [ \varepsilon ( { k}x + { l}y + { m}z - { c}t)]- \mathop { \lim } \limits_{\eta=0,\xi \to - \infty } [ \varepsilon ( { k}x + { l}y + { m}z - { c}t ) ] , \\ \end{array}\ ] ] the phase parameters , @xmath116 and @xmath117 , therefore , present the overall phase - shifts of solitary structures labeled @xmath118 and @xmath119 in the head - on collision . finally , making use of eqs .
( [ phase - x ] ) , ( [ phase - y ] ) and ( [ stretch ] ) , we obtain the following simplified expressions @xmath120{\varphi _ { \eta 0}}{\delta _ \eta } , \\
\delta { q_0 } = - { \varepsilon ^2}\left[\frac{{1 + 2r_0 ^ 2(1 - { \alpha ^ { - 1/3 } } ) - { \alpha ^ { - 5/3}}}}{{2r_0 ^ 2\left ( { \sqrt { 1 + r_0 ^ 2 } + { \alpha ^ { - 1/3}}\sqrt { 1 + r_0 ^ 2{\alpha ^{4/3 } } } } \right)}}\right]{\varphi _ { \xi 0}}{\delta _ \xi } .
\\ \end{array}\ ] ]
in this section we evaluate the nonlinear wave dynamics using typical values of mass - density for a degenerate white dwarf .
figure 1 depicts the variation of amplitude and width of a solitary excitations with respect to various parameters in a magnetized degenerate plasma with relativistically degenerate electrons and positrons .
it is remarked from figs .
1(a ) and 1(b ) that , the increase of the mass - density enhances both the amplitude and width of the excitations for fixed other plasma parameters .
it is also indicated that the only compressive solitons are present in this plasma for all given parameters . however , regarding the relative positron density variations , one observes that , the soliton amplitude / width increases / decreases as the relative positron density increases .
moreover , as indicated by figs 1(c ) , the amplitude of soliton is minimum when the magnetic field is parallel to the propagation direction and increases as the the field direction approaches to the normal , @xmath121 position .
the soliton width , on the other hand , shows ( fig .
1(d ) ) a maximum value at field - angles defined by @xmath122^{\frac{1}{2}}},\ ] ] where the maximum value of this quantity itself is @xmath123 .
this is in complete agreement with the cases presented in ref .
@xcite treated with a classical model for ultra - relativistic and ordinary electron - positron - ion plasmas .
it is from eq .
( [ gamma ] ) remarked that this maximum width field - angle is not affected by the mass - density of plasma .
however , the value of the maximum width increases as the plasma gets denser . figures 1(e ) and 1(f ) reveal that , while the amplitude increases monotonically with increases in the relative positron number - density , the width has a maximum value which enhances and tend to lower @xmath124-values as the plasma mass - density increases .
figure 2 shows variations of the head - on collision phase - shift for interacting solitons .
it is observed that , the head - on collision phase - shift is always positive indicating that the post - collision parts of solitons always move ahead of initial trajectories @xcite .
the value of collision phase - shift increases with increase of the plasma mass - density , as it is concluded from fig .
it is also observed from fig .
2(b ) that this parameter is increased as the magnetic field direction moves from the direction parallel to propagation to normal position , however , this effect gets more pronounced when the mass - density of plasma is higher . on the other hand , the variation of collisional phase - shift with respect to the relative positron concentration , presented in fig .
2(c ) is much similar to that of the width of soliton shown in fig .
furthermore , fig .
2(d ) indicates that the increase in the relative magnitude of magnetic field , @xmath125 , has also significant effect on the collision of solitons , so that , the higher is the field strength the lower is the shift in the soliton trajectory in the head - on collisions .
* using the extended * poincar - lighthill - kuo ( plk ) reductive perturbation method we studied the oblique propagation and quasi - elastic head - on collisions of solitary structures in a fully degenerate magnetized electron - positron - ion plasma with relativistic ultra - cold electrons and positrons in the framework of quantum hydrodynamics model .
it is observed that , plasma parameters such as mass - density , normalized magnetic field strength , its angle with respect to the soliton propagation and the relative positron number - density play important roles on the propagation as well as collision dynamics of solitary structures in a fermi - dirac plasma .
current findings can elucidate new aspects of nonlinear dynamics in dense astrophysical objects * such as white dwarfs and may as well be applicable * to the laboratory produced inertial - confined fusion plasmas .
@xmath126 @xmath130 - \left\ { { 1 - \alpha - { n_i } + 3r_0^ { - 2}\left [ { \sqrt { 1 + r_0 ^ 2 } + { \alpha ^ { - 1/3}}\sqrt { 1 + r_0 ^ 2{\alpha ^{4/3 } } } } \right ] + } \right .
\\ \left . { \frac{{3r_0^ { - 4}}}{2}\left [ { ( 1 + 2r_0 ^ 2 ) - { \alpha ^ { - 5/3}}(1 + 2r_0 ^ 2{\alpha ^{4/3 } } ) } \right ] } \right\ } + \ldots = 0 .
\\ \end{array}\ ] ] m. akbari - moghanjoughi , phys .
plasmas , * 17 * , 092304(2010 ) .
f. haas , l. g. garcia , j. goedert , and g. manfredi , phys .
plasmas * 10 * , 3858(2003 ) .
shukla p k , eliasson b , `` nonlinear aspects of quantum plasma physics '' phys .
usp . * 51 * 53(2010 ) .
g. manfredi , fields inst .
commun . * 46 * 263(2005 ) .
m. akbari - moghanjoughi , phys .
plasmas , * 17 * , 052302(2010 ) .
m. bonitz , d. semkat , a. filinov , v. golubnychyi , d. kremp , d.o .
gericke , m.s .
murillo , v. filinov , v. fortov , w. hoyer , j. phys . a , * 36 * 5921(2003 ) .
s. chandrasekhare , science , * 226 * 4674(1984 ) .
a. a. mamun and p. k. shukla , phys .
a , * 374 * , 4238(2010 ) . m. akbari - moghanjoughi , phys .
plasmas , * 17 * , 072101(2010 ) .
m. akbari - moghanjoughi , astrophys .
space sci . ,
( 2010 ) doi 10.1007/s10509 - 010 - 0499 - 3 .
n. j. zabusky and m. d. kruskal phys .
lett . * 15 * 240(1965 ) .
h. washimi and t. taniuti , phys .
* 17 * 996(1966 ) .
m. oikawa and n. yajima , j. phys .
37 * 1093(1973 ) .
shapiro , s. l. and teukolsky , s. a. , black holes , white dwarfs and neutron stars , ( wiley , newyork ) , ( 1983 ) .
m. j. rees , in : g. b. gibbons , s. w. hawking , s. siklas ( eds . ) , the very early universe , cambridge univ .
press , cambridge , ( 1983 ) .
h. r. miller , p. j. witta , active galetic nuclei , springer - verlag , berlin , p.202 , ( 1987 ) .
p. goldreich , w. h. julian , astrophys .
j. , * 157 * , 869(1969 ) .
f. c. michel , rev .
phys . * 54 * , 1(1982 ) . l. o. silva , r. bingham , j.m .
dawson , j. t.mendonca , and p. k. shukla , phys .
83 * 2703(1999 ) . c. m. surko , m. levelhal , w. s. crane , a. passne , f. wysocki , t. j. murphy , j. strachan , and w.l .
rowan , rev .
instrum . * 57 * 1862(1986 ) . c. m. surko , t. murphy , phys . fluids , * b2 * 1372(1990 ) .
r. g. greeves and c. m. surko , phys .
* 75 * 3847(1995 ) .
v. i. berezhiani , d. d. tskhakaya and p. k. shukla , phys .
a , * 46 * 6608(1992 ) .
s. chandrasekhar , `` an introduction to the study of stellar structure '' , chicago , ill . , ( the university of chicago press ) , ( 1939 ) p. 360 .
a. rasheed , g. murtaza and n. l. tsintsadze , phys .
e. , * 82 * 016403(2010 ) .
nam c. lee , phys .
plasmas , * 16 * 042316(2009 ) .
mouloud tribeche and soufiane boukhalfa , astrophys .
space sci . ,
( 2010 ) doi : 10.1007/s10509 - 010 - 0549-x a. jeffery , t. kawahawa , asymptotic method in monlinear wave theory , ( pitman , london , 1982 ) .
masayuki oikawa and nobuo yajima , phys .
jpn . , * 34 * , 1093(1973 ) .
e. infeld , g. rowlands , nonlinear waves , solitons and chaos , cambridge univ . press , cambridge , ( 1990 ) .
nobumasa sugimoto and tsunehiko kakutani , j. phys .
. jpn . * 43 * , 1469(1977 ) .
a. esfandyari - kalejahi , m. mehdipoor and m. akbari - moghanjoughi , phys .
plasmas , * 16 * 052309(2009 ) . m. akbari - moghanjoughi , phys . lett .
a , * 374 * , 1721(2010 ) .
( color online ) variations of soliton amplitude ( left column of figure ) and width ( right column of figure ) with respect to magnetic field direction @xmath60 , relative positron concentrations , @xmath124 , and normalized mass - density of plasma .
the dash sizes in all plots are appropriately related to the values of the varied parameter .
the values of @xmath131 and @xmath132 are used for all plots .
( color online ) variations of the head - on collision phase - shift with respect to with respect to magnetic field direction @xmath60 , relative positron concentrations , @xmath124 , and normalized mass - density of plasma , when other parameters are fixed .
the values of @xmath131 and @xmath132 are used for all plots .
the dash sizes in all plots are appropriately related to the values of the varied parameter . | oblique propagation and head - on collisions of solitary structures is studied in a dense magnetized plasma comprised of relativistic ultra - cold electrons and positrons and positive dynamic ions using conventional extended multi - scales technique , in the ground of quantum hydrodynamics model .
the variations of head - on collision phase - shift as well as the characteristic soliton amplitude and width is evaluated numerically in terms of other plasma parameters such as mass - density , normalized magnetic field strength , its angle with respect to the soliton propagation and the relative positron number - density .
the relevance of current investigations , with appropriate plasma parameters for the astrophysical dense magnetized objects such as white - dwarfs , is addressed . |
the current picture of disk galaxy formation and evolution has as its basis the dissipative infall of baryons within a dominant dark halo potential well ( white & rees 1978 ) .
the collapse and spin - up of the baryons , with angular momentum conservation , can provide an explanation for many of the observed properties of disks , with the standard initial conditions of baryonic mass fraction @xmath4 and dark halo angular momentum parameter @xmath5 ( fall & efstathiou 1980 ; gunn 1982 ; jones & wyse 1983 ; dalcanton , spergel & summers 1997 ; hernandez & gilmore 1998 ; mo , mao & white 1998 ; van den bosch 1998 ) .
galaxies such as the milky way which have an old stellar population in the disk , must , within the context of a hierarchical - clustering scenario , evolve through only quiescent merging / accretion , so as to avoid excessive heating and disruption of the disk ( ostriker 1990 ) .
further , the merging processes with significant substructure cause angular momentum transport to the outer regions , which must somehow be suppressed to allow the formation of extended disks as observed ( e.g. zurek , quinn & salmon 1988 ; silk & wyse 1993 ; navarro & steinmetz 1997 ) . thus here
we adopt the simplified picture that disk galaxies form from smooth gaseous collapse to centrifugal equilibrium , within a steady dark halo potential .
we discuss where appropriate below how this may be modified to take account of subsequent infall , or earlier star formation .
our model incorporates the adiabatic response of the dark halo to the disk infall , and we provide new , more general , analytic solutions for the density profile , given a wide range of initial density profiles and angular momentum distributions .
we provide new insight into the ` disk - halo ' conspiracy within the context of this model , demonstrating how an imperfect conspiracy is improved by the disk - halo interaction .
we explicitly include subsequent viscous evolution of the gas disk to provide the exponential profile of the stellar disk , and develop analytic expressions that illustrate the process .
the resulting radial inflow builds up the central regions of the disk and we investigate the properties of ` bulges ' that may form as a consequence of instabilities of the central disk .
we derive new constraints on the characteristic redshift of disk star formation .
we obtain a simple relation connecting the initial conditions , such as spin parameter and baryonic mass fraction , to the efficiency of viscous evolution and star formation .
in this section we shall derive the mass profiles of disk and halo after the collapse of the baryons .
we shall follow earlier treatments of disk galaxy formation ( e.g. mo , mao & white 1998 ) by assuming that the virialized dark halo , mixed with baryonic gas , is ` formed ' or at least assembled at redshift @xmath0 .
this virialized halo has a limiting radius @xmath6 within which the mean density is @xmath7 , and contains a baryonic mass fraction @xmath2
. then [ rh_h ] = v_200 10 h(z_f ) ; m_tot = v_200 ^2g = v_200 ^3 10 g h(z_f ) , where @xmath8 is the value of the hubble parameter at redshift @xmath0 , @xmath9 is the total mass within virialized radius @xmath6 , and @xmath10 is the circular velocity at @xmath6 .
the baryonic gas cools and settles into a disk , causing the dark halo to contract adiabatically ( blumenthal _ et al .
the specific angular momentum distribution of the gas is assumed to be conserved during these stages .
we shall include below the subsequent re - arrangement of the disk due to angular momentum transport .
this we investigate by variation of the disk angular momentum distribution function , choosing an appropriate analytic functional form .
let @xmath11 and @xmath12 respectively denote the fraction of the total baryonic mass , and total dark mass , that is contained within radius @xmath13 , and denote the baryonic mass angular momentum distribution function by m_d ( < j ) = f ( j / j_max ) , where @xmath14 is the maximum specific angular momentum of the disk .
we will be requiring that the functional form , @xmath15 , vary as the disk evolves , and it is convenient to introduce the notation @xmath16 and define c_f 1-_0 ^ 1 f ( ) d , which represents the area above the angular momentum distribution function curve @xmath17 for @xmath18 .
we will mimic the effects of viscous evolution by decreasing the value of @xmath19 in our evolving disk models in section 4 below . in terms of this parameter
the total disk angular momentum is : @xmath20 with @xmath21 .
thus @xmath22 is the average specific angular momentum of the disk material .
the specific angular momentum of the disk material is assumed to follow that of the dark halo , but in general will not be a simple analytic function ( e.g. quinn & binney 1992 ) . for illustration , we adopt an analytic monotonic increasing function @xmath23 containing a free parameter @xmath24 , with @xmath25 .
we require the initial angular momentum distribution to be scale free , representing the angular momentum distribution of the virialized halo , and will adopt @xmath26 .
we shall vary the value of the parameter @xmath24 to mimic the effects of viscous evolution on the angular momentum distribution .
the total energy is : e_tot = - = - , where @xmath27 is a constant of order unity , depending on the dark halo density profile , and since it is constant for any specific halo model , we can take @xmath28 without loss of generality .
the spin parameter @xmath1 is by definition j_tot | e_tot |^1/2 g^-1 m_tot^-5/2 .
thus the mean specific angular momentum of the disk material may be expressed as c_f j_max = v_200 r_200 .
assuming spherical symmetry , the rotationally - supported disk has mass profile given by m_d = f ( j / j_max ) = f ( ) = f ( ) , while the initial virialized halo mass profile is g(r_ini ) = m_h(r_ini ) = m_ini(r_ini ) /m_tot
, with @xmath29 .
the above equations describe the disk and halo just upon the settling of the gas disk to the mid - plane , prior to the subsequent adiabatic compression of the halo
. a self - consistent calculation of the modified disk and halo density profiles may be made by consideration of the adiabatic invariance of the angular action , @xmath30 , together with the assumption of no shell crossing ( cf .
1986 ) .
suppose a dark matter particle initially at @xmath31 finally settles at @xmath32 , the radius within which the dark halo mass fraction is @xmath33 .
then under adiabatic invariance the disk mass profile , @xmath34 , and the halo mass profile , @xmath33 , are related by : @xmath35 where @xmath36 is the inverse function of @xmath37 , the initial virialized halo mass profile .
further manipulation of these relations is simplified by introduction of the parameter @xmath38 , given by = .
generally @xmath38 is a quantity that is closely related to the overall disk collapse factor . from equations ( 8) and ( 12 )
, we have @xmath39 where @xmath40 is the normalized specific angular momentum .
again , @xmath41 , and @xmath42 corresponds to the maximum specific angular momentum of the disk , which occurs at the edge of the disk , equivalently at the disk cutoff radius .
the fraction of the dark matter contained within the disk thus has a maximum value , @xmath43 , given by @xmath44 in the above equation , and for radii with @xmath45 , @xmath46 . to illustrate the physical meaning of these parameters ,
consider the rigid singular isothermal halo , for which @xmath47 .
then from equation ( 15 ) , @xmath48 which corresponds to the disk cutoff radius .
thus in this case , remembering that @xmath49 is the normalized radius , @xmath50 is the disk collapse factor .
up to now we know the mass profile of disk and halo after collapse , in terms of the normalized specific angular momentum @xmath40 , as given in equations ( 14 ) and ( 15 ) , for given forms of the angular momentum distribution function , @xmath51 , and initial virialized dark halo mass profile , @xmath52 .
next we shall obtain the relation between @xmath40 and radius @xmath49 , to complete the derivation of the mass profiles of disk and halo after collapse .
returning to a general halo density profile , the total mass contained within the radius corresponding to @xmath33 is m ( m_h ) = m_tot ( ( 1-f ) m_h + f m_d ) . from equations ( 9 ) - ( 12 ) and ( 16 ) , we have @xmath53 introducing the radius variable @xmath54 and coefficient @xmath55 as @xmath56 we may finally derive the functional dependences on @xmath40 of the disk mass @xmath34 , of the halo mass @xmath33 , and of the radius @xmath54 : @xmath57 thus @xmath40 can be thought of as a normalized radius . as we shall see later
, @xmath55 is a measure of the compactness of the final collapsed disk due to the competition between the spin parameter @xmath1 and the baryonic mass fraction @xmath2
. these equations ( 21 ) - ( 23 ) can be used to derive disk and halo properties for a free choice of virialized halo profile @xmath37 and angular momentum distribution function @xmath17 . within the disk cutoff radius , with @xmath58 , the disk surface density , circular velocity and the disk - to - dark mass ratio as function of radius @xmath40 or @xmath49
have the generic forms : @xmath59 where @xmath60 and @xmath61 is defined similarly . the circular velocity at radii beyond the disk cutoff , but within the halo , is given by : @xmath62 armed with these relations , one may now look at various initial virialized halo density profiles and angular momentum distributions , and determine the allowed parameter space from observed properties of disk galaxies .
it is convenient to adopt power - law approximations for the initial virialized halo mass profile and angular momentum distributions , such that @xmath63 and @xmath64 for small @xmath40 .
figure 1 shows the location of various fiducial models in the plane of these power law indices @xmath65 and @xmath66 ; the value @xmath67 corresponds to the singular isothermal sphere , the value @xmath68 corresponds to the hernquist ( 1990 ) and to the navarro , frenk & white ( 1997 ) profiles , while the value @xmath69 corresponds to the non - singular isothermal sphere with a constant - density core .
these profiles span the range of dark - halo profiles suggested by theory , and plausibly consistent with observations .
the shaded region is the allowed parameter space for these models , constrained by the surface - density profile , rotation curve and disk - to - halo central mass ratio .
the line abc is the maximum angular - momentum index consistent with a disk surface density profile in the central regions that declines with increasing radius , while the line def is the minimum angular - momentum index consistent with a finite value of the central circular velocity .
the line bf denotes the maximum values of @xmath66 consistent with a non - zero central disk - to - halo mass ratio .
this power - law approximation has an initial virialized halo density profile at small radius as @xmath70 ( seen by solution of ( 22 ) for the form of @xmath52 ) . solving equations ( 21 ) - ( 23 ) above gives the corresponding profile after adiabatic infall . in the central region , where the disk dominates the gravitational potential ( i.e. @xmath71 ) , the halo density profile is @xmath72 .
note that in the region where the dark halo dominates the gravitational potential ( i.e. @xmath73 ) , the halo density profile is essentially unaffected by the disk , as expected .
for the case @xmath74 , the central halo density profile is unaffected since the disk mass density profile and the initial virialized halo density profile have the same dependence on @xmath40 .
the viable models within the shaded region have @xmath75 , so that the final halo density profile in the central , disk - dominated region should be steeper than its initial virialized profile in this region , not surprisingly . thus the final halo profile for these models ranges from @xmath76 to @xmath77 .
it is interesting to note that @xmath76 , the outcome of an initial virialized halo with a constant density core responding to the settling of a disk with angular momentum index @xmath78 , corresponds to the de - projected de vaucouleurs central density profile .
thus provided the virialized halo does not have a declining density profile with decreasing radius , which is unphysical , the final dark halo can not have a constant density core , at least in the very central region where the disk dominates , but should be cuspy .
the singular isothermal sphere provides a virialized halo density profile that is the most tractable analytically , and we can obtain some important scaling relations without having to specify the angular momentum distribution @xmath17 ; aspects of the analysis of this profile should hold in general , and provide insight .
the final disk and halo mass profiles are given by solution of : @xmath79 where @xmath58 and @xmath55 and @xmath38 , the parameters describing the compactness of the collapsed disk and its collapse factor , are defined above in equations ( 20 ) and ( 13 ) .
the collapse factor , defined as the ratio of the pre - collapse radius @xmath6 to the cutoff disk radius @xmath80 at @xmath81 ( to be distinguished from the final disk scale length ) is = ( 1-f)(1+c_0)= ( 1-f + ) , where @xmath19 , defined in equation ( 3 ) , is a measure of the shape of the angular momentum distribution , and small values of @xmath19 mean steeply - rising angular momentum distributions .
typical angular momentum distributions for disks are shown in figure 2 ( a , b ) and discussed below . here
one should just note that a value @xmath82 is reasonable . with this , and @xmath83 and @xmath84 , we obtain @xmath85 and @xmath86 .
the collapse factor as defined here is then about a factor of 9 . for fixed angular momentum distribution @xmath19
, the collapse factor depends not only on @xmath87 but also on compactness @xmath88 .
this is consistent with the results of previous studies of the collapse factor in two extreme cases ( jones & wyse 1983 ; peebles 1993 ) : if the final disk is so self - gravitating that the value of @xmath3 corresponds to @xmath89 , then the collapse factor is @xmath90 ; on the other hand if the final disk is sufficiently far from self - gravitating , with small @xmath3 and @xmath91 , the collapse factor is @xmath92 .
the above collapse factor relationship is valid over the entire range of values for the parameters @xmath1 and @xmath2 ( and is the most general relation derived to date ) .
it should be noted that varying the normalized angular momentum distribution by varying @xmath93 changes the derived collapse factor ; this is investigated further below .
comparison between the sizes of observed disks and those predicted from such collapse calculations provides a constraint on the redshift at which the collapse happened ( e.g. mo , mao & white 1998 ) .
our disk cutoff radius for non - trivial @xmath17 may be expressed in terms of the initial conditions as : r_c= , where @xmath0 is the ` formation ' or assembly redshift , at which the halo is identified to have given mass and circular velocity , with no mass infall after this epoch .
one can see from this relation that both @xmath1 and @xmath3 are equally important ; previous determinations considered the baryonic mass fraction fixed ( dalcanton , spergel & summers 1997 ; mo , mao & white 1998 ) .
the explicit inclusion of the parameter @xmath19 allows us to take account of disk evolution , as gas is transformed to stars . adopting a disk cutoff radius at three disk scale - lengths ( e.g. van der kruit 1987 ) , and choosing specific values of the present - day stellar disk scale - length @xmath94kpc , @xmath95 , @xmath96 kms@xmath97 and an einstein - de - sitter universe with hubble constant @xmath98 , the above relation gives the formation redshift @xmath99 for @xmath100 and @xmath101 ; @xmath102 for @xmath103 and @xmath104 ; @xmath105 for @xmath100 and @xmath106 ; @xmath107 for @xmath103 and @xmath108 .
consistent with previous calculations , for fixed formation redshift smaller values of @xmath1 can lead to smaller disk size ; we have here explicitly demonstrated that larger @xmath3 can also lead to this result .
lower values of @xmath19 also lead to higher redshift of formation .
thus for no viscous evolution i.e. there is no angular momentum redistribution during the evolution of the galactic disk and @xmath19 has a time - independent value , the formation redshift @xmath0 determined by assuming fixed initial @xmath1 and @xmath2 and fixed disk size @xmath80 is smaller than would be determined if @xmath19 could be decreased ( to mimic say viscous evolution ) .
lower values of @xmath19 for given total angular momentum content imply a larger disk scale - length ; the effect of viscous evolution is to re - arrange the disk material so as to increase the disk scale - length .
typical values of viscosity parameters lead to a factor of 1.5 increase in disk scale - length in a hubble time . as we demonstrate below
, a smaller value for @xmath19 corresponds to larger bulge - to - disk ratio .
this result is consistent with the results of mo , mao & white ( 1998 ) : halo and disk formation redshift can be pushed to higher value when a bulge is included .
these trends are general , and not tied to the specific model of the halo density profile .
one should bear in mind that the old stars in the local thin disk of the milky way have ages of at least 10 gyr , and may be as old as the oldest stars in the galaxy ( edvardsson _ et al .
_ 1993 ) ; the age distribution of stars at other locations of the galactic disk is very poorly - determined , but it is clear that a non - trivial component of the thin disk was in place at early times ( at redshift @xmath109 for the cosmologies considered above ) .
a common assumption in previous work is that the mass angular momentum distribution of the disk is that of a solid - body , rotating uniform density sphere , @xmath110 ( mestel 1963 ) . for this distribution , @xmath111 , and
one derives a low redshift of formation for a galaxy like the milky way ( mo , mao & white 1998 ) , which has difficulties with the observations .
analytic solutions to equations ( 29 ) - ( 31 ) can be obtained by assuming a simple monotonic increasing function @xmath23 containing one parameter @xmath24 with @xmath25 . in order to avoid the situation where one obtains a trivial collapse factor due to a very small amount of disk material at very large radius , we restrict the shape of the mass angular momentum distribution function @xmath17 to avoid too shallow an asymptotic slope as @xmath51 approaching 1 when @xmath40 increases ( see figure 2 ) . from figure 1 , the singular isothermal halo at point f requires @xmath112 for @xmath113 .
a simple form consistent with this is @xmath114 with @xmath18 and @xmath115 .
the angular momentum parameter @xmath116 .
figure 2a shows this angular momentum distribution function with different values of the parameter @xmath24 , compared with the mass angular momentum distribution of a solid - body , rotating uniform density sphere , @xmath110 ( mestel 1963 ) .
the normalized total angular momentum @xmath19 corresponds to the area above each curve . within the disk , where @xmath58 , the galactic disk surface density , circular velocity and the disk - to - dark mass ratio as a function of radius are : @xmath117 v_c ( ) = v_200(1-f)[1+c_0(1+b - b ) ] ,
= , + where = . from equation ( 27)-(28 ) , the circular velocity at radii beyond the disk cutoff , but within the halo , is : v_c(r ) = .
these results are plotted in figure 3 ( a , b , c ) for @xmath95 , @xmath103 , @xmath58 .
the different curves correspond to @xmath118 .
larger values of @xmath24 yield larger disk cut - off radii ; note that the circular velocities for points beyond the cut - off radius of a given model may be obtained by forming the envelope of the values for the cut - off radius for larger values of @xmath24 .
varying the value of the parameter @xmath24 changes the angular momentum distribution function similar to the effects of viscous evolution ; the ratio of disk mass to dark halo mass increases at small radius with increasing @xmath24 , which can be interpreted as due to radial inflow of disk material . for initial virialized halo profiles other than the singular isothermal sphere , the collapse factor has a more general form : = ( 1-f)(m_hc + c_0 ) , where @xmath43 is the solution of @xmath119 and is the mass fraction of the dark halo that is contained within the cut - off radius of the disk .
the trend of the dependence of the collapse factor on the values of @xmath1 and @xmath2 remains the same as found above for the singular isothermal halo .
the range of viable models represented by the shaded region in figure 1 can be investigated by the appropriate virialized halo profile @xmath37 and angular momentum distribution function @xmath23 corresponding to the points e , b , d and c. the distribution of surface density , circular velocity and disk - to - halo mass ratio for these models , varying parameter @xmath24 , are shown in figures 47 , by choosing fixed @xmath95 , @xmath103 ( and halo core size @xmath120 if the halo has a core radius ) .
model e : the results of a model corresponding to point e in figure 1 are shown in figure 4 ; this has a hernquist halo profile @xmath121 and @xmath122 .
the different curves correspond to @xmath118 .
model b : the results of a model corresponding to point b in figure 1 are shown in figure 5 ; this has a hernquist halo profile @xmath121 and @xmath123 .
this @xmath23 is shown in figure 2b . for @xmath124 , @xmath125 ; the @xmath126 index in the second term is determined by the requirement that @xmath23 should be a monotonic increasing function of @xmath40 for all values of @xmath115 .
the different curves correspond to @xmath127 .
model d : the results of a model corresponding to point d in figure 1 are shown in figure 6 ; this has a non - singular isothermal halo with a constant density core , @xmath128 , and @xmath122 . again
the different curves correspond to @xmath118 .
model c : the results of a model corresponding to point c in figure 1 are shown in figure 7 ; this has a non - singular isothermal halo with constant density core , @xmath128 , and @xmath123 . again
the different curves correspond to @xmath127 .
as can be seen from the figures , the different halo profiles and angular momentum distributions produce disks with a variety of surface density profiles and rotation curves . as in figure 3b , the circular velocities for points beyond the cut - off radius of a given model may be obtained by forming the envelope of the values for the cut - off radius for larger values of @xmath24 . thus if the disk is very compact , from equations ( 27)-(28 ) or equation ( 38 ) , we find that the circular velocity beyond the edge of the disk tends to decreases with radius .
the location of the edge of the disk depends on both @xmath1 and @xmath2 ( in addition to @xmath24 ) .
thus we have shown that rotation curves should show an imperfect disk halo ` conspiracy ' if the disk is too compact or too massive .
this is consistent with observations ( casertano & van gorkom 1991 ) .
further , these figures demonstrate that with increasing @xmath24 , the inner rotation curves become flat , and the transition between disk - dominated and halo - dominated sections of the rotation curve becomes more and more smooth with increasing @xmath24 . as discussed above , and illustrated in figure 2 , a higher value of @xmath24 corresponds to a flatter specific angular momentum distribution function , and increasing @xmath24 mimics the effects of viscous evolution in transporting angular momentum .
this indicates that viscous evolution can help the creation of an apparent disk - halo ` conspiracy ' . as can be seen from equations ( 21 ) - ( 26 ) , for given @xmath55 ( the compactness parameter defined in equation ( 20 ) ) , or @xmath3 ratio , within a given model of @xmath129 and virialized dark halo profile @xmath37 , the normalized properties of the disks formed for @xmath130 are very similar . in particular the disk surface density profile , rotation curve and disk - to - halo mass ratio profile scale similarly with @xmath40 .
for example , in the case of the singular isothermal sphere ( model f ) , equations ( 34 ) - ( 37 ) show that for @xmath130 the normalized disks are identical for given @xmath3 ratio
. thus @xmath3 must be an important factor in distinguishing one disk from another .
the overall normalization of the surface density is @xmath90 , so that again @xmath3 and @xmath1 enter separately and are both important .
note that here we are not insisting that the surface density profile of the gas disk so formed be exponential , unlike previous work ( mo , mao & white 1998 ) .
we shall however appeal to viscous evolution tied to star formation to provide a stellar exponential disk .
we now turn to this .
in this paper we aim to link viscous evolution within disks and the hubble sequence of disk galaxies .
one of the motivations for invoking viscous disks is that if the timescale of angular momentum transport via viscosity is similar to that of star formation , a stellar exponential disk is naturally produced independent of the initial gaseous disk surface density profile ( silk & norman 1981 ; lin & pringle 1987 ; saio & yoshii 1990 ; firmani , hernandez & gallagher 1996 ) .
angular momentum transport and associated radial gas flows ( both inwards and outwards ) can also , as shown above , provide a tight ` conspiracy ' between disk and halo rotation curves , and , as demonstrated below , provide a higher phase space density in bulges as compared to disks .
the star formation rate per unit area in a disk can be represented by a modified schmidt law involving the dynamical time and the gas density ( wyse 1986 ; wyse & silk 1989 ) .
we shall use the form of the global star formation rate per unit area , @xmath131 , of kennicutt ( 1998 ) , based on his observations of the inner regions of nearby large disk galaxies : _ = _
gas_gas , where @xmath132 is total gas surface density , @xmath133 is the dynamical time at the edge of the gas disk , and the normalization constant , related to the efficiency of star formation , has the value @xmath134 ( kennicutt 1998 ) .
note that observations of the star - formation rates in the outer regions of disk galaxies suggest that it is actually _ volume _ density that should enter the schmidt law , rather than surface density , and since many ( if not all ) gas disks flare in their outer regions , equation ( 40 ) will over estimate the star formation rate ( ferguson _ et al .
this is beyond the scope of the present model , but should be borne in mind and will be incorporated in our future work . for our models here ,
the edge of the initial gas disk is where @xmath42 , @xmath135 and thus @xmath136 . for halo formation redshift @xmath0 ,
we can obtain the relationship between global star formation timescale and the galaxy initial conditions in the general form using equation ( 1 ) , ( 25 ) and ( 39 ) : @xmath137 where again @xmath43 is the solution of @xmath119 , and is the fraction of the dark halo mass that is contained within the cut - off radius of the disk .
the parameters @xmath38 and @xmath55 are defined in equations ( 13 ) and ( 20 ) ; @xmath2 is the baryonic mass fraction in the initial density perturbation . in the case of the singular isothermal halo
, this relation has a simple form : t_*^-1= 10 h(z_f ) ( 1-f)^2(1+c_0)^2 . the gas consumption timescale will be longer than the characteristic star formation timescale due to the gas returned by stars during their evolution and death . for a standard stellar initial mass function , @xmath138 ( e.g. kennicutt _ et al .
_ 1994 ) .
this modified schmidt law is based on observations of the inner regions of nearby large disk galaxies , and simple theoretical principles .
assuming it holds at all epochs allows one to estimate the properties of present - day disks from the initial conditions of earlier sections in this paper .
let us assume that the dark halo is fully virialized at redshift @xmath0 . in keeping with the spirit of hierarchical clustering ,
let us allow for some star formation that could have taken place in the disk , from an earlier redshift @xmath139 , and that the total mass of the system could increase until @xmath0 ( although to maintain the thin disk , this accretion and merging must be only of low mass , low density systems ) .
so for any time @xmath140 or redshift @xmath141 between @xmath139 and @xmath0 , @xmath142 where @xmath143 is defined in equation ( 1 ) and @xmath144 is mass locked up into stars . in an einstein - de - sitter universe , @xmath145 and @xmath146 .
we have = a - , where @xmath147 and @xmath148 , and @xmath38 and @xmath55 are expressed in terms of @xmath19 , @xmath1 and @xmath2 through equations ( 13 ) and ( 20 ) in section 2.1 .
hence , from equation ( 1 ) , @xmath149 . identifying
the dark halo to have fixed @xmath10 , independent of redshift , leads to @xmath150 also being a constant , and thus the total mass grows as @xmath151 .
this differs from the standard solution of infall onto a point , @xmath152 ( gunn & gott 1972 ) .
the solution to the above equation is then m_g = t , where @xmath153 corresponds to the redshift @xmath139 of the onset of star formation .
thus at the halo formation redshift @xmath0 , the disk gas fraction is f_g(z_f ) = , since @xmath154 depends on @xmath19 , @xmath1 and @xmath2 , i.e. @xmath155 , the value of the constant @xmath154 may be evaluated for @xmath95 and various reasonable values of @xmath2 and @xmath19 as : for @xmath103 , @xmath156 , @xmath157 ; for @xmath100 , @xmath156 , @xmath158 ; for @xmath103 , @xmath159 , @xmath160 ; and for @xmath100 , @xmath159 , @xmath161 . for fixed @xmath1 and @xmath2 , small values of @xmath154
correspond to small values of @xmath19 , and hence flatter specific angular momentum distributions .
thus for @xmath162 , @xmath163 , typically @xmath164 .
we assume that after @xmath0 , there is no further infall , and that the gas in the disk will be consumed with characteristic timescale @xmath165 .
thus the gas fraction of a typical disk galaxy at the present time is f_g = f_g(z_f ) , where @xmath166 is the time interval between the halo formation redshift @xmath0 and present time @xmath167 .
assuming an einstein - de - sitter universe , we have @xmath168 and @xmath169 .
thus a typical value for the present gas fraction of disk galaxies is : ( ) = b . with the approximation @xmath170
, we have ( 1+z_f)^3/2 = 1 + . for the milky way galaxy ,
if we adopt @xmath171 for the atomic @xmath172 gas , and @xmath173 for the molecular @xmath174 gas ( blitz 1996 ; dame 1993 ) , then with an estimate of @xmath175 for the total baryonic mass of the milky way depending on stellar exponential scale length ( dehnen & binney 1998 ) , we obtain gas fraction , the total gas mass includes @xmath176 helium mass , @xmath177 or even higher depending on the mass model of the galaxy .
as mentioned earlier , for fixed @xmath1 and @xmath2 , small values of @xmath154 correspond to small values of @xmath19 , and hence flatter specific angular momentum distributions . for small values of the parameter @xmath154 ,
say @xmath178 , we obtain @xmath179 , while for large values of @xmath154 , @xmath180 , we obtain @xmath181 .
the larger value of @xmath0 is preferred , given what we know of the age distribution of stars in the local thin disk ( e.g. edvardsson 1993 ) .
the effect of viscous evolution is equivalent to choosing small @xmath19 , i.e. small values of @xmath154 .
so the inclusion of viscous evolution can give relatively higher halo formation redshift , which is consistent with the results from the constraint on the redshift of formation that we obtained from considerations of the size of the disk .
it should be noted that for fixed halo ` formation ' redshift @xmath0 , the star formation timescale @xmath182 .
again , both @xmath1 and @xmath3 are important .
as we discussed in the previous section , the structure of the normalized disk depends strongly on @xmath3 while the overall normalization depends strongly on @xmath1 for fixed @xmath3 .
as we show later , the bulge - to - disk ratio also depends on the importance of @xmath1 and @xmath3 .
thus many aspects of the hubble sequence of disk galaxies star formation timescale , disk gas fraction , bulge - to - disk ratio depend on both @xmath1 and @xmath3 .
the star formation timescale derived above is independent of @xmath10 , which at first sight is surprising , given that the hubble sequence of disk galaxies has been interpreted as a sequence of star formation timescales , relative to collapse times ( sandage 1986 ) , and observations show that the hubble type of a disk galaxies is broadly correlated with the disk luminosity ( lake & carlberg 1988 ; de jong 1995 ) .
however , @xmath10 is not an easily - observed quantity .
the present - day luminosity of a disk in our model can be written as l = = , with @xmath183 the current value of the mass - to - light ratio .
estimation of the predicted tully - fisher relation depends on what model parameter we use for the width of the hi line .
obviously if we identify this with @xmath10 we have a reasonable relationship provided that the coefficient @xmath184 is constant , i.e. @xmath185/h(z_f ) = \gamma_{tf}$ ] .
this then requires that there should be a correlation between @xmath2 and @xmath0 , in the sense that for large @xmath2 , @xmath0 is large .
then we can see from equation ( 42 ) that this leads to the star formation timescale being small , implying more efficient viscous evolution , and leading to larger b / t ratio .
large @xmath0 may be correlated with small @xmath10 in the context of hierarchical - clustering cosmologies , and in that case , a short star formation time and large b / t ratio should be correlated with low luminosity , which is not consistent with observation .
however , an interpretation that is compatible with observations is that high n - sigma fluctuations for fixed @xmath10 can form high luminosity disks with large b / t ratio .
the tully - fisher relationship is not a simple relation between luminosity and @xmath10 , but depends on where the circular velocity @xmath186 is measured ( courteau 1997 ) .
the relationship between @xmath10 and @xmath186 obviously depends on the details of the halo density profile and the angular momentum distribution . from the rotation curves in fig . 3 to fig .
7 , we can see that it is appropriate to choose for our estimate of @xmath187 the circular velocity at the cutoff radius of disk , adopted as 3 three scale lengths . from equation ( 35 ) , we have @xmath188 , where @xmath189 , as defined in equation ( 20 ) , is the compactness of disk .
now the predicted tully - fisher relation is l = . requiring that the coefficient @xmath190 in the tully - fisher relation be constant , i.e. @xmath191/h(z_f)(1+c_0)^3(1-f)^3 $ ] , then gives , from equation ( 42 ) , @xmath192 .
so for small @xmath55 , the star formation timescale is large , which can cause less efficient viscous evolution .
so less efficient viscous evolution and small @xmath3 will lead to small b / t ratio .
also small @xmath55 is correlated with large @xmath0 from the constancy of the coefficient in the tully - fisher relation .
similarly large @xmath0 may be correlated with small @xmath10 in the context of hierarchical clustering cosmology .
so small @xmath55 and small @xmath10 will lead to small @xmath187 and lower disk luminosity .
thus this version of the predicted tully - fisher relation appears fully compatible with the observations .
however , one should bear in mind that we have adopted a fixed constant of proportionality @xmath193 in the star formation law , and this may well vary with global potential well depth ( white & frenk 1991 ) or local potential well depth ( silk & wyse 1993 ) .
a further test of the model is the relation between disk scale and circular velocity , and its variation with redshift ; observations indicate that @xmath194 is smaller at high redshift , @xmath195 ( vogt _ et al .
_ 1996 ; simard _ et al .
_ 1999 ) . in our model
there is little change in total mass between redshifts of unity and the present , and so this evolution of disk size can not be due to the halo mass growth , as had been proposed by mao , mo & white ( 1998 ) . instead in our model this is due to the different and changing scale lengths of gas and stars . in our model , due to early star formation , the gas fraction at @xmath0 is about @xmath196 , typically @xmath197 .
it is natural that the gas component of the disk will have a larger scale length than the stellar component .
the scale length of the gas component can increase with time due to viscous evolution while the scale length of stellar component can also increase with time due to the non - linear local star formation law ( saio & yoshi 1990 ) .
so the stellar scale length at high redshift , when the gas fraction is large , should be much smaller than the stellar scale length at present time , when gas fraction is lower .
the study of detailed evolution of gas and stellar component is beyond the scope of this paper ; but the prediction of stellar size evolution of our model is qualitatively consistent with the observations . the measured distribution of @xmath194 for the local disk galaxy sample is approximately peaked at @xmath198 ( @xmath199 in kpc , @xmath187 in kms@xmath97 ) with a spread from -2 to -1 ( mao , mo & white 1998 ; courteau 1996 ) .
our estimation of the disk size or scale length in section 2 is valid for galaxies at the present time , when the gas fraction is small .
then assuming the disk cut off radius is three scale lengths , from equation ( 41 ) , @xmath200 , and the present @xmath201 is an indication of the galactic global star formation timescale ; further , from the constancy of @xmath190 , the coefficient of the tully - fisher relation in equation ( 52 ) , we have r_d / v_c = t_*/3= _ tf _ * .
thus this ratio is also an indication of the disk compactness . adopting the b - band mass - to - light ratio of our local disk @xmath202 , and using the luminosity of our galaxy @xmath203 and @xmath204 kms@xmath97 to estimate @xmath205
, we can obtain that @xmath206 .
obviously @xmath207 roughly corresponds to the predicted lower limit of the local sample , which is consistent with observations . the distribution of @xmath2 from 0.05 to 0.2 and the distribution of @xmath1 from 0.03 to 0.12 will cause the value of the compactness parameter @xmath55 to spread from 0.1 to 2 approximately . adopting typical values
@xmath103 , @xmath208 , @xmath101 , the peak will be located at @xmath209 with @xmath210 , which is again consistent with observations . the spread of @xmath194 is simply caused by the spread in the value of compactness parameter @xmath55 .
galaxies classification schemes based on morphology are the basic first step in understanding how galaxies form and evolve ( van den bergh 1998 ) .
the bulge - to - disk luminosity ratio is one of the three basic classification criterion for the hubble sequence ( sandage 1961 ) .
however , the relation between bulge - to - disk ratio and hubble type has a fair amount of scatter , some of which must be related to the difficulty of a decomposition of the light profile into bulge and disk , and the bulge - to - disk ratio is dependent on the band - pass used to define the luminosity ( de jong 1996 ) .
the current observational data show that bulges can be diverse and heterogeneous ( wyse , gilmore & franx 1997 ) .
some share properties of disks and some are more similar to ellipticals .
models of bulge formation can be classified into several categories : the bulge is formed from early collapse of low angular momentum gas , with short cooling time and efficient star formation ( eggen _ et al . _ 1962 ; larson 1976 who invoked viscosity to transport angular momentum away from the proto - bulge ; van den bosch 1998 ) ; the bulge is formed from merging of disk galaxies ( toomre & toomre 1972 ; kauffmann , white & guiderdoni 1993 ) ; the bulge is formed from the disk by secular evolution process after bar instability ( combes _ et al .
_ 1990 ; norman , sellwood & hasan 1996 ; sellwood & moore 1999 ; avila - reese & firmani 1999 ) ; the bulge is formed from early dynamical evolution of massive clumps formed in disk ( noguchi 1999 ) .
however it is speculated that large bulges , which tend to have de vaucouleur s law surface brightness profiles , share formation mechanisms with ellipticals , while smaller bulges , which tend to be better fit by an exponential profile , are formed from their disks through bar dissolution . it should be noted that the significantly higher phase space density of bulges when compared to disks suggests that gaseous inflow should play a part in the instability ( wyse 1998 ) , just as invoked in earlier sections of this paper , for other reasons . here
we only consider this latter case , the formation of small bulges from the disk .
early studies of disk instabilities showed that toomre s local stability criterion @xmath211 is also sufficient for global stability to axisymmetric modes ( hohl 1971 ; binney & tremaine 1987 ) .
it is known that the bar instability requires a similar condition ( hockney & hohl 1969 ; ostriker & peebles 1973 ) .
efstathiou , lake & negroponte ( 1982 ) used n - body techniques to study the global bar instability of a pure exponential disk embedded in a dark halo and proposed a simple instability criterion for the stellar disk , based on the disk - to - halo mass ratio .
however , it has been argued recently that there is no such simple criterion for bar instability ( christodoulou , shlosman & tohline 1995 ; sellwood & moore 1999 ) .
further , recent n - body simulations ( sellwood & moore 1999 ; norman , sellwood & hasan 1996 ) show that every massive disk form a bar during the early stages of evolution , but later the bar is destroyed by the formation of a dense central object , once the mass of that central concentration reaches several percent of the total disk mass . this can be understood in terms of the linear mode analysis work and nonlinear processes of swing amplifier and feed back loops ( goldreich & lynden - bell 1965 ; julian & toomre 1966 ) .
toomre ( 1981 ) argued that the bar - instability can be inhibited in two ways : one is that a large dark halo mass fraction can reduce the gain of the swing amplifier , while the other is that feedback through the centre can be shut off by a inner lindblad resonance ( ilr ) .
the dense central object can destroy the bar via the second of these mechanisms .
however , most of these studies assume that the dark halo has a constant density core . on the contrary , we have shown that dark halo profile can not have a constant density core after adiabatic infall , if one starts from physical initial conditions
. it will be interesting to study bar formation under different disk - halo profiles in addition to the well - studied harmonic core .
we will for simplicity adopt the simple criterion of efstathiou , lake & negroponte ( 1982 ) , interpreted to determine the size of a bar - unstable region , with the radial extent of the bar , @xmath212 defined by m_d(r_b)/m_h(r_b ) , with the value of the parameter @xmath213 chosen to fit observations . as we have argued in section 2 , @xmath3 is the important quantity determining the overall normalization of the disk surface density . from n - body simulations ( warren _
_ 1992 ; barnes & efstathiou 1987 ; cole & lacey 1996 ; steinmetz & bartelmann 1995 ) , the distribution of @xmath1 can be fit by a log - normal distribution : p()d= exp , where @xmath214 and @xmath215 ( this result is fairly independent of the slope of the power spectrum of density fluctuations ) .
what of the possible range of the baryonic fraction @xmath2 ? some previous studies suggested that @xmath216 as an explanation of the disk - halo ` conspiracy ' ( fall & efstathiou 1980 ; jones & wyse 1983 ; ryden & gunn 1987 ; hernandez & gilmore 1998 ) .
some interpretations of the observed tully - fisher relationship suggest that indeed @xmath2 is not invariant ( mcgaugh & de blok 1998 ) . here
we shall assume that the distribution of @xmath2 , similar to @xmath1 , is log - normal , centreed at @xmath217 with @xmath218 .
so @xmath2 is mainly within the range @xmath219 .
we generated a monte carlo sample of disks , with fixed halo @xmath10 and halo formation redshift @xmath0 , but with values of @xmath2 and @xmath1 following the above distributions .
then for given virialized dark halo profiles @xmath37 and angular momentum distribution @xmath23 in figure 1 , we can calculate the star formation timescale from equation ( 41 ) .
the parameter @xmath24 represents the efficiency of viscous evolution ; on the assumption that the viscous timescale is equal to the star formation timescale , we can use a simple linear correlation between the value of @xmath24 and the star formation timescale . then the value of parameter @xmath24 can be obtained . from equations ( 21 ) , ( 22 ) , ( 23 ) , ( 26 ) , ( 50 ) and ( 54 ) , one can calculate the bulge - to - disk ratio and final disk gas fraction .
thus we can plot b / t ratio versus disk gas fraction , for the whole monte carlo sample , confining the parameter space of @xmath1 and @xmath2 to give small bulges with @xmath220 .
larger values of b / t would correspond to such an unstable disk that the exercise is invalid .
disks that are too stable , having low @xmath3 or large @xmath1 , will evolve little and probably end up as low surface brightness systems .
figure 8a shows model f , representing the singular isothermal halo .
assuming the typical values @xmath208 and @xmath103 , then the choice of @xmath221 is required to allow the existence of bulges .
this is just the value of @xmath213 used in the global bar instability criterion ( efstathiou , lake & negroponte 1982 ) . for this model
the disk - to - halo mass ratio varies little with radius ( as shown in figure 3a ) , so the b / t ratio is strongly dependent on @xmath222 .
only a small range of values of @xmath222 is allowed , so as to not over - produce either bulge - less disks or completely unstable disks .
the relation between b / t ratio and gas fraction for this model is given in figure 8b ; there a general trend in the observed sense , with large scatter .
the different curves in this plot correspond to different values of @xmath222 , which sets the overall trend .
figure 9 ( a , b ) shows the equivalent plots of model e , representing a hernquist profile halo model , which is probably a more realistic case . here the the disk - to - halo mass ratio varies strongly with radius when approaching the centre , and the allowed parameter space for values of @xmath1 and of @xmath2 that allowing the formation of bulges can be large .
overly - unstable disks are denoted by asterisk symbols in figure 9a , and low surface - brightness disks are denoted by cross symbols .
the relation between b / t ratio and gas fraction is similar to that for the isothermal halo .
the milky way bulge is reasonably well - fit by an exponential profile , with a scale - length approximately one - tenth that of the disk ( kent _ et al . _ 1991 ) .
the morphology of the bulge is consistent with some triaxiality ( blitz & spergel 1991 ; binney _ et al .
perhaps the milky way is a system in which the bulge has formed from the disk , through a bar instability ?
observations show no evidence for a significant young or even intermediate - age stellar population in the field population of the galactic bulge ( feltzing & gilmore 1999 ) , despite their being ongoing star formation in the inner disk .
this implies that if the bulge were formed from the disk through bar dissolution , only one such episode is allowed , and this should have happened at high redshift . in the context of the present model , the lower star formation rates and longer viscosity timescales of later times act to stabilize the system .
however , it remains to be seen if the observed relative frequencies of bars , bulges and central mass concentration is consistent with the models of bar dissolution .
in the context of hierarchical clustering cosmology , the dark halo of a disk galaxy can be formed by quiescently merging small sub - halos into the primary dark halo , or by smoothly accreting matter into the dark halo .
we derive the generic solution to the adiabatic infall model of disk galaxy formation pioneered by many authors ( mestel 1963 ; fall & efstathiou 1980 ; gunn 1982 ; faber 1982 ; jones & wyse 1983 ; ryden & gunn 1987 ; dalcanton , summer & spergel 1997 ; mo , mao & white 1998 ; hernandez & gilmore 1998 ) .
through exploring the allowed parameter space of dark halo profile and angular momentum distribution function , we show that the central halo density profile should be cuspy , with the power law index ranging from @xmath223 to @xmath224 , in the central regions where the disk mass dominates . using a modified schmidt law of global star formation rate ,
we derive a simple scaling relationship between the disk gas fraction and the assembly redshift .
we explicitly allow a distribution in the values of the baryonic mass fraction , @xmath2 , in addition of the distribution in values of the spin parameter @xmath1 .
these two are found to play different role in determining the structural properties of the final disk , the star formation properties and bulge - to - disk ratio .
we mimic viscous evolution of disks by varying the specific angular momentum distribution of the disk , to redistribute angular momentum as a function of time .
we derive a consistent picture of the formation of galaxies like the milky way , with old stars in the disk . under the assumption that the viscous evolution timescale is equal to the star formation timescale
, we can further combine the @xmath1 and @xmath2 with the efficiency of angular momentum redistribution caused by viscosity .
assuming that small bulges are formed from their disks through bar dissolution , we can use the global bar instability condition to obtain bulge - to - total ratio , and explore the dependence on @xmath2 , @xmath1 and the viscous evolution efficiency .
the inclusion of viscous evolution has the merits of addressing several important issues : the conspiracy between disk and halo , the formation of the exponential profile of stellar disk , the high phase space density of bulges .
we have presented an analytic treatment , to illustrate these points and identify areas of particular need for more work .
we acknowledge support from nasa , atp grant nag5 - 3928 .
bz thanks colin norman , jay gallagher for helpful comments .
rfgw thanks all at the center for particle astrophysics , uc berkeley , for their hospitality during the early stages of this work .
blitz , l. , 1996 , in latter , w.b . ,
radford , s.j.e . , jewell , p.r . ,
mangum , j.g . &
bally , j.,eds , proc iau symp .
170 , co : twenty - five years of millimeter - wave spectroscopy .
kluwer academic publishers , p11 . | we investigate a model of disk galaxies whereby viscous evolution of the gaseous disk drives material inwards to form a proto - bulge .
we start from the standard picture of disk formation through the settling of gas into a dark halo potential well , with the disk initially coming into centrifugal equilibrium with detailed conservation of angular momentum .
we derive generic analytic solutions for the disk - halo system after adiabatic compression of the dark halo , with free choice of the input virialized dark halo density profile and of the specific angular momentum distribution .
we derive limits on the final density profile of the halo in the central regions .
subsequent viscous evolution of the disk is modelled by a variation of the specific angular momentum distribution of the disk , providing analytic solutions to the final disk structure .
the assumption that the viscous evolution timescale and star formation timescale are similar leads to predictions of the properties of the stellar components .
focusing on small ` exponential ' bulges , ones that may be formed through a disk instability , we investigate the relationship between the assumed initial conditions , such as halo ` formation ' , or assembly , redshift @xmath0 , spin parameter @xmath1 , baryonic fraction @xmath2 , and final disk properties such as global star formation timescale , gas fraction , and bulge - to - disk ratio .
we find that the present properties of disks , such as the scale length , are compatible with a higher initial formation redshift if the re - distribution by viscous evolution is included than if it is ignored .
we also quantify the dependence of final disk properties on the ratio @xmath3 , thus including the possibility that the baryonic fraction varies from galaxy to galaxy , as perhaps may be inferred from the observations .
= cmcsc10 200v_200 a&a , # 1to 0pt#1 galaxies : formation galaxies : structure galaxies : spiral cosmology : theory dark matter |
a lattice path matroid is a special type of transversal matroid whose bases can be thought of as lattice paths in the region of the plane delimited by two fixed bounding paths .
these matroids , which were introduced and studied from an enumerative perspective in @xcite , have many attractive structural properties that are not shared by arbitrary transversal matroids ; this paper focuses on such properties .
the definition of lattice path matroids is reviewed in section [ sec : back ] , where we also give some elementary properties of their bases and make some remarks on connectivity and automorphisms .
section [ sec : lpm ] proves basic results that are used throughout the paper ; for example , we show that the class of lattice path matroids is closed under minors , duals , and direct sums , we determine which lattice path matroids are connected , and we describe circuits and connected flats .
the next section discusses generalized catalan matroids , a minor - closed , dual - closed subclass of lattice path matroids that has particularly simple characterizations .
section [ sec : fundamental ] introduces special connected flats called fundamental flats that we use to characterize lattice path matroids and to show that the bounding paths can be recovered from the matroid . in section [ sec : trans ] , we describe the maximal presentation of a lattice path matroid , and we use this result to give a geometric description of these matroids as well as a polynomial - time algorithm for recognizing lattice path matroids within the class of transversal matroids . we also contrast lattice path matroids with fundamental transversal matroids and bicircular matroids . section [ sec : high ] treats higher connectivity .
the final section introduces another minor - closed , dual - closed class of lattice path matroids , the notch matroids , and characterizes this class by excluded minors .
we assume familiarity with basic matroid theory ( see , e.g. , @xcite ) .
we follow the notation and terminology of @xcite , with the following additions . a flat @xmath0 of a matroid @xmath1 is _ connected _ if the restriction @xmath2 is connected .
a flat @xmath0 is _ trivial _ if @xmath0 is independent ; otherwise @xmath0 is _
nontrivial_. the flats in a collection @xmath3 of flats are _ incomparable _ , or _ mutually incomparable _ , if no flat in @xmath3 contains another flat in @xmath3 .
the _ nullity _
, @xmath4 , of a set @xmath0 is denoted by @xmath5 .
recall that a matroid @xmath1 of rank @xmath6 is a _ paving matroid _ if every flat of rank less than @xmath7 is trivial .
most matroids in this paper are transversal matroids ( see @xcite ) .
recall that for a transversal matroid @xmath1 , a _ presentation _ of @xmath1 is a multiset @xmath8 of subsets of the ground set @xmath9 such that the bases of @xmath1 are the maximal partial transversals of @xmath10 . as is justified by the following lemma ( see @xcite ) , we always consider presentations of rank-@xmath6 transversal matroids by set systems of size @xmath6 .
[ lem : minsets ] let @xmath8 be a presentation of a rank-@xmath6 transversal matroid @xmath1 .
if some basis of @xmath1 is a transversal of @xmath11 , with @xmath12 , then @xmath11 is also a presentation of @xmath1 .
we use @xmath13 $ ] to denote the interval @xmath14 of integers , and , similarly , @xmath15 $ ] to denote the interval @xmath16 of integers .
this section starts by reviewing the definition and basic properties of lattice path matroids from @xcite .
the notation established in this section is used throughout the paper .
also included are the basic results about matroid connectivity that we use later . unless otherwise stated , all lattice paths in this paper start at the point @xmath17 and use steps @xmath18 and @xmath19 , which are called _ east _ and _ north _ , respectively .
paths are usually represented as words in the alphabet @xmath20 .
we say that a lattice path @xmath21 has a _
@xmath22 corner at @xmath23 _ if step @xmath23 of @xmath21 is north and step @xmath24 is east .
an _ @xmath25 corner at @xmath26 _ is defined similarly .
a corner can also be specified by the coordinates of the point where the north and east steps meet .
a _ lattice path matroid _ is , up to isomorphism , a matroid of the type @xmath27 $ ] that we now define .
let @xmath21 and @xmath28 be lattice paths from @xmath17 to @xmath29 with @xmath21 never going above @xmath28 .
let @xmath30 be the set of all lattice paths from @xmath17 to @xmath29 that go neither above @xmath28 nor below @xmath21 .
for @xmath31 with @xmath32 , let @xmath33 be the set @xmath34 thus , @xmath35 is a sequence of intervals in @xmath36 $ ] , and both the left endpoints and the right endpoints form strictly increasing sequences ; the left and right endpoints of @xmath33 correspond to the positions of the @xmath31-th north steps in @xmath28 and @xmath21 , respectively .
the matroid @xmath27 $ ] is the transversal matroid on the ground set @xmath36 $ ] that has @xmath37 as a presentation .
we call @xmath37 the _ standard presentation _ of @xmath27 $ ] . note that @xmath27 $ ] has rank @xmath6 and nullity @xmath38
. figure [ draw ] shows a lattice path matroid of rank @xmath39 and nullity @xmath40 .
the intervals in the standard presentation are @xmath41 $ ] , @xmath42 $ ] , @xmath43 $ ] , and @xmath44 $ ] .
( section [ ssec : geom ] explains how to find a geometric representation of a lattice path matroid . )
4.5truein a feature that enriches the subject of lattice path matroids is the variety of ways in which these matroids can be viewed . on the one hand , the theory of transversal matroids provides many useful tools . on the other hand ,
the following theorem from @xcite gives an interpretation of the bases that leads to attractive descriptions of many matroid concepts ( see , e.g. , ( * ? ? ?
* theorem 5.4 ) on basis activities ) .
[ thm : bases ] the map @xmath45 is a bijection from @xmath30 onto the set of bases of @xmath27 $ ] .
we use @xmath46 to denote the class of all lattice path matroids .
we call the pair @xmath47 a _ lattice path presentation _ of @xmath27 $ ] , or , if there is no danger of confusion , a _ presentation _ of @xmath27
$ ] . unless we say otherwise , all references to an order on the ground set @xmath36 $ ] of @xmath27 $ ] are to the natural order @xmath48 .
however , this order is not inherent in the matroid structure ; the elements of a lattice path matroid typically can be linearly ordered in many ways so as to correspond to the steps of lattice paths . also , a lattice path matroid of rank @xmath6 and nullity @xmath38 need not have @xmath36 $ ] as its ground set .
these comments motivate the following definition .
[ def : lpo ] a linear ordering @xmath49 of the ground set of a lattice path matroid @xmath1 is a _ lattice path ordering _ if the map @xmath50 is an isomorphism of @xmath1 onto a lattice path matroid of the form @xmath27 $ ] . for some purposes it is useful to view lattice path matroids from the following perspective , which does not refer to paths .
lattice path matroids are the transversal matroids @xmath1 for which @xmath9 can be linearly ordered so that @xmath1 has a presentation @xmath51 where @xmath52 $ ] is an interval in @xmath9 and the endpoints of these intervals form two chains , @xmath53 and @xmath54 .
the _ incidence function _ of a presentation @xmath51 of a transversal matroid is given by @xmath55 for subsets
@xmath0 of @xmath9 .
if no other presentation is mentioned , the incidence function of the matroid @xmath27 $ ] of rank @xmath6 and nullity @xmath38 is understood to be that associated with the standard presentation .
for this incidence function and for any element @xmath56 in @xmath36 $ ] , the set @xmath57 is an interval in @xmath58 $ ] ; if @xmath59 , then @xmath60 and @xmath61 .
an independent set @xmath62 in a lattice path matroid @xmath27 $ ] is a partial transversal of @xmath63 .
typically there are many ways to match @xmath62 with @xmath35 .
the next two results show that @xmath62 can always be matched in a natural way .
the following lemma , which is crucial in the proof of theorem [ thm : bases ] , is from @xcite [ lem : basisnoncross ] assume @xmath64 is a basis of a lattice path matroid @xmath27 $ ] with @xmath65 . then @xmath66 is in @xmath33 for all @xmath31 .
corollary [ cor : indepnoncross ] follows by extending the given independent set @xmath62 to a basis and applying lemma [ lem : basisnoncross ] .
[ cor : indepnoncross ] assume @xmath62 is an independent set of a lattice path matroid @xmath27 $ ] with @xmath67 ; let @xmath62 be @xmath68 with @xmath69 and let @xmath70 be @xmath71 with @xmath72 .
then @xmath73 is in @xmath74 for all @xmath75 with @xmath76 .
we now gather several results on matroid connectivity that are relevant to parts of the paper .
the first result ( * ? ? ?
* theorem 7.1.16 ) gives a fundamental link between connectivity and the operation of parallel connection .
[ lem : discon ] if @xmath1 is connected and @xmath77 is the direct sum @xmath78 , then @xmath1 is the parallel connection @xmath79 of @xmath80 and @xmath81 . in lemma [ lem : discon ] , since @xmath1 is connected , both @xmath82 and @xmath83 are connected .
recall that the rank @xmath84 of a parallel connection whose basepoint is not a loop is @xmath85 .
these observations give the following lemma .
[ lem : smallerflats ] if @xmath1 is connected , @xmath56 is not parallel to any element of @xmath1 , and @xmath86 is disconnected , then there is a pair @xmath87 of nontrivial incomparable connected flats of @xmath1 with @xmath88 and @xmath89 .
the following useful lemma is easy to prove by using separating sets .
assume that @xmath0 is a connected flat of a connected matroid @xmath1 , that @xmath56 is in @xmath0 , and that @xmath90 is connected
. then @xmath91 is connected .
the cyclic flats of a matroid @xmath1 ( that is , the flats @xmath92 for which @xmath93 has no isthmuses ) , together with their ranks , determine the matroid ( * ? ? ?
* proposition 2.1 ) .
as we show next , in the loopless case it suffices to consider nontrivial connected flats .
note that nontrivial connected flats are cyclic , but cyclic flats need not be connected .
thus , the next result is a mild refinement of ( * ? ? ?
* proposition 2.1 ) , and essentially the same idea proves both results . [
lem : cyclicgen ] the circuits of a loopless matroid @xmath1 ( and hence @xmath1 itself ) are determined by the nontrivial connected flats and their ranks .
note that if @xmath94 is an @xmath31-circuit , then @xmath95 is a connected flat of rank @xmath96 .
thus , the circuits can be recovered inductively as follows : the @xmath97-circuits are the @xmath97-subsets of nontrivial rank-@xmath98 flats ; the @xmath99-circuits are the @xmath99-subsets of @xmath9 that contain no @xmath97-circuit and are subsets of connected lines , and so on .
[ cor : genauto ] the automorphisms of a loopless matroid are the permutations of the ground set that are rank - preserving bijections of the collection of nontrivial connected flats .
this section treats the basic structural properties of lattice path matroids that play key roles throughout this paper .
some of these properties are shared by few other classes of matroids ; for instance , every nontrivial connected lattice path matroid has a spanning circuit .
other properties , such as the closure of the class of lattice path matroids under minors and duals , while shared by many classes of matroids , do not hold for the larger class of transversal matroids .
some of the properties are more technical and their significance will become apparent only later in the paper .
the topics treated are fairly diverse , so we divide the material into subsections that focus in the following issues : minors , duals , and direct sums ; connectivity and spanning circuits ; the structure of circuits and connected flats . the class of transversal matroids , although closed under deletions and direct sums , is closed under neither contractions nor duals .
in contrast , we have the following result for lattice path matroids .
[ thm : minors ] the class @xmath46 is closed under minors , duals , and direct sums .
2.75truein figure [ sum ] illustrates the obvious construction to show that @xmath46 is closed under direct sums . for closure under duality , note that , from theorem [ thm : bases ] , a basis of the dual of @xmath27 $ ] ( i.e. , the complement of a basis of @xmath27 $ ] ) corresponds to the east steps in a lattice path ; the east steps of a lattice path are the north steps of the lattice path obtained by reflecting the entire diagram about the line @xmath100
this idea is illustrated in figure [ dual ] .
2.7truein for closure under minors , it suffices to consider single - element deletions .
let @xmath56 be in the lattice path matroid @xmath101 $ ] on @xmath36 $ ] with standard presentation @xmath37 .
note that @xmath102 is a presentation of @xmath91 ; from this presentation , we will obtain one that shows that @xmath91 is a lattice path matroid .
some set @xmath33 is @xmath103 if and only if @xmath56 is an isthmus of @xmath1 ; in this case , discard the empty set @xmath104 from the presentation above to obtain the required presentation of @xmath91 .
thus , assume @xmath56 is not an isthmus of @xmath1 .
the sets @xmath105 are intervals in the induced linear order on @xmath36-x$ ] . in only two cases
will the least elements or the greatest elements ( or both ) fail to increase strictly : ( a ) @xmath56 is the least element of the interval @xmath33 and @xmath106 is the least element of @xmath107 , and ( b ) @xmath108 and @xmath56 are the greatest elements of @xmath109 and @xmath110 , respectively . assume case ( a ) applies .
any basis of @xmath91 ( that is , any basis of @xmath1 that does not contain @xmath56 ) that contains @xmath106 can , by lemma [ lem : basisnoncross ] , be matched with @xmath111 so that @xmath106 is not matched to @xmath107 .
thus , the set system obtained by replacing @xmath107 by @xmath112 is also a presentation of @xmath91 .
the same argument justifies replacing @xmath113 by @xmath114 if @xmath115 is the least element of @xmath113 , and so on .
case ( b ) is handled similarly .
the result is a presentation of @xmath91 by intervals in which the least and greatest elements increase strictly , so @xmath91 is a lattice path matroid .
single - element deletions and contractions can be described in terms of the bounding paths of @xmath101 $ ] as follows .
an isthmus is an element @xmath56 for which some @xmath33 is @xmath103 ; to delete or contract @xmath56 , eliminate the corresponding common north step from both bounding paths .
a loop is an element that is in no set @xmath33 ; to delete or contract a loop , eliminate the corresponding common east step from @xmath21 and @xmath28 .
now assume @xmath56 is neither a loop nor an isthmus . the upper bounding path for @xmath91 is formed by deleting from @xmath28 the first east step that is at or after step @xmath56 ; the lower bounding path for @xmath91 is formed by deleting from @xmath21 the last east step that is at or before step @xmath56 .
this is shown in figure [ delete ] , where the dashed steps in the middle diagram indicate the steps that bases of @xmath91 must avoid .
dually , the upper bounding path for the contraction @xmath86 is formed by deleting from @xmath28 the last north step that is at or before step @xmath56 ; the lower bounding path for @xmath86 is formed by deleting from @xmath21 the first north step that is at or after step @xmath56 .
3.8truein corollary [ cor : restint ] treats restrictions of lattice path matroids to intervals .
the lattice path interpretation of this result is illustrated in figure [ fundint ] on page .
[ cor : restint ] let @xmath1 be the lattice path matroid @xmath27 $ ] on the ground set @xmath36 $ ] .
let @xmath0 be the initial segment @xmath116 $ ] and @xmath117 be the final segment @xmath118 $ ] of @xmath36 $ ] .
let the @xmath31-th step of @xmath28 end at the point @xmath119 and let the @xmath75-th step of @xmath21 end at @xmath120 .
* the bases of the restriction @xmath2 correspond to the lattice paths that go from @xmath17 to @xmath119 and go neither below @xmath21 nor above @xmath28 .
* the bases of the restriction @xmath121 correspond to the lattice paths that go from @xmath120 to @xmath29 and go neither below @xmath21 nor above @xmath28 . * if @xmath122 , then the bases of @xmath123 correspond to the lattice paths that go from @xmath120 to @xmath119 and go neither below @xmath21 nor above @xmath28 .
we close this section by noting that although @xmath124 is a lattice path matroid , its truncation is not transversal .
it follows that @xmath46 is not closed under the following operations : truncation , free extension , and elongation .
we begin with a rare property .
[ thm : spanning ] a connected lattice path matroid @xmath27 $ ] on at least two elements has a spanning circuit . let @xmath27 $ ] have rank @xmath6 , let @xmath110 be @xmath125 $ ] for @xmath126 , and let @xmath94 be the set @xmath127
. showing that each set @xmath128 , for @xmath56 in @xmath94 , is a basis shows that @xmath94 is a spanning circuit . that @xmath129 and @xmath130 are bases is clear .
since @xmath27 $ ] is not a direct sum of two matroids , @xmath131 must be in @xmath33 for @xmath132 , from which it follows that each set @xmath133 , with @xmath134 , is a basis .
it will be useful to single out the following immediate corollary of theorem [ thm : spanning ] .
[ cor : connflat ] if @xmath0 is a nontrivial connected flat of a matroid @xmath1 and @xmath2 is a lattice path matroid , then @xmath0 is @xmath95 for some circuit @xmath94 of @xmath1 .
the next theorem determines which lattice path matroids are connected .
one implication follows from the description of direct sums and the other from the construction of the spanning circuit in the proof of theorem [ thm : spanning ] .
[ thm : conncrit ] a lattice path matroid @xmath27 $ ] of rank @xmath6 and nullity @xmath38 is connected if and only if @xmath21 and @xmath28 intersect only at @xmath17 and @xmath29 .
the parallel connection of two @xmath99-point lines , which has only one spanning circuit , shows that there may be elements of a connected lattice path matroid that are in no spanning circuit .
there are several ways to identify the elements of connected lattice path matroids that are in spanning circuits .
the next result identifies these elements via the standard presentation .
[ thm : spanningvar ] an element @xmath56 of a nontrivial connected lattice path matroid @xmath27 $ ] of rank @xmath6 is in a spanning circuit of @xmath27 $ ] if and only if @xmath56 is in at least two of the sets @xmath111 , or @xmath56 is in @xmath135 or @xmath136 .
assume @xmath56 is in @xmath33 and @xmath107 .
let @xmath94 be @xmath137 where @xmath110 is @xmath125 $ ] . by connectivity , we have @xmath138 and @xmath139 .
an argument like that in the proof of theorem [ thm : spanning ] shows that @xmath94 is a spanning circuit .
similar ideas show that @xmath56 is in a spanning circuit of @xmath27 $ ] if @xmath56 is in @xmath135 or @xmath136 .
assume @xmath57 is @xmath140 with @xmath141 . note that the basepoint is in no spanning circuit of a parallel connection of matroids of rank two or more ,
so to complete the proof we need only show that @xmath27 $ ] is a parallel connection of two lattice path matroids , each of rank at least two , with basepoint @xmath56 .
thus , by lemma [ lem : discon ] , we need to show that @xmath27/x{\backslash}x$ ] is disconnected where @xmath0 is the set of loops of @xmath27/x$ ] .
this statement follows from the lattice path description of contraction along with the observations that @xmath142 contains only elements less than @xmath56 while @xmath107 contains only elements greater than @xmath56 .
the following characterizations of the elements that are in spanning circuits use structural properties rather than presentations .
[ cor : spanningvar ] let @xmath56 be in a nontrivial connected lattice path matroid @xmath1 .
* no spanning circuit contains @xmath56 if and only if @xmath1 is a parallel connection of two lattice path matroids , each of rank at least two , with basepoint @xmath56 . * some spanning circuit contains @xmath56 if and only if @xmath143 is connected , where @xmath0 is the set of loops of @xmath86 .
part ( a ) follows from the proof of theorem [ thm : spanningvar ] .
if @xmath56 is in a spanning circuit @xmath94 of @xmath1 , then @xmath128 is a spanning circuit of @xmath86 , so @xmath143 is connected .
conversely , if @xmath56 is in no spanning circuit of @xmath1 , then , by part ( a ) , @xmath1 is a parallel connection , with basepoint @xmath56 , of matroids of rank at least two , so @xmath143 is disconnected .
our first goal in this section is to characterize the circuits of lattice path matroids .
this is done in theorem [ thm : cinterval ] , the proof of which uses the following well - known elementary result about the circuits of arbitrary transversal matroids .
this lemma follows easily from hall s theorem .
[ lem : circform ] let @xmath144 be the incidence function of a presentation of a transversal matroid @xmath1 . if @xmath94 is a rank-@xmath26 circuit of @xmath1 , then @xmath145 is @xmath26 , as is @xmath146 for any @xmath56 in @xmath94 .
[ thm : cinterval ] let @xmath147 be a set in the lattice path matroid @xmath27 $ ] ; assume @xmath148 .
let @xmath149 be @xmath150 , where @xmath151 .
then @xmath94 is a circuit of @xmath27 $ ] if and only if * @xmath152 , * @xmath153 , * @xmath154 , and * @xmath155 for @xmath75 with @xmath156 .
furthermore , if @xmath94 is a circuit , then @xmath157 for @xmath158 .
it is immediate to check that if conditions ( @xmath98)(@xmath39 ) hold , then @xmath94 is dependent and every @xmath26-subset of @xmath94 is a partial transversal and so is independent ; thus @xmath94 is a circuit . for the converse ,
assume @xmath94 is a circuit .
assertion ( @xmath98 ) follows from lemma [ lem : circform ] , which also gives the equalities @xmath159 .
since @xmath160 is independent and @xmath161 is @xmath26 , it follows from corollary [ cor : indepnoncross ] that @xmath162 is in @xmath74 for @xmath163 . a similar argument using @xmath164 shows that @xmath162 is in @xmath165 for @xmath166 .
this proves assertions ( @xmath97)(@xmath39 ) . to prove the last assertion , assume
there were an @xmath23 not in @xmath149 with @xmath167 . from statement ( @xmath39 )
, we have that @xmath162 is in both @xmath74 and @xmath165 .
the inequalities @xmath168 imply that @xmath162 is in @xmath169 , which contradicts the assumption that @xmath23 is not in @xmath149 . by lemma
[ lem : circform ] , if @xmath56 is parallel to some element , then @xmath170 . by property ( @xmath39 ) of theorem
[ thm : cinterval ] , at most two elements @xmath56 in a circuit of a lattice path matroid can satisfy the equality @xmath170 .
this observation proves the next result .
[ cor : cinterval ] at most two elements in any circuit of a lattice path matroid are in nonsingleton parallel classes .
the following result gives two useful properties of connected flats .
[ thm : finterval ] let @xmath27 $ ] have rank @xmath6 and nullity @xmath38 .
any nontrivial connected flat @xmath0 of @xmath27 $ ] is an interval in @xmath36 $ ] and @xmath171 is an interval of @xmath172 elements in @xmath58 $ ] .
the second assertion follows from corollary [ cor : connflat ] and theorem [ thm : cinterval ] . for the first statement , let @xmath171 be @xmath173 $ ] and assume @xmath174 with @xmath175 . that @xmath75 is in @xmath0 follows from the inequalities @xmath176 theorem [ thm : finterval ] has many implications for the connected flats of lattice path matroids , of which we mention four .
[ cor : finterval ] assume @xmath27 $ ] has rank @xmath6 .
* for @xmath177 , there are at most @xmath178 nontrivial connected flats of rank @xmath179 in @xmath27 $ ] .
in particular , @xmath27 $ ] has at most two connected hyperplanes and at most @xmath7 connected lines . *
a flat of positive rank of @xmath27 $ ] is covered by at most two connected flats . *
the nontrivial connected flats of @xmath27 $ ] that are not contained in a fixed connected hyperplane @xmath180 of @xmath27 $ ] are linearly ordered by inclusion . *
if @xmath180 and @xmath181 are connected hyperplanes of @xmath27 $ ] , then every nontrivial connected flat of @xmath27 $ ] is contained in at least one of @xmath180 and @xmath181 . the matroid @xmath182 $ ] , which is a parallel connection of @xmath7 three - point lines in which elements have been added parallel to the `` joints '' and the `` ends '' , shows that all upper bounds in parts ( i ) and ( ii ) of corollary [ cor : finterval ] are optimal . the next result is another corollary of theorem [ thm : cinterval ] .
[ cor : initfin ] let @xmath94 be the circuit @xmath183 of @xmath27 $ ] with @xmath184 . if @xmath56 is not in @xmath94 and @xmath185 is a circuit of @xmath27 $ ] for some subset @xmath186 of @xmath94 , then @xmath186 is either an initial segment @xmath187 or a final segment @xmath188 of @xmath94 .
the result follows from lemma [ lem : circform ] and this simple corollary of theorem [ thm : cinterval ] : for any proper subset @xmath0 of @xmath94 that is neither an initial nor final segment of @xmath94 , the inequality @xmath189 holds .
we conclude this section with a result we will use to show that certain matroids are not lattice path matroids .
[ thm : notlpm ] assume a rank-@xmath6 matroid @xmath1 has two nontrivial connected flats @xmath0 and @xmath190 such that * @xmath191 , * @xmath192 , and * @xmath193 is a proper subset of the ground set @xmath9 of @xmath1 .
then @xmath1 is not a lattice path matroid .
assume , to the contrary , that @xmath1 is @xmath27 $ ] .
fix @xmath56 in @xmath194 and @xmath195 in @xmath196 . by theorem [ thm : finterval ] , along with assumptions ( @xmath98 ) and ( @xmath97 ) , up to switching @xmath0 and @xmath190 we would have @xmath197 $ ] and @xmath198 $ ] for some @xmath26 and @xmath199 with @xmath200 . the inequality @xmath201 would give @xmath202 , so @xmath195 would be in @xmath203 . the inequality @xmath59 would give @xmath204 , so @xmath195 would be in @xmath205 . that these conclusions contradict the hypothesis proves the lemma
our next aim is to characterize lattice path matroids ; this will be done in section [ sec : fundamental ] .
this section focuses on an important subclass of @xmath46 that has particularly simple characterizations and many interesting properties .
the _ @xmath144-th catalan matroid _ @xmath206 is @xmath207 $ ] .
generalized catalan matroid _ is , up to isomorphism , a matroid of the form @xmath208 $ ] . for generalized catalan matroids , the notation @xmath27 $ ] is simplified to @xmath209 $ ] .
we use @xmath210 to denote the class of generalized catalan matroids .
generalized catalan matroids have arisen in different contexts with a corresponding variety of names and perspectives .
we gather here the references currently known to us .
crapo ( * ? ? ?
* section 8) introduced these matroids to show that there are at least @xmath211 nonisomorphic matroids of rank @xmath6 on @xmath144 elements .
his perspective was rediscovered in ( * ? ? ?
* theorem 3.14 ) : generalized catalan matroids are precisely the matroids that are obtained from the empty matroid by repeatedly applying the operations of adding an isthmus and forming the free extension ( this result is generalized in theorem [ thm : extend ] below ) . by using `` nested '' presentations ,
welsh @xcite proved that crapo s lower bound on the number of matroids holds within the smaller class of transversal matroids .
these matroids arose again in @xcite in connection with matroids defined in terms of integer - valued functions on finite sets .
they were studied further in @xcite , where they were called schubert matroids and were shown to have the rapid mixing property . in @xcite they were rediscovered and related to shifted complexes , and so acquired the name shifted matroids .
the link that was established in @xcite between generalized catalan matroids and an enumerative problem known as the tennis ball problem influenced the techniques used in @xcite to solve that problem . in @xcite , under the name of freedom matroids , generalized catalan matroids were used to construct a free algebra of matroids .
catalan matroids have rich enumerative properties ( see @xcite ) .
their name comes from the fact that the number of bases of @xmath206 is the catalan number @xmath212 ; several other invariants of @xmath206 are also catalan numbers .
although there is only one catalan matroid of each rank , these matroids generate the entire class @xmath210 , in the sense of the following theorem .
the smallest minor - closed class of matroids that contains all catalan matroids is @xmath210 .
it follows from the lattice path interpretation of deletion and contraction given after the proof of theorem [ thm : minors ] that @xmath210 is closed under minors . to see that any generalized catalan matroid @xmath209 $ ] is a minor of a catalan matroid , simply insert east and north steps into @xmath28 so that the result is a catalan matroid @xmath213 $ ] . from @xmath213
$ ] , delete the elements that correspond to the added east steps and contract the elements that correspond to the added north steps ; by the lattice path interpretation of these operations , the resulting minor of @xmath213 $ ] is @xmath209 $ ] .
it is easy to see that @xmath210 , in addition to being closed under minors , is closed under duals and ( unlike @xmath46 ) free extension ; therefore @xmath210 is closed under truncation and elongation .
however , @xmath210 is not closed under direct sums . by theorem [ thm : conncrit ] ,
a generalized catalan matroid with at least two elements is connected if and only if it has neither loops nor isthmuses .
the rest of this section focuses mainly on connected generalized catalan matroids since some results are slightly easier to state with this restriction and , by what we just noted , there is essentially no loss of generality .
the feature that makes generalized catalan matroids easy to characterize is the structure of the connected flats , as described in the following lemma .
[ lem : flatform ] assume @xmath209 $ ] has rank @xmath6 , nullity @xmath38 , and neither loops nor isthmuses .
let the @xmath25 corners of @xmath28 be at steps @xmath214 with @xmath72 .
the proper nontrivial connected flats of @xmath209 $ ] are the initial segments @xmath215\subset[i_2]\subset\cdots\subset[i_k]$ ] of @xmath36 $ ] . the rank ( resp .
nullity ) of @xmath216 $ ] is the number of north ( resp .
east ) steps among the first @xmath217 steps of @xmath28 .
the lemma follows easily once we show that any proper nontrivial connected flat @xmath92 of @xmath209 $ ] is an initial segment of @xmath36 $ ] . by theorem [ thm : finterval ]
, @xmath92 is an interval , say @xmath218 $ ] , in @xmath36 $ ] . by corollary [ cor : restint ] , the restriction of @xmath209 $ ] to @xmath219 $ ] is @xmath220 $ ] where @xmath221 consists of the first @xmath222 steps of @xmath28 . since @xmath222 is not an isthmus of @xmath209|f$ ] , it is not an isthmus of @xmath220 $ ] , so the @xmath222-th step of @xmath28 must be east .
let @xmath220 $ ] have rank @xmath26 .
note that @xmath223 $ ] is a spanning circuit of @xmath220 $ ] that is contained in @xmath92 and has closure @xmath219 $ ] .
thus , @xmath92 is the initial segment @xmath219 $ ] . the following result ( which is essentially lemma 2 of @xcite ) is an immediate corollary of lemmas [ lem : cyclicgen ] and [ lem : flatform ] . [ cor : chainflats ] a connected matroid is a generalized catalan matroid if and only if its nontrivial connected flats are linearly ordered by inclusion .
the following excluded - minor characterization of @xmath210 from @xcite is not difficult to prove from corollary [ cor : chainflats ] and the results in section [ sec : lpm ] .
let @xmath224 be the truncation @xmath225 to rank @xmath144 of the direct sum of two @xmath144-circuits .
thus , @xmath224 is the paving matroid of rank @xmath144 whose only nontrivial proper flats are two disjoint circuit - hyperplanes whose union is the ground set .
it follows that @xmath224 is isomorphic to @xmath226 $ ] and , by corollary [ cor : chainflats ] , that @xmath224 is not in @xmath210 .
[ thm : oxprow ] a matroid is in @xmath210 if and only if it has no minor isomorphic to @xmath224 for any @xmath227 .
while the structure of the connected flats of arbitrary connected lattice path matroids is not as simple as that for generalized catalan matroids ( corollary [ cor : chainflats ] ) , this structure is still easy to describe .
we analyze this structure in this section and we use it to characterize connected lattice path matroids . we also show that if @xmath27 $ ] is connected , then the paths @xmath21 and @xmath28 are determined , up to a @xmath228 rotation , by any matroid isomorphic to @xmath27 $ ] .
the flats of central interest for these results are those we define now .
let @xmath0 be a connected flat of a connected matroid @xmath1 for which @xmath229 and @xmath230 .
we say that @xmath0 is a _ fundamental flat of @xmath1 _ if for some spanning circuit @xmath94 of @xmath1 the intersection @xmath231 is a basis of @xmath0 .
the first lemma shows how fundamental flats of lattice path matroids reflect the order of the elements .
[ lem : cflemma ] assume @xmath27 $ ] is connected and has rank @xmath6 and nullity @xmath38 .
let @xmath0 be a connected flat of @xmath27 $ ] with @xmath229 and @xmath232 .
then @xmath0 is a fundamental flat of @xmath27 $ ] if and only if @xmath0 is an initial or final segment of @xmath36 $ ] .
let @xmath33 be @xmath233 $ ] for @xmath234 .
if @xmath0 is an initial segment @xmath235 $ ] of @xmath36 $ ] , then the spanning circuit @xmath236 constructed in the proof of theorem [ thm : spanning ] , has the property that @xmath231 is a basis of @xmath0 .
similarly , for a final segment @xmath0 of @xmath36 $ ] , a spanning circuit with the required property is @xmath237 .
conversely , assume @xmath94 is a spanning circuit of @xmath27 $ ] and @xmath238 is a basis of @xmath0 ; say @xmath94 is @xmath239 with @xmath240 . by theorem [ thm : finterval ] , it suffices to show that either @xmath98 or @xmath241 is in @xmath0 .
let @xmath56 be in @xmath242 and let @xmath243 be the unique circuit in @xmath244 . by corollary [ cor : initfin ]
, @xmath243 has the form @xmath245 or @xmath246 .
we will show that in the first case , @xmath98 is in @xmath0 ; a similar argument gives @xmath241 in @xmath0 in the second case .
thus , let @xmath243 be @xmath245 .
note that @xmath247 is @xmath248 and @xmath98 is in @xmath249 .
note also that @xmath250 is an independent set and , by lemma [ lem : circform ] applied to @xmath243 , we have @xmath251 .
thus , @xmath252 it follows that @xmath98 is in @xmath253 , so @xmath98 is in @xmath0 , as claimed . hence , to determine the fundamental flats of @xmath27 $ ] , it suffices to know which initial and final segments of @xmath36 $ ] are connected flats .
note that the initial segment @xmath235 $ ] of @xmath36 $ ] is a proper nontrivial connected flat , and hence a fundamental flat , if and only if the upper path @xmath28 has an @xmath25 corner at @xmath23 .
similarly , the final segment @xmath254 $ ] of @xmath36 $ ] is a fundamental flat of @xmath27 $ ] if and only if @xmath21 has a @xmath22 corner at @xmath255 .
these observations prove the following theorem .
[ thm : shape ] assume @xmath27 $ ] is connected and has rank @xmath6 and nullity @xmath38 .
let the @xmath25 corners of @xmath28 be at @xmath256 , with @xmath257 , and the @xmath22 corners of @xmath21 be at @xmath258 , with @xmath259 .
the fundamental flats of @xmath27 $ ] are @xmath215\subset [ i_2]\subset\cdots\subset [ i_h]$ ] and @xmath260\subset \cdots \subset [ j_2,m+r]\subset [ j_1,m+r]$ ] .
corollary [ cor : twoch ] follows immediately from theorem [ thm : shape ] .
note that for generalized catalan matroids , the fundamental flats are precisely the flats given in lemma [ lem : flatform ] , so they form one chain under inclusion .
[ cor : twoch ] the fundamental flats of a connected matroid @xmath1 in @xmath261 form two chains under inclusion ; no set in one chain contains a set in the other chain .
furthermore , for each pair @xmath262 of incomparable fundamental flats , * if @xmath263 , then @xmath264 , and * if @xmath265 , then @xmath266 . while a connected lattice path matroid of rank @xmath6 has at most @xmath178 connected flats of rank @xmath179 ( corollary [ cor : finterval ] ) ,
it has at most two fundamental flats of any given rank .
theorem [ thm : shape ] and the lattice path interpretation of duality give the next result .
[ cor : critcomp ] for any lattice path matroid @xmath1 , the fundamental flats of the dual @xmath267 are the set complements , @xmath268 , of the fundamental flats @xmath92 of @xmath1 .
a key observation that follows from theorem [ thm : shape ] is that although which flats are fundamental is independent of the order of the elements that is inherent in any particular lattice path presentation of a lattice path matroid , such a presentation makes it easy to identify the fundamental flats .
conversely , the chains of fundamental flats give the bounding paths .
more precisely , the paths @xmath21 and @xmath28 associated with @xmath27 $ ] are determined by the @xmath22 corners of @xmath21 and the @xmath25 corners of @xmath28 , and these corners are determined by the ranks and nullities of the fundamental flats .
typically there are two possible pairs of paths , according to which chain of fundamental flats contains the least element of the ground set .
these observations give the following theorem , which is one of the main results of this section .
[ thm : shape2 ] the bounding paths @xmath21 and @xmath28 of a connected lattice path matroid @xmath27 $ ] are determined by the matroid structure , up to a @xmath228 rotation .
that is , the only matroids @xmath269 $ ] isomorphic to @xmath27 $ ] are @xmath27 $ ] and @xmath270 $ ] where @xmath271 is @xmath272 . theorem [ thm : shape ] and its corollaries ( including theorem [ thm : shape2 ] ) show that a connected lattice path matroid is determined by its fundamental flats and their ranks .
the next several results further develop this idea .
the following theorem describes all connected flats of a connected lattice path matroid in terms of its fundamental flats .
[ thm : cflats ] let @xmath1 be the connected lattice path matroid @xmath27 $ ] of rank @xmath6 and nullity @xmath38 and let @xmath273 and @xmath274 be the chains of fundamental flats of @xmath1 .
the proper nontrivial connected flats of @xmath1 are * @xmath275 , and * the intersections @xmath276 for which the inequality @xmath277 holds .
a nontrivial connected flat of the form @xmath276 has rank @xmath278 .
the flats @xmath279 , being fundamental , are connected . the element @xmath98 is in either @xmath280 or @xmath281
; we may assume it is in @xmath280 .
for part ( ii ) , we use corollary [ cor : restint ] to find a lattice path presentation that shows that @xmath276 is connected . using the notation in that corollary ,
let @xmath0 be @xmath282 , so the point @xmath119 on @xmath28 is @xmath283 ; let @xmath117 be @xmath284 , so the point @xmath120 on @xmath21 is @xmath285 .
the inequality in part ( ii ) along with part ( c ) of corollary [ cor : restint ] give a presentation of @xmath286 ( illustrated in figure [ fundint ] ) that , together with the fact that @xmath21 and @xmath28 meet only at @xmath17 and @xmath29 , implies that @xmath276 is connected and nontrivial .
3.6truein now assume @xmath0 is a proper nontrivial connected flat . by theorem [ thm :
finterval ] , @xmath0 is an interval , say @xmath218 $ ] , in @xmath36 $ ] . as in the proof of lemma [ lem : flatform ] , it follows that the @xmath287-th step of @xmath21 and the @xmath222-th step of @xmath28 are east steps .
since @xmath0 is a flat , both @xmath288 and @xmath289 exceed @xmath172 , so step @xmath290 of @xmath21 and step @xmath291 of @xmath28 , if there are such steps , are north steps . from these observations and
theorem [ thm : shape ] , it follows that @xmath0 is of the form @xmath282 , @xmath284 , or @xmath276 .
we need to show that if @xmath276 is connected , then the inequality @xmath292 holds .
this inequality follows by viewing @xmath286 as a restriction of @xmath293 and using the path presentations of these matroids given in corollary [ cor : restint ] .
indeed , from the lattice path diagrams ( figure [ fundint ] ) it follows that @xmath286 is either free or connected , and the latter holds precisely when @xmath285 is strictly to the left of @xmath283 .
lastly , let the connected flat @xmath0 be @xmath276
. from lattice path diagrams , we get @xmath294 from which the last assertion follows .
it follows from theorem [ thm : cflats ] that any intersection of connected flats is either a fundamental flat or an intersection of two fundamental flats . from this observation and the second paragraph of the proof
, it follows that a nonempty intersection of connected flats is either connected or trivial . despite what the last part of theorem [ thm : cflats ] might suggest , it is easy to construct examples in which the fundamental flats of lattice path matroids are not modular .
the image , under an automorphism , of a fundamental flat of any matroid is also fundamental .
this observation , corollary [ cor : genauto ] , and theorem [ thm : cflats ] give the following result .
[ cor : lpmauto ] the automorphisms of a connected lattice path matroid are the permutations of the ground set that are rank - preserving bijections of the collection of fundamental flats .
the proof of the second main result of this section , theorem [ thm : lpmchar ] , uses the following basic notions about ordered sets .
strict partial order _ is an irreflexive , transitive relation .
thus , strict partial orders differ from partial orders only in whether each element is required to be unrelated , or required to be related , to itself . given a strict partial order @xmath295 on @xmath296 , elements @xmath56 and @xmath195 of @xmath296 are _ incomparable _ if neither @xmath59 nor @xmath201 holds .
_ weak orders _ are strict partial orders in which incomparability is an equivalence relation
. thus , linear orders are weak orders in which the incomparability classes are singletons .
two weak orders @xmath297 and @xmath298 on @xmath296 are _ compatible _ if whenever elements @xmath56 and @xmath195 of @xmath296 are comparable in both @xmath297 and @xmath298 , and @xmath299 , then @xmath300 . [ lem : weak ] any two compatible weak orders have a common linear extension .
let @xmath297 and @xmath298 be compatible weak orders on @xmath296 and let the relation @xmath295 on @xmath296 be defined as follows : @xmath301 if either @xmath302 or @xmath300 .
it is easy to check that @xmath295 is a weak order .
the lemma follows since @xmath295 , like any strict partial order , can be extended to a linear order .
we now turn to the second main result of the section .
this theorem shows that the properties we developed above for the fundamental flats and the connected flats of connected lattice path matroids characterize these matroids .
[ thm : lpmchar ] a connected matroid @xmath1 is a lattice path matroid if and only if the following properties hold . * the fundamental flats form at most two disjoint chains under inclusion , say @xmath273 and @xmath274 . * if @xmath303 , then @xmath304 . *
the proper nontrivial connected flats of @xmath1 are precisely the following sets : * * @xmath275 , and * * intersections @xmath276 for which the inequality @xmath277 holds . *
the rank of the flat @xmath276 of item _
( iii : b ) _ is @xmath305 by theorem [ thm : shape ] , lemma [ lem : flatform ] , and corollary [ cor : chainflats ] , @xmath1 is a generalized catalan matroid if and only if properties ( i)(iv ) hold where there is at most one chain of fundamental flats . by theorems [ thm : shape ] and [ thm : cflats ] , the fundamental flats of a lattice path matroid that is not a generalized catalan matroid satisfy properties ( i)(iv ) with neither chain of fundamental flats being empty . hence we need only prove the converse in the case that neither chain of fundamental flats is empty .
assume @xmath1 has rank @xmath6 and nullity @xmath38 . to show that @xmath1 is a lattice path matroid
, we construct lattice paths @xmath21 and @xmath28 and an isomorphism of @xmath1 onto @xmath27 $ ] . to show that @xmath21 stays strictly below @xmath28 except at @xmath17 and @xmath29
, we will use the following statements about fundamental flats . *
if @xmath306 , then @xmath307 . * if @xmath308 , then @xmath309 . to prove statement ( a ) , note that we have the inequality @xmath310 by semimodularity and property ( ii ) . since @xmath1 has no loops , @xmath311 is positive , so the desired inequality follows . to prove statement ( b )
, first recall that @xmath312 is nondecreasing , i.e. , if @xmath313 , then @xmath314 .
since @xmath282 and @xmath284 are disjoint , we have @xmath315 .
thus , if @xmath316 , then we have @xmath317 . if @xmath318
, then @xmath319 is disconnected and we have the equality @xmath320 . since @xmath1 is connected , we have @xmath321 , which gives the desired inequality .
let lattice paths @xmath21 and @xmath28 from @xmath17 to @xmath29 be given as follows .
* the @xmath22 corners of @xmath21 are at the points @xmath322 for @xmath75 in @xmath323 $ ] .
* the @xmath25 corners of @xmath28 are at the points @xmath324 for @xmath31 in @xmath235 $ ] .
note that @xmath21 stays strictly below @xmath28 except at the endpoints if and only if for every @xmath22 corner @xmath325 of @xmath21 and every @xmath25 corner @xmath326 of @xmath28 , at least one of the inequalities @xmath327 and @xmath328 holds .
these inequalities are those in statements ( a ) and ( b ) , so @xmath21 stays strictly below @xmath28 except at @xmath17 and @xmath29 . to construct an isomorphism of @xmath1 onto @xmath27 $ ] , we define a linear order on @xmath9 that we use to map @xmath9 onto @xmath36 $ ] , the ground set of @xmath27 $ ] .
we first define two relations @xmath329 and @xmath330 on @xmath9 .
let @xmath331 and @xmath332 be @xmath9 .
define @xmath329 as follows : @xmath333 for @xmath334 if there is an integer @xmath31 in @xmath235 $ ] with @xmath335 and @xmath336 .
note that @xmath329 is a weak order whose incomparability classes are @xmath280 and the set differences @xmath337 .
define @xmath330 similarly : @xmath338 for @xmath334 if there is an integer @xmath75 in @xmath323 $ ] with @xmath339 and @xmath340 .
thus , @xmath330 is also a weak order and the incomparability classes are @xmath281 and the differences @xmath341 .
note that if we had @xmath342 and @xmath343 , then there would be fundamental flats @xmath282 and @xmath284 that both contain @xmath56 and not @xmath195 , contrary to hypothesis ( ii ) .
thus , the weak orders @xmath329 and @xmath330 are compatible , so by lemma [ lem : weak ] there is a linear order , say @xmath344 , of @xmath9 that extends both @xmath329 and @xmath330 .
let @xmath345 $ ] be given by @xmath346 . by construction
, @xmath347 is a bijection of @xmath9 onto @xmath36 $ ] that is a rank - preserving bijection of the fundamental flats of @xmath1 onto the fundamental flats of @xmath27 $ ] .
furthermore , by assumptions ( iii ) and ( iv ) and theorem [ thm : cflats ] , @xmath347 is a rank - preserving bijection of the set of connected flats of @xmath1 onto those of @xmath27 $ ] . by lemma [ lem : cyclicgen ]
, it follows that @xmath347 is an isomorphism of @xmath1 onto @xmath27 $ ] ; thus , @xmath1 is a lattice path matroid .
we close this section by giving a pair of six - element matroids that have the same collection of fundamental flats , yet only one of which is in @xmath46 ; thus , conditions ( i ) and ( ii ) in theorem [ thm : lpmchar ] are not enough to characterize lattice path matroids .
the uniform matroid @xmath348 is a lattice path matroid with no fundamental flats since the bounding paths are @xmath349 and @xmath350 .
the prism ( the matroid @xmath351 of figure [ b22 ] on page ) is not a lattice path matroid ( condition ( iii ) of theorem [ thm : lpmchar ] fails ) and , since it has no spanning circuits , it too has no fundamental flats .
the aspects of lattice path matroids treated in this section relate to important topics in the theory of transversal matroids .
we start by characterizing the set systems that are maximal presentations of lattice path matroids .
this result plays a key role in an algorithm for determining whether a transversal matroid is in @xmath46 . by combining the result on maximal presentations with brylawski s affine representation of transversal matroids ,
we get a geometric description of lattice path matroids .
we conclude the section by comparing @xmath46 with the dual - closed class of fundamental transversal matroids and the minor - closed class of bicircular matroids .
two types of presentations are of interest in this section .
a presentation @xmath352 of a transversal matroid @xmath1 is _ minimal _ if the only presentation @xmath353 of @xmath1 with @xmath354 contained in @xmath355 for all @xmath31 is @xmath10 .
the presentation @xmath10 is _ maximal _ if the only presentation @xmath353 of @xmath1 with @xmath355 contained in @xmath354 for all @xmath31 is @xmath10 .
it is well known that while each transversal matroid has a unique maximal presentation , it typically has many minimal presentations .
( see , e.g. , @xcite . )
standard presentations of lattice path matroids are minimal .
let @xmath37 be the standard presentation of the matroid @xmath27 $ ] and let @xmath356 be any presentation of @xmath27 $ ] with @xmath357 for @xmath358 . to prove the theorem
, we must show the inclusion @xmath359 for @xmath234 .
let @xmath56 be in @xmath33 .
let @xmath360 consist of the least elements of @xmath361 , the greatest elements of @xmath362 , and @xmath56 .
thus , @xmath360 is a basis of @xmath27 $ ] .
note that for @xmath360 to be a transversal of @xmath356 , the element @xmath56 must be in @xmath363 , as needed . with the following result of bondy @xcite
, we will get a simple description , in terms of intervals , of the maximal presentation of a lattice path matroid .
[ lem : max ] given a presentation @xmath51 of a rank-@xmath6 transversal matroid @xmath1 , the maximal presentation of @xmath1 is @xmath364 where @xmath365 is the set of isthmuses of the deletion @xmath366 .
together with lemma [ lem : max ] , the following result from @xcite implies that from any presentation of a transversal matroid , the maximal presentation can be found in polynomial time in the size of the ground set
. this observation will be important in the algorithm for recognizing lattice path matroids among transversal matroids .
[ lem : poly ] the maximal size of a matching in a bipartite graph can be found in polynomial time in the number of vertices .
the discussion below focuses on matroids that have no isthmuses .
this restriction is justified by noting that the isthmuses of a transversal matroid are in all sets in the maximal presentation , and so are easy to deal with .
let @xmath37 be the standard presentation of the lattice path matroid @xmath101 $ ] on @xmath36 $ ] , where @xmath1 has no isthmuses .
let @xmath33 be @xmath233 $ ] .
theorem [ thm : cinterval ] implies that each connected component of @xmath367 is a subset of either @xmath368 $ ] or @xmath369 $ ] .
thus , the set of isthmuses of @xmath367 is the union of the sets @xmath370 and @xmath371 of isthmuses of the restrictions of @xmath1 to @xmath368 $ ] and @xmath369 $ ] , respectively .
corollary [ cor : restint ] implies that @xmath370 and @xmath371 are given as follows : @xmath372 @xmath373 this proves the following theorem .
4.0truein [ thm : max ] let @xmath37 be the standard presentation of the lattice path matroid @xmath27 $ ] that has no isthmuses . the maximal presentation of @xmath27 $ ] is @xmath356 where @xmath363 is @xmath374 and @xmath370 and @xmath371 are given by eqs .
( [ eq : isth1])([eq : isth2 ] ) .
the sets in the maximal presentation of a lattice path matroid have a simple graphical interpretation , as figure [ maximal ] illustrates . while there are no containments among intervals in the standard presentation
, this figure shows that there may be containments ( even equalities ) among intervals in the maximal presentation .
theorem [ thm : charint ] , which characterizes the multisets of intervals in @xmath36 $ ] that are maximal presentations of lattice path matroids , uses the following notation . for an indexed multiset @xmath375 of nonempty intervals in
@xmath36 $ ] with @xmath376 $ ] , write @xmath377 if either @xmath378 or @xmath379 .
thus , two intervals are unrelated if and only if they are equal .
for arbitrary multisets of intervals , both @xmath377 and @xmath380 may hold ; in contrast , if @xmath375 is the maximal presentation of a lattice path matroid , then @xmath381 is a weak order .
if @xmath381 is a weak order , then we assume that the set system @xmath375 is indexed so that we can have @xmath377 only for @xmath382 . in this case , let @xmath383 be @xmath384 and let @xmath385 be @xmath386 .
[ thm : charint ] a set system @xmath375 of nonempty intervals in @xmath36 $ ] is the maximal presentation of a rank-@xmath6 lattice path matroid on @xmath36 $ ] that has no isthmuses if and only if * the relation @xmath381 is a weak order , * for all pairs @xmath387 and @xmath388 , neither @xmath389 nor @xmath390 is @xmath98 , and * @xmath391 for every @xmath31 . for the maximal presentation of a lattice path matroid
@xmath27 $ ] with no isthmuses , properties ( i)(iii ) follow from theorem [ thm : max ] . for the converse ,
note that removing from @xmath387 its least @xmath392 elements and its greatest @xmath393 yields the standard presentation of a lattice path matroid that , by property ( iii ) , has no isthmuses and for which @xmath375 is , by theorem [ thm : max ] , the maximal presentation .
when treating algorithmic questions about matroids , it is usual to assume that a matroid is given by an _ independence oracle _ , that is , a subroutine that outputs , in constant time , whether a subset of the ground set is independent . while there are algorithms that recognize transversal matroids within the class of all matroids ( see @xcite ) , jensen and korte @xcite have shown that there is no polynomial - time algorithm to decide if a matroid is transversal from an independence oracle .
the same proof as in @xcite shows that there is no such algorithm to decide whether a matroid is a lattice path matroid .
transversal matroids are more conveniently specified by set systems than by independence oracles .
this section gives a polynomial - time algorithm that , given a set system , decides whether the corresponding transversal matroid is a lattice path matroid .
we start with some simplifications .
a presentation @xmath10 of @xmath1 can be represented by a bipartite graph @xmath394 $ ] in the obvious way ( * ? ? ?
* section 1.6 ) .
therefore , by lemma [ lem : poly ] , the isthmuses of a transversal matroid can be identified and deleted in polynomial time .
if @xmath1 has no isthmuses , then the connected components of @xmath1 come from those of @xmath394 $ ] .
these observations and theorem [ thm : minors ] justify focusing on connected transversal matroids . as noted in section [ ssec : pres ] , the maximal presentation can be found from any presentation in polynomial time , so we focus on maximal presentations .
the key to the recognition algorithm below is to efficiently recover lattice path orderings from the maximal presentation .
we begin with some observations that relate these notions .
assume @xmath352 is the maximal presentation of the connected lattice path matroid @xmath27 $ ] on the ground set @xmath36 $ ] and let @xmath144 be the incidence function of @xmath10 .
let @xmath395 be the equivalence classes of the relation on @xmath36 $ ] in which @xmath56 and @xmath195 are related if and only if @xmath396 .
each set @xmath397 is an interval in @xmath36 $ ] .
we may assume that @xmath395 are indexed so that @xmath398 for any elements @xmath399 with @xmath400 in @xmath397 .
since @xmath27 $ ] is connected , we have @xmath401 for @xmath31 with @xmath402 . any permutation @xmath403 of @xmath36 $ ] with @xmath404 for @xmath405 is clearly an automorphism of @xmath27 $ ] , so the linear order @xmath406 is a lattice path order , as is @xmath407 .
relative to any of these linear orders , the sets in @xmath10 are intervals and the properties in theorem [ thm : charint ] hold .
these lattice path orderings of @xmath36 $ ] are essentially equivalent to the orderings @xmath408 and @xmath409 of @xmath395 .
observe that @xmath395 and @xmath410 are the only permutations @xmath411 of @xmath412 that satisfy the following property .
_ _ _ _ _ _ * ( p ) * _ _ _ _ _ _ _ thus , to determine whether a transversal matroid @xmath1 with a given presentation is a lattice path matroid , carry out the following steps .
* detect and delete the isthmuses .
* determine the connected components .
* find the maximal presentation for each connected component .
* for each component , find the classes defined above relative to the maximal presentation .
* for each component , determine whether there is a linear order of these classes that satisfies property ( p ) . *
if there is such a linear order of these classes for each component , then use the criterion in theorem [ thm : charint ] to determine whether , with respect to any corresponding linear order of a component , the intervals in the maximal presentation of that component are those of a maximal presentation of a lattice path matroid .
if , in step ( 5 ) , there is no suitable order for some connected component , then @xmath1 is not a lattice path matroid .
if there is such an order for each connected component , then @xmath1 is a lattice path matroid if and only if step ( 6 ) yields only positive results .
each of these steps can be done in polynomial time in the size of the ground set , so we get the following result .
whether a transversal matroid is a lattice path matroid can be determined from any presentation in polynomial time in the size of the ground set .
brylawski @xcite ( see also ( * ? ? ?
* proposition 12.2.26 ) ) gave a geometric description of arbitrary transversal matroids .
this section applies his result to lattice path matroids .
let @xmath1 be a transversal matroid on the set @xmath413 with presentation @xmath51 .
brylawski showed that @xmath1 can be realized geometrically as follows .
start with the free matroid @xmath414 on a set @xmath415 disjoint from @xmath9 .
for @xmath31 from @xmath98 to @xmath26 , form @xmath416 from @xmath417 by taking the principal extension of @xmath417 defined by the flat @xmath418 , with the element added being @xmath400 .
the matroid @xmath1 is @xmath419 .
thus , a rank-@xmath6 matroid is transversal if and only if it can be realized by placing the elements freely on the faces of the @xmath6-simplex .
the next theorem , which is illustrated in figure [ draw ] , shows how lattice path matroids can be constructed by successively adding isthmuses and loops , and by taking principal extensions by certain flats . to motivate this result ,
consider a lattice path matroid @xmath27 $ ] that has rank @xmath6 and nullity @xmath38 in which @xmath241 is neither a loop nor an isthmus .
let @xmath420 be the length of the longest final segment of north steps in @xmath21 . by theorem [ thm : max ] , the sets of the maximal presentation of @xmath27 $ ] that contain @xmath241 are the last @xmath420 ( those arising from @xmath421 ) .
by brylawski s result , @xmath241 is added freely to the flat spanned by @xmath422 in the notation above ; note that this flat is also spanned by the last @xmath420 elements of @xmath423 $ ] , since they are independent in @xmath27{\backslash}(m+r)$ ] .
thus , we have the following result .
[ thm : extend ] a matroid @xmath1 is a lattice path matroid if and only if the ground set can be written as @xmath413 so that each restriction @xmath424 is formed from @xmath417 by either * adding @xmath400 as an isthmus , * adding @xmath400 as a loop , or * adding @xmath400 via the principal extension of @xmath417 generated by the closure of an independent set of the form @xmath425 for some @xmath23 with @xmath426 .
we have seen that the class of lattice path matroids is closed under taking both minors and duals . while @xcite develops a dual - closed , minor - closed class of transversal matroids that properly contains @xmath46 , and while there are infinitely many dual - closed , minor - closed classes contained in @xmath46 ( see sections [ sec : gcat ] and [ sec : notch ] for two such classes ) , few other known classes of transversal matroids are either dual - closed or minor - closed . in this section
, we make some remarks about two important classes of transversal matroids , each of which has one of these properties .
fundamental transversal matroids ( called principal transversal matroids in @xcite ) were introduced by bondy and welsh @xcite and they play an important role in the study of transversal matroids . a transversal matroid @xmath1 is a _ fundamental transversal matroid _ if it can be represented on the simplex with an element of @xmath1 at each vertex of the simplex .
thus , transversal matroids are the restrictions of fundamental transversal matroids . while the class @xmath3 of fundamental transversal matroids is closed under neither deletion nor contraction , it is well - known and not hard to prove that @xmath3 is dual - closed .
the class @xmath3 is much larger than @xmath46 : brylawski @xcite showed that there are on the order of @xmath427 simple fundamental transversal matroids on @xmath144 elements , for some constant @xmath428 ; in contrast , @xmath429 is an upper bound on the number of lattice path matroids on @xmath144 elements since there are @xmath429 pairs of paths of length @xmath144 ( see @xcite for a formula for the number of connected lattice path matroids ) . both @xmath3 and @xmath46
contain all transversal matroids of rank two .
however , a fundamental transversal matroid of rank three or more can not have a pair of disjoint connected hyperplanes , but such hyperplanes can occur in lattice path matroids , such as the matroid @xmath430 of theorem [ thm : oxprow ] . on the other hand
, the number of connected hyperplanes of a fundamental transversal matroid , such as the @xmath144-whirl @xmath431 , can exceed two ( see corollary [ cor : finterval ] ) .
let us call a matroid _ bitransversal _ if both the matroid and its dual are transversal .
it is easy to prove that the class of bitransversal matroids is closed under direct sums , free extensions , and free coextensions .
hence by starting with the union of the classes @xmath46 and @xmath3 , and using these three operations , we can construct a larger class of bitransversal matroids ; let @xmath432 denote this class .
for instance , the free extension @xmath433 of @xmath434 is in @xmath432 but not in @xmath435 .
there are bitransversal matroids , such as the identically self - dual matroids of ( * ? ? ?
* section 4 ) , that are not in @xmath432 .
the problem of characterizing all bitransversal matroids , which was posed by welsh , currently remains open ( see ( * ? ? ?
* problem 14.7.4 ) ) .
bicircular matroids @xcite form another important class of transversal matroids .
the notion of a bicircular matroid we consider is a mild extension of that in @xcite ( as originally defined , bicircular matroids have no loops ) .
a transversal matroid @xmath1 is _ bicircular _ if it has a presentation @xmath10 so that each element of @xmath1 is in at most two sets in @xmath10 ( counting multiplicity ) .
thus , bicircular matroids are the transversal matroids that have a representation on the simplex in which all nonloops are on vertices or lines of the simplex .
it follows that minors of bicircular matroids are bicircular . on the other hand ,
the class of bicircular matroids is not dual - closed : the prism ( the matroid @xmath351 of figure [ b22 ] ) is bicircular , but its dual ( the matroid @xmath436 in the same figure ) is not transversal . among the matroids that are both bicircular and lattice path matroids are all transversal matroids of rank two as well as iterated parallel connections of rank-@xmath97 uniform matroids , @xmath437 and
@xmath438 , where the basepoint used to construct @xmath416 is not in @xmath439 . a bicircular matroid , unlike a lattice path matroid
, can have more than two connected hyperplanes .
also , while most uniform matroids are not bicircular ( for instance , @xmath440 is bicircular if and only if @xmath441 ) , all uniform matroids are in @xmath46 .
thus , the class of bicircular matroids differs significantly from @xmath46 in all ranks greater than two .
in this section , we show how to find the connectivity @xmath442 of a lattice path matroid in a simple way from the path presentation of @xmath1 .
we also show that at least one exact @xmath442-separation of @xmath1 is given by a fundamental flat and its complement .
we start by recalling the relevant definitions ; for more information on higher connectivity , see ( * ? ? ?
* chapter 8) . for a positive integer @xmath26 , a _
@xmath26-separation _ of a matroid @xmath1 is a partition of the ground set into two sets @xmath0 and @xmath117 , each with at least @xmath26 elements , such that the inequality @xmath443 holds .
a @xmath26-separation for which the equality @xmath444 holds is an _
exact @xmath26-separation_. the _ connectivity _ , or _ tutte connectivity _
, @xmath442 of @xmath1 is the least positive integer @xmath26 such that @xmath1 has a @xmath26-separation ; if there is no such @xmath26 , then @xmath442 is taken to be @xmath445 .
the connectivity of uniform matroids is well known ( see ( * ? ? ?
* corollary 8.1.8 ) ) , so we consider only lattice path matroids that are not uniform . also , as justified by theorem [ thm : conncrit ] , we focus exclusively on lattice path matroids that are connected .
let @xmath1 be a connected lattice path matroid , say @xmath27 $ ] , that is not uniform .
let the integer @xmath446 be defined as follows : @xmath447 figure [ lambda ] ( a ) illustrates a lattice path matroid @xmath1 in which the relevant values of @xmath75 are @xmath448 ( for which @xmath449 is @xmath99 ) and @xmath450 ( for which @xmath449 is @xmath39 ) , so @xmath446 is @xmath99 .
the main result of this section , theorem [ thm : conthm ] , is that the connectivity @xmath442 of @xmath1 is @xmath446 .
several lemmas enter into the proof of this result .
the first lemma reflects the equality @xmath451 that holds for any matroid .
4.25truein [ lem : lambdadual ] the number @xmath446 is invariant under duality , that is , @xmath452 .
recall that the lattice path diagram for the dual of @xmath27 $ ] is obtained by reflecting the lattice path diagram for @xmath27 $ ] about the line @xmath100 ( figure [ dual ] ) .
equivalently , the dual of @xmath27 $ ] is @xmath453 $ ] where @xmath454 and @xmath455 are obtained from @xmath21 and @xmath28 by switching east and north steps .
let @xmath144 and @xmath456 be the incidence functions of the standard presentations of @xmath27 $ ] and @xmath453 $ ] , respectively . note that @xmath21 has a @xmath22 corner at @xmath75 if and
only if @xmath454 has an @xmath25 corner at @xmath75 ; also , @xmath28 has an @xmath25 corner at @xmath457 if and only if @xmath455 has a @xmath22 corner at at @xmath457 .
thus , the lemma follows once we show the following statements : if @xmath28 has an @xmath25 corner at @xmath457 , then @xmath458 ; if @xmath21 has a @xmath22 corner at @xmath75 , then @xmath459 .
these assertions hold since we can pair off the relevant east and north steps that share a lattice point , as suggested in figure [ lambda ] ( b ) .
recall that in a matroid of connectivity at least @xmath144 with at least @xmath460 elements , circuits and cocircuits have at least @xmath144 elements ( * ? ? ?
* proposition 8.1.6 ) .
the next lemma will be used to show that circuits and cocircuits of a lattice path matroid @xmath1 have at least @xmath446 elements .
[ lem : nbd ] every element of @xmath1 is in at least @xmath461 sets in the maximal presentation of @xmath1 .
let @xmath1 be @xmath27 $ ] and let @xmath144 and @xmath456 be the incidence functions of its standard and maximal presentations .
let steps @xmath462 , @xmath463 , and @xmath464 be , respectively , the first east step of @xmath28 , the last north step of @xmath28 , and the first north step of @xmath21 . by the symmetry given by the order - reversing isomorphism of @xmath27 $ ] onto
@xmath270 $ ] ( see theorem [ thm : shape2 ] ) , it suffices to prove ( a ) if @xmath465 , then @xmath466 ( b ) if @xmath467 , then @xmath468 and ( c ) if @xmath469 , then @xmath470 .
theorem [ thm : max ] and the observation that @xmath471 is at least @xmath461 prove part ( a ) .
part ( c ) is trivial .
the proof of part ( b ) uses the following easily - verified statements .
* if the @xmath75-th and @xmath472-st steps of @xmath28 are east , then @xmath473 is either @xmath474 or @xmath475 , so we have @xmath476 . * if the @xmath75-th step of @xmath28 is north , then @xmath477 is either @xmath0 or @xmath478 where @xmath0 is @xmath479 , so @xmath480 .
first assume that steps @xmath481 of @xmath28 are east and that step @xmath24 is north .
thus , @xmath28 has an @xmath25 corner at @xmath23 . statements ( i ) and
( ii ) give the inequalities @xmath482 since @xmath483 , we have @xmath468 .
finally , if the @xmath31-th step of @xmath28 is north , a similar application of statement ( ii ) completes the proof of part ( b ) . from lemmas [ lem : circform ] and [ lem : nbd ] ,
the rank of any circuit of @xmath1 is at least @xmath461 .
the next lemma follows from this observation and lemma [ lem : lambdadual ] .
the generalized catalan matroid @xmath484 $ ] shows that @xmath1 can have circuits of rank @xmath461 .
[ lem : indlem ] any set of @xmath461 element of @xmath36 $ ] is independent in both @xmath1 and @xmath267 .
circuits of @xmath1 have at least @xmath446 elements , as do circuits of @xmath267 .
we now prove that @xmath446 is the connectivity of the lattice path matroid @xmath1 .
[ thm : conthm ] let @xmath1 be a connected lattice path matroid of rank @xmath6 and nullity @xmath38 , say @xmath27 $ ] , that is not uniform .
the connectivity @xmath442 of @xmath1 is @xmath446 , where @xmath446 is @xmath485 furthermore , at least one exact @xmath446-separation of @xmath1 consists of some fundamental flat and its complement .
we first show that @xmath1 has an exact @xmath446-separation that consists of a fundamental flat and its complement .
assume first that @xmath446 is @xmath449 where @xmath21 has a @xmath22 corner at @xmath75 .
let @xmath0 and @xmath117 be @xmath486 $ ] and @xmath118 $ ] , respectively .
thus , @xmath117 is a fundamental flat of @xmath1 .
note that both @xmath0 and @xmath117 have at least @xmath449 elements .
it follows from the path presentations of restrictions given in corollary [ cor : restint ] that @xmath172 is @xmath487 , that is , @xmath488 , so @xmath262 is an exact @xmath446-separation of @xmath1 .
similarly , if @xmath446 is @xmath449 where @xmath28 has an @xmath25 corner at @xmath457 , then @xmath489 $ ] and @xmath490 $ ] give an exact @xmath446-separation of @xmath1 . it remains to show that @xmath1 has no @xmath23-separation for any positive integer @xmath23 less than @xmath446 .
let @xmath23 be such an integer and assume @xmath0 and @xmath117 partition @xmath36 $ ] , where both @xmath0 and @xmath117 have at least @xmath23 elements .
we need to prove the inequality @xmath491 if an element @xmath195 in @xmath0 is in the closure of @xmath117 , and if @xmath0 has more than @xmath23 elements , then we have @xmath492 , @xmath493 , and @xmath494 .
thus , it suffices to prove inequality ( [ eqn : conngoal ] ) when @xmath495 is @xmath23 or @xmath117 is a nontrivial flat of @xmath1 . by lemma [ lem : indlem ]
, each nontrivial connected component of the restriction @xmath121 to a flat @xmath117 of @xmath1 has more than @xmath23 elements ; with an argument similar to the one above , it follows that if @xmath117 is a nontrivial flat of @xmath1 , then we may assume @xmath117 is connected .
assume @xmath495 is @xmath23 . by lemma [ lem : indlem ] , @xmath0 is an independent set that does not contain a cocircuit , so @xmath117 spans @xmath1 .
thus , @xmath496 is @xmath497 .
now assume @xmath117 is a nontrivial connected flat of @xmath1 .
if @xmath117 is a fundamental flat , then inequality ( [ eqn : conngoal ] ) follows as in the first paragraph .
if @xmath117 is not a fundamental flat , then , by theorem [ thm : cflats ] , @xmath117 is the intersection of two incomparable fundamental flats , say @xmath498 and @xmath499 where @xmath500 and @xmath360 partition @xmath0 .
we may assume @xmath98 is in @xmath500 , so @xmath241 is in @xmath360 .
since @xmath501 is a fundamental flat and @xmath360 is the complement of @xmath501 , we have @xmath502 .
thus , since @xmath446 exceeds @xmath23 , to prove inequality ( [ eqn : conngoal ] ) , it suffices to prove @xmath503 , that is , @xmath504 observe that @xmath505 is @xmath506 ; the inequality @xmath507 is obvious and the inequality @xmath508 follows by matching each set @xmath33 , for @xmath31 in @xmath509 , with its first element , which must be in @xmath500 , and each set in @xmath110 , for @xmath75 in @xmath510 , with its last element , which must be in @xmath360 .
a similar argument gives the equality @xmath511 . from theorem [ thm :
finterval ] , we also have @xmath512 and @xmath513 .
thus , inequality ( [ eqn : conngoala ] ) is equivalent to @xmath514 note that @xmath506 is @xmath515 . substituting this and the analogous formula for @xmath516 into inequality ( [ eqn : conngoal1 ] ) and simplifying gives that this inequality is equivalent to the inequality @xmath517 , which clearly holds .
thus , inequality ( [ eqn : conngoal ] ) holds , as needed to complete the proof . as the matroid @xmath518 of figure [ small ] shows ,
not every exact @xmath446-separation of a lattice path matroid @xmath1 has a fundamental flat as one of the sets .
there are infinitely many minor - closed , dual - closed classes of transversal matroids within the class of lattice path matroids .
one way to define such classes is to impose certain requirements on the bounding paths ; for example , the lower bounding path of a generalized catalan matroid must have the form @xmath519 . in this section
we introduce the minor - closed , dual - closed class of notch matroids , which is defined by special forms for the bottom bounding path .
we relate notch matroids to generalized catalan matroids via circuit - hyperplane relaxations .
the main result is the characterization of notch matroids by excluded minors .
we include some remarks on the excluded minors for lattice path matroids .
as figure [ notshp ] illustrates , notch matroids are either in @xmath210 or their lattice path presentations differ from those of generalized catalan matroids by the `` notch '' in the lower right corner .
it follows from the lattice path descriptions of minors and duals , along with theorem [ thm : shape2 ] , that the class @xmath521 of notch matroids is minor - closed and dual - closed .
note that @xmath521 , like its subclass @xmath210 , is not closed under direct sums .
in contrast to @xmath210 , the class @xmath521 is not closed under any of the following operations , as can be seen from the matroid @xmath522 of figure [ small ] : free extension , truncation , and the dual operations .
the first lemma gives a basic property that @xmath521 shares with @xmath210 .
note that a connected notch matroid either is in @xmath210 or has a circuit - hyperplane relaxation in @xmath210 .
not every matroid that has a circuit - hyperplane relaxation in @xmath210 is a notch matroid ; for instance , the matroids @xmath523 and @xmath524 of figure [ a3 ] each have two circuit - hyperplane relaxations that are in @xmath210 , yet neither is a lattice path matroid since condition ( ii ) of theorem [ thm : lpmchar ] fails . however , we have the following result .
[ thm : chrelax ] a connected matroid in @xmath525 is a notch matroid if and only if it has a circuit - hyperplane . relaxing any circuit - hyperplane of a lattice path matroid yields a generalized catalan matroid .
the last @xmath6 elements of a connected notch matroid @xmath520 $ ] obviously form a circuit - hyperplane .
for the converse , assume that @xmath180 is a circuit - hyperplane of the rank-@xmath6 , nullity-@xmath38 matroid @xmath101 $ ] .
since @xmath180 is an @xmath6-circuit of @xmath1 , by theorem [ thm : finterval ] the set @xmath526 is an interval of @xmath7 elements in @xmath58 $ ] ; we may assume that @xmath526 is @xmath527 $ ] . since @xmath180 is a flat
, @xmath180 is an interval of @xmath6 elements in the ground set @xmath36 $ ] of @xmath1 , so @xmath36 $ ] consists of an initial interval , the interval @xmath180 , and a final interval @xmath117 .
since @xmath180 is a hyperplane , @xmath117 must be empty , so @xmath180 consists of the last @xmath6 elements of @xmath36 $ ] . from these conclusions , it is immediate that @xmath1 is a notch matroid .
the last assertion follows from part ( iii ) of corollary [ cor : finterval ] .
[ lem : relax ] let @xmath528 be @xmath209 $ ] , a connected rank-@xmath6 , nullity-@xmath38 matroid in @xmath210 .
if the basis @xmath360 of @xmath528 is mapped onto the final segment @xmath529 $ ] by some automorphism of @xmath528 , then there is a unique matroid @xmath1 in which @xmath360 is a circuit - hyperplane and from which @xmath528 is obtained by relaxing @xmath360 .
furthermore , @xmath1 is in @xmath521 .
the incomparable flats @xmath0 and @xmath117 show that @xmath1 is not in @xmath210 , so @xmath1 has a circuit - hyperplane , say @xmath180 .
either @xmath0 or @xmath117 must be @xmath180 since @xmath180 can not properly contain either @xmath0 or @xmath117 and , by part ( iii ) of corollary [ cor : finterval ] , nontrivial connected flats that are not contained in @xmath180 are comparable .
we may assume that @xmath1 has no isthmuses and that @xmath0 and @xmath117 are incomparable .
from lemma [ lem : incomplemma ] , either @xmath0 or @xmath117 , say @xmath0 , is a circuit - hyperplane of @xmath1 .
part ( iii ) of corollary [ cor : finterval ] implies that @xmath117 and @xmath186 are comparable .
we turn to the excluded - minor characterization of @xmath521 .
let @xmath530 and @xmath531 denote the sets of excluded minors for @xmath521 and @xmath46 , respectively .
we first discuss the matroids in @xmath530 that are not lattice path matroids and so are in @xmath532 . in each case , we show that the matroids are not in @xmath46 ; it is easy to check that all their proper minors are in @xmath521 , so we omit this part . among the self - dual matroids in @xmath532
are the 3-wheel @xmath533 and the 3-whirl @xmath534 , which are shown in figure [ ww ] .
since all @xmath99-point lines of @xmath533 and @xmath534 are fundamental flats , condition ( i ) of theorem [ thm : lpmchar ] fails , so @xmath533 and @xmath534 are not in @xmath46 . for @xmath535 ,
let @xmath536 be the rank-@xmath144 paving matroid with only two nontrivial hyperplanes , @xmath537 and @xmath538 , and with only one point , @xmath195 , in neither circuit - hyperplane ( figure [ a3 ] ) .
the two circuit - hyperplanes violate condition ( ii ) of theorem [ thm : lpmchar ] , so @xmath536 is not in @xmath46 .
note that @xmath536 is self - dual .
we next consider two doubly - indexed families in @xmath539 that are related by duality ; three of these matroids are shown in figure [ b22 ] .
let @xmath144 and @xmath26 be integers with @xmath540 .
let @xmath541 be the truncation @xmath542 to rank @xmath144 of the direct sum of two @xmath144-circuits and a @xmath26-circuit .
the three disjoint circuits are fundamental flats of @xmath541 , so condition ( i ) of theorem [ thm : lpmchar ] shows that @xmath541 is not in @xmath46 .
the dual @xmath543 of @xmath541 is the rank-@xmath544 paving matroid @xmath543 for which the ground set can be partitioned into sets @xmath545 with @xmath546 and @xmath547 so that the only nontrivial hyperplanes are @xmath548 , @xmath549 , and @xmath550 .
the remaining matroids in @xmath532 , two of which are shown in figure [ d4 ] , form two infinite families that are related by duality .
recall that @xmath551 denotes the free extension of @xmath1 by the point @xmath195 . for @xmath535 , let @xmath552 be the rank-@xmath144 matroid @xmath553 that @xmath552 is not in @xmath46 for @xmath554 follows since the two @xmath555-circuits , as well as their union , are fundamental flats of @xmath552 , contrary to condition ( i ) of theorem [ thm : lpmchar ] . in the dual @xmath556 of @xmath552 ,
the element @xmath195 is parallel to an element @xmath56 , and the deletion @xmath557 is a rank-@xmath144 paving matroid whose only nontrivial hyperplanes are two circuit - hyperplanes that intersect in @xmath56 .
( the matroids @xmath522 and @xmath518 , which are shown in figure [ small ] , are lattice path matroids . ) we have proven the easy part of the following theorem ; the more substantial part of this result follows from the excluded - minor characterization of notch matroids , which is given in theorem [ thm : notchchar ] .
we now turn to the excluded - minor characterization of notch matroids .
the excluded minors are those in theorem [ thm : exlexn ] together with the three types of lattice path matroids illustrated in figure [ large ] and the four matroids in figure [ small ] .
* @xmath560 and @xmath561 , * the three - wheel , @xmath533 , and the three - whirl , @xmath534 , * @xmath536 for @xmath535 , * @xmath541 and @xmath543 for @xmath144 and @xmath26 with @xmath559 , * @xmath552 for @xmath535 , * @xmath556 for @xmath535 , * for @xmath554 , the rank-@xmath144 matroid @xmath562 , * for @xmath227 , the rank-@xmath144 matroid @xmath563 , and * for @xmath535 , the rank-@xmath144 matroid @xmath564 . to make the proof of theorem [ thm : notchchar ] less verbose , we will use abbreviations such as the following : from theorem [ thm : notlpm ] applied to @xmath1 , @xmath565 , @xmath566 , and @xmath195 , we get @xmath567 . by this
we mean that the matroid @xmath1 and the flats @xmath565 and @xmath566 satisfy the hypotheses of theorem [ thm : notlpm ] , with the point @xmath195 showing the validity of the third condition .
the remarks before theorem [ thm : exlexn ] show that of the matroids in the theorem , only @xmath522 , @xmath518 , and those in items ( @xmath98 ) and ( @xmath40)(@xmath568 ) are in @xmath46 .
the presentations of these matroids , illustrated in figures [ large ] and [ small ] , make it clear that they are not in @xmath521 .
it is easy to check that all proper minors of these matroids are in @xmath521 .
note that @xmath569 is self - dual , and that @xmath570 and @xmath571 are dual to each other .
the proof that theorem [ thm : notchchar ] gives all excluded minors is intricate , so we first outline the argument .
part ( [ thm : notchchar].1 ) proves that the disconnected excluded minors are @xmath560 , @xmath572 , @xmath573 , @xmath574 , and @xmath575 .
the rest of the proof revolves around three properties a connected excluded minor @xmath1 may have : in ( [ thm : notchchar].2 ) , we show that if @xmath1 has property ( a ) , then @xmath1 is @xmath552 for some @xmath535 .
part ( [ thm : notchchar].3 ) gives a key property of all connected excluded minors . in ( [ thm : notchchar].4 ) , we show that if property ( b ) but not ( a ) holds , then @xmath1 is one of the matroids in items ( @xmath97 ) and ( @xmath39 ) .
part ( [ thm : notchchar].5 ) shows that if only property ( c ) holds , then @xmath1 is one of the matroids in items ( @xmath578)(@xmath568 ) . if none of the properties holds , then for any mutually incomparable connected flats @xmath411 , we have @xmath579 , and if @xmath26 is @xmath97 , then at least one of @xmath565 or @xmath566 is a circuit - hyperplane . since restrictions to proper subsets of circuit - hyperplanes
are free , it follows that relaxing a circuit - hyperplane of such an excluded minor yields a matroid @xmath528 in which the connected flats are linearly ordered by inclusion , that is , @xmath528 is in @xmath210 .
the proof of theorem [ thm : notchchar ] is completed in ( [ thm : notchchar].6 ) by showing that the only rank-@xmath144 excluded minor that has a circuit - hyperplane relaxation in @xmath210 is @xmath536 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].1 ) * if @xmath1 is disconnected , then @xmath1 is one of @xmath560 , @xmath572 , @xmath573 , @xmath574 , and @xmath575 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ now assume @xmath1 has exactly two components , @xmath581 and @xmath582 . being proper minors of @xmath1 , both @xmath581 and @xmath582 are notch matroids .
observe that if @xmath583 , then , by theorem [ thm : spanning ] and corollary [ cor : cinterval ] , there is an element @xmath56 for which @xmath584 is connected .
dually , if @xmath585 , then @xmath586 is connected for some @xmath195 .
assume @xmath581 is @xmath587 . from lattice path presentations and from the statements
@xmath588 and @xmath589 , it follows that @xmath590 and @xmath591 are both at least @xmath97 . similarly , if @xmath83 is a connected minor of @xmath582 for which @xmath592 and @xmath593 are both @xmath97 , then @xmath594 .
these observations , together with those in the last paragraph , imply that @xmath590 and @xmath591 are both @xmath97 . from lattice path presentations , we see that only two connected lattice path matroids have rank and nullity @xmath97 , namely @xmath595 and @xmath596 , so @xmath1 is either @xmath575 or @xmath572 . now assume @xmath597 with @xmath598 .
since @xmath599 , the nullity of @xmath582 is at least @xmath97 .
arguments like those in the last paragraph imply that @xmath26 is @xmath99 , that @xmath591 is @xmath97 , and that @xmath590 is @xmath98 ; therefore @xmath582 is @xmath600 , so @xmath1 is @xmath574 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].2 ) * if @xmath1 has nontrivial incomparable connected flats @xmath565 and @xmath566 with @xmath601 ,
then @xmath1 is @xmath552 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ choose such a pair of flats @xmath577 so that @xmath602 is as small as possible .
lemma [ lem : incomplemma ] applied to @xmath603 , @xmath565 , and @xmath566 implies that either @xmath565 or @xmath566 is a circuit - hyperplane of @xmath603 .
assume @xmath603 is disconnected .
this disconnected notch matroid has neither loops nor isthmuses , so one component , say @xmath565 , has rank @xmath98 and the other , @xmath566 , has nullity @xmath98 ; thus , @xmath565 is a parallel class and @xmath566 is a circuit .
if @xmath604 and @xmath605 , then @xmath606 , @xmath607 , and @xmath566 contradict lemma [ lem : incomplemma ] . if @xmath608 and @xmath609 , then @xmath610 , @xmath611 , and @xmath612 contradict lemma [ lem : incomplemma ]
thus , @xmath613 .
since @xmath1 has neither @xmath436 nor @xmath560 as a proper minor , @xmath565 and @xmath566 are the only nontrivial parallel classes of @xmath1 .
let @xmath56 and @xmath195 be in @xmath614 . by lemma [ lem : incomplemma ] , the rank-@xmath98 flats @xmath615 and @xmath616
are hyperplanes of @xmath86 , so @xmath617 is @xmath99 .
it follows that @xmath618 , and so @xmath1 , is one of the excluded minors @xmath619 or @xmath522 ; since @xmath1 is connected , @xmath1 is @xmath522 .
* @xmath1 is simple , * @xmath565 and @xmath566 are disjoint circuits , and @xmath620 is a flat of @xmath1 , * @xmath621 contains only two elements , say @xmath56 and @xmath195 , * the only nonspanning circuits of @xmath622 are @xmath565 and @xmath566 , * @xmath623 , so both @xmath565 and @xmath566 are circuit - hyperplanes of @xmath622 , and * the only circuits of @xmath1 that contain @xmath56 and @xmath195 are spanning circuits . to prove statement ( i ) , note that since @xmath603 is connected , and since @xmath565 and @xmath566 are incomparable flats , neither @xmath565 nor @xmath566 is a parallel class .
if elements @xmath56 and @xmath195 of @xmath1 were parallel , then @xmath606 , @xmath607 , and @xmath624 ( which may be @xmath565 and @xmath566 ) would contradict lemma [ lem : incomplemma ] . for statement ( ii )
, we first show that both @xmath625 and @xmath626 are connected for any @xmath56 in @xmath627 .
if , say , @xmath625 were disconnected , then by lemma [ lem : smallerflats ] , there would be nontrivial incomparable connected flats @xmath500 and @xmath360 of @xmath628 with @xmath629 .
since @xmath1 is simple , @xmath630 exceeds @xmath98 , so the flats @xmath500 and @xmath360 of @xmath1 would contradict the choice of @xmath565 and @xmath566 as minimizing the sum @xmath631 . since @xmath625 and @xmath626 are connected , @xmath86 , @xmath632 , and @xmath633 contradict lemma [ lem : incomplemma ] .
thus , @xmath565 and @xmath566 are disjoint .
the connected notch matroids @xmath628 and @xmath634 have spanning circuits ; this observation and the minimality of @xmath1 show that @xmath565 and @xmath566 are circuits . for any @xmath56 in @xmath635 ,
the deletion @xmath91 is connected , so @xmath91 , @xmath565 , and @xmath566 would violate lemma [ lem : incomplemma ] . thus , @xmath636 is @xmath637 , so statement ( ii ) holds .
let @xmath195 be in @xmath621 .
the contraction @xmath638 has neither loops nor isthmuses . by lemma [ lem : incomplemma ] ,
at least one of @xmath639 and @xmath640 is a circuit - hyperplane of the notch matroid @xmath638 , so @xmath641 is @xmath642 . for @xmath606 , @xmath565 , and @xmath566 to not contradict lemma [ lem : incomplemma ] , @xmath606 must have an isthmus . from these conclusions , statement ( iii ) follows .
assume @xmath94 is a nonspanning circuit of @xmath622 other than @xmath565 and @xmath566 .
recall that either @xmath565 or @xmath566 , say @xmath565 , is a circuit - hyperplane of @xmath622 .
thus , @xmath565 and @xmath95 are incomparable and @xmath643 spans the flat @xmath637 .
let @xmath644 be in the difference @xmath645 of circuits .
note that @xmath646 is connected . that @xmath646 , @xmath565 , and @xmath647 contradict lemma [ lem : incomplemma ] proves statement ( iv ) .
statement ( v ) follows since if @xmath648 and @xmath644 is in @xmath565 , then @xmath610 , @xmath649 , and @xmath566 would contradict lemma [ lem : incomplemma ] . from statements ( i ) and ( v ) we have @xmath554 .
assume @xmath56 and @xmath195 are in a nonspanning circuit @xmath94 .
at least one of @xmath565 and @xmath566 is not contained in @xmath95 , so we may assume that @xmath565 and @xmath95 are incomparable .
let @xmath644 be in the difference @xmath645 of circuits .
note that @xmath650 is a connected hyperplane of @xmath646 since @xmath554 , so @xmath646 is connected .
lemma [ lem : incomplemma ] applied to @xmath646 , @xmath565 and @xmath651 implies that @xmath651 must be a circuit - hyperplane of @xmath646 , so @xmath95 is a hyperplane of @xmath1 .
note that @xmath95 is either @xmath651 or @xmath652 , that is , either @xmath94 or @xmath653 , so @xmath654 .
thus , if @xmath655 , then @xmath95 is @xmath656 .
however , if @xmath95 is @xmath656 and @xmath657 is in @xmath565 , then @xmath658 , @xmath659 , @xmath656 contradict lemma [ lem : incomplemma ] . therefore @xmath566 and @xmath95 are incomparable . by switching @xmath565 and @xmath566 if necessary
, we may assume @xmath660 . since @xmath661 , we have @xmath662 ; however , there are at least two elements , say @xmath663 and @xmath664 , in @xmath665 , that is , in @xmath666 , so by theorem [ thm : notlpm ] , @xmath667 is not a lattice path matroid , contrary to the minimality of @xmath1 .
thus , statement ( vi ) holds , so @xmath1 is @xmath552 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].3 ) * if @xmath0 is a proper nontrivial connected flat of @xmath1 and the element @xmath56 of @xmath0 is not parallel to any element , then @xmath668 is a connected flat of @xmath86 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if @xmath668 were a disconnected flat of @xmath86 , then , by lemma [ lem : smallerflats ] applied to @xmath2 , we would have @xmath669 for some nontrivial incomparable connected flats @xmath670 of @xmath2 . since @xmath565 and @xmath566 would also be flats of @xmath1 , by ( [ thm : notchchar].2 )
, @xmath1 would be @xmath552 . that @xmath552 has no such flat @xmath0 and element @xmath56 provides the contradiction that proves the result .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].4 ) * if @xmath1 has three mutually incomparable connected flats @xmath671 , then @xmath1 is @xmath533 , @xmath534 , @xmath541 , or @xmath672 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the minimality of @xmath1 and lemma [ lem : no3incomp ] imply that the ground set of @xmath1 is @xmath673 and that any pair @xmath674 of parallel elements can be in only one of @xmath671 .
if an element @xmath56 were in @xmath675 , then by ( [ thm : notchchar].3 ) , @xmath86 , @xmath632 , @xmath633 , and @xmath676 would contradict lemma [ lem : no3incomp ] , so @xmath677 .
note that @xmath1 is not @xmath552 , so we have @xmath678 for @xmath679 .
first assume @xmath680 .
there are at least two points @xmath56 and @xmath195 in @xmath681 , so if @xmath682 were nonempty , then @xmath606 , @xmath565 , and @xmath683 would contradict theorem [ thm : notlpm ] .
thus , @xmath684 .
similarly @xmath685 .
the minimality of @xmath1 implies that @xmath565 , @xmath566 , and @xmath683 are circuits .
let @xmath686 be @xmath687 . since @xmath688 is @xmath144 , for any @xmath56 in @xmath689 the notch matroid @xmath91 has no isthmuses ; thus , from lemma [ lem : incomplemma ] , either @xmath690 or @xmath691 is a circuit - hyperplane of @xmath91 and so of @xmath1 .
it follows that at least two of @xmath671 , say @xmath565 and @xmath566 , are circuit - hyperplanes of @xmath1 .
let @xmath692 be @xmath26 .
note that @xmath1 is @xmath541 if @xmath565 , @xmath566 , and @xmath683 are the only nonspanning circuits of @xmath1 .
if @xmath94 were another nonspanning circuit , then for any @xmath644 in the difference @xmath693 of circuits , the flat @xmath651 would be contained in neither of the hyperplanes @xmath565 and @xmath566 of @xmath646 , contrary to part ( iv ) of corollary [ cor : finterval ] .
thus , @xmath1 is @xmath541 .
now assume @xmath694 for all sets @xmath695 . we claim that @xmath565 , @xmath566 , and @xmath683 are hyperplanes and the union of any two contains all but at most one point of @xmath1 .
let @xmath686 be @xmath687 and let @xmath56 be in @xmath696 .
the equality @xmath697 and theorem [ thm : notlpm ] give the inequality @xmath698 , so the second claim holds . to see that @xmath689 is a hyperplane , note that lemma [ lem : no3incomp ] applied to @xmath86 , @xmath699 , @xmath700 , and @xmath701 implies that there is a containment among at least two of these sets . of the two possible containments
, we may assume @xmath702 .
thus , @xmath703 .
this containment , the inequality @xmath704 , and that @xmath691 is connected imply that @xmath705 is @xmath673 , so @xmath689 is a hyperplane of @xmath1 .
if @xmath56 and @xmath195 are in @xmath696 , then @xmath56 is in the nontrivial connected hyperplanes @xmath706 and @xmath707 of the notch matroid @xmath638 , so , by theorem [ thm : notlpm ] , @xmath708 is @xmath709 .
thus , if @xmath710 , then @xmath711 .
assume @xmath712 is @xmath98 .
since @xmath565 is connected and at most one point of @xmath565 is in neither @xmath627 ( one point ) nor @xmath682 ( a flat ) , there is one point in @xmath713 .
similarly , there is one point in @xmath714 .
these conclusions , and that in the last paragraph , give the equality @xmath715 .
therefore @xmath565 , @xmath566 , and @xmath683 are @xmath99-point lines .
it follows easily that @xmath1 is either @xmath533 or @xmath534 .
assume @xmath710 for @xmath716 .
thus , @xmath717 . let @xmath56 and @xmath195 be in @xmath718 .
lemma [ lem : incomplemma ] applied to @xmath638 , @xmath719 , and @xmath720 implies that either @xmath719 or @xmath707 is a circuit - hyperplane of @xmath638 . since , in addition , @xmath690 and @xmath691 are connected hyperplanes of @xmath1 , either @xmath690 or @xmath691 is a circuit - hyperplane of @xmath1 .
it follows that at least two hyperplanes , say @xmath565 and @xmath566 , are circuit - hyperplanes of @xmath1 .
assume @xmath721 .
that @xmath565 and @xmath566 are circuit - hyperplanes of @xmath1 gives the equality @xmath722 . to prove that @xmath1 is @xmath672
, we need only show that the only proper nontrivial connected flat @xmath0 other than @xmath565 and @xmath566 is @xmath683 .
clearly @xmath0 is incomparable to the circuit - hyperplanes @xmath565 and @xmath566 .
as we deduced for @xmath671 , we get @xmath723 , so @xmath724 . since @xmath682 and @xmath725 are independent , both @xmath726 and @xmath727 are nonempty . with this
, the claim in the third paragraph shows that @xmath0 is a hyperplane . since @xmath724
, it follows that @xmath0 is @xmath683 , as needed . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].5 ) * if @xmath1 has no circuit - hyperplane and is not @xmath552 , then @xmath1 is one of @xmath556 , @xmath570 , @xmath728 , or @xmath569 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ since @xmath1 is not a generalized catalan matroid , there is a pair @xmath565 , @xmath566 of incomparable connected flats .
since @xmath1 is not @xmath552 , part ( [ thm : notchchar].2 ) gives the equality @xmath729 for any such pair of flats .
assume there were an element @xmath56 in @xmath621 .
since @xmath641 is @xmath144 , the deletion @xmath91 would have no isthmuses .
therefore either @xmath565 or @xmath566 would be a circuit - hyperplane of @xmath91 and so of @xmath1 .
since @xmath1 has no circuit - hyperplane , the equality @xmath730 follows . *
each element in @xmath627 is parallel to another element of @xmath1 , * @xmath627 contains just two elements , say @xmath56 and @xmath195 , and at least one of @xmath632 and @xmath633 , say @xmath632 , is a circuit - hyperplane of @xmath91 , * @xmath731 is a circuit , * @xmath732 , and * the nonspanning circuits of @xmath1 are @xmath632 , @xmath607 , @xmath633 , @xmath624 , and @xmath733
. assume statement ( i ) failed for some @xmath56 in @xmath627 . from ( [ thm : notchchar].3 ) and lemma [ lem : incomplemma
] , either @xmath632 or @xmath633 , say @xmath734 , would be a circuit - hyperplane of @xmath86 .
it follows that @xmath565 would be a circuit - hyperplane of @xmath1 .
this contradiction to the hypotheses of ( [ thm : notchchar].5 ) proves statement ( i ) .
it follows that for each @xmath735 , the deletion @xmath91 is a connected notch matroid , so by lemma [ lem : incomplemma ] , either @xmath632 or @xmath633 , say @xmath632 , is a circuit - hyperplane of @xmath91 .
since the circuit @xmath632 of @xmath91 can not contain parallel elements , statement ( ii ) follows . by ( [ thm : notchchar].3 )
the minor @xmath736 is connected , so by part ( b ) of corollary [ cor : spanningvar ] there is a spanning circuit @xmath737 of @xmath634 that contains @xmath195 .
lemma [ lem : incomplemma ] and the minimality of the excluded minor @xmath1 imply that @xmath566 is @xmath738 , so statement ( iii ) holds . for statement ( iv ) ,
note that if @xmath739 and @xmath740 , then @xmath610 , @xmath741 , and @xmath742 contradict lemma [ lem : incomplemma ] .
statement ( v ) follows from part ( iv ) of corollary [ cor : finterval ] since each of the notch matroids @xmath91 and @xmath606 has two circuit - hyperplanes .
now assume any two incomparable nontrivial connected flats are disjoint .
we showed that the union of any two such flats is @xmath9 .
let @xmath577 be such flats .
it follows that all nonspanning circuits of @xmath1 span either @xmath628 or @xmath634 , so @xmath1 is @xmath743 ; also , @xmath628 and @xmath634 are uniform matroids .
if @xmath565 is not a circuit and @xmath56 is in @xmath565 , then @xmath91 is a connected notch matroid in which @xmath566 is not a circuit - hyperplane , so @xmath734 is a circuit - hyperplane of @xmath91 ; it follows that @xmath628 is @xmath744 .
assume that @xmath565 is a circuit , and so not a hyperplane of @xmath1 ; let @xmath56 be in @xmath566 .
note that @xmath565 and @xmath633 are incomparable connected flats of the notch matroid @xmath86 , which has no isthmuses . since @xmath566 is not a circuit - hyperplane of @xmath1
, it follows that @xmath633 can not be a circuit - hyperplane of @xmath86 .
therefore by lemma [ lem : incomplemma ] , @xmath565 is a circuit - hyperplane of @xmath86 .
thus , @xmath628 is @xmath745 . in this manner
, we see that there are , up to switching @xmath565 and @xmath566 , three possibilities : @xmath628 and @xmath634 are both @xmath745 ; @xmath628 is @xmath745 and @xmath634 is @xmath744 ; both @xmath628 and @xmath634 are @xmath744 .
these possibilities give , respectively , @xmath570 , @xmath569 , and @xmath728 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * ( [ thm : notchchar].6 ) * if relaxing some circuit - hyperplane @xmath94 of @xmath1 gives a generalized catalan matroid @xmath528 , then @xmath1 is @xmath536 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * there is at least one element @xmath195 in @xmath748 . *
the ground set of @xmath1 is @xmath749 ; also @xmath750 . *
the circuit @xmath243 is a hyperplane of @xmath1 . *
the only nonspanning circuits of @xmath1 are @xmath94 and @xmath243 .
let the chain of proper nontrivial connected flats of @xmath528 be @xmath751 .
if @xmath752 were empty , then , by corollary [ cor : lpmauto ] , there would be an automorphism of @xmath528 that maps @xmath94 to a final segment ; by lemma [ lem : relax ] we would get the contradiction that @xmath1 is a notch matroid .
thus , @xmath753 is not empty , which gives statement ( i ) . among all circuits that intersect @xmath94 ,
choose @xmath243 with smallest cardinality .
the closure @xmath754 is one of the connected flats @xmath691 , and by the choice of @xmath243 , the basis @xmath94 of @xmath528 is disjoint from @xmath690 for @xmath382 . to prove statement ( ii )
we must show that @xmath94 does not contain the complement of @xmath691 ; if this were false , then by corollary [ cor : lpmauto ] and lemma [ lem : relax ] we would get , as before , that @xmath1 is a notch matroid . by theorem [ thm : notlpm ] , @xmath755 is not a lattice path matroid .
this observation and the minimality of @xmath1 prove the first part of statement ( iii ) .
the second part holds since if @xmath756 and @xmath757 , then , by theorem [ thm : notlpm ] , @xmath86 would not be a lattice path matroid .
let @xmath758 be @xmath56 . to prove statement ( iv ) , first note that @xmath759 is a uniform matroid since , by the choice of @xmath243 , any nonspanning circuit @xmath186 of @xmath759 would be disjoint from @xmath94 , which gives the contradiction that the circuit @xmath243 properly contains the circuit @xmath186 . since @xmath759 is a uniform matroid that consists of @xmath243 and a subset of @xmath94 , and since , by statement ( iii ) , any circuit @xmath760 with @xmath761 that intersects @xmath94 contains just one element of @xmath94
, it follows that @xmath762 is @xmath56 , so @xmath243 is closed . if @xmath243 is not a hyperplane of @xmath1 , then there is an element @xmath644 in @xmath763 , so @xmath195 is not in @xmath764 .
however , for such a @xmath644 , theorem [ thm : notlpm ] applied to @xmath610 , @xmath765 , @xmath766 , and @xmath195 shows that @xmath610 is not in @xmath46 , contrary to @xmath1 being an excluded minor for @xmath521 .
since @xmath243 is a circuit - hyperplane of @xmath1 and of the generalized catalan matroid @xmath528 , it follows that @xmath243 is the only nonspanning circuit of @xmath528 , so @xmath94 and @xmath243 are the only nonspanning circuits of @xmath1 , as needed to complete the proof .
figure [ otherex ] shows two excluded minors for @xmath46 that are not among those given in theorem [ thm : exlexn ] .
presently we do not know whether these two matroids complete the list of excluded minors for the class of lattice path matroids .
we close by noting that a lattice path matroid is graphic if and only if it is the cycle matroid of an outerplanar graph in which each inner face shares edges with at most two other inner faces .
one implication follows since @xmath533 and @xmath351 ( i.e. , the cycle matroids of the two excluded minors , @xmath767 and @xmath768 , for outerplanar graphs ) are excluded minors for lattice path matroids , as is @xmath436 , which is the cycle matroid of the graph formed by adding an edge parallel to each edge of @xmath769 .
the other implication follows since by adding edges any graph of the stated type can be extended to a graph of this type in which each face is bounded by at most three edges , and the cycle matroids of such graphs , which are certain parallel connections of @xmath99-point lines , are easily seen to be lattice path matroids . | this paper studies structural aspects of lattice path matroids , a class of transversal matroids that is closed under taking minors and duals . among the basic topics treated are direct sums , duals , minors , circuits , and connected flats .
one of the main results is a characterization of lattice path matroids in terms of fundamental flats , which are special connected flats from which one can recover the paths that define the matroid .
we examine some aspects related to key topics in the literature of transversal matroids and we determine the connectivity of lattice path matroids .
we also introduce notch matroids , a minor - closed , dual - closed subclass of lattice path matroids , and we find their excluded minors . |
quantum chromodynamics is a cp - violating theory , implying that the neutron should have a large electric dipole moment , far in excess of experimental limits @xcite .
one elegant solution of this `` strong cp problem '' was proposed by peccei and quinn ( pq ) who postulated a new global u(1 ) symmetry that is spontaneously broken at some large energy scale , allowing for the dynamical restoration of the cp symmetry @xcite .
weinberg @xcite and wilczek @xcite realized that an inevitable consequence of this mechanism is a new pseudoscalar boson , the axion , which is the nambu - goldstone boson of the pq symmetry .
the axion solution would open a new low - energy window to high - energy physics .
string theory provides another motivation in that axion - like fields typically appear so that plausibly one of them could play the role of the qcd axion @xcite .
the pq symmetry is explicitly broken at low energies by instanton effects so that the axion acquires a small mass , @xmath0 where @xmath1 is the axion decay constant or pq scale that governs all axion properties , @xmath2 mev is the pion decay constant , and @xmath3 is the mass ratio of up and down quarks .
we will follow the previous axion literature and assume the canonical value @xmath4 @xcite , although it could vary in the range 0.30.6 @xcite .
axions were not found in early searches , ruling out `` standard axions '' where @xmath1 would have been related to the electroweak scale @xmath5 .
`` invisible '' axion models with @xmath6 were quickly constructed @xcite and quickly recognized to be far from inconsequential because they can provide the dark matter of the universe @xcite and can be searched in realistic experiments @xcite .
the properties of axions are closely related to those of neutral pions .
one generic property is a two - photon interaction that plays a key role for most searches , @xmath7 here , @xmath8 is the electromagnetic field - strength tensor , @xmath9 its dual , and @xmath10 and @xmath11 the electric and magnetic fields , respectively . the coupling constant is @xmath12 where , @xmath13 and @xmath14 , respectively , are the electromagnetic and color anomaly of the axial current associated with the axion field . in grand
unified models , e.g. the dfsz model @xcite , axions couple to ordinary quarks and leptons , implying @xmath15 , whereas in the ksvz model @xcite @xmath16 . while these cases are often used as generic examples , in general @xmath17 is not known so that for fixed @xmath1 a broad range of @xmath18 values is possible @xcite .
still , barring fine - tuned cancelations , @xmath18 scales from the corresponding pion interaction by virtue of eq .
( [ eq : axionphoton ] ) . taking the model - dependent factors to be of order unity
defines the `` axion line '' in the @xmath19-@xmath18 plane .
axions transform into photons and vice versa in external @xmath20 fields @xcite , a phenomenon similar to neutrino oscillations @xcite .
this effect can have intriguing applications , although the relevant range of @xmath18 and @xmath19 typically is far away from the axion line because for a given @xmath18 the implied @xmath19 usually suppresses conversion effecs by energy - momentum conservation .
therefore , beginning with `` arions '' @xcite , the axion relation between @xmath18 and @xmath19 was often taken to be relaxed , postulating nearly massless axion - like particles for which `` alps '' has recently become standard coinage . while cast @xcite ( solar axions ) and admx @xcite ( galactic dark matter ) are the only experiments that probe alps on the axion line , we will briefly review other searches , notably the puzzling signal in the pvlas experiment @xcite . section [ sec : axionphoton ] is devoted to searches for axions and axion - like particles based on their two - photon interaction and the relevant astrophysical limits . in sec .
[ sec : astrolimits ] other astrophysical limits are summarized . section [ sec : cosmology ] is devoted to the possible cosmological role of axions before turning to a summary in sec .
[ sec : summary ] .
particles with a two - photon vertex , including neutral pions and gravitons besides the hypothetical axions , can transform into photons in external electric or magnetic fields , an effect first discussed by primakoff in the early days of pion physics @xcite .
therefore , stars produce these particles from thermal photons in the fluctuating electromagnetic fields of the stellar plasma @xcite . calculating the solar axion flux is straightforward except for the proper inclusion of screening effects @xcite .
the transition rate for a photon of energy @xmath13 into an axion of the same energy ( recoil effects are neglected ) is @xcite @xmath21\,,\ ] ] where @xmath22 is the temperature ( natural units with @xmath23 are used ) .
the screening scale in the debye - hckel approximation is @xmath24 where @xmath25 is the electron density and @xmath26 that of the @xmath27-th ion of charge @xmath28 . near the solar center @xmath29 kev and @xmath30
is nearly constant throughout the sun .
the axion flux at earth , based on a standard solar model , is well approximated by @xmath31 where @xmath13 is in kev and @xmath32 .
the integrated flux parameters are @xmath33 the maximum of the distribution is at @xmath34 kev , the average energy is @xmath35 . this flux can be searched with the inverse process where an axion converts into a photon in a macroscopic @xmath20 field , the `` axion helioscope '' technique @xcite .
one would look at the sun through a `` magnetic telescope '' and place an x - ray detector at the far end . the conversion can be coherent over a large propagation distance and is then pictured as a particle oscillation effect @xcite .
the conversion probability is @xmath36 where @xmath37 is the path length and @xmath38 the axion - photon momentum difference ; in vacuum it is @xmath39 . for @xmath40
the oscillation length exceeds @xmath37 .
for @xmath41 we have @xmath42 , implying an x - ray flux of @xmath43 for @xmath44 this rate is reduced by the momentum mismatch
. a low-@xmath45 gas would provide a refractive photon mass @xmath46 so that @xmath47 . for @xmath48 the maximum rate can thus be restored @xcite .
early helioscope searches were performed in brookhaven @xcite and tokyo @xcite .
solar axions could also transform in electric crystal fields , but the limits obtained by solax @xcite , cosme @xcite , and dama @xcite are less restrictive and require an axion luminosity exceeding the helioseismological constraint @xmath49 @xcite .
the first helioscope that can actually reach the `` axion line '' is the cern axion solar telescope ( cast ) , using a refurbished lhc test magnet with @xmath50 and two pipes of length @xmath51 .
the magnet is mounted with @xmath52 vertical movement , allowing for observation of the sun for 1.5 h at both sunrise and sunset .
cast operated for about 6 months in 2003 .
the non - observation of a signal above background in all three detectors leads to the exclusion range shown in fig .
[ fig : cast ] . in the mass range
@xmath53 ev the new limit is @xcite @xmath54 the data taken in 2004 do not show any apparent signal above background , leading to limits comparable to the sensitivity prospected in fig .
[ fig : cast ] . in the ongoing phase ii
, a variable - pressure helium filling provides an effective photon mass .
the vapor pressure of he@xmath55 at 1.8 k , the magnet s operating temperature , allows one to reach @xmath19 up to 0.26 ev . in the next phase
, he@xmath56 will be used because its higher vapor pressure allows one to reach 0.8 ev .
yet higher masses require an isolating gas cell in the bore where he@xmath56 at 5.4 k would allow one to reach @xmath19 of 1.4 ev . for @xmath57 ,
a competitive method to search for solar axions could be to look with an x - ray satellite at the sun when it is shadowed by the earth and search for x - rays from the axion - photon conversion in the earth s magnetic field @xcite .
a related idea is to look for high - energy @xmath58-rays , e.g. with the glast satellite , that `` pass through the sun '' by magnetic conversion and back conversion @xcite .
cast can detect axions only if they actually escape from the sun .
their mean free path ( mfp ) against the primakoff process is the inverse of eq .
( [ eq : gamma - ag ] ) . as an example we consider 4 kev axions and
note that at the solar center @xmath59 kev and @xmath60 kev .
the axion mfp is then @xmath61 , or about @xmath62 of the radius of the visible universe .
therefore , @xmath18 would have to be more than @xmath63 larger than the cast limit for axions to be re - absorbed in the sun . even in this extreme case
they are not harmless because they would carry the bulk of the energy flux that otherwise is carried by photons .
the mfp of low - mass particles that are trapped in the sun should be shorter than that of photons ( about 10 cm near the solar center ) to avoid a dramatic modification of the solar structure @xcite .
this requirement is so extreme that for anything similar to axions the possibility of re - absorption is not a serious possibility .
the most restrictive astrophysical limit on @xmath18 arises from globular cluster stars .
a globular cluster is a gravitationally bound system of stars that formed at the same time and thus differ primarily in their mass .
a globular cluster provides a homogeneous population of stars , allowing for detailed tests of stellar - evolution theory .
the stars surviving since formation have masses somewhat below @xmath64 . in a color - magnitude diagram , where one plots essentially the surface brightness vs. the surface temperature , stars appear in characteristic loci , allowing one to identify their state of evolution .
the stars on the horizontal branch ( hb ) have reached helium burning where their core ( about @xmath65 ) generates energy by fusing helium to carbon and oxygen with a core - averaged energy release of about @xmath66 . on the other hand ,
the average primakoff energy loss is approximately @xmath67 .
the main effect would be an acceleration of the consumption of helium and thus a reduction of the hb lifetime by a factor @xmath68 , i.e. , by about 30% for @xmath69 .
number counts in 15 globular clusters @xcite reveal typically 100 hb stars in the used fields of view .
this is compared with the number of red giants ( evolutionary phase before helium ignition , but after core - exhaustion of hydrogen ) and shows that the hb lifetime in any one cluster is established within about 2040% .
note that the primakoff losses are much smaller for red giants so that their lifetime is not reduced by primakoff emission .
compounding the results of all 15 clusters , the helium - burning lifetime agrees with expectations to within about 10% @xcite .
of course , with modern data these results likely could be improved .
either way , a reasonably conservative limit is ( figs . [
fig : cast ] and [ fig : exclusion ] ) @xmath70 it is comparable to the cast limit eq .
( [ eq : castlimit ] ) , but applies for higher masses because the relevant temperature is about 10 kev so that significant threshold effects only begin at about @xmath71 kev . for qcd axions
the coupling increases with mass so that the limit reaches to even larger masses . in the helium - burning core , convection and semi - convection dredges helium fuel to the burning site near the core s center so that 2530% of all helium is burnt during the hb phase .
therefore , while the standard theoretical predictions depend on a phenomenological treatment of convection , there is limited room for additional energy supply , even if the treatment of convection were grossly incorrect .
large - scale magnetic fields are ubiquitous in astrophysics so that photon - axion conversions could be of interest . in practice
, the energy - momentum mismatch typically prevents this effect from being important anywhere near the `` axion line . '' on the other hand , `` axion - like particles '' ( alps ) with smaller masses may well exist and could then have intriguing consequences .
examples are the polarization of radio galaxies @xcite and quasars @xcite , the diffuse x - ray background @xcite , or ultra - high energy cosmic rays if they have a photon component @xcite .
one interesting cosmological consequence is the dimming of distant sources by photon - alp conversion in intergalactic @xmath20 fields @xcite , for a review see ref .
the apparent dimming of distant sne ia is usually attributed to accelerated cosmic expansion .
all things considered , the required dimming can not be caused by photon - alp conversion , but this effect may still figure for a detailed interpretation of the cosmic equation of state implied by current and future sn ia data .
preventing excessive suppression of photon - alp conversion in this context requires @xmath72@xmath73 ev , i.e. , the alp mass should be smaller than the photon plasma frequency in the intergalactic medium .
for such small masses , the absence of @xmath58-rays from sn 1987a implies @xmath74 , valid for @xmath75 ev @xcite .
the alps would have been emitted in sn 1987a by the primakoff effect and converted in the galactic magnetic field .
a similar argument applied to the nearby red giant betelgeuse yields a slightly less restrictive limit @xcite , but of course depends on fewer assumptions .
alp - photon conversion in stellar magnetic fields is also possible , notably in the magnetic fields of sun spots @xcite or the strong magnetic fields of pulsars @xcite , although at present no new limit or positive signature is available from these considerations .
magnetic fields induce vacuum birefringence in that electromagnetic radiation with polarization parallel ( @xmath76 ) or perpendicular ( @xmath77 ) to @xmath20 has a different refractive index induced by a one - loop qed correction to maxwell s equations @xcite
. this vacuum cotton - mouton effect has never been measured , but can be searched in the laboratory by propagating a linearly polarized laser beam in a strong @xmath20 field and measure a small ellipticity that develops @xcite .
alps , axions , neutral pions , and other particles with a two - photon vertex also contribute to this effect in that the photon mixes with them in a transverse @xmath20 field @xcite .
if sufficiently light , these particles also cause vacuum dichroism , i.e. , the @xmath76 and @xmath77 polarization states are absorbed differently due to photon - alp conversion .
the observable effect is a rotation of the plane of polarization .
a huge effect of this sort has been reported by the pvlas collaboration @xcite , where the amplitude of the observed dichroism is roughly @xmath78 times larger than the expected qed ellipticity amplitude .
an instrumental origin has not been identified , but further tests and modifications of the experiment will be performed . in pvlas ,
the superconducting magnet is rotated to modulate the signal because a static dichroism or birefringence is overwhelmed by residual instrumental effects .
it has been proposed that the @xmath20 field rotation mixes sidebands to the primary laser frequency with a separation corresponding to the rotation frequency @xcite .
while such effects would in principle occur , the quantitative treatment in this paper vastly overestimates the effect ; not even the use of electromagnetic units is correct .
the slow magnet rotation does not cause a @xmath78 fold enhancement of the qed effect .
still , the exciting possibility that an alp has been detected is damped by the extreme requirements on its properties @xcite @xmath79 with this coupling strength , the sun s alp luminosity would exceed its photon emission by more than a factor of @xmath80 and thus could live only for 1000 years .
a number of models have been proposed to circumvent this problem in that the alp properties and interactions could strongly depend on the environment in the sun or on energy @xcite .
it has become clear that to square the particle interpretation of pvlas with the astrophysical limits is a tall order .
therefore , barring the identification of an instrumental effect that causes the pvlas signature , the best hope is to test this hypothesis directly , notably in a new generation of `` shining light through walls '' experiments @xcite .
besides their generic interaction with photons , axions also interact with fermions @xmath81 by a derivative interaction of the form @xmath82 where @xmath83 is a model - dependent numerical factor and @xmath84 the fermion mass .
the pseudoscalar form is equivalent when only a single axion is attached to the fermion line in a given feynman graph .
subtleties for axion - nucleon interactions in a nuclear medium arise because of the axion - pion mixing @xcite .
the pseudoscalar form suggests to quantify the interaction in terms of a dimensionless yukawa coupling @xmath85 and associated `` fine structure constant '' @xmath86 . in stellar plasmas ,
axions can be emitted by the compton process @xmath87 , bremsstrahlung @xmath88 , bound - free and free - bound transitions , or pair annihilation @xmath89 . for the sun ,
@xmath90 with 25% due to the compton process @xcite .
helioseismology conservatively implies that a new energy - loss channel can not exceed 20% of @xmath91 @xcite , implying @xmath92 .
avoiding excessive white - dwarf cooling @xcite or a delay of helium ignition in globular - cluster stars @xcite implies @xmath93 . in the dfsz model
we have @xmath94 , assuming three families of fermions .
for @xmath95 the exclusion range is shown in fig .
[ fig : exclusion ] . near the limit
, axion emission can affect the cooling speed and thus the period decrease of zz ceti stars ( pulsating white dwarfs ) , although a previously measured anomalous period decrease of the star g117-b15a has disappeared in the light of more recent analyses @xcite .
the axion - electron interaction provides extremely restrictive limits , but in hadronic axion models this coupling does not exist at tree level .
axions are a qcd phenomenon , so their interactions with pions , nucleons and photons are far more generic than those with electrons .
unfortunately , the astrophysical arguments leading to the electron coupling can not be trivially recycled because axions couple to the nucleon spin and thus at low energies do not interact with @xmath96 particles .
therefore , processes of the sort @xmath97 do not occur , although they provide limits on the scalar or vector coupling of new bosons @xcite . however , the sun largely consists of hydrogen so that axions are produced by the compton - like process @xmath98 .
the cross section is @xmath99 with @xmath84 the fermion mass , i.e. , the energy loss rate caused by protons relative to electrons is suppressed by a factor @xmath100 . taking the proton density in the sun to be 1/2 that of electrons , using @xmath101 for the solar axion luminosity from the electron compton process , and observing that helioseismology allows at most a 20% exotic energy loss , we find @xmath102 for the axion - proton interaction .
much more restrictive limits arise from the observed neutrino burst of supernova ( sn ) 1987a @xcite .
the burst lasted for about 10 s , in agreement with the usual picture of neutrino trapping and slow diffusive energy loss of the collapsed sn core .
excessive axion losses would have shortened the burst duration . on the other hand ,
if axions interact so strongly that their mfp is similar to that of neutrinos , they will not much affect the neutrino signal .
the results from different numerical simulations and treatments agree reasonably well @xcite and lead to the `` consolidated '' exclusion range shown in fig .
[ fig : exclusion ] .
early papers found more restrictive limits , but they overestimated the emission rate .
the main process is nucleon - nucleon bremsstrahlung .
axions couple to the nucleon spin so that only spin - nonconserving interactions contribute , i.e. , the nuclear tensor force .
if it is modeled by one - pion exchange , the interaction rate is so large that axions emitted from different collisions would destructively interfere , leading to a suppression of the landau - pomeranchuk - migdal type @xcite .
later it was noted that one - pion exchange may be a poor model for the relevant conditions and the bremsstrahlung rate was calculated on the basis of measured nuclear phase shifts , leading to a much smaller emission rate @xcite .
however , this result applies only in the low - energy limit .
either way , the `` naive '' bremsstrahlung rate used in early studies is reduced .
since the sn 1987a limits suffer from these and other nuclear - physics uncertainties , but also from the sparse experimental data , it is not easily possible to quantify an objective error estimate .
if axions interact so strongly that they are trapped more efficiently than neutrinos , still a significant flux will be emitted .
it interacts with oxygen in the water cherenkov detectors that have recorded the sn 1987a signal and could have mimicked some of the signal @xcite .
this argument excludes another range of axion parameters ( fig .
[ fig : exclusion ] ) .
axions are produced in the hot thermal plasma of the early universe and , since they have a mass , contribute to the cosmic dark matter . before the qcd confinement transition at @xmath103 ,
the relevant processes involve quarks and gluons , e.g. , the gluonic primakoff effect caused by the axions s generic two - gluon vertex @xcite .
after confinement , the most generic interaction process involves pions , @xmath104 @xcite . note that thermal pions begin disappearing only at @xmath105 mev , long after the qcd epoch .
axions decouple after the qcd epoch if @xmath106 gev , i.e. , for @xmath107 ev .
thermal axions provide a cosmic hot dark matter component fully analogous to neutrinos .
therefore , the usual neutrino mass limits from cosmic large - scale structure data can be adapted to the axion case @xcite . relying only on the pion process for thermalization , for hadronic axions
@xmath108 ev ( 95% cl ) , although this limit probably could be somewhat improved with a more recent set of cosmological data . for comparison , we note that this limit corresponds to @xmath109 ev ( 95% cl ) .
axions or axion - like particles with a two - photon vertex decay into two photons with a rate @xmath110 ^ 2\,\frac{m_a^5}{m_\pi^2 f_\pi^2 } \nonumber\\ & = & 1.1\times10^{-24}~{\rm s}^{-1}\,\left(\frac{m_a}{\rm ev}\right)^5 \,,\end{aligned}\ ] ] where the first expression is general , the second applies to axions , and the numerical one assumes @xmath4 and the hadronic case @xmath16 . comparing with the age of the universe of @xmath111 s reveals that axions decay on a cosmic time scale for @xmath112 ev . the decay photons would cause a variety of observable consequences @xcite , depending on the axion mass , so that cosmology alone entirely rules out axions with @xmath113 ev ( fig .
[ fig : exclusion ] ) .
the low - mass end is covered by the hot dark matter argument , seamlessly connecting to high masses to avoid excess radiation from axion decays @xcite .
hot dark matter axions would be partially trapped in galaxies and galaxy clusters .
searching for a decay line @xcite provides direct limits on a range of axion masses marked `` telescope '' in fig .
[ fig : exclusion ] .
the main cosmological interest in axions derives , of course , from their possible role as the dominant cold dark matter component .
almost immediately after `` invisible '' axion models had been constructed it was recognized that they can be abundantly produced by the `` misalignment mechanism '' @xcite .
after the spontaneous breaking of the pq symmetry at some high energy scale , the axion field relaxes to a location somewhere in the mexican hat potential . near the qcd epoch , instanton effects explicitly break the pq symmetry , the very effect that causes the dynamical pq symmetry restoration .
this `` tilting of the mexican hat '' causes the axion field to roll toward the cp - conserving minimum , but this motion does not stop there . rather , coherent oscillations of the axion field are excited that ultimately represent a condensate of cold dark matter .
the cosmic matter density in this homogeneous field mode is @xcite @xmath114 where @xmath115 is the initial `` misalignment angle '' relative to the cp - conserving position .
if the pq symmetry breaking takes place after inflation , @xmath116 will take on different values in different patches of the universe .
the approximate average contribution is @xcite @xmath117 comparing with the cosmic cold dark matter density of @xmath118 this implies that axions with @xmath119ev would provide the dark matter ( fig .
[ fig : exclusion ] ) .
however , this number sets only a crude scale of the expected mass for axion dark matter .
apart from the overall particle - physics uncertainties entering this result , the cosmological sequence of events crucially matters . assuming axions make up the cold dark matter of the universe , significantly smaller masses are possible if inflation took place after the pq transition and the initial value @xmath116 was relatively small .
conversely , if the pq transition takes place after inflation , there are additional sources for nonthermal axion production , notably the formation and decay of cosmic strings and domain walls associated with the breaking of the pq symmetry @xcite .
therefore , the mass of cosmic dark matter axions could be significantly larger than the @xmath120ev scale .
for a recent review of axion cosmology we refer to ref .
@xcite .
if axions are the dark matter of our galaxy , they are accessible to direct searches by the axion `` haloscope '' technique .
a microwave resonator is placed in a strong magnetic field and thus can be driven by the galactic axion field due to the two - photon coupling @xcite . after a series of pilot experiments ,
the admx experiment in livermore has reached the axion line in a narrow range of masses @xcite .
the project will be upgraded in several stages so that it should be able to cover the range @xmath121ev within the upcoming decade .
therefore , axion dark matter can be detected in a realistic range of parameters @xcite . in the exciting event that dark matter axions are found , intriguing opportunities for observational axion cosmology open up .
the fine energy resolution that is possible with a microwave resonator allows one to study the detailed axion velocity distribution that may reveal narrow peaks from the phase - space folding that occurred during galaxy formation @xcite .
axions remain a well motivated solution of the strong cp problem .
ongoing and near - future searches for solar axions by cast and for galactic cold dark matter axions by admx cover realistic parameters and thus have a chance to detect these elusive particles .
if cast were to detect solar axions , they would contribute a small hot dark matter component of the universe so that these two experimental directions are complementary .
there remains a range of axion parameters , roughly corresponding to @xmath122 mev , where axions are not excluded , yet no detection strategy has been proposed . perhaps we will not have to worry about this gap if axions show up in one of the accessible search ranges .
unfortunately , the pvlas signature likely has nothing to do with the strong cp problem or dark matter .
the particle interpretation is in severe conflict with astrophysical arguments so that plausible theoretical models are hard to construct .
however , the effect is so large that it can be tested relatively easily at several forthcoming `` shining light through walls '' experiments .
even if no new particle is confirmed , a strong push is made towards measuring the qed - implied birefringence of a magnetized vacuum .
such a measurement would be an astounding achievement in itself .
partial support by the deutsche forschungsgemeinschaft under grant sfb 375 and the european union , contract rii3-ct-2004 - 506222 ( ilias project ) , is acknowledged .
99 c. a. baker _
et al . _ , `` an improved experimental limit on the electric dipole moment of the neutron , '' phys . rev .
* 97 * ( 2006 ) 131801 [ hep - ex/0602020 ] .
r. d. peccei and h. r. quinn , `` cp conservation in the presence of pseudoparticles , '' phys .
* 38 * ( 1977 ) 1440 .
r. d. peccei and h. r. quinn , `` constraints imposed by cp conservation in the presence of pseudoparticles , '' phys .
d * 16 * ( 1977 ) 1791 .
s. weinberg , `` a new light boson ?
, '' phys .
* 40 * ( 1978 ) 223
. f. wilczek , `` problem of strong p and t invariance in the presence of instantons , '' phys .
* 40 * ( 1978 ) 279 .
p. svrcek and e. witten , `` axions in string theory , '' jhep * 0606 * ( 2006 ) 051 [ hep - th/0605206 ] .
j. gasser and h. leutwyler , `` quark masses , '' phys .
* 87 * ( 1982 ) 77 .
h. leutwyler , `` the ratios of the light quark masses , '' phys .
b * 378 * ( 1996 ) 313 [ hep - ph/9602366 ] .
w. m. yao _ et al .
_ [ particle data group ] , `` review of particle physics , '' j. phys .
g * 33 * ( 2006 ) 1 .
j. e. kim , `` weak interaction singlet and strong cp invariance , '' phys .
* 43 * ( 1979 ) 103 .
m. a. shifman , a. i. vainshtein and v. i. zakharov , `` can confinement ensure natural cp invariance of strong interactions ?
, '' nucl .
b * 166 * ( 1980 ) 493 .
a. r. zhitnitsky , `` on possible suppression of the axion hadron interactions , '' sov .
j. nucl .
* 31 * ( 1980 ) 260 [ yad .
* 31 * ( 1980 ) 497 ] .
m. dine , w. fischler and m. srednicki , `` a simple solution to the strong cp problem with a harmless axion , '' phys .
b * 104 * ( 1981 ) 199 .
j. e. kim , `` light pseudoscalars , particle physics and cosmology , '' phys .
* 150 * ( 1987 ) 1 .
j. preskill , m. b. wise and f. wilczek , `` cosmology of the invisible axion , '' phys .
b * 120 * ( 1983 ) 127 .
l. f. abbott and p. sikivie , `` a cosmological bound on the invisible axion , '' phys .
b * 120 * ( 1983 ) 133 .
m. dine and w. fischler , `` the not - so - harmless axion , '' phys .
b * 120 * ( 1983 ) 137 .
r. l. davis , `` cosmic axions from cosmic strings , '' phys .
b * 180 * ( 1986 ) 225 .
p. sikivie , `` experimental tests of the invisible axion , '' phys .
* 51 * ( 1983 ) 1415 ; ( e ) bid .
* 52 * ( 1984 ) 695 .
r. bradley _ et al .
_ , `` microwave cavity searches for dark - matter axions , '' rev .
phys . *
75 * ( 2003 ) 777 .
s. j. asztalos _ et al .
_ , `` an improved rf cavity search for halo axions , '' phys . rev .
d * 69 * ( 2004 ) 011101 [ astro - ph/0310042 ] .
l. d. duffy _ et al .
_ , `` a high resolution search for dark - matter axions , '' phys .
d * 74 * ( 2006 ) 012006 [ astro - ph/0603108 ] .
s. l. cheng , c. q. geng and w. t. ni , `` axion - photon couplings in invisible axion models , '' phys .
d * 52 * ( 1995 ) 3132 [ hep - ph/9506295 ] .
g. raffelt and l. stodolsky , `` mixing of the photon with low mass particles , '' phys .
d * 37 * ( 1988 ) 1237 .
a. a. anselm and n. g. uraltsev , `` a second massless axion ?
, '' phys .
b * 114 * , 39 ( 1982 ) .
a. a. anselm and n. g. uraltsev , `` long range ` arion ' field in the radiofrequency band , '' phys
. lett .
b * 116 * , 161 ( 1982 ) .
k. zioutas _ et al . _
( cast collaboration ) , `` first results from the cern axion solar telescope ( cast ) , '' phys .
* 94 * ( 2005 ) 121301 [ hep - ex/0411033 ] .
e. zavattini _ et al .
_ [ pvlas collaboration ] , `` experimental observation of optical rotation generated in vacuum by a magnetic field , '' phys .
* 96 * ( 2006 ) 110406 [ hep - ex/0507107 ] .
h. primakoff , `` photo - production of neutral mesons in nuclear electric fields and the mean life of the neutral meson , '' phys .
* 81 * ( 1951 ) 899 .
d. a. dicus , e. w. kolb , v. l. teplitz and r. v. wagoner , `` astrophysical bounds on the masses of axions and higgs particles , '' phys .
d * 18 * ( 1978 ) 1829 .
g. g. raffelt , `` astrophysical axion bounds diminished by screening effects , '' phys .
d * 33 * ( 1986 ) 897 .
t. altherr , e. petitgirard and t. del ro gaztelurrutia , `` axion emission from red giants and white dwarfs , '' astropart .
* 2 * ( 1994 ) 175 [ hep - ph/9310304 ] .
g. g. raffelt , `` plasmon decay into low mass bosons in stars , '' phys . rev .
d * 37 * ( 1988 ) 1356 .
k. van bibber , p. m. mcintyre , d. e. morris and g. g. raffelt , `` a practical laboratory detector for solar axions , '' phys .
d * 39 * ( 1989 ) 2089 .
d. m. lazarus , g. c. smith , r. cameron , a. c. melissinos , g. ruoso , y. k. semertzidis and f. a. nezrick , `` a search for solar axions , '' phys . rev .
* 69 * ( 1992 ) 2333 .
s. moriyama , m. minowa , t. namba , y. inoue , y. takasu and a. yamamoto , `` direct search for solar axions by using strong magnetic field and x - ray detectors , '' phys .
b * 434 * ( 1998 ) 147 [ hep - ex/9805026 ] .
y. inoue , t. namba , s. moriyama , m. minowa , y. takasu , t. horiuchi and a. yamamoto , `` search for sub - electronvolt solar axions using coherent conversion of axions into photons in magnetic field and gas helium , '' phys .
b * 536 * ( 2002 ) 18 [ astro - ph/0204388 ] .
f. t. avignone _ et al .
_ ( solax collaboration ) , `` experimental search for solar axions via coherent primakoff conversion in a germanium spectrometer , '' phys . rev .
* 81 * ( 1998 ) 5068 .
[ astro - ph/9708008 ] .
a. morales _ et al .
_ ( cosme collaboration ) , `` particle dark matter and solar axion searches with a small germanium detector at the canfranc underground laboratory , '' astropart .
* 16 * ( 2002 ) 325 [ hep - ex/0101037 ] .
r. bernabei _ et al .
_ , `` search for solar axions by primakoff effect in nai crystals , '' phys .
b * 515 * ( 2001 ) 6 .
h. schlattl , a. weiss and g. raffelt , `` helioseismological constraint on solar axion emission , '' astropart
. phys . * 10 * ( 1999 ) 353 [ hep - ph/9807476 ] .
h. davoudiasl and p. huber , `` detecting solar axions using earth s magnetic field , '' phys .
* 97 * ( 2006 ) 141302 [ hep - ph/0509293 ] .
m. fairbairn , t. rashba and s. troitsky , `` shining light through the sun , '' astro - ph/0610844 . g. g. raffelt and g. d. starkman , `` stellar energy transfer by kev mass scalars , '' phys .
d * 40 * ( 1989 ) 942 .
a. buzzoni , f. fusi pecci , r. buonanno and c. e. corsi `` helium abundance in globular clusters : the r - method '' astron .
astrophys .
* 128 * ( 1983 ) 94 .
g. g. raffelt , `` particle physics from stars , '' ann
. rev .
* 49 * ( 1999 ) 163 [ hep - ph/9903472 ] .
d. harari and p. sikivie , `` effects of a nambu - goldstone boson on the polarization of radio galaxies and the cosmic microwave background , '' phys .
b * 289 * , 67 ( 1992 ) .
d. hutsemekers , r. cabanac , h. lamy and d. sluse , `` mapping extreme - scale alignments of quasar polarization vectors , '' astron .
astrophys . * 441 * , 915 ( 2005 ) [ astro - ph/0507274 ] .
y. n. gnedin , m. y. piotrovich and t. m. natsvlishvili , `` pvlas experiment : some astrophysical consequences , '' astro - ph/0607294 . s. v. krasnikov , `` new astrophysical constraints on the light pseudoscalar photon coupling , '' phys .
lett . *
76 * , 2633 ( 1996 ) .
d. s. gorbunov , g. g. raffelt and d. v. semikoz , `` axion - like particles as ultrahigh - energy cosmic rays ?
, '' phys .
d * 64 * , 096005 ( 2001 ) [ hep - ph/0103175 ] . c. cski , n. kaloper , m. peloso and j. terning , `` super - gzk photons from photon axion mixing , '' jcap * 0305 * , 005 ( 2003 ) [ hep - ph/0302030 ] . c. cski , n. kaloper and j. terning
( ckt i ) , `` dimming supernovae without cosmic acceleration , '' phys . rev .
* 88 * , 161302 ( 2002 ) [ hep - ph/0111311 ] .
y. grossman , s. roy and j. zupan , `` effects of initial axion production and photon axion oscillation on type ia supernova dimming , '' phys .
b * 543 * , 23 ( 2002 ) [ hep - ph/0204216 ] .
b. a. bassett and m. kunz , `` cosmic acceleration versus axion photon mixing , '' astrophys . j. * 607 * , 661 ( 2004 ) [ astro - ph/0311495 ] .
l. ostman and e. mrtsell , `` limiting the dimming of distant type ia supernovae , '' jcap * 0502 * , 005 ( 2005 ) [ astro - ph/0410501 ] . c. cski , n. kaloper and j. terning , `` exorcising w @xmath123 -1 , '' annals phys .
* 317 * , 410 ( 2005 ) [ astro - ph/0409596 ] .
a. mirizzi , g. g. raffelt and p. d. serpico , `` photon axion conversion as a mechanism for supernova dimming : limits from cmb spectral distortion , '' phys .
d * 72 * , 023501 ( 2005 ) [ astro - ph/0506078 ] .
a. mirizzi , g. g. raffelt and p. d. serpico , `` photon axion conversion in intergalactic magnetic fields and cosmological consequences , '' astro - ph/0607415 .
j. w. brockway , e. d. carlson and g. g. raffelt , `` sn 1987a gamma - ray limits on the conversion of pseudoscalars , '' phys .
b * 383 * , 439 ( 1996 ) [ astro - ph/ 9605197 ] .
j. a. grifols , e. mass and r. toldr , `` gamma rays from sn 1987a due to pseudoscalar conversion , '' phys .
lett . * 77 * , 2372 ( 1996 ) [ astro - ph/9606028 ] .
e. d. carlson , `` pseudoscalar conversion and x - rays from stars , '' phys .
b * 344 * ( 1995 ) 245 .
e. d. carlson and l. s. tseng , `` pseudoscalar conversion and x - rays from the sun , '' phys .
b * 365 * ( 1996 ) 193 [ hep - ph/9507345 ] .
a. dupays , c. rizzo , m. roncadelli and g. f. bignami , `` looking for light pseudoscalar bosons in the binary pulsar system j0737 - 3039 , '' phys .
lett . * 95 * ( 2005 ) 211302 [ astro - ph/0510324 ] .
d. lai and j. heyl , `` probing axions with radiation from magnetic stars , '' astro - ph/0609775 .
w. heisenberg and h. euler , `` consequences of dirac s theory of positrons , '' z. phys .
* 98 * ( 1936 ) 714 [ physics/0605038 ]
. s. l. adler , `` photon splitting and photon dispersion in a strong magnetic field , '' annals phys .
( n.y . ) * 67 * ( 1971 ) 599 .
e. iacopini and e. zavattini , `` experimental method to detect the vacuum birefringence induced by a magnetic field , '' phys .
b * 85 * ( 1979 ) 151 .
l. maiani , r. petronzio and e. zavattini , `` effects of nearly massless , spin zero particles on light propagation in a magnetic field , '' phys .
b * 175 * ( 1986 ) 359 .
j. t. mendona , j. dias de deus and p. castelo ferreira , `` higher harmonics in non - linear vacuum from qed effects without low mass intermediate particles , '' phys .
* 97 * ( 2006 ) 100403 [ hep - ph/0606099 ] .
r. cameron _ et al .
_ , `` search for nearly massless , weakly coupled particles by optical techniques , '' phys .
d * 47 * ( 1993 ) 3707 .
e. mass and j. redondo , `` evading astrophysical constraints on axion - like particles , '' jcap * 0509 * ( 2005 ) 015 [ hep - ph/0504202 ] .
e. mass and j. redondo , `` compatibility of cast search with axion - like interpretation of pvlas results , '' phys .
* 97 * ( 2006 ) 151802 [ hep - ph/0606163 ] .
i. antoniadis , a. boyarsky and o. ruchayskiy , `` axion alternatives , '' hep - ph/0606306 .
r. n. mohapatra and s. nasri , `` reconciling the cast and pvlas results , '' hep - ph/0610068 .
j. jckel , e. mass , j. redondo , a. ringwald and f. takahashi , `` the need for purely laboratory - based axion - like particle searches , '' hep - ph/0610203 .
j. redondo , `` a model solving the pvlas - cast puzzle , '' hep - ph/0610213 .
r. rabadan , a. ringwald and k. sigurdson , `` photon regeneration from pseudoscalars at x - ray laser facilities , '' phys .
* 96 * ( 2006 ) 110407 [ hep - ph/0511103 ] .
u. ktz , a. ringwald and t. tschentscher , `` production and detection of axion - like particles at the vuv - fel : letter of intent , '' hep - ex/0606058 .
a. lindner , `` search for axion - like particles at desy : the alps project , '' talk given at _
axions at the institute for advanced study _
( 2022 oct .
2006 , ias , princeton , new jersey ) , http://www.sns.ias.edu/@xmath124axions g. cantatore , `` probing the quantum vacuum with polarized light : results and outlook from the pvlas experiment , '' talk given at _
axions at the institute for advanced study _
( 2022 oct .
2006 , ias , princeton , new jersey ) , http://www.sns.ias.edu/@xmath124axions a. v. afanasev , o. k. baker and k. w. mcfarlane , `` production and detection of very light spin - zero bosons at optical frequencies , '' hep - ph/0605250 .
a. afanasev , `` lipss project : a search for photon regeneration at optical frequencies , '' talk given at _
axions at the institute for advanced study _
( 2022 oct .
2006 , ias , princeton , new jersey ) , http://www.sns.ias.edu/@xmath124axions p. pugnat , `` laser - based experiments in high magnetic field for qed test and axion search at cern '' talk given at _
axions at the institute for advanced study _
( 2022 oct .
2006 , ias , princeton , new jersey ) , http://www.sns.ias.edu/@xmath124axions m. carena and r. d. peccei , `` the effective lagrangian for axion emission from sn 1987a , '' phys .
d * 40 * ( 1989 ) 652 .
g. g. raffelt , `` axion constraints from white dwarf cooling times , '' phys . lett .
b * 166 * ( 1986 ) 402 .
j. isern and e. garca - berro ,
`` white dwarf stars as particle physics laboratories , '' nucl .
proc . suppl .
* 114 * ( 2003 ) 107 .
j. a. grifols and e. mass , `` constraints on finite range baryonic and leptonic forces from stellar evolution , '' phys .
b * 173 * ( 1986 ) 237 .
j. a. grifols , e. mass and s. peris , `` energy loss from the sun and red giants : bounds on short range baryonic and leptonic forces , '' mod .
a * 4 * ( 1989 ) 311 .
j. r. ellis and k. a. olive , `` constraints on light particles from supernova 1987a '' phys .
b * 193 * ( 1987 ) 525 .
g. raffelt and d. seckel , `` bounds on exotic particle interactions from sn 1987a '' phys .
lett . * 60 * ( 1988 ) 1793
. m. s. turner , `` axions from sn 1987a , '' phys . rev .
* 60 * ( 1988 ) 1797 .
r. mayle , j. r. wilson , j. r. ellis , k. a. olive , d. n. schramm and g. steigman , `` constraints on axions from sn 1987a '' phys .
b * 203 * ( 1988 ) 188 .
r. mayle , j. r. wilson , j. r. ellis , k. a. olive , d. n. schramm and g. steigman , `` updated constraints on axions from sn 1987a , '' phys .
b * 219 * , 515 ( 1989 ) .
a. burrows , m. s. turner and r. p. brinkmann , `` axions and sn 1987a , '' phys .
d * 39 * ( 1989 ) 1020 .
a. burrows , m. t. ressell and m. s. turner , `` axions and sn 1987a : axion trapping , '' phys .
d * 42 * , 3297 ( 1990 ) .
h. t. janka , w. keil , g. raffelt and d. seckel , `` nucleon spin fluctuations and the supernova emission of neutrinos and axions , '' phys .
* 76 * ( 1996 ) 2621 [ astro - ph/9507023 ] .
w. keil , h. t. janka , d. n. schramm , g. sigl , m. s. turner and j. r. ellis , `` a fresh look at axions and sn 1987a , '' phys .
d * 56 * , 2419 ( 1997 ) [ astro - ph/9612222 ] . c. hanhart , d. r. phillips and s. reddy , `` neutrino and axion emissivities of neutron stars from nucleon nucleon scattering data , '' phys .
b * 499 * , 9 ( 2001 ) [ astro - ph/0003445 ] .
j. engel , d. seckel and a. c. hayes , `` emission and detectability of hadronic axions from sn 1987a , '' phys .
lett . * 65 * ( 1990 ) 960 . m. s. turner , `` thermal production of not so invisible axions in the early universe , '' phys .
* 59 * ( 1987 ) 2489 , ( e ) ibid .
* 60 * ( 1988 ) 1101 .
e. mass , f. rota and g. zsembinszki , `` on axion thermalization in the early universe , '' phys .
d * 66 * ( 2002 ) 023004 [ hep - ph/0203221 ] .
s. chang and k. choi , `` hadronic axion window and the big bang nucleosynthesis , '' phys .
b * 316 * ( 1993 ) 51 [ hep - ph/9306216 ] .
s. hannestad and g. raffelt , `` cosmological mass limits on neutrinos , axions , and other light particles , '' jcap * 0404 * ( 2004 ) 008 [ hep - ph/0312154 ]
. s. hannestad , a. mirizzi and g. raffelt , `` new cosmological mass limit on thermal relic axions , '' jcap * 0507 * ( 2005 ) 002 [ hep - ph/0504059 ] .
e. masso and r. toldra , `` new constraints on a light spinless particle coupled to photons , '' phys .
d * 55 * ( 1997 ) 7967 [ hep - ph/9702275 ] .
m. a. bershady , m. t. ressell and m. s. turner , `` telescope search for multi - ev axions , '' phys .
* 66 * ( 1991 ) 1398
. m. t. ressell , `` limits to the radiative decay of the axion , '' phys . rev .
d * 44 * ( 1991 ) 3001 .
p. sikivie , `` axion cosmology , '' astro - ph/0610440
. m. s. turner , `` cosmic and local mass density of invisible axions , '' phys .
d * 33 * ( 1986 ) 889 .
k. van bibber , `` microwave cavity searches for dark - matter axions '' talk given at _
axions at the institute for advanced study _
( 2022 oct .
2006 , ias , princeton , new jersey ) , + http://www.sns.ias.edu/@xmath124axions | the axion solution of the strong cp problem provides a number of possible windows to physics beyond the standard model , notably in the form of searches for solar axions and for galactic axion dark matter , but in a broader context also inspires searches for axion - like particles in pure laboratory experiments .
we briefly review the motivation for axions , astrophysical limits , their possible cosmological role , and current searches for axions and axion - like particles . |
tight dynamical bounds on the graviton mass can be obtained using gw observations in space .
this is due to two reasons : ( 1 ) the larger mass of observable bh binaries , as compared to ground - based gw observations ; ( 2 ) the statistical increase in the bound that would result from observing _
populations _ of individually resolved binaries .
to illustrate point ( 1 ) , in this section we review existing work on massive graviton bounds from gw observations of individual binaries with lisa . in section [ sec :
pops ] we will present the first attempt to quantify the statistical improvement of bounds on @xmath5 that could be achievable in reality by observing several events in the context of hierarchical models of massive bh formation .
will @xcite first pointed out that interesting bounds on @xmath5 could come from a careful monitoring of the phase of gws emitted by binaries of compact objects , such as bhs and/or neutron stars . in hypothetical massive graviton theories , the gw damping formulae and dispersion relation
would be modified . as a consequence ,
the gw phasing @xmath23 would acquire an additional term : _ mg(f)=_gr(f)- _
g ( mf)^-1 , where @xmath24 $ ] and @xmath25 is a distance parameter , similar to ( but not quite the same as ) the luminosity distance @xmath26 ( here and below we assume a @xmath27cdm model with @xmath28 km@xmath29s@xmath30mpc@xmath30 , @xmath31 , @xmath32 ) . using the restricted post - newtonian approximation and neglecting binary spins , will estimated that stellar - mass binary inspirals to be observed with ligo would yield a bound @xmath33 km , only slightly better than solar system bounds .
for an `` optimal '' lisa system , i.e. an equal - mass bh binary of total mass @xmath34 at @xmath35 gpc ( @xmath36 ) , the bound would be four orders of magnitude better : @xmath37 .
these initial estimates were refined in various papers .
will and yunes @xcite showed that bounds from binary bh inspirals at @xmath35 gpc would range between @xmath38 km and @xmath39 km for @xmath11 in the range @xmath40 .
they found that the bound on @xmath5 is proportional to @xmath41 ( where @xmath42 is the lisa armlength ) and to ( lisa acceleration noise)@xmath43 , and that it scales in the following way : _
g()^1/2 , where @xmath44 is the `` chirp mass '' , @xmath45 ( in hz@xmath30 ) sets the scale of the noise power spectral density ( psd ) of the detector , and @xmath46 is a characteristic `` knee '' frequency where the psd has a minimum .
distance dependence is weak because the effect of the massive graviton and measurement errors both grow with distance .
berti _ et al . _
@xcite performed a more detailed monte carlo calculation of lisa s parameter estimation capabilities .
they still considered restricted post - newtonian inspiral waveforms , but they used cutler s model @xcite to take into account the motion of the detector ( all previous analyses assumed a _ static _
lisa constellation ) . in this paper
we will do the same . in cutler s model , lisa
can be seen as one ( two ) independent `` ligo - like '' michelson interferometers depending on whether 4 ( 5/6 , respectively ) laser links are available between the three satellites forming the constellation . for the `` optimal binary '' ( a nonspinning , equal - mass binary with @xmath34 at @xmath47gpc ) , monte carlo averages over the binary position and orientation yield a bound @xmath48 km when using two ( one ) michelsons , in excellent agreement with will s original results . for masses in the range @xmath49 , the mean bound using two michelsons at @xmath47gpc is in the range @xmath50 km . later work used mostly cutler s model and investigated the effect of more complex inspiral waveforms .
arun and will @xcite studied the effect of higher harmonics and pn amplitude corrections .
they found that bounds on @xmath5 improve by a factor of a few for systems of total mass @xmath51 , and that this improvement is more significant for binaries with large mass ratios . if one ignores spin precession , adding spins to the waveforms generally introduces degeneracies between the binary parameters , degrading parameter estimation accuracy @xcite .
stavridis and will @xcite showed that modulations induced by spin precession can break these degeneracies .
for the `` optimal binary '' , for example , the average bound including spin precession is @xmath52 km , basically the same as in the case of nonspinning binaries .
yagi and tanaka @xcite included _ both _ spin precession and eccentricity . performing simulations for a @xmath53 binary at 3 gpc
, they found an average bound @xmath54 km .
this is again consistent with @xcite to within a factor 2 .
more recently , keppel and ajith @xcite revisited this problem using phenomenological waveforms for nonspinning bh binaries that include the merger / ringdown phase .
their analysis of lisa bounds is similar to the original work by will @xcite in that it ignores the motion of the detector , using an effective non - sky - averaged noise psd .
this simplification should not affect their main conclusions : ( i ) for the `` optimal binary '' , the bound on @xmath5 _ improves by about one order of magnitude _
, up to @xmath55 km ; ( ii ) a comparable bound is obtained also for binaries of larger mass , up to @xmath56 . at @xmath35 gpc ,
best _ bound using the full merger is @xmath57 km for @xmath58 , to be compared with a best bound @xmath59 km for @xmath60 if we consider only inspiral waves .
a similar order - of - magnitude improvement is expected for earth - based detectors .
del pozzo _
_ @xcite recently revisited graviton mass bounds using bayesian inference ( see also @xcite for similar work including other possible alternatives to general relativity ) .
they focused on ground - based observations of sources within 150 mpc and found results consistent with will s original analysis @xcite : for example , their fig .
6 shows that binary observations with advanced ligo would yield typical bounds @xmath61 km .
in this study we consider quasicircular , nonspinning , restricted post - newtonian inspiral waveforms .
we consider an observation time of two years .
we take into account weak lensing errors on the redshift following @xcite , and we compute the individual graviton mass bounds by generalizing the fisher matrix formalism described in @xcite , which takes into account correlations with the other waveform parameters .
the results of refs .
@xcite suggest that our bounds will be ( i ) very close to the bounds we would obtain for spinning , precessing inspirals , and ( ii ) about one order of magnitude worse than the bounds achievable if we used `` full '' merger waveforms .
all studies of massive graviton bounds so far analyzed isolated systems , such as will s `` optimal binary '' of mass @xmath62 at @xmath63 .
unfortunately , hierarchical models of massive bh formation and evolution predict that _ typical _ systems observable by space - based interferometers would have masses smaller than this , and be located at redshift @xmath64 or higher ( see e.g. @xcite and fig . 2 of @xcite ) .
the main uncertainties in these models concern the seeding mechanism and the role of accretion in bh growth @xcite . in this work
we ignore spins in the gravitational waveforms .
accretion mostly influences the spin magnitude @xcite , so we will focus on the role of seeding . following work by the lisa parameter estimation taskforce @xcite , we will consider two `` extreme '' scenarios : small seeds , efficient accretion ( se ) and large seeds , efficient accretion ( le ) .
these models being extreme , we expect that our results should bracket the constraints that would be obtained using other population models .
non sky - averaged noise power spectral density @xmath65 for new lisa c2 ( black , solid line ) , new lisa c5 ( blue , dash - dotted line ) and classic lisa ( red , dashed line).,width=321 ] here we consider the `` classic lisa '' design along with two different designs for the proposed esa - led space - based detector , that we will call by the working name of `` new lisa '' .
`` classic lisa '' consists of three spacecraft forming an equilateral triangle with laser power @xmath66 w , telescope diameter @xmath67 m and armlength @xmath68 m , trailing 20@xmath69 behind the earth at an inclination of @xmath70 with respect to the ecliptic .
the authors are members of a science performance task force that is considering several different lisa - like configurations with different characteristics and sensitivities .
the configurations that we call new lisa c2 ( c5 , respectively ) consist of three spacecraft forming an equilateral triangle with armlength @xmath71 m ( @xmath72 m ) , laser power @xmath66 w and telescope diameter @xmath67 m
( @xmath73 m ) .
`` new lisa '' should be deployed 10@xmath69 behind the earth , gradually drifting to @xmath74 behind the earth in 5 years .
the _ non sky - averaged _ noise power spectral densities for all three configurations are shown in fig .
[ fig : noise ] ; they are related to the sky - averaged power spectral density by @xmath75 ( see @xcite for a discussion of sky averaging ) .
these curves include galactic confusion noise , estimated using methods similar to @xcite ( which in turn was based on @xcite ) . in our study
we consider the noise power spectral density to be infinite below a cutoff frequency @xmath76 hz . as shown in @xcite , bounds on @xmath5 drop significantly at masses @xmath77 if the noise can not be trusted below @xmath78 hz .
this assumption has a mild effect on our results , because most binaries in our models have mass lower than this .
distribution of bounds on the graviton compton wavelength for individual observations .
we consider 1000 realizations of the massive bh population and three different detector designs : new lisa c2 ( black , thick line ) , new lisa c5 ( blue , medium line ) and classic lisa ( red , dashed line).,width=321 ] for each model we consider 1000 realizations of the universe .
each of these realizations typically produces @xmath79 events observable with signal - to - noise ratio ( snr ) larger than 8 . the distribution of bounds on @xmath5 resulting from individual observations is shown in fig .
[ individual ] .
the top half of table [ table3 ] shows that the mean bound over individual observations is @xmath80 km for new lisa c2 , and only slightly better for the other designs .
this conclusion is quite robust , in the sense that numbers vary only mildly for different seeding mechanisms and different detector characteristics .
c|cc|cc + detector & se , 1 mich .
& le , 1 mich . &
se , 2 mich .
& le , 2 mich .
+ classic lisa & 4.26(2.60 ) & 6.83(5.77 ) & 4.87(2.72 ) & 9.13(7.72 ) + new lisa c2 & 3.03(2.44 ) & 3.62(3.27 ) & 3.60(2.80 ) & 4.76(4.29 ) + new lisa c5 & 3.41(2.53 ) & 4.63(4.13 ) & 4.02(2.84 ) & 6.15(5.48 ) + + detector & se , 1 mich . &
le , 1 mich . &
se , 2 mich . &
le , 2 mich .
+ classic lisa & 4.93(4.87 ) & 5.67(5.59 ) & 6.51(6.45 ) & 7.50(7.37 ) + new lisa c2 & 2.29(2.25 ) & 2.73(2.71 ) & 3.09(3.04 ) & 3.66(3.64 ) + new lisa c5 & 3.10(3.07 ) & 3.64(3.62 ) & 4.16(4.12 ) & 4.85(4.82 ) + distribution of _ combined _ bounds over 1000 realizations of the mbh population .
linestyles are as in fig .
[ individual].,width=321 ] in most alternative theories , deviations from general relativity can be parametrized by one or more global parameters ( such as @xmath5 ) which are the same for every system .
it is natural to expect that one can obtain better constraints on these parameters , as well as other universal constants , by combining multiple observations ( see e.g. @xcite ) . assuming that estimates for individual sources are independent and gaussian posteriors for each source , consistent with the fisher matrix approximation , the width @xmath81 of the combined posterior on @xmath82 is given by @xmath83 , where @xmath84 is the width of the posterior for the @xmath85th source .
the bound on @xmath5 can thus be obtained by adding the individual bounds in quadrature .
the results are shown in fig . [
combined ] and in the bottom half of table [ table3 ] .
the combined bound obtained from the whole bh population is about one order of magnitude better than the average bound obtained from typical observations .
a rough estimate would suggest that @xmath86 identical sources should provide a bound @xmath87 times better than the bound from a single source .
our combined bound is typically about 3 times better than the bound from the best event , but the median bound is typically an order of magnitude worse than the best , and hence @xmath88 times worse than the combined bound .
a typical realisation has @xmath2 events , so our analysis shows that we can beat the @xmath89 extrapolation from the median bound by a considerable margin .
if 5/6 links ( two michelsons ) are available instead of 4 links ( one michelson ) , the bound typically improves by a factor @xmath90 .
we assessed the capability of future space - based interferometers , such as `` classic lisa '' and the proposed esa - led `` new lisa '' , to constrain the mass of the graviton by combining observations of a _ population _ of massive bh binaries .
we found that : ( 1 ) by using a population of merging bh binaries we can obtain a bound on @xmath5 that is @xmath91 times better than the mean bound on individual observations ; ( 2 ) quite independently of the detector s design and of details of the massive bh formation models , the combined bound from inspiral observations will be @xmath92 .
this figure is likely to underestimate the bound achievable in practice by about one order of magnitude , as we have ignored the merger and ringdown portion of the waveform @xcite , but further work is required to confirm this expectation . in conclusion ,
space - based observations of a population of merging bhs should set bounds in the range @xmath93 $ ] on the graviton compton wavelength , depending on details of the detector and on the specific waveform model used to set the bounds .
this is comparable to the ( static and model - dependent ) bounds from cosmological - scale observations quoted in table [ bounds ] but it is very different in nature , because gravitational radiation tests the dynamical regime of einstein s general relativity . * _ acknowledgments .
is supported by nsf grant no .
phy-0900735 and by nsf career grant no .
j.g.s work is supported by the royal society .
the authors wish to thank martin elvis and the `` new lisa '' science performance task force for discussions , marta volonteri for sharing her bh formation models , and the aspen centre for physics ( where this work was started ) for providing a very stimulating environment .
t. damour , i. i. kogan , a. papazoglou , phys .
rev . * d67 * , 064009 ( 2003 ) ; a. s. goldhaber , m. m. nieto , rev .
phys . *
82 * , 939 - 979 ( 2010 ) . m. v. bebronne , p. g. tinyakov , phys .
rev . * d76 * , 084011 ( 2007 ) ; d. bessada , o. d. miranda , class .
* 26 * , 045005 ( 2009 ) ; s. dubovsky , r. flauger , a. starobinsky , i. tkachev , phys .
rev . * d81 * , 023523 ( 2010 ) ; s. basilakos , m. plionis , m. e. s. alves , j. a. s. lima , phys .
rev . * d83 * , 103506 ( 2011 ) .
m. volonteri , f. haardt , p. madau , astrophys .
j. * 582 * , 559 - 573 ( 2003 ) . m. c. begelman , m. volonteri , m. j. rees , mon . not .
soc . * 370 * , 289 - 298 ( 2006 ) .
l. s. finn , p. j. sutton , phys .
rev . * d65 * , 044022 ( 2002 ) . c. talmadge , j. p. berthias , r. w. hellings , e. m. standish , phys .
lett . * 61 * , 1159 - 1162 ( 1988 ) .
s. r. choudhury , g. c. joshi , s. mahajan , b. h. j. mckellar , astropart .
* 21 * , 559 - 563 ( 2004 ) .
k. lee , f. a. jenet , r. h. price , n. wex , m. kramer , astrophys .
j. * 722 * , 1589 - 1597 ( 2010 ) . c. cutler , w. a. hiscock , s. l. larson , phys .
* d67 * , 024015 ( 2003 ) .
b. kocsis , z. haiman , k. menou , astrophys .
j. * 684 * , 870 - 887 ( 2008 ) . k. nakamura _ et al .
_ [ particle data group collaboration ] , j. phys .
g * g37 * , 075021 ( 2010 ) .
c. m. will , n. yunes , class .
* 21 * , 4367 ( 2004 ) .
e. berti , a. buonanno , c. m. will , phys .
rev . * d71 * , 084025 ( 2005 ) .
e. berti , a. buonanno , c. m. will , class .
* 22 * , s943-s954 ( 2005 ) . c. cutler , phys .
rev . * d57 * , 7089 - 7102 ( 1998 ) .
[ gr - qc/9703068 ] .
k. g. arun , c. m. will , class .
* 26 * , 155002 ( 2009 ) .
a. stavridis , c. m. will , phys .
rev . * d80 * , 044002 ( 2009 )
. k. yagi , t. tanaka , phys .
rev . * d81 * , 064008 ( 2010 ) .
d. keppel , p. ajith , phys .
rev . * d82 * , 122001 ( 2010 ) .
w. del pozzo , j. veitch , a. vecchio , phys .
rev . * d83 * , 082002 ( 2011 ) .
n. yunes , f. pretorius , phys .
* d80 * , 122003 ( 2009 ) .
n. cornish , l. sampson , n. yunes , f. pretorius , [ arxiv:1105.2088 [ gr - qc ] ] .
k. yagi and t. tanaka , prog .
phys . * 123 * , 1069 ( 2010 ) .
k. yagi , n. tanahashi and t. tanaka , phys .
d * 83 * , 084036 ( 2011 ) .
y. wang , d. e. holz , d. munshi , astrophys .
j. * 572 * , l15-l18 ( 2002 ) .
[ astro - ph/0204169 ] .
a. sesana , j. gair , e. berti , m. volonteri , phys .
rev . * d83 * , 044036 ( 2011 ) .
j. r. gair , a. sesana , e. berti , m. volonteri , class .
quant . grav .
* 28 * , 094018 ( 2011 ) .
e. berti , m. volonteri , astrophys . j. * 684 * , 822 - 828 ( 2008 ) . k. g. arun , s. babak , e. berti , n. cornish , c. cutler , j. gair , s. a. hughes , b. r. iyer _ et al .
_ , class .
* 26 * , 094027 ( 2009 ) .
s. e. timpano , l. j. rubbo , n. j. cornish , phys .
rev . * d73 * , 122001 ( 2006 ) . b. f. schutz , nature * 323 * , 310 - 311 ( 1986 ) . | space - based gravitational - wave ( gw ) detectors , such as lisa or a similar esa - led mission , will offer unique opportunities to test general relativity .
we study the bounds that space - based detectors could realistically place on the graviton compton wavelength @xmath0 by observing multiple inspiralling black hole ( bh ) binaries .
we show that while observations of _ individual _ inspirals will yield mean bounds @xmath1 km , the _ combined _ bound from observing @xmath2 events in a two - year mission is about ten times better : @xmath3 km ( @xmath4 ev ) .
the bound improves faster than the square root of the number of observed events , because typically a few sources provide constraints as much as three times better than the mean .
this result is only mildly dependent on details of bh formation and detector characteristics .
the bound achievable in practice should be one order of magnitude better than this figure ( and hence almost competitive with the static , model - dependent bounds from gravitational effects on cosmological scales ) , because our calculations ignore the merger / ringdown portion of the waveform .
the observation that an ensemble of events can sensibly improve the bounds that individual binaries set on @xmath5 applies to any theory whose deviations from general relativity are parametrized by a set of global parameters .
the formulation of gravitational theories with nonzero mass for the graviton that are consistent with cosmological observations is an important open problem .
attempts to construct such theories led to well - known conceptual difficulties , such as the so - called van dam - veltman - zakharov ( vdv - z ) discontinuity @xcite , due to the fact that the helicity-0 component of the graviton does not decouple from matter when the putative mass of the graviton @xmath6 . to circumvent pathologies related to the vdv - z discontinuity
, various versions of lorentz - violating massive graviton theories have been proposed in recent years @xcite .
massive graviton signatures in the cmb and possible constraints on @xmath7 from cosmological observations are an active area of research ( see e.g. @xcite ) .
in this paper we are interested in hypothetical massive graviton theories as `` straw men '' for alternative theories of gravity in which the propagation speed of gravity differs from that of electromagnetic waves , leading to a modified dispersion relation .
therefore we will adopt a phenomenological point of view and ask the following question : _ if _ the graviton mass were nonzero , what upper bounds on @xmath7 could we set by gravitational - wave ( gw ) observations of inspiralling compact binaries with future space - based detectors ? using @xmath0 , upper limits on the graviton mass @xmath7 ( in ev ) can be expressed as lower limits on its compton wavelength @xmath5 ( in km ) : _ g [ km]m_g[ev]= 1.2410 ^ -9 .
our analysis will show that , in hierarchical models of massive bh formation @xcite , the bound on @xmath5 from gw observations of a population of merger events is about one order of magnitude better than the mean bound from gw observations of individual mergers .
.[bounds ] graviton mass bounds . for proposed methods we quote the _
best _ achievable bounds . in the notation of the main text ,
a dagger @xmath8 denotes static bounds ; a number sign @xmath9 , dynamical bounds ; an asterisk @xmath10 , bounds that could be achieved by comparing gw and electromagnetic observations . [ cols="^,^,^,^",options="header " , ] to put our results in context , in table [ bounds ] we present a _
non - exhaustive _ summary of current and proposed bounds on @xmath5 that do not rely solely on gw observations .
these bounds can be roughly divided into three classes . * _ i ) static bounds . _ * if the graviton has nonzero mass , it is reasonable to expect that the newtonian gravitational potential will be modified to the yukawa form in the nonradiative near zone of any body of mass @xmath11 : @xmath12 .
talmadge _
et al . _
@xcite investigated deviations from kepler s third law for the inner planets of the solar system . by translating the uncertainties in these measurements in terms of accelerations of test bodies resulting from a yukawa potential , will @xcite found that the strongest bound on @xmath5 comes from the very nearly keplerian orbit of mars : @xmath13 km .
some bounds on @xmath7 quoted by the particle data group ( pdg ) @xcite similarly assume yukawa corrections to the newtonian potential in the weak - field limit .
the strongest bounds are naturally obtained from observations on the largest ( cosmological ) scales . as early as 1974 ,
goldhaber and nieto assumed that the graviton mass would produce a yukawa - type correction to the standard newtonian potential and argued that the evidence for bound clusters and tidal interactions between galaxies should imply a range for gravity at least as large as a few mpc , so that @xmath14 km @xcite .
however , the need to include dark matter to explain galactic rotation curves indicates that there are complications in the nature of gravity on those scales that are not necessarily well characterized by a yukawa - type potential .
an even stronger bound @xmath15 km comes from weak gravitational lensing , because no distortions are observed in the measured values of the variance of the power spectrum @xcite .
because of uncertainties in the amount and dynamics of dark matter in the universe ( and in the absence of a consistent massive graviton theory compatible with cosmology ) these bounds should be regarded as model - dependent and viewed with some caution .
* _ ii ) dynamical bounds .
_ * all bounds listed in part i ) are static , in the sense that they do not probe features related to the _ propagation _ of the gravitational interaction when @xmath16 .
the best available dynamical bounds come from the ( indirect ) observations of gws from binary pulsars , that are in excellent agreement with general relativity @xcite .
finn and sutton @xcite observed that the consistency of the orbital decay rates of binary pulsars psr b1534 + 12 and b1913 + 16 with general relativistic predictions yields a `` dynamical '' bound @xmath17 km .
a recent idea is to set bounds on @xmath7 using pulsar timing arrays @xcite .
the best bound achievable in the near future would be @xmath18 km , but this figure could worsen by an order of magnitude depending on the number of pulsars used for the test , timing accuracy and observation time . * _ iii ) comparisons of gravitational and electromagnetic observations . _ * if @xmath16 , the modified dispersion relation for gws would result in different arrival times of gws and electromagnetic waves emitted by the same astrophysical source . cutler _
et al . _
@xcite proposed to correlate electromagnetic observations and future space - based gw observations of white dwarf binaries .
the best bound that could be obtained in this way is @xmath19 km , but realistic bounds would probably be worse by about one order of magnitude .
kocsis _
et al . _
proposed to correlate lisa observations of gws from massive bh binary mergers with their possible electromagnetic counterparts @xcite .
an intrinsic limitation of this method is related to timing uncertainties in the gw burst , which are comparable to the dynamical timescale for the coalescing binary during merger , and can be estimated as the inverse of the orbital frequency at the innermost stable circular orbit . for binaries in the mass range @xmath20 at redshift @xmath21
, this uncertainty leads to a _
best _ bound @xmath22 km , worse than bounds coming from gw observations _ alone _ ( as we will see below ) .
systematic , model - dependent uncertainties in the electromagnetic counterpart will further weaken graviton mass bounds achievable in this way . |
here , we consider the following problem : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ given a prime @xmath2 , recover an unknown polynomial @xmath5 over a finite field @xmath6 of @xmath2 elements from several approximations to the values @xmath1 , computed at several points @xmath3 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ several problems of this type are related to the so - called _ hidden number problem _ introduced by boneh and venkatesan @xcite , and have already been studied intensively due to their cryptographic relevance , see the survey @xcite .
usually the evaluation points @xmath3 are chosen from the whole field @xmath6 .
however , here , motivated by the links and possible applications to the recently introduced himmo key distribution scheme @xcite , we concentrate on a very different case , which has never been discussed in the literature .
namely , in the settings relevant to himmo , the values of @xmath3 are taken from a short interval rather than from the whole field @xmath6 .
this case requires a careful adaptation of existing algorithms of @xcite and also establishing new number theoretic results about the frequency of small residues of polynomial values evaluated at a small argument , which are based on some ideas from @xcite .
we remark the polynomial recovery problem as studied in this paper arises in a collusion attack on a single node in the himmo system .
collusion attacks on the complete himmo system lead to recovery problems involving several polynomials reduced modulo several distinct unknown integers , which we believe to be much more difficult . for a prime @xmath2
we denote by @xmath6 the field of @xmath2 elements .
we identify the elements of @xmath6 with the set @xmath7 , so we can talk about their bits ( for example most and least significant bits ) and also approximations .
we consider the _ noisy polynomial interpolation problem _ of finding an unknown polynomial @xmath8,\ ] ] of degree @xmath9 from approximations to values of @xmath1 ( treated as elements of the set @xmath7 ) at polynomially many points @xmath10 selected uniformly at random .
more precisely , for integers @xmath11 and @xmath12 we denote by @xmath13 the distance between @xmath11 and the closest multiple of @xmath14 , that is , @xmath15 then , these approximations can , for example , be given as integers @xmath16 with @xmath17 for some `` precision '' @xmath18 for values of @xmath3 that are chosen uniformly at random from a certain set @xmath19 of `` samples '' . in cryptographic applications ,
the approximations are usually given by strings of @xmath11 most or least significant bits of @xmath1 , where , if necessary , some leading zeros are added to make sure that @xmath1 is represented by @xmath20 bits , where @xmath20 is the bit length of @xmath2 .
it is clear that giving the most significant bits is essentially equivalent to giving approximations of the type with an appropriate @xmath18 .
we now observe that giving the least significant bits can be recast as a question of this type as well . indeed , assume we are given some @xmath11-bit integer @xmath21 so that @xmath22 for some ( unknown ) integer @xmath23 .
we now have @xmath24 so , if @xmath25 is the multiplicative inverse of @xmath26 in @xmath6 then , setting @xmath27 , we obtain an inequality of the type with the polynomial @xmath28 instead of @xmath29 and @xmath30 .
so from now on we concentrate on the case where the approximations to @xmath1 are given in the form of .
the case of linear polynomials corresponds to the hidden number problem introduced by boneh and venkatesan @xcite .
the case of general polynomials , including sparse polynomials of very large degree , has been studied in @xcite ( see also @xcite for a survey of several other problems of similar types ) .
we note that if the polynomial @xmath5 belongs to some family @xmath31 of polynomials ( such as polynomials of degree @xmath32 or sparse polynomials with @xmath32 monomials ) , it is crucial for the algorithms of @xcite to have some uniformity results for every non - zero polynomial of @xmath31 on the values of @xmath3 chosen uniformly at random from @xmath19 .
we also note that a multiplicative analogue of this problem has been studied in @xcite .
we recall that in the settings of all previous works , typically the set @xmath19 is the whole field @xmath6 . in turn , this enables the use of such a powerful number theoretic technique as the method of exponential sums and thus the use of the weil bound ( see ( * ? ? ?
* theorem 5.38 ) ) for dense polynomials and the bound of cochrane , pinner and rosenhouse @xcite for sparse polynomials .
in particular , in @xcite a polynomial time algorithm is designed that works with very large values of @xmath18 , namely , up to @xmath33 , provided that @xmath5 is a sparse polynomial of degree @xmath34 and with @xmath35 monomials ( of known degrees ) , where @xmath36 is some absolute constant .
the analysis of the algorithm of @xcite is based on the method of exponential sums , the bound of @xcite and some ideas related to the waring problem .
we also note that if @xmath19 is an interval of length at least @xmath37 for some fixed @xmath38 , where , as before @xmath39 , then one can use the method of @xcite augmented with the bounds of exponential sums obtained via the method of vinogradov , see the striking results of wooley @xcite
. however , here we are interested in shorter intervals of the form @xmath40,\ ] ] which appear naturally in the construction of @xcite .
thus instead of the method of exponential sums , we use some ideas from @xcite .
we remark that for the noisy interpolation to succeed , we always assume that some low order the coefficients @xmath41 of @xmath5 are known , where @xmath42 is such that @xmath43 for some fixed @xmath38 .
clearly such a condition is necessary as if @xmath44 then the approximations of the type are not likely to be sufficient to distinguish between @xmath45 from approximations at random @xmath46 $ ] .
furthermore , it is clear that without loss of generality we can always assume that @xmath47 note that in the case when the test set @xmath19 is the whole field @xmath6 we only need to request that the constant coefficient @xmath48 of @xmath5 is known ( which is also , always assumed to be zero , see @xcite ) .
the above example , which shows the limits of _ interpolation _ of @xmath5 from the information given by , motivates the following question of _ approximation _ to @xmath5 , which is also more relevant to attacking himmo .
namely , instead of finding a polynomial @xmath5 we ask whether we can find a polynomial @xmath49 , such that @xmath1 and @xmath50 are close to each other for all @xmath46 $ ] so the approximations do not allow to distinguish between @xmath5 and @xmath49 . in the terminology of @xcite both @xmath5 and @xmath49 lead to the same keys and thus the attacker can use @xmath49 instead of @xmath5 .
finally , we note that the algorithmic problems considered in this paper have led us to some new problems which are of intrinsic number theoretic interest .
generally , our approach follows that of @xcite which in turn is based on the idea of boneh and venkatesan @xcite .
thus lattice algorithms , namely the algorithm for the closest vector problem , see section [ sec : lat cvp ] , and a link between this problem and polynomial approximations , see section [ sec : lap poly ] , play a crucial role in our approach .
however , the analysis of our algorithm requires very different tools compared with those used in @xcite . namely we need to establish some results about the frequency of small polynomial values at small arguments , which we derive in section [ sec : polyval ] , closely following some ideas from @xcite .
it also clear that there is a natural limit for such estimates on the frequency of small values as any polynomial with small coefficients takes small values at small arguments ( it is certainly easy to quantify the notion of `` small '' in this statement ) .
in fact our bound of lemma [ lem : nfij-1 ] is nontrivial up to exactly this limit .
rather unexpectently , in section [ sec : except poly ] we show that there are several other types of polynomials which satisfy this property : that is , they take small values at small arguments even if the coefficients are quite large ( certainly this may only happen beyond the range of the bound of lemma [ lem : nfij-1 ] which applies to all polynomials ) .
we call such polynomials _ exceptional_. then , after recalling in section [ sec : prep ] some result from analytic number theory , in section [ sec : coeff except ] we give some results describing the structure of the coefficients of exceptional polynomials . note the presence of exceptional polynomials makes noisy polynomial interpolation impossible for a wide class of instances ( which is actually good news for the security of himmo )
. so it is important to understand the structure and the frequency of exceptional polynomials .
the results of section [ sec : coeff except ] provide some partial progress toward this goal , yet many important questions are still widely open .
our main results are presented in sections [ sec : poly int ] and [ sec : poly app ] . in section [ sec : poly int ] we treat the noisy polynomial interpolation problem , where we actually want to recover the hidden polynomial @xmath5 .
however , from the cryptographic point of view it is enough to recover another polynomial @xmath51 which for all or most of small arguments takes values close to those taken by @xmath5 . here
the existence of exceptional polynomials becomes of primal importance .
we use the results of section [ sec : coeff except ] to obtain an upper bound on the number of such polynomials @xmath51 .
furthermore , in sections [ sec : flat poly ] and [ sec : oscill poly ] we also give two interesting explicit constructions of such exceptional polynomials , which we illustrate by some concrete numerical examples . in section [ sec : poly app ] we treat the approximate recovery problem and give a formula that predicts whether an approximate recovery is likely to be successful , depending on the number of randomly chosen observation points .
we compare its predictions with numerical experiments for parameter values suitable for himmo . throughout the paper , the implied constants in the symbols ` @xmath52 ' , ` @xmath53 ' and ` @xmath54 ' may occasionally , where obvious , depend on the degrees of the polynomials involved and on the real parameter @xmath55 and are absolute , otherwise .
we recall that the notations @xmath56 , @xmath57 and @xmath58 are all equivalent to the assertion that the inequality @xmath59 holds for some constant @xmath36 .
the letter @xmath2 always denotes a prime number , while the letters @xmath60 , @xmath42 , @xmath61 , @xmath14 and @xmath9 ( in lower and upper cases ) always denote positive integer numbers .
for a polynomial @xmath62\ ] ] and two intervals @xmath63 of @xmath64 and @xmath65 consecutive integers , respectively , we denote by @xmath66 the number of values @xmath67 for which @xmath68 ( where the elements of the intervals @xmath69 and @xmath70 are embedded into @xmath6 under the natural reduction modulo @xmath2 ) . for two intervals @xmath69 and @xmath70 of the same length ,
various bounds on @xmath66 are given in @xcite .
it is easy to see that the argument of the proof of ( * ? ? ?
* theorem 1 ) allows us to estimate @xmath66 for intervals of @xmath69 and @xmath70 of different lengths as well . to present this result , for positive integers @xmath42 , @xmath61 and @xmath64 ,
we denote by @xmath71 the number of solutions to the system of equations @xmath72 with @xmath73 next , we define @xmath74 as the smallest integer @xmath75 such that for @xmath76 there exists a constant @xmath77 depending only on @xmath42 and @xmath61 and such that @xmath78 holds as @xmath79 . presently , the strongest available upper bound @xmath80 is due to wooley @xcite , see also ( * ? ? ?
* theorem 1.1 ) and ( * ? ? ? * theorem 1.1 ) ; note that , in particular , these bounds also imply the existence of @xmath74 which is a very nontrivial fact .
it is now easy to see that the proof of ( * ? ? ?
* theorem 1 ) can be generalised to lead to the following bound .
[ lem : nfij-1 ] let @xmath81 $ ] be a polynomial of degree @xmath61 . then for intervals @xmath82 and @xmath83 with @xmath84 we have @xmath85 as @xmath86
note that kerr ( * ? ? ?
* theorem 3.1 ) gives a more general form of lemma [ lem : nfij-1 ] that applies to multivariate polynomials and also to congruences modulo a composite number .
we also note , that several other results from @xcite can be extended from the case @xmath87 to the general case as well , but this does not affect our main result .
clearly the bound of lemma [ lem : nfij-1 ] is nontrivial provided that @xmath88 and also @xmath89 for a fixed @xmath38 and a sufficiently large @xmath64 .
it is easy to see that this conditions can not be substantially improved as it is clear that if @xmath90 then @xmath91 for any polynomial @xmath81 $ ] .
also , if @xmath92 then @xmath91 for @xmath93 , @xmath94 and @xmath95 . the above example may suggest that unless the coefficients of @xmath96 are small , the value of @xmath97 can be nontrivially estimated even for @xmath98 .
we now show that this is false and in fact there are many other polynomials , which we call _ exceptional _ , which have the same property .
let us fix integers @xmath99 , @xmath100 , @xmath101 and @xmath102 .
we now consider a polynomial @xmath103,\ ] ] where @xmath104 , \qquad i = 0 , \ldots , \ell.\ ] ] in particular @xmath105 for some integers @xmath106 , \qquad i = 0 , \ldots , \ell.\ ] ] then for @xmath107 with an integer @xmath108 $ ] we have @xmath109 hence @xmath110 $ ] for such values of @xmath3 , which implies the inequality @xmath111 for the above example .
taking @xmath11 to be small , we see that @xmath96 has quite large coefficients , however no nontrivial bound on @xmath97 is possible for @xmath102 .
in fact , in sections [ sec : flat poly ] and [ sec : oscill poly ] we give examples of polynomials with large coefficients that remain small for all small values of the arguments , that is , polynomials for which @xmath91 , for some rather short intervals @xmath70 compared with the length of @xmath69 . in the above examples ,
the ratios of the coefficients and @xmath2 are chosen to be very close to rational numbers with small numerators and denominators .
alternatively , this can be written as the condition that the coefficients of @xmath96 satisfy @xmath112 for some small integers @xmath113 , @xmath114 .
below we present several results that demonstrate that this property indeed captures the class of exceptional polynomials @xmath96 with large values of @xmath97 adequately , at least in the qualitative sense .
first we need to recall some number theoretic results .
the following result is well - known and can be found , for example , in ( * ? ? ?
* chapter 1 , theorem 1 ) ( which is a more precise form of the celebrated erds turn inequality ) .
[ lem : et small int ] let @xmath115 be a sequence of @xmath64 points of the unit interval @xmath116 $ ] .
then for any integer @xmath117 , and an interval @xmath118 \subseteq [ 0,1]$ ] , we have @xmath119\ } - h(\beta - \alpha)\\ \ll \frac{h}{r } + & \sum_{r=1}^r \(\frac{1}{r } + \min\{\beta - \alpha , 1/r\}\ ) \left|\sum_{t=1}^h \exp(2 \pi i r \gamma_t)\right| . \end{split}\ ] ] to use lemma [ lem : et small int ] we also need an estimate on exponential sums with polynomials , which is essentially due to weyl , see ( * ? ? ?
* proposition 8.2 ) .
let @xmath120 denote the distance between a real @xmath121 and the closest integer ( which can be considered as a modulo 1 version of the notation @xmath122 ) .
[ lem : weyl ] let @xmath123 $ ] be a polynomial of degree @xmath124 with the leading coefficient @xmath125 . then @xmath126 finally , we need the following version of several similar and well known inequalities , see , for example , ( * ? ? ?
* section 8.2 ) .
[ lem : sum|| ] for any integers @xmath35 , @xmath127 , @xmath64 and @xmath128 with @xmath129 and a real @xmath130 with @xmath131 we have @xmath132 first we estimate the contribution from @xmath133 trivially as @xmath134 . for the remaining @xmath135
we note that @xmath136 as @xmath137 therefore , the contribution from @xmath138 can be estimated as @xmath139 since @xmath140 and @xmath141 this simplifies as @xmath142 which concludes the proof .
our first result uses in an essential way the argument of the proof of ( * ? ? ?
* theorem 3 ) , [ lem : nfij-2 ] let @xmath103\ ] ] be a polynomial of degree @xmath61 .
assume that for some @xmath143 and intervals @xmath93 and @xmath144 with @xmath84 we have @xmath145 then @xmath112 for some integers @xmath146 as we have mentioned , we follow the proof of ( * ? ? ?
* theorem 3 ) .
let @xmath147 .
since @xmath148 , there exist @xmath149 pairs @xmath150 such that @xmath151 lie in an interval @xmath152 of length @xmath153 and @xmath154 .
now , we consider the system of congruences @xmath155 the determinant @xmath127 of this system is the determinant of a vandermonde matrix : @xmath156 note that @xmath157 .
thus , we have that @xmath158 where @xmath159 and @xmath160 is the determinant of the matrix obtained from the vandermonde matrix after removing the @xmath161-th row and the @xmath162-th column , @xmath114 , @xmath163 .
it is easy to see that for each @xmath164 , the determinant @xmath160 is a polynomial in @xmath61 variables @xmath165 , @xmath166 , @xmath167 , of degree @xmath168 , which vanishes when @xmath169 for distinct @xmath20 and @xmath11 .
thus @xmath170 where @xmath171 is a polynomial ( that does not depend on @xmath96 or @xmath2 ) of degree @xmath172 .
therefore , we have @xmath173 this estimate and together imply the bound @xmath174 on the other hand , it is clear that @xmath175 then , comparing with and we conclude the proof
. we note that if @xmath11 and @xmath176 are fixed the expressions given by the example and those of lemma [ lem : nfij-2 ] differ by a factor @xmath177 and it is certainly natural to try to reduce this gap .
we now obtain yet another estimate on @xmath97 that depends only on the behaviour of the leading coefficient .
we recall the definition of @xmath74 used in lemma [ lem : nfij-1 ] . since for @xmath178 we can use the bound of lemma [ lem : nfij-1 ] , we assume that @xmath179 then the argument of the proof of ( * ? ? ?
* theorem 5 ) leads to the following result : [ lem : nfij-3 ] let @xmath103\ ] ] be a polynomial of degree @xmath61 .
assume that for some @xmath143 and intervals @xmath93 and @xmath144 with @xmath84 satisfying we have @xmath180 then either @xmath181 or @xmath182 for some integer @xmath183 let @xmath147 .
we apply lemma [ lem : et small int ] to the sequence of fractional parts @xmath184 , @xmath67 , with @xmath185 without loss of generality , we can assume that @xmath186 as otherwise the bound is trivial .
then we have @xmath117 and thus @xmath187 for @xmath188 , we derive @xmath189 therefore , by lemma [ lem : weyl ] , we have @xmath190 now , separating from the sum over @xmath191 the terms with @xmath192 , ( giving a total contribution @xmath193 ) , we obtain @xmath194 where @xmath195 hence , recalling the choice of @xmath196 , we derive @xmath197 the hlder inequality implies the bound @xmath198 collecting together the terms with the same value of @xmath199 and using the well - known bound on the divisor function , we conclude that @xmath200 let us set @xmath201 note that our assumption implies that
. @xmath202 by the dirichlet approximation theorem we can write @xmath203 for some integers @xmath35 and @xmath204 .
note that this immediately implies that @xmath205 .
hence @xmath206 for some integer @xmath207 with @xmath208 it now remains to estimate @xmath176 or @xmath127 .
we apply lemma [ lem : sum|| ] with @xmath209 , and we derive : @xmath210 which after the substitution in implies @xmath211 noting that the last term , up to the factor @xmath212 coincides with the second term , we now obtain @xmath213 hence we either have @xmath214 or @xmath215 which concludes the proof .
note that the dependence on @xmath176 in the bound on @xmath127 of lemma [ lem : nfij-3 ] is worse that that of lemma [ lem : nfij-2 ] .
however , since we are mostly interested in large values of @xmath176 , for example , the extreme case @xmath216 is of special interest , lemma [ lem : nfij-3 ] limits the possible values of @xmath217 as @xmath218 , which is stricter than the bound @xmath219 that follows from the estimates of lemma [ lem : nfij-2 ] on @xmath127 and @xmath220 .
as in @xcite , and then in @xcite , our results rely on some lattice algorithms .
we therefore review some relevant results and definitions , we refer to @xcite for more details and the general theory .
let @xmath221 be a set of @xmath11 linearly independent vectors in @xmath222 .
the set of vectors @xmath223 is called an @xmath11-dimensional full rank lattice .
the set @xmath224 is called a _ basis _ of @xmath225 .
the volume of the parallelepiped defined by the vectors @xmath226 is called _ the volume _ of the lattice and denoted by @xmath227 .
typically , lattice problems are easier when the euclidean norms of all basis vectors are close to @xmath228 .
one of the most fundamental problems in this area is the _
closest vector problem _ , cvp : given a basis of a lattice @xmath225 in @xmath229 and a target vector @xmath230 , find a lattice vector @xmath231 which minimizes the euclidean norm @xmath232 among all lattice vectors .
it is well know that cvp is * np*-hard when the dimension @xmath233 ( see @xcite for references ) .
however , its approximate version admits a deterministic polynomial time algorithm which goes back to the lattice basis reduction algorithm of lenstra , lenstra and lovsz @xcite , see also @xcite for possible improvements and further references .
we remark that all lattices that appear in this paper are of fixed dimension , so instead of using the above approximate cvp algorithms , we can simply use the following result of micciancio and voulgaris @xcite .
let @xmath234 denote the standard euclidean norm in @xmath222 .
[ lem : cvp ] for any fixed @xmath11 , there exists a deterministic algorithm which , for given a lattice @xmath225 generated by an integer vectors @xmath235 and a vector @xmath236 , in time polynomial in @xmath237 finds a lattice vector @xmath238 satisfying @xmath239 for @xmath240 , integer @xmath241 , we set @xmath242 and we denote by @xmath243 the @xmath244-dimensional lattice generated by the rows of the following @xmath245-matrix @xmath246 the following result is a generalization of several previous results of similar flavour , see @xcite .
[ lem : latt - pol ] let @xmath247 be fixed and let @xmath2 be a sufficiently large prime .
assume that some real numbers @xmath248 and integer numbers @xmath249 , @xmath250 satisfy @xmath251 then there is an integer @xmath32 , depending only on @xmath9 and @xmath252 , such that if @xmath253\ ] ] and @xmath254 $ ] are chosen uniformly and independently at random , then with probability at least @xmath255 the following holds .
for any vector @xmath256 with @xmath257 all vectors @xmath258 satisfying @xmath259 are of the form @xmath260 let @xmath261 denote the set of @xmath262 polynomials @xmath263\ ] ] with @xmath264 .
we see from lemma [ lem : nfij-1 ] ( taking into account that @xmath74 is a monotonically non - decreasing function of @xmath61 ) that for any polynomial @xmath265 the probability @xmath266 that @xmath267 for @xmath46 $ ] selected uniformly at random is @xmath268 provided that @xmath2 is large enough , where @xmath269 depends only on @xmath252 and @xmath9
. therefore , for any @xmath265 , @xmath270~:~\dmod{f(t_i)-g(t_i)}{p } > 2d\right ] = 1 - \rho\(g\)^d\ge 1 -p^{-d \eta } , \end{split}\ ] ] where the probability is taken over @xmath240 chosen uniformly and independently at random . since @xmath271 , we obtain @xmath272~:~\dmod{f(t_i)-g(t_i)}{p } >
2d\right]&\\ \ge 1 - p^{n -\eta d } > 1 & - p^{-1 } \end{split}\ ] ] for @xmath273 , provided that @xmath2 is large enough .
we now fix some integers @xmath274 with @xmath275 } \dmod{f(t_i)-g(t_i)}{p } > 2d.\ ] ] let @xmath276 be a lattice point satisfying . clearly , for a vector @xmath277 , there are some integer @xmath278 such that @xmath279 suppose that @xmath280 for some @xmath281 and let @xmath51 be given by
. in this case we have @xmath282 } \dmod{g(t_i ) - s_i}{p } \\ & \ge \max_{i \in [ 1,d ] } \ ( \dmod{f(t_i ) - g(t_i)}{p } - \dmod{s_i - f(t_i)}{p}\ ) > 2d - \delta \ge d \end{split}\ ] ] that contradicts .
now , if @xmath283 , @xmath281 , then for all @xmath284 we have @xmath285 since otherwise there is @xmath286 $ ] such that @xmath287 and thus @xmath288 , which contradicts again .
hence @xmath289 is of the desired form .
as we have seen , the condition holds with probability exceeding @xmath255 and the result follows .
we are ready to prove the main result .
we follow the same arguments as in the proof of ( * ? ? ?
* theorem 1 ) that have also been used for several other similar problems , see @xcite .
[ thm : polappr ] let @xmath290 be fixed and let @xmath2 be a sufficiently large prime .
assume that a real @xmath291 and integers @xmath292 , @xmath250 satisfy @xmath293 there exists an integer @xmath32 , depending only on @xmath9 and @xmath55 and a deterministic polynomial time algorithm @xmath294 such that for any polynomial @xmath253\ ] ] given integers @xmath254 $ ] and @xmath295 with @xmath257 its output satisfies @xmath296 } \left [ \ca\(t_1 , \ldots , t_d ; s_1 , \ldots , s_d \ )
= ( a_k , \ldots , a_n)\right ] \ge 1- 1/p\ ] ] if @xmath274 are chosen uniformly and independently at random from @xmath297 $ ] .
we refer to the first @xmath32 vectors in the matrix as @xmath2-vectors and we refer to the other @xmath298 vectors as power - vectors .
let us consider the vector @xmath299 where @xmath300 multiplying the @xmath161th power - vector of the matrix by @xmath301 and subtracting a certain multiple of the @xmath161th @xmath2-vector , @xmath281 , we obtain a lattice point @xmath302 and thus @xmath303 therefore , @xmath304 now we can use lemma [ lem : cvp ] to find in polynomial time time a lattice vector @xmath305 such that @xmath306 hence , we conclude from that @xmath307 applying lemma [ lem : latt - pol ] with @xmath308 and , say @xmath309 ) we see that @xmath310 with probability at least @xmath255 , and therefore the coefficients of @xmath5 can be recovered in polynomial time . as we have mentioned , it is impossible to recover a `` complete '' polynomial of degree @xmath9 from the approximations for @xmath46 $ ] for a small value of @xmath60 .
however , it is interesting to estimate the number of possible `` false '' candidates @xmath311 $ ] , which for many @xmath312 $ ] take values close to those of @xmath5 .
namely , for a polynomial @xmath313 $ ] , and real positive parameter @xmath314 we define @xmath315 as the set of polynomials @xmath311 $ ] with @xmath316 for at least @xmath317 values of @xmath46 $ ] . for example , if for some fixed @xmath55 we have @xmath318 the by lemma [ lem : nfij-1 ] we have @xmath319 unless @xmath320 .
so we now consider slightly smaller values of @xmath60 .
[ thm : polappr - imposible ] let @xmath290 be fixed and let @xmath2 be a sufficiently large prime . assume real @xmath291 and integers @xmath321 , @xmath322 satisfy @xmath323 for some fixed @xmath38 . then for @xmath324 we have @xmath325 for every @xmath326 , we apply lemma [ lem : nfij-3 ] to estimate the number of possible values of the leading coefficient of @xmath327 as @xmath328
. then we also use lemma [ lem : nfij-2 ] to estimate the number of possible values for the other coefficients of @xmath327 as @xmath329 which concludes the proof .
the case of @xmath330 , when the recovery of @xmath5 from the approximations with @xmath46 $ ] is impossible regardless of the number of queries and the complexity of algorithm is certainly of special interest . we denote @xmath331 .
under the conditions of @xmath60 and @xmath18 of theorem [ thm : polappr - imposible ] we have @xmath332 clearly if the coefficients of @xmath311 $ ] are sufficiently close to those of @xmath5 then @xmath333 .
this argument easily leads to the lower bound @xmath334 we now show that this is not necessary and in particular the set @xmath335 contains many polynomials that `` visually '' look very differently from @xmath5 .
let us take any integer @xmath127 and non - zero polynomial @xmath336 $ ] of degree @xmath337 and with coefficients in the range @xmath338 $ ] such that @xmath339 for examples if @xmath127 is a square - free integer , we can construct such a polynomial via the chinese remainder theorem from polynomials of the form @xmath340 $ ] for each prime divisor @xmath341 .
alternatively , given an integer @xmath342 , for any @xmath162 with @xmath343 one can consider take @xmath344 with an integer @xmath345 that is relatively prime to @xmath127 .
we note that the function @xmath346 , defined as the smallest @xmath162 with the property is called the _
smarandache function _ and has been extensively studied in the literature , see @xcite and references therein .
now , for any sufficiently small integers @xmath347 we define the coefficients @xmath348 from the congruences @xmath349 , @xmath350 , and consider the polynomial @xmath351.\ ] ] it is easy to see that the coefficients of @xmath352 can be rather large compared to @xmath2 .
furthermore , for every @xmath46 $ ] as the fractions @xmath353 , @xmath354 , take integer values of size @xmath355 , we have @xmath356 where @xmath357 varying the parameter @xmath127 , the polynomials @xmath358 and the integers @xmath347 , @xmath350 , ( and also introducing a small non - zero constant coefficient @xmath359 ) one can get a large family of such polynomials @xmath96 that remain `` flat '' on short intervals . in particular , one can choose @xmath127 to be a product of many small primes .
this argument can easily be made more precise with explicit constants , however we instead present an illustrative example of such a polynomial .
let us fix the prime @xmath360 we also set @xmath361 , @xmath362 and for @xmath363 , choose integers @xmath364 such that @xmath365 : @xmath366 and pick numbers @xmath367 randomly such that @xmath368 , for example : @xmath369 and let @xmath370 the plot on figure [ fig : flat ] , shows that @xmath371 ( where @xmath18 is indicated by the horizontal line ) for @xmath372 : versus @xmath373 for @xmath374 .
, scaledwidth=80.0% ] .
the plot in figure [ fig : flat2 ] shows that @xmath375 stays flat and close to zero for @xmath376 close to 0 and , after some transition period , appears to behave completely randomly for @xmath377 . versus @xmath373 for @xmath378 .
, scaledwidth=80.0% ] a different type of polynomial with large coefficients that is small in an interval can be constructed if @xmath379 .
thus this class is interesting only if the degree @xmath9 becomes a growing parameter , which is actually the case in the settings of himmo .
let @xmath380 $ ] be polynomial of degree @xmath9 that , after reduction modulo @xmath2 takes values in the interval @xmath381 $ ] for @xmath382 $ ] .
this means that @xmath383 for integer valued functions @xmath32 and @xmath384 , where for @xmath382 $ ] , @xmath385 $ ] .
let @xmath386 denote the discrete difference operator , that is , for any function @xmath387 , @xmath388 .
we have , since @xmath5 is a polynomial of degree @xmath9 : @xmath389 so @xmath390 . on the other hand , since @xmath391 $ ] for @xmath382 $ ] , it holds that @xmath392 $ ] for @xmath393 $ ] . for @xmath394
it follows that @xmath395 for all @xmath393 $ ] , so that after @xmath396-fold integration we have @xmath397 this means that @xmath32 and @xmath384 are both polynomials of degree at most @xmath9 on the interval @xmath297 $ ] . for @xmath398 there are more possibilities . for instance ,
if @xmath399,\ ] ] we obtain , by integrating @xmath400 defining @xmath401 and @xmath402 we have a decomposition @xmath403 , where , for integer @xmath376 , these three functions are integer valued . by choosing @xmath404 to be close to @xmath405 and , for @xmath406 , @xmath407 close to the residues of @xmath408 modulo @xmath2 , we obtain a polynomial @xmath5 of degree @xmath9 that , after reduction modulo @xmath2 , lies in @xmath381 $ ] for arguments in the interval @xmath297 $ ] . for the prime @xmath2 given by , with @xmath409 , and @xmath410 where @xmath411 denotes the closest integer ( and defined arbitrary in case of a tie )
, we obtain the graph in figure [ fig : doubleflat ] .
note that the left part of the graph in figure [ fig : doubleflat ] looks like a plot of two curves , while in fact this is one curve , which exhibits a highly oscillatory behaviour . versus @xmath373 . ] .
using lattice reduction and rounding techniques , we are able to find a polynomial @xmath412 satisfying @xmath413 for @xmath414 , the set of observation points .
the question is whether @xmath415 approximates @xmath1 also in all or at least some , non - observed points @xmath416\setminus\{t_1,\ldots , t_d\}$ ] .
the lattice used in this technique is spanned by the rows of @xmath417 where the first @xmath32 columns correspond to the evaluation of the polynomial @xmath412 in the points @xmath418 , and the last @xmath396 columns to its coefficients , scaled such that all coordinates of the wanted lattice point lie in an interval of length @xmath419 . in the remainder of this section we focus on the first @xmath32 columns , that is , the @xmath32-dimensional lattice @xmath225 of which the @xmath420 rows of the matrix @xmath421 are an overcomplete basis .
if @xmath422 , a straightforward basis transformation that corresponds to lagrange interpolation on @xmath423 eliminates @xmath396 rows and transforms this basis into @xmath424 from which it is clear that @xmath425 ( see section [ sec : lat cvp ] for the definition of @xmath227 ) .
we consider the case that @xmath394 , so that , as shown in section [ sec : oscill poly ] all functions @xmath412 for which @xmath426 for all @xmath416 $ ] satisfy @xmath427 for sufficiently small @xmath428 .
these differences correspond to the @xmath396-dimensional lattice @xmath429 in @xmath430 of which the rows of the matrix @xmath431 form a basis .
the volume of the lattice corresponding to these approximating polynomials is given by @xmath432 the determinant can be expanded as a sum of squares of vandermonde determinants , yielding @xmath433 where the set @xmath434 runs over all subsets of @xmath435 with @xmath396 elements .
the lattice technique returns a polynomial @xmath412 such that the point @xmath436 lies in the lattice @xmath225 .
if this pointis also a lattice point of @xmath429 , then @xmath412 approximates @xmath5 in all points of @xmath297 $ ] , and the attack succeeds .
so , from the attacker point of view , it is good if all lattice vectors of @xmath225 that do not lie in @xmath429 are long , that is , have the infinity norm greater than @xmath18 , since then those points lie outside the hypercube around the target vector .
we define the lattice @xmath437 , obtained by orthogonally projecting all points of @xmath225 onto the hyperplane through the origin that is orthogonal to the basis vectors of @xmath429 .
obviously it holds that @xmath438 if @xmath437 does not have short vectors , its volume is large and the attack succeeds . if its volume is small , @xmath437 has short vectors and , depending on the number of independent short vectors and their lengths , the attack is likely to fail .
this leads us to define the function @xmath439 } } \)^{1/(d - n-1 ) } \).\ ] ] it compares an esimate of the length of the shortest vector in @xmath437 , based on the minkowski bound , to @xmath18 , in an average sense .
the cross - over between @xmath437 having short vectors or not is at @xmath440 .
the factor @xmath441 in the definition of the function @xmath442 is included to account for the orientations of @xmath443 and of @xmath437 with respect to the coordinate axes , and the averaging is done in such a way that it can be explicitly evaluated if @xmath32 is much smaller that @xmath60 but is not too small : @xmath444 = \left ( \prod_{i=1}^n \frac{1}{i ! } \right)^2 \times \\ & \quad \frac{1}{(2h+1)^d } \sum_{t_1,\ldots , t_d =- h}^h \sum_{\substack{s\subseteq\{1,\ldots , d\ } \\ \ # s = n+1 } } \prod_{1\leq k < \ell \leq n+1 } ( t_{s(k ) } - t_{s(\ell)})^2 \\ & = \left ( \prod_{i=1}^n \frac{1}{i ! } \right)^2 \frac{\binom{d}{n+1}}{{(2h+1)}^{n+1 } } \sum_{t_1,\ldots , t_{n+1}=-h}^h \prod_{1\leq k < \ell \leq n+1 } ( t_k - t_\ell)^2 \\ & \approx \left ( \prod_{i=1}^n \frac{1}{i ! } \right)^2
\binom{d}{n+1 } ( 2h)^{n(n+1 ) } \\ & \qquad \idotsint\limits_{[0,1]^{n+1 } } \prod_{1\leq k<\ell\leq n+1 } ( x_k - x_\ell)^2 \,\mathrm{d}x_1\cdots\mathrm{d}x_{n+1}.\end{aligned}\ ] ] the integral of the square of the vandermonde determinant is equal to @xmath445 times the determinant of the @xmath446 hilbert matrix @xmath447 , which has been calculated by hilbert himself @xcite and is equal to @xmath448 we thus obtain @xmath449 we are mainly interested in parameter values corresponding to himmo , that is , when we have a polynomial of degree @xmath9 and @xmath450-bit keys , we choose @xmath2 to have precisely @xmath451 bits and let @xmath452 .
we are interested only in reconstruction in intervals of length at most @xmath453 . note that in this case @xmath442 becomes independent of @xmath2 . figure [ fig : reconstruction ] shows the reconstriction error @xmath454 versus @xmath455 for a randomly chosen polynomial @xmath5 of degree @xmath409 , @xmath456 , @xmath2 a random 112-bit prime and @xmath457 and @xmath458 , respectively .
the @xmath32 observation points are uniformly chosen in the interval @xmath459 . for @xmath457
, the error appears to be random for all @xmath376 , whereas for @xmath460 the error is close to @xmath405 for small @xmath376 . versus @xmath455 for @xmath409 , @xmath461 , @xmath2 a random 112-bit prime and @xmath457 and @xmath458 , respectively .
the @xmath32 observation points are uniformly chosen in the interval @xmath462.,title="fig : " ] versus @xmath455 for @xmath409 , @xmath461 , @xmath2 a random 112-bit prime and @xmath457 and @xmath458 , respectively .
the @xmath32 observation points are uniformly chosen in the interval @xmath462.,title="fig : " ] figure [ fig : graphs ] shows graph of @xmath463 as a function of @xmath32 , showing that @xmath442 becomes positive at @xmath460 .
this suggests that an approximate reconstruction is likely to be successful for @xmath460 , completely in line with our experimental results . as a function of @xmath32 for large @xmath2 . ] for larger degree @xmath464 , and @xmath361 , @xmath2 a @xmath465-bit prime , @xmath466 , the function @xmath442 becomes positive for @xmath467 , which may be too large for a successful lattice attack on the present computers .
but in a shorter interval , @xmath468 , the function @xmath442 becomes positive at @xmath469 , see figure [ fig : s69 - 75 ] . as a function of @xmath32 for large @xmath2.,scaledwidth=80.0%
] the experimental results in figures [ fig : reconstructionsshort69 ] , [ fig : reconstructionsshort72 ] and [ fig : reconstructionsshort75 ] show how the reconstruction improves as @xmath32 increases from @xmath470 to @xmath471 .
for @xmath472 , the reconstruction is close to zero only in the observation points ( note that the points near the horizontal line @xmath473 are also considered to be close to zero ) . for @xmath474 also in a sizeable fraction of the not - observed points .
finally for @xmath475 in nearly all points of an interval of length @xmath476 . with @xmath477 , @xmath478 , @xmath479 for @xmath480.,scaledwidth=80.0% ] with @xmath477 , @xmath478 , @xmath479 for @xmath481 . , scaledwidth=80.0% ] with @xmath477 , @xmath478 , @xmath479 for @xmath482.,scaledwidth=80.0% ]
finally , we note that for fairly small values of @xmath450 , for example , @xmath483 , @xmath442 is negative for all @xmath484 if @xmath9 is large enough , as shown for @xmath485 in figure [ fig : plotssmallb ] .
this suggests that approximate reconstruction in the full interval @xmath486 $ ] can not work , no matter how many observation points are used . as a function of @xmath32 . ]
clearly in practical implementations of our algorithm , using a certain approximate version of lemma [ lem : cvp ] is more natural . one can easily check that the algorithm and the result of theorem [ thm : polappr ] remain valid in this case as well . in the settings of himmo with @xmath450-bit keys and @xmath450-bit identifiers we have @xmath487 ,
@xmath488 and @xmath489 , see @xcite , so theorem [ thm : polappr ] does not apply .
this can be considered as an indirect confirmation of the strength of himmo : recovery of the complete polynomial is impossible . from studying as a function of @xmath32 for various values of @xmath450
, we can obtain indications if an approximate recovery in an interval @xmath297 $ ] is likely to succeed , also for shorter intervals , @xmath490 , that do not contain all possible identifiers .
we also recall that there is a different approach to the hidden number problem , due to akavia @xcite .
this approach does not use any lattice algorithms but rather examine the fourier coefficients of the `` hidden '' linear function .
however , the properties of these coefficients ( large values near the origin and a smooth decay away from the origin ) do not hold for non - linear polynomial functions , where all fourier coefficients are expected to be of about the same size . d. boneh and r. venkatesan , ` hardness of computing the most significant bits of secret keys in diffie
hellman and related schemes ' , _ lect .
notes in comp .
_ , springer - verlag , berlin , * 1109 * ( 1996 ) , 129142 .
chang , j. cilleruelo , m. z. garaev , j. hernndez , i. e. shparlinski and a. zumalacrregui , ` points on curves in small boxes and applications ' , _ preprint _ , 2011 ( available from http://arxiv.org/abs/1111.1543 ) .
o. garcia - morchon , r. rietman , l. tolhuizen , d. gmez - prez , j. gutierrez and s. merino del pozo , ` an ultra - lightweight id - based pairwise key establishment scheme aiming at full collusion resistance ' , _ eprint archive _ , report 618 , 2012 ( available from http://eprint.iacr.org/2012/618 ) .
i. e. shparlinski , ` playing `` hide - and - seek '' with numbers : the hidden number problem , lattices and exponential sums ' , _ proc .
symp . in appl .
soc . , providence , ri , * 62 * ( 2005 ) , 153177 . | motivated by a recently introduced himmo key distribution scheme , we consider a modification of the _ noisy polynomial interpolation problem _ of recovering an unknown polynomial @xmath0 $ ] from approximate values of the residues of @xmath1 modulo a prime @xmath2 at polynomially many points @xmath3 taken from a short interval .
= = = # 1#1 [ theorem]lemma [ theorem]claim [ theorem]corollary [ theorem]proposition [ theorem]open question # 1([#1 ] )
+ @xmath4 ) # 1#1 # 1#1 # 1#1 # 1 # 1#1_p # 1#1_m # 1#2|#1|_#2 # 1#1 # 1_m#1 # 1_p#1 # 1#1 # 1#1 |
it has been demonstrated that detection of at least one of the higgs bosons of the minimal supersymmetric standard model ( mssm ) is possible either at lep2 or at the lhc throughout all of the standard @xmath3 parameter space ( for a recent review , see ref .
@xcite ) . here
, we reconsider this issue in the context of the next - to - minimal supersymmetric standard model ( nmssm ) @xcite in which there is one higgs singlet superfield in addition to the two higgs doublet superfields of the mssm .
( the nmssm higgs sector is taken to be cp - conserving . )
we will demonstrate that there are regions of parameter space for which none of the nmssm higgs bosons can be detected at either lep2 or the lhc .
this result should be contrasted with the nlc no - lose theorem @xcite , according to which at least one of the cp - even higgs bosons ( @xmath4 ) for a cp - even ( cp - odd ) higgs boson . ] of the nmssm
will be observable in the @xmath5 production mode .
however , we do find that the parameter regions for which higgs boson observability is not possible at lep2 or the lhc represent a small percentage of the total possible parameter space .
many detection modes are involved in establishing the lhc no - lose theorem for the mssm .
a more than adequate set is : 1 ) @xmath5 at lep2 ; 2 ) @xmath6 at lep2 ; 3 ) @xmath7 at lhc ; 4 ) @xmath8 at lhc ; 5 ) @xmath9 at lhc ; 6 ) @xmath10 at lhc ; 7 ) @xmath11 at lhc .
additional lhc modes that have been considered include : a ) @xmath12 ; b ) @xmath13 ; c ) @xmath14 ; d ) @xmath15 .
because of the more complicated higgs self interactions , b ) and c ) can not be reliably computed in the nmssm without additional assumptions .
the higgs mass values for which mode a ) is kinematically allowed can be quite different than those relevant to the mssm and thus there are uncertainties in translating atlas and cms results for the mssm into the present more general context . finally , mode d ) is currently of very uncertain status and might turn out to be either more effective or less effective than current estimates .
thus , to be conservative , we excluded from our considerations any choice of nmssm parameters for which the modes a)-d ) might be relevant . even over this restricted region of parameter space , we shall demonstrate that nmssm parameter choices can be found such that there are no observable higgs signatures at either lep2 or the lhc .
in order to specify a point in nmssm parameter space , we have adopted the following procedure . *
employ a basis in which only the first neutral higgs field has a vev : @xmath16 . in this basis ,
the @xmath17 elements of the higgs mass - squared matrix ( denoted @xmath18 below ) take the simple form @xmath19 where @xmath20 appears in the superpotential in the term @xmath21 , @xmath22 , and @xmath23 are the radiative corrections ( which are independent of @xmath20 and @xmath24 , but depend on @xmath25 and @xmath26 we take @xmath27 ) .
we note that there are enough parameters in the nmssm model superpotential and soft - supersymmetry - breaking terms that the @xmath28 entries can have arbitrary values .
( specific planck scale boundary conditions could restrict these latter @xmath29 entries and thereby impose restrictions on the allowed parameter space beyond those described below ; such boundary conditions will not be imposed here . ) * pick a value for @xmath25 and a value for @xmath30 , where @xmath31 .
the crucial ingredient in limiting the scan is the upper limit of @xmath32 @xcite obtained by requiring that @xmath20 remain perturbative during evolution from scale @xmath33 to the planck scale . *
pick values for the angles @xmath34 , @xmath35 , and @xmath36 that appear in the matrix @xmath37 which diagonalizes the cp - even higgs mass - squared matrix via @xmath38 : @xmath39 where @xmath40 , and so forth .
it is useful to note that @xmath41 ^ 2-v_{11}^2 \mhi^2 \over 1 - v_{11}^2 } \label{mhiilim } \\ \mhiii^2&\leq & { [ \mhi^{\rm max}]^2-v_{11}^2 \mhi^2-v_{12}^2\mhii^2 \over 1-v_{11}^2-v_{12}^2 } \ , .
\label{mhiiivalue}\end{aligned}\ ] ] * pick a value @xmath42 , and compute @xmath43 the lower limit on @xmath20 is given by @xmath44\,,\ ] ] which is obtained by noting that @xmath45 . if @xmath46 then use @xmath47 .
it is consistent to consider only those @xmath48 values such that @xmath49 .
further restrictions are imposed on the @xmath50 as follows .
first , we require that @xmath51 , in which case the decays @xmath52 , @xmath53 and @xmath54 are all kinematically disallowed .
( if kinematically allowed , such decays are model dependent and could be dominant ; their experimental accessibility would have to be evaluated . )
second , we require that @xmath55 so that the decays @xmath56 are forbidden . * the cp - odd mass - squared matrix takes the form @xmath57 where the unspecified entries may take on any value given the parameter freedom of the model . for simplicity , we assume that only one cp - odd scalar , the @xmath4 ( which must have @xmath58 ) , is possibly light and that the other is heavy and , therefore , unobservable . in principle
, we could scan @xmath59 .
however , we impose three additional restrictions on @xmath60 as follows . in order to avoid the presence of the model - dependent
, possibly dominant @xmath61 decays , we restrict the scan to @xmath62 .
in particular , this implies that no @xmath60 scan is possible if @xmath63 .
we also impose the restrictions : @xmath64 , so that @xmath65 decays are forbidden ; and @xmath66 , which implies that the model - dependent decays @xmath67 are absent .
we emphasize that there may be parameter choices , for which no higgs bosons of the nmssm are observable , that lie outside the restricted portion of parameter space that we search .
our goal here is not to fully delineate all problematical parameter choices , but rather to demonstrate the existence of parameters for which it is guaranteed that no nmssm higgs boson can be found without increased lep2 energy and/or luminosity , or increased lhc luminosity or lhc detector improvements .
in order to assess the observability of modes 1)-7 ) we need the couplings of the @xmath68 and @xmath4 .
those required are : @xmath69 v_{1i}\label{zzh}\\ & & z\h_i \a : { g\over 2c_w } v_{2i}\label{zha}\\ & & t\anti t\h_i : { g\mt\over 2\mw } ( v_{1i}+v_{2i}\cotb ) \label{tth } \\ & & b\anti b\h_i : { g\mb\over 2\mw } ( v_{1i}-v_{2i}\tanb ) \label{bbh } \\ & & t\anti t\a , b\anti b\a : { g\mt\over 2\mw}\cotb,~~{g\mb\over 2\mw}\tanb \label{ttbba}\end{aligned}\ ] ] as already noted , we do not search parameter regions in which the very model - dependent higgs self - couplings would be needed . within the domain of parameter space that we search ,
we evaluate the potential of modes 1)-7 ) as follows . for the lep2 modes 1 ) and 2 ) ,
we require 30 and 50 events , respectively , for @xmath1 , before any cuts , branching ratios , or efficiency factors . for the lhc modes 3)-7 ) , we require @xmath70 statistical significance for @xmath2 .
the individual mode treatments are as follows .
* for the @xmath71 and @xmath72 modes , 3 ) and 4 ) , we compute the number of events as compared to predictions for the sm higgs boson , and then compute the resulting statistical significance assuming scaling proportional to the signal event rate . the most optimistic sm higgs statistical significances for the @xmath73 and @xmath74 channels as a function of higgs mass are those from cms @xcite , fig . 4 ( @xmath73 ) and fig . 8 ( @xmath75 ) , and tables 35 and 36 ( @xmath76 ) of ref .
we increase these @xmath77 statistical significances by a factor of @xmath78 for @xmath2 and then apply the nmssm corrections .
* for the @xmath79 detection mode 5 ) we employ the @xmath2 contours , fig . 76 , of ref . @xcite .
we note that when @xmath79 is kinematically allowed , the @xmath80 decays are forbidden for the @xmath81 values we consider here .
thus , the @xmath82 decays are exactly as in the mssm and the mssm results can be employed ` as is ' when the @xmath70 contour is specified as a function of @xmath83 and @xmath25 .
* for the @xmath84 and @xmath85 final states we refer to the @xmath77 statistical significances quoted for the mssm model @xmath86 process at @xmath87 in table 34 of ref .
@xcite and the input @xmath88 from fig . 22 ( @xmath87 results ) of ref . @xcite . from these results
we compute a standard statistical significance for @xmath89 , @xmath90 , and @xmath2 .
statistical significances in the nmssm model are obtained for the @xmath91 and @xmath4 by multiplying these standard statistical significances by the appropriate @xmath92 enhancement factor or by @xmath93 and by the computed @xmath94 branching ratio of the higgs boson in question .
recall that we do not search parameter regions for which the @xmath94 branching ratios would be uncertain due to higgs pair decay channels being kinematically allowed .
* finally , we assume that mode 7 ) is only relevant for the @xmath4 ( as in the mssm ) .
however , we can not directly use the discovery region shown for @xmath95 in fig .
53 of ref .
@xcite since @xmath96 decays deplete the @xmath94 branching ratio for @xmath97 .
thus , we use an optimistic limit for this mode s @xmath2 region of viability ; @xmath98 is assumed to be achieved in this mode for @xmath99 if @xmath100 .
if none of the higgs bosons @xmath68 , @xmath4 or @xmath101 are observable as defined above we declare a parameter point in our search to be a `` point of unobservability '' or a `` bad point '' .
we now summarize our results . we find that if @xmath102 then all parameter points that are included in our search are observable for @xmath81 values up to the maximum allowed ( @xmath103 for @xmath32 , after including radiative corrections ) . for such low @xmath25 ,
the lhc @xmath73 and @xmath74 modes allow detection if lep2 does not . for high @xmath104
, the parameter regions where points of unobservability are found are also of very limited extent , disappearing as the @xmath105 and/or @xmath85 lhc modes allow detection where lep2 does not .
however , significant portions of searched parameter space contain points of unobservability for moderate @xmath25 values . that such @xmath25 values should be the most ` dangerous ' can be anticipated from the mssm results .
it is well - known ( see , for example , ref .
@xcite ) that there is a wedge of mssm parameter space at moderate @xmath25 and with @xmath106 and @xmath107 masses above about @xmath108 for which the only observable mssm higgs boson is the light sm - like @xmath109 , and that it can only be seen in the @xmath73 mode at the lhc ( @xmath110 at lep2 ) . by choosing @xmath81 and @xmath60 in the nmssm so that @xmath111 and @xmath112 are close to or above the @xmath113 of lep2 , then , by analogy , at moderate @xmath25 we would need to rely on the @xmath114 modes
. however , in the nmssm , parameter choices are possible for which all the @xmath115 couplings are reduced relative to sm strength .
this reduction will suppress the @xmath73 couplings coming from the @xmath116-boson loop .
all the @xmath71 widths can be sufficiently smaller than the somewhat enhanced @xmath117 widths so that the @xmath73 branching ratios are _ all _ no longer of useful size . to illustrate , we shall discuss results for @xmath118 , @xmath119 and @xmath87 ( for which @xmath120 , @xmath121 and @xmath122 , respectively ) and @xmath123 . * in fig .
[ tanb5 ] , we display for @xmath119 both the portions of @xmath124 parameter space that satisfy our search restrictions , and the regions ( termed `` regions of unobservability '' ) within the searched parameter space such that , for _ some _ choice of the remaining parameters ( @xmath20 and @xmath60 ) , no higgs boson will be detected using any of the techniques discussed earlier .
value such that there is a choice of @xmath20 and @xmath60 for which no higgs boson is observable , there are generally other choices of @xmath20 and @xmath60 for which at least one higgs boson _ is _ observable .
] relatively large regions of unobservability within the searched parameter space are present .
* at @xmath118 , a similar picture emerges .
the search region that satisfies our criteria is nearly the same ; the regions of unobservability lie mostly within those found for @xmath119 , and are about 50% smaller . * for @xmath87 ,
the regions of unobservability comprise only a very small portion of those found for @xmath119 .
this reduction is due to the increased @xmath117 couplings of the @xmath91 and @xmath4 , which imply increased @xmath125 production cross sections .
as these cross sections become large , detection of at least one of the @xmath126 in the @xmath127 final state becomes increasingly difficult to avoid .
for values of @xmath104 , depends sensitively on @xmath81 .
] we find that one or more of the @xmath126 should be observable regardless of location in @xmath128 parameter space ( within the somewhat restricted search region that we explore ) .
another perspective on the parameter space and the location of points of unobservability is provided in fig .
[ v11mh2mh3 ] .
there , we display for @xmath119 and @xmath123 the regions searched in the @xmath129 , @xmath130 and @xmath131 parameter spaces , and the portion thereof in which the remaining model parameters can be chosen such that no higgs boson is observable .
the @xmath129 plot shows that higgs boson unobservability is possible for any value of @xmath132 and for all values of @xmath133 up to the bound of eq .
( [ mhiilim ] ) , so long as @xmath134 . for
@xmath135 , the region of @xmath133 for which higgs boson unobservability is possible does not include the highest @xmath133 values .
the @xmath130 plot shows that unobservability is possible only if @xmath136 , the @xmath137 coupling is reduced relative to sm strength by @xmath138 , implying that @xmath139 is difficult to detect in the @xmath140 mode .
the @xmath141 plot shows that unobservability is possible for almost all @xmath142 values so long as @xmath143 . for @xmath144 ,
the @xmath145 must be detected in the relatively weak @xmath146 modes ; both are typically somewhat suppressed at moderate ( or large ) @xmath25 by a @xmath147 coupling that is smaller than sm - strength and by an enhanced @xmath117 decay width that diminishes the @xmath148 branching fractions . throughout the regions displayed in fig .
[ v11mh2mh3 ] where choices for the remaining model parameters can make observation of any of the higgs bosons impossible , there are other choices for the remaining parameters such that at least one higgs boson _ is _ observable .
the mass @xmath123 is typical of the ` intermediate ' values that yield the largest regions of unobservability . if @xmath149 , then discovery of one of the @xmath91 at lep2 is almost certain .
as @xmath150 , then discovery of at least one higgs boson at the lhc becomes possible over most of parameter space , as we now describe .
as @xmath150 , @xmath151 .
, then eqs . ( [ mhiilim ] ) and ( [ mhiiivalue ] ) imply that @xmath152 as @xmath153 . in this limit
we have @xmath154 by orthogonality of @xmath37 .
unless @xmath155 , there is an inconsistency which can only be avoided by simultaneously taking @xmath151 .
] since @xmath156 , this means either @xmath157 or @xmath158 .
however , only if @xmath158 can all the higgs bosons be unobservable . if @xmath159 is not near 0 , @xmath160 must be , in which case @xmath161 and @xmath162 and the @xmath163 has completely sm - like couplings [ see eqs .
( [ zzh])-([ttbba ] ) ] , and for @xmath164 ( @xmath165 at @xmath119 ) @xmath163 will be detectable in the @xmath73 final state . if @xmath158 , then any value of @xmath160 is possible , but ( again ) @xmath157 would make @xmath163 sm - like and observable ; in addition , @xmath166 ( @xmath167 ) yields @xmath168 and @xmath169 implying that @xmath145 would be sm - like and observable ( in the @xmath73 or @xmath170 modes ) . thus , the only ` dangerous ' region is @xmath158 and @xmath171 , for which , eq . ( [ mhiilim ] ) implies @xmath172 so that both @xmath173 and @xmath174 would have to be found in the @xmath73 mode .
channel , the resolution is such that extreme degeneracy , @xmath175 , is required before we must combine signals . ]
if the value of @xmath176 is such that neither @xmath177 nor @xmath178 is small , then both @xmath179 and @xmath180 can be substantial , and the @xmath73 mode can be suppressed for both @xmath181 and @xmath182 by a combination of @xmath183 coupling suppression ( to diminish @xmath184 production ) and @xmath185 coupling enhancement ( as natural for moderate or large @xmath25 ) . the latter enhances the @xmath117 partial width and diminishes the @xmath186 branching ratio .
the moderate @xmath187 value makes it possible to have the required @xmath185 coupling enhancement without it being so large as to make the @xmath188 mode observable in @xmath185 production .
it is useful to present details on what goes wrong at a typical point of unobservability . for @xmath119 and @xmath123 , no higgs boson
can be observed for @xmath189 if @xmath190 , @xmath191 , @xmath192 , and @xmath193 ( for which @xmath194 , @xmath195 , @xmath196 , and @xmath197 ) .
the corresponding @xmath37 matrix entries are : @xmath198 from the @xmath199 , and the value of @xmath25 , we compute ( relative to the sm values ) @xmath200 where @xmath201 or @xmath202 .
note that @xmath139 has very small couplings to @xmath203 .
the manner in which this point escapes discovery is now apparent .
first , the minimum values required for the @xmath92 values for @xmath91 observability in the @xmath94 mode are : 53 ( @xmath204 ) ; 32 ( @xmath205 ) ; 35 ( @xmath206 ) .
the actual values all lie below those required .
observation of the @xmath4 at @xmath189 ( without adding in the much smaller overlapping @xmath163 signal ) would require @xmath207 .
regarding the other discovery modes , @xmath163 and @xmath145 are both in the mass range for which the @xmath73 mode is potentially viable and the @xmath139 is potentially detectable in the @xmath140 channel .
however , the suppressed @xmath208 couplings imply smallish @xmath209 production rates for @xmath68 .
relative to a sm higgs of the same mass we have : @xmath210 ( note that these strengths are not simply the @xmath211 magnitudes due to enhanced @xmath212-quark loop contributions which interfere with the @xmath213-quark loop contributions at amplitude level . )
further , the enhanced higgs decay rate to @xmath214 and the reduced @xmath116-loop contributions to the @xmath73 coupling suppress the @xmath73 branching ratios of @xmath163 and @xmath145 relative to sm expectations .
we find : @xmath215 suppression sufficient to make @xmath163 and @xmath145 invisible in the @xmath73 mode . the suppressed @xmath137 coupling and the enhanced @xmath216 decays are sufficient to suppress @xmath217 much below sm expectations : @xmath218 such that the @xmath74 signal has a significance of only @xmath219 , even though a sm higgs of this mass would yield a @xmath220 signal . in short ,
there is enough flexibility due to the addition of the singlet higgs field ( which has no couplings to sm fermions and vector bosons ! ) for _ all _ the higgs bosons to escape detection for certain choices of model parameters , provided @xmath25 is moderate in size .
moderate @xmath25 implies that @xmath186 decays for light higgs are suppressed , while at the same time @xmath84 production is not adequately enhanced for detection of the @xmath221 mode .
the regions of nmssm parameter space where no higgs boson can be detected will expand if full @xmath2 ( @xmath1 ) luminosity is not available at the lhc ( lep2 ) or efficiencies are smaller than anticipated .
conversely , these `` regions of unobservability '' could decrease substantially ( perhaps disappear ) with improved efficiency ( due to the expanded calorimeter option discussed in ref .
@xcite ) in the @xmath222 final state or higher luminosity . these issues will be pursued elsewhere .
we have explicitly neglected supersymmetric ( susy ) decay modes of the higgs bosons in our treatment . if these decays are important , the regions of unobservability found without using the susy final states will increase in size .
however , higgs masses in the regions of unobservability are typically modest in size ( @xmath223 ) , and as susy mass limits increase with lep2 running this additional concern will become less relevant . of course , if susy decays are significant , detection of the higgs bosons in the susy modes might be possible , in which case the regions of unobservability might decrease in size .
assessment of this issue is dependent upon a specific model for soft susy breaking and will not be pursued here . finally ,
although we can not establish a no - lose theorem for the nmssm higgs bosons at lep2 and the lhc ( in contrast to the no - lose theorems applicable to the nlc higgs search with @xmath224 ) , the regions of complete higgs boson unobservability appear to constitute a small fraction of the total model parameter space .
it would be interesting to see whether or not these regions of unobservability correspond to unnatural choices for the planck scale supersymmetry - breaking parameters .
gunion , a. stange , and s. willenbrock , `` weakly - coupled higgs bosons '' , in _ electroweak symmetry breaking and new physics at the tev scale _ , edited by t.l .
barklow , s. dawson , h.e .
haber , and j.l .
siegrist ( world scientific , singapore , 1996 ) . | we demonstrate that there are regions of parameter space in the next - to - minimal ( two - higgs - doublet , one - higgs - singlet superfield ) supersymmetric extension of the sm for which none of the higgs bosons are observable either at lep2 with @xmath0 and an integrated luminosity of @xmath1 or at the lhc with @xmath2
. # 1 # 2 # 3 mod .
phys .
lett . *
# 1 * , # 2 ( # 3 ) # 1 # 2 # 3 nucl .
phys . *
# 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys .
lett . *
# 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys . rep .
* # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys .
rev .
* # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys . rev .
lett . *
# 1 * , # 2 ( # 3 ) # 1 # 2 # 3 rev .
mod .
phys . *
# 1 * , # 2 ( # 3 ) # 1 # 2 # 3 z. phys .
* # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 int .
j. mod .
phys . *
# 1 * , # 2 ( # 3 ) versim#1#2 /#1#1 - 6pt/ |
there have been remarkable progresses in particle physics in the last century .
new discoveries have changed our view on elementary particles and cosmology . in our current standpoint ,
goals of the elementary particle physics are not restricted to finding fundamental constitutes of matter and fundamental interactions .
we have to address questions concerning the vacuum state of the universe and the structure of space and time to understand the law of elementary particles .
these issues are also closely related to questions like how the universe have begun and evolved to today .
current understanding of particle physics is based on the standard model ( sm ) , which is a gauge theory of quarks and leptons .
the gauge groups are su(3 ) @xmath1 su(2 ) @xmath1 u(1 ) : the su(3 ) group describes the strong interaction , and the su(2 ) and u(1 ) groups correspond to the electromagnetic and the weak interactions .
the sm has been tested experimentally for the last three decades , and especially the gauge interaction among fermions and bosons has been studied very precisely at lep and slc experiments in 1990 s .
the gauge structure of the sm is now well established experimentally .
the next important steps of the particle physics are : * to establish the mechanism of the electroweak symmetry breaking and the particles mass generation . * to find the physics scenario beyond the electroweak scale .
i would like to discuss how @xmath0 linear collider ( lc ) experiments can play essential roles for these purposes .
studies of research potentials at an @xmath0 lc have been carried out in world - wide , and its outcome is found in literatures , especially in recent reports @xcite .
higgs fields are introduced to break electroweak symmetry and generate masses for elementary particles .
although only one higgs doublet field is necessary for this purpose , little is known about the structure of the higgs sector .
we do not know how many higgs fields exists , whether there are fields other than weak doublet fields , and what is the mass of the higgs particle .
we have little information on couplings between the higgs field and fermions / gauge bosons .
we need the lc experiment to answer these questions .
the higgs boson mass is the most important parameter in the higgs sector . since particle mass is generated from interaction with the higgs field in the sm , the mass of the particle carries information on the strength of its interaction to the higgs field
this is also true for the higgs boson mass itself .
in particular , the higgs - boson mass in the sm is expressed as @xmath2 , where @xmath3 is the higgs self - coupling constant ( @xmath4 ) . in general
, a light higgs boson corresponds to a weakly - coupled scenario for the mechanism of the electroweak symmetry breaking .
supersymmetry ( susy ) , grand unified theory , and string unification are such examples . on the other hand ,
a heavy higgs boson is a signal that some strong dynamics is behind the symmetry breaking .
the current experimental lower bond on the higgs boson mass in the standard model is 114 gev from lep experiments at 95 % cl . on the other hand , from global analysis of the sm
, we can derive the 95 % cl upper bound as 193 gev @xcite .
although this bound is valid for the sm higgs boson , a weakly - coupled scenario is favored for the dynamics of the electroweak symmetry breaking .
if we introduce some theoretical assumption , a possible mass range of the higgs boson can be derived .
for instance , if the sm is assumed to be valid without any change , except for a possible see - saw mechanism for neutrino mass generation , the upper and lower bounds of the higgs boson mass are determined as a function of the cut - off scale @xmath5 of the theory . for @xmath6 gev of the planck scale ,
the lower and upper bounds are about 130 gev and 180 gev , respectively .
the higgs sector of the minimal supersymmetric standard model ( mssm ) consists of two higgs doublets .
a remarkable feature of this model is that the upper bound of the lightest cp - even higgs boson can be derived without reference to the cutoff scale . for reasonable choice of parameters in the model ,
especially the stop mass and mixing parameters , the bound is given by 130 gev @xcite .
for other models like general two higgs doublets and susy models with extended higgs sectors , we can derive the mass range of a higgs boson whose properties are similar to the sm higgs boson , if the cutoff scale is specified @xcite .
the mass range is shown in figure 1 for various models in the case of @xmath6 gev .
we can see that at least one higgs boson must exist near or less that 200 gev under this theoretical assumption .
the higgs boson search is currently undertaken at fermilab tevatron . with an integrated luminosity of 15 fb@xmath7 , it will become evident whether or not a higgs boson exists for a relatively small mass range like 120 gev .
one of main goals of the lhc experiment , which is scheduled to start in 2007 , is the discovery of a higgs boson , irrespectively of its mass .
the higgs boson discovery is relatively easy if its mass exceeds 200 gev , because the @xmath8 mode is available . for a smaller higgs boson mass ,
the higgs boson search involves other modes such as two photon and @xmath9 decay modes , and @xmath10 production .
it is believed that the discovery of a sm higgs boson is possible at lhc experiments with a low integrated luminosity of an order of 10 fb@xmath7 .
in addition , recent studies of prospects of the higgs physics at lhc show that useful information on higgs coupling constants will be obtained in the era of a high luminosity run @xmath11 , especially for a relatively light higgs boson .
higgs boson production processes through a vector boson fusion play an essential role for this purpose @xcite .
goals of higgs physics at lcs are to establish the mass - generation mechanism for elementary particles and to clarify the dynamics of the electroweak symmetry breaking . with an integrated luminosity of 500 fb@xmath7 ,
over 100,000 light higgs bosons can be produced at a lc , and precise information on various couplings related to the higgs boson will be obtained from measurements of production cross section and decay branching ratios . the lc is a higgs factory in this sense .
* production of higgs bosons * for a lc with center - of - mass energy of up to 500 gev , we can detect the sm higgs boson up to 400 gev .
the discovery of a higgs boson only requires an integrated luminosity of a few fb@xmath7 , if the higgs boson mass is less than 200 gev .
if we consider models beyond the sm , the production cross section and decay modes may be different from the sm .
for example , the production cross section of the lightest cp - even higgs boson can be reduced for the mssm and the next - to - minimal supersymmetric standard model ( nmssm ) , where an additional gauge - singlet higgs field is introduced to the mssm .
however we can show quantitatively that at least one of higgs bosons has a production cross section in the @xmath12 precess which is smaller only by a factor of two or three than that of the sm higgs boson .
therefore , discovery of a higgs boson is guaranteed in these models @xcite . *
spin and parity of the higgs boson * in order to confirm that the observed particle has right properties for the higgs boson , we first have to determine its spin and parity .
the lc experiment is necessary for unambiguous determination of these basic quantum numbers .
we can distinguish the cp property of a spin-0 particle from the angular distribution of the production angle .
furthermore , an energy scan at the production threshold region provides enough information to discriminate the spin and parity of the observed particles @xcite .
an integrated luminosity of several ten s fb@xmath7 are necessary for these determinations
. * mass - generation mechanism * in the sm , all quarks , leptons and gauge bosons are massless , if there is no electroweak symmetry breaking .
masses of these particles are generated through interactions with the higgs field , which is supposed to condense in the vacuum .
since the higgs boson is a physical fluctuation mode around the vacuum expectation value , we need to determine the higgs - fermion , higgs - w - w , higgs - z - z coupling constants , and compare them with values predicted from the masses of elementary particles , in order to test the mass generation mechanism . from the production cross sections of @xmath13 and @xmath14 processes and higgs decay branching ratios , these coupling constants can be determined precisely . for a higgs boson mass of less than 140 gev , the @xmath15 , @xmath16 , @xmath17 , @xmath18
coupling constants can be determined at accuracy of a few % level with an integrated luminosity of 500 fb@xmath7 at @xmath19 gev @xcite .
we can also derive the total width of the higgs boson with an error of 5 - 10 % .
the top yukawa coupling constant can be determined from @xmath20 production process , and the center - of - mass energy larger than 700 gev is preferable for this purpose . in figure 2 ,
the mass and coupling constant relation is plotted for various particles . in general , the mass and coupling constant have a relation such as @xmath21 in the sm , where @xmath22 is the vacuum expectation value of the higgs field , and the @xmath23 is a dimensionless coupling constant which can be obtained from a three - point coupling measurement . in this figure , the expected statistical error of each coupling measurement at lc experiments
is shown for charm , @xmath24 , bottom , w , z , and top quark . for the higgs boson ,
the error is derived from the measurement of the triple higgs - boson coupling constant .
* higgs self - coupling constant * the first information about the higgs potential will be obtained from the measurement of the triple higgs - boson coupling constant . for this purpose
, we need to study double higgs production processes , namely , @xmath25 and @xmath26 . in both of these processes ,
in addition to the diagram involving the triple higgs - boson vertex , there are diagrams which only depend on vertexes with gauge bosons and higgs boson(s ) .
the production cross section is not simply proportional to the square of the self - coupling constant . for @xmath27 gev
, the @xmath25 has much larger cross section than the @xmath26 process . for a higher energy , the production cross section of the latter process increases , which is an important future of the ww fusion process , whereas that for the former process decreases as @xmath28 . for @xmath29 1 tev ,
the ww fusion process become more important for determination of the triple higgs - boson coupling constant .
it is expected that the sensitivity of the coupling measurement is about 20 ( 7 ) % for 1 ( 5 ) ab@xmath7 at @xmath30 500 ( 3000 ) gev @xcite . the center - of mass energy dependence of the estimated sensitivity on the coupling constant
is shown in figure 3 , assuming 100 % efficiency for 1 ab@xmath7 @xcite . for @xmath29 1 tev ,
the sensitivity is further improved by using initial electron ( and positron ) polarization , because the production cross section of the ww fusion process is increased by a factor of 2 ( 4 ) with the electron ( electron and positron ) polarization .
the study of the higgs sector could also provide important information concerning possible new physics scenario .
in particular , the mssm has several characteristic features in the higgs sector .
first , the higgs sector is a special type of a two higgs doublet model .
the physical higgs modes contain two cp - even @xmath31 , one cp - odd @xmath32 , and one pair of charged higgs bosons @xmath33 .
there is a theoretical upper bound on the lighter cp - even higgs boson @xmath34 , which is about 130 gev .
since the form of the higgs potential is restricted by the requirement of susy , the higgs sector is parametrized by two parameters at the tree level , usually taken to be the cp - odd higgs boson mass @xmath35 and the ratio of the two vacuum expectation values @xmath36 .
although the radiative correction to the higgs potential introduces additional parameters like the stop mass , higgs boson masses and couplings are very restrictive , and these features are useful in discriminating the mssm from other models .
if only one of cp - even higgs bosons is found at an earlier stage of the @xmath0 lc experiment and the lhc experiment , measurements of the coupling constants related to the higgs boson can provide useful information on model parameters .
for instance , the ratios of the branching ratios such as @xmath37 , @xmath38 can is deviated from the sm prediction , if the heavy higgs boson is not too heavy @xcite .
indirect information on the heavy higgs boson mass may be obtained for @xmath39 gev , as shown in figure 4 . on the other hand
, @xmath40 should be the same at the tree level in the mssm as the sm prediction , but the deviation can be large due to susy loop corrections to the bottom yukawa coupling constant , especially for a large @xmath41 @xcite .
if the center - of mass energy is large enough , heavy higgs bosons can be produced directly .
contours of constant production cross sections ( 0.1 fb@xmath7 ) for various processes in the case of @xmath30 500 and 1500 gev are shown in the parameter space of @xmath42 and @xmath41 for the mssm in figure 5 @xcite .
the discovery limit is essentially determined by the @xmath43-@xmath44 and @xmath45-@xmath46 pair production threshold except for limited regions of large and small @xmath41 .
however , heavy higgs bosons can be seen in different modes depending on parameter regions .
this will be useful to test the mssm and to determine parameters of the model , especially @xmath41 .
the photon - photon collider option of the lc can be especially important for the heavy higgs searches .
this is because that the s - channel single higgs production is possible through the photon - photon - higgs boson coupling induced at one - loop level .
since the energy of the back - scattered photon can be 80 % of the original electron beam , the heavy higgs - boson mass reach can increase up to about 400 gev for the 500 gev collider .
the discovery potential of the heavy higgs boson at a photon - photon collider was investigated for@xmath47 630 gev , and it was shown that the photon - photon collider can cover the `` wedge '' of the mssm parameter up to @xmath48 gev , where only one light higgs boson is found at lhc @xcite .
the photon - photon collider is also useful to distinguish the cp property of the higgs boson using transverse polarization and angular correlation of decay products @xmath49 , for instance ) @xcite .
the mass reach is , however , reduced , if we utilize the transverse polarization .
although the sm is very successful so far , there are strong motivations to search for physics beyond it .
first , the sm only deals with three of the four fundamental interactions , and the gravity is not included .
unification of all gauge interactions as well as gauge interactions with gravity can be addressed only in the context of physics beyond the sm .
second , the discovery of the neutrino oscillation indicates existence of new particles and/or interactions .
third , some of fundamental problems in cosmology such as what is the dark matter , and how the unbalance between baryon and anti - baryon arose require understanding of the physics scenario beyond the electroweak scale .
there are many theoretical proposals for physics beyond the sm .
we can expect some signals at the tev scale , if introduction of new physics is motivated to explain the separation between the weak scale and the planck scale .
susy and a scenario with large extra - dimensions are typical examples . in both case
, we will be led to a conceptual change on space - time , if it is true .
since the three gauge coupling constants determined at lep and slc experiments turned out to be consistent with the prediction of supersymmetric grand unified theory , unified models based on susy has been a prime candidate of the physic beyond the sm . in susy models ,
a superpartner is introduced for each of an ordinary particle .
new particles include scalar partners of quarks and leptons ( squarks and sleptons ) , fermionic superpartners ( gluino , charginos , and neutralinos ) .
the mass spectrum of these particles depends on susy breaking scenarios .
therefore , the experimental determination of the spectrum will point to a specific scenario . for susy search , lhc experiments will provide a crucial test . in a typical mass spectrum such as one in the minimal supergravity model , colored susy particles up to 2 tev can be discovered .
in addition , a light higgs boson should exist below about 130 gev for the mssm , which is an important check point for this model .
roles of the lc experiment are not restricted to discovery of new susy particles .
masses , quantum numbers , and various couplings of susy particles will be measured in a good accuracy .
determination of these quantities without relying on some specific model of susy breaking is very important to test and establish a new symmetry principle of nature . in many cases of susy studies
, beam polarization can be a very useful tool .
this is because susy is a symmetry between boson and fermion , so that spin information is important .
furthermore , the concept of chirality is extended to the squark and slepton sectors in susy , so that the initial electron polarization is very useful to distinguish left - handed or right - handed superparticles .
there are many issues to be addressed on susy at lcs
. * determination of mass and spin of susy particles from decay energy distributions , production angle distributions , and threshold scans in pair production processes . * reconstruction of chargino and neutralino mass matrices from production cross sections and angular distributions with possible effects of cp violation @xcite .
* determination of the selectron - electron - bino coupling through @xmath50 for a test of a susy relation ( figure 6 ( a ) ) @xcite . * search for lepton flavor violation in slepton pair production @xcite .
* test of the gaugino mass gut relation ( figure 6 ( b ) ) @xcite .
these measurements play important roles in determination of susy lagrangian in a model - independent way , and proving susy . in order to obtain a whole picture of a susy model , it is most likely that information from lhc and lcs has to be combined .
for instance , the neutralino mass determined at a lc will provide an important input for the lhc analysis . combining colored susy particle masses from lhc and slepton / chargino / neutralino masses from a lc
, we may be able to figure out the origin of susy breaking @xcite .
recently , a scenario with a large extra - dimensions is proposed , motivated by string theory @xcite .
if there are extra - space dimensions where only gravity can propagate , the fundamental scale of gravity where all interactions are unified can be close to the weak scale .
the smallness of the gravity can be attributed to the existence of the large extra - space , and the planck scale of @xmath51 gev is not a real physical scale . in this case
there is no fundamental problem of hierarchy between the planck and weak scales .
an interesting aspect of this scenario is that we can test this idea from collider experiments .
fairly model - independent signals of this scenario are emissions of kaluza - klein graviton modes ( @xmath52 ) and virtual graviton exchange in @xmath53 processes
. the sensitivity of lc experiments can be comparable to the lhc search for extra - dimensions .
the cross section of the graviton emission process depends on the fundamental scale and the number of extra - dimensions . for a larger number of extra - dimensions , the cross section rises more quickly .
experiments with different collider energies are useful to determine the number of dimensions .
figure 7 illustrates how we can determine the number of the extra - dimensions from experiments at lcs with two different center - of - mass energies @xcite .
if a signal is seen at one beam energy , the energy upgrade is useful to figure out the structure of extra - space . by determination of the fundamental scale and the number of extra - dimension
, we can also derive the size of extra - space .
the information of the number of the extra - dimensions can be also obtained by the analysis of the differential cross section with a fixed center - of - mass energy .
if some signals of the extra - space are available within the energy of the lc , thresholds of signals of strong gravity or string may be open at lhc .
these signals include production of black holes and string resonance states . in order to establish a new picture of space - time
, both lhc and lcs will be necessary .
the studies on the top quark and the w and z bosons are guaranteed at the lc experiment .
these studies provides fundamental parameters of the particle physics . in addition
, we may be able to look for new physics effects from precise determination of anomalous coupling constants involving these particles .
the top quark is the heaviest particle observed so far .
its production and decay are only studies at tevatron .
an @xmath0 collider can provide a unique opportunity to scan the threshold region .
we can determine the top quark mass at the level of accuracy of 100 mev and less , and the width to a few % @xcite .
the mass determinaton is improved by an order of magnitudes or more compared to lhc .
the width measurement will be practically only possible at the threshold scan of an @xmath0 collider .
an example of the studies at the top production threshold is shown in figure 8 .
the anomalous coupling measurements can be carried out at the threshold region as well as at the open - top - production region @xcite .
anomalous couplings of the top and photon / z boson include the top electric dipole moment which is induced by cp violation outside the sm .
the accuracies of electronic dipole moments determinations are @xmath54-@xmath55 e cm , for instance . at lhc ,
information on the gluon anomalous coupling will be available .
triple and quadratic anomalous couplings of gauge bosons will be also measured precisely at lcs .
although constraints on these coupling constants are already obtained by lep and tevatron experiments , the higher energy and beam polarization are advantages of the lc experiment for this study .
these measurements are particularly important when a light higgs boson is not found at the lhc and lc experiments , in which case something is missing in our current understanding of the electroweak symmetry breaking .
sensitivities to triple gauge boson coupling constants @xmath56 and @xmath57 defined in @xmath58 are @xmath59-@xmath60 at lc experiments @xcite .
current understanding of elementary particle physics is based on the sm , in which the gauge principle and the electroweak symmetry breaking are two basic ingredients .
nature of gauge interactions has been studied experimentally in full details last twenty years , but we know little about the electroweak symmetry breaking and the higgs mechanism .
the most important next step of the high energy physics is to clarify the dynamics of electroweak symmetry breaking and the mass generation mechanism of elementary particles . for this purpose ,
an @xmath0 lc is necessary .
the lc with the center - of - mass energy of up to 500 gev and an integrated luminosity of 500 fb@xmath7 can be considered as a higgs factory .
we can determine the coupling constants related to the mass generation mechanism very precisely . in order to determine the
top yukawa coupling constant and the higgs self - coupling constant with good precisions , a higher center - of - mass energy is required .
the lc experiments will also play essential roles in finding and studying new physics signals , such as susy and the scenario with a large extra - dimensional space . for these analysis ,
the lhc and linear collider experiments can be complementary to each other .
the clean experimental environment , availability of the beam polarization , capability of energy scan at the lc is very useful for these studies .
in addition to direct search for new signals , precise studies on the sm processes including the top quark and the w boson provide alternative ways to look for new physics effects .
the high energy physics has been developed by concurrent running of hadron and lepton machines . since the end of the lep operation , there is no energy frontier @xmath0 machine .
there are very strong physics cases to construct the @xmath0 lc which can operate concurrently with the lhc experiment .
the physics of the tev energy scale may or may not be what we are expecting , but whatever it may be , it will be necessary for us to have both machines to elucidate the physics at the electroweak scale and beyond .
99 t. abe _ et al . _
[ american linear collider working group collaboration ] , `` linear collider physics resource book for snowmass 2001 , '' in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ , arxiv : hep - ex/106055 , hep - ex/106055 , hep - ex/106057 , hep - ex/106058 .
y. okada , m. yamaguchi and t. yanagida , prog .
* 85 * ( 1991 ) 1 ; phys .
b * 262 * ( 1991 ) 54 ; j. r. ellis , g. ridolfi and f. zwirner , phys .
b * 257 * ( 1991 ) 83 ; h. e. haber and r. hempfling , phys .
* 66 * ( 1991 ) 1815 .
m. carena , s. mrenna and c. e. wagner , phys . rev .
d * 60 * ( 1999 ) 075010 ; k. s. babu and c. f. kolda , phys .
b * 451 * ( 1999 ) 77 ; h. e. haber , m. j. herrero , h. e. logan , s. penaranda , s. rigolin and d. temes , phys .
d * 63 * ( 2001 ) 055004 ; j. guasch , w. hollik and s. penaranda , phys .
b * 515 * ( 2001 ) 367 .
j. f. gunion and j. g. kelly , phys .
b * 333 * ( 1994 ) 110 ; m. kramer , j. h. kuhn , m. l. stong and p. m. zerwas , z. phys .
c * 64 * ( 1994 ) 21 ; g. j. gounaris and g. p. tsirigoti , phys . rev .
d * 56 * ( 1997 ) 3030 [ erratum - ibid .
d * 58 * ( 1998 ) 059901 ] ; s. y. choi and j. s. lee , phys .
d * 62 * ( 2000 ) 036005 ; e. asakawa , j. i. kamoshita , a. sugamoto and i. watanabe , eur .
j. c * 14 * ( 2000 ) 335 ; k. hagiwara , nucl .
instrum .
a * 472 * ( 2001 ) 12 ; e. asakawa , in _ physics and experiments with future linear @xmath61 colliders ( lcws 2000 ) _ , fermilab , batavia , illinois , 24 - 28 oct 2000 , arxiv : hep - ph/0101234 .
s. y. choi , a. djouadi , m. guchait , j. kalinowski , h. s. song and p. m. zerwas , eur .
j. c * 14 * ( 2000 ) 535 ; s. y. choi , j. kalinowski , g. moortgat - pick and p. m. zerwas , eur .
j. c * 22 * ( 2001 ) 563 . | physics at an @xmath0 linear collider is described as a theoretical introduction to linear collider workshop 2002 .
kek - th-884 + april 2003 + |
an unambiguous determination of pairing symmetry in unconventional superconductors is crucial to understand the pairing mechanism of superconductivity .
strong evidences suggesting @xmath3-wave pairing symmetry in the high-@xmath4 cuprates have been provided using several phase - sensitive probes @xcite including tunneling spectroscopy via andreev surface bound states ( abs s ) .
@xcite the tunneling spectroscopy via abs s enables us to detect the sign change in the pair potential as well as its nodal structure .
@xcite this state , which originates from the interference effect in the effective pair potential of the @xmath5-wave symmetry through reflection at a surface or an interface , have significant influences on several charge transport properties @xcite .
the existence of abs s , which manifests itself as a distinct conductance peak at zero - bias in the tunneling spectrum ( zero - bias conductance peak , referred to as zbcp ) , has been actually observed not only in the high-@xmath4 cuprates @xcite but also in ruthenates @xcite , heavy fermion systems @xcite , and more recently @xmath6 .
@xcite in this context , it is of great interest to investigate whether the zbcp due to the abs s can be observed in organic superconductors @xcite such as @xmath0-(bedt - ttf)@xmath7 and @xmath8 @xcite .
the tunneling spectroscopy via abs s can be used to determine the pairing symmetry if one can prepare well - treated surfaces with arbitrary orientations in the superconducting plane . for high-@xmath4 cuprates , which has a @xmath3 pair potential ,
it is theoretically shown that the zbcp should be observed most prominently for ( 110 ) surfaces or interfaces .
moreover , it has been clarified that the zbcp may be observed due to atomic - scale roughness even in the ( 100 ) surfaces .
@xcite in fact , iguchi et al .
@xcite have measured the zbcp for ag / ybco ramp - edge junctions with various orientations , where the injection direction varies continuously from ( 100 ) to ( 110 ) interfaces .
the height of the zbcp has shown to vary according as the misorientation angle from @xmath9-axis within the plane .
as regards organic superconductors such as @xmath0-(bedt - ttf)@xmath7 , the pairing symmetry of the pair potentials still remains to be a controversial problem .
it has indeed become an issue of great interest whether @xmath0-(bedt - ttf)@xmath7 has a @xmath10-wave pair potential similar to high-@xmath4 cuprates .
there is now a body of accumulating experimental evidences suggesting that @xmath0-(bedt - ttf)@xmath7 have anisotropy in the pair potential .
@xcite earlier theories support @xmath3-wave pairing , @xcite while a recent thermal conductivity measurement suggests @xmath11-wave pairing .
@xcite concerning this issue , two of the present authors have theoretically shown that a @xmath11-like pairing may slightly dominate over @xmath3 pairing when the dimerization of the bedt - ttf molecules is not so strong .
@xcite according to previous studies , @xcite if the pairing symmetry of @xmath0-(bedt - ttf)@xmath7 is @xmath10-wave , abs is expected to be created at surfaces for arbitrary injection orientations .
however , a scanning tunneling microscopy ( stm ) experiment for @xmath0-(bedt - ttf)@xmath7 by arai et al.@xcite showed the absence of zbcp for arbitrary injection angle from the @xmath12-axis in the @xmath13-plane , which is in contrast with the case of the high-@xmath4 cuprates .
the presence / absence of the zbcp of @xmath10-wave superconductors is sensitive to several factors : ( i ) roughness effect of surfaces or interfaces , ( ii ) random impurity scattering effect near the interfaces , ( iii ) the shape of the fermi surface , and ( iv ) the degradation of surfaces .
the disappearance of the zbcp in @xmath10-wave superconductors due to reason ( i ) has been studied previously@xcite .
depending on the shape of the fermi surface and the geometry of the surface , the atomic - size wave nature of the zero energy abs s ( zes ) , @xmath14 , the oscillatory behavior of the wave function of zes induces an interference effect which locally destroys the zbcp .
in fact , it is by no means easy to make well - oriented cleavage surfaces in organic materials , so this point may be important .
as regards point ( ii ) , asano @xmath15 @xmath16 @xcite have shown , both from analytical and numerical calculation beyond quasiclassical approximations @xcite , that impurity scattering near the interface in the high-@xmath17 cuprates can induce a splitting or a disappearance of the zbcp .
as for point ( iii ) , we have recently studied the disappearance of zbcp due to the warping of quasi-1d fermi surface as in @xmath18 .
@xcite the results indicate that the abs s are sensitive to the shape of the fermi surface .
however , most of the theoretical studies on tunneling spectroscopy via abs s up to date have been performed for single band systems .
it has not been clarified how the multiplicity of the fermi surface influences the zbcp .
motivated by this point , here we investigate the surface density of states in systems having multiple fermi surfaces , where we focus on two organic superconductors as case studies , namely , @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 and ( tmtsf)@xmath1clo@xmath2 . the fermi surface of @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 , which has been determined by shubnikov - de haas experiment , @xcite consists of two portions separated by small gaps .
the fermi surface of ( tmtsf)@xmath1clo@xmath2 is also separated by a small gap , which is due to anion ordering . in this paper , we extend our previous studies @xcite on anisotropic triangular lattice by taking into account these multiplicity of the fermi surface .
the organization of the paper is as follows .
the formulation of calculating the tunneling spectrum on anisotropic triangular lattice is presented in sec .
[ sec:02 ] . in sec .
[ sec:03 ] , results of the numerical calculations are discussed in detail . finally , we summarize the paper in sec .
[ sec:04 ] .
in the present study , we start from an extended hubbard model given by @xmath19 where @xmath20 creates a hole with spin @xmath21 at site @xmath22 . as a model for @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 , each site corresponds to bedt - ttf molecule dimmers .
we consider five kinds of hopping integrals , @xmath23 , @xmath24 , @xmath25 , @xmath26 , and @xmath27 in the @xmath28 plane on the anisotropic triangular lattice as shown in fig .
[ fig:01](a ) . in order to reproduce the shape of fermi surface for @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 and @xmath29
, we adopt the values of ( i ) @xmath30 , @xmath31 , @xmath32 , @xcite and ( ii ) @xmath33 , @xmath34 , @xmath35 , @xcite respectively .
two subchains in the @xmath36 direction alternatively have the site energy @xmath37 in the @xmath38 direction .
@xcite the chemical potential @xmath39 is determined such that the band in @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 [ @xmath29 ] is half - filled [ quarter - filled ] .
the effective attraction @xmath41 is assumed to act on a pair of electrons . by solving the mean - field equation for a unit cell with @xmath42 sites in the @xmath36 direction and two sites in the @xmath38 direction
, we obtain the eigenenergy @xmath43 . in terms of the eigenenergy @xmath43 and the wave functions @xmath44 , @xmath45 , the bogoliubov - de gennes equation for the ( 100 ) surface in the @xmath28 plane is given by @xmath46 with @xmath47 where we define @xmath48 and @xmath49 . as for plausible pairing symmetries in @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1
, we consider @xmath3-wave pairing given by , @xmath50 and @xmath11-like pairing given by @xmath51 with @xmath52 in accord with ref . .
the pairing in real space is shown in fig .
[ fig:01](b ) . here , we select @xmath53 and @xmath54 , where @xmath55 is a bulk value .
the upper and lower panels of fig .
[ fig:02 ] show the @xmath3 and @xmath11-like pair potentials in momentum space along with the fermi surfaces . for @xmath34 ,
the fermi surface is continuously connected . in the actual @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 , however , bedt - ttf dimmers are further dimerized so that @xmath56 , which leads to a splitting of the fermi surface at around @xmath57 . in @xmath58 , the orientational order of the anions @xmath59 doubles the unit cell , leading again to a splitting of the fermi surface .
@xcite although the pairing symmetry for @xmath60 remains to be undetermined , here we assume singlet @xmath10-wave pairing as an example in which the multiband effect is prominent . in this case @xmath10-wave is a pairing separated by two lattice spacings given by @xmath61 the anion potential @xmath62 is estimated from experimental measurement of angle dependent magnetoresistance .
@xcite in order to compare our theory with stm experiments , we assume that the stm tip is metallic with a constant density of states , and that tunneling occurs only to the site nearest to the tip . this has been shown to be valid through the study of tunneling conductance of unconventional superconductors @xcite .
the tunneling conductance spectrum is then given at low temperatures by the normalized surface density of states @xcite , @xmath63.\end{aligned}\ ] ] here @xmath64 denotes the surface density of states for the superconducting state while @xmath65 the bulk density of states in the normal state .
in this section , we present the calculation results .
first , let us focus on the model for @xmath0-(bedt - ttf)@xmath7 .
we examine the case of the tunneling spectrum at ( 100 ) surface on the @xmath28 plane as shown in fig .
[ fig:01 ] . as seen in fig .
[ fig:04 ] , in the case of @xmath34 , where the fermi surface is elliptical but continuous , there exists a distinct peak at zero energy , which resembles those obtained in previous theories assuming round shape fermi surface . the zep arises because incident and reflected ( including oblique incidence ) quasiparticles normal to the surface feel opposite signs of the pair potential , which results in a formation of the abs .
if we turn on the multiband effect by letting @xmath66 , the zep is found to split into two .
this is reminiscent of the zep splitting originating from broken time reversal symmetry states.@xcite we have further studied the @xmath67 dependence of the zep splitting . in fig .
[ fig:05 ] , the width of the zep splitting @xmath68 is plotted as functions of @xmath67 for @xmath3 and @xmath11-like pairings .
@xmath68 for the @xmath11-like pair potential is almost proportional to @xmath67 , and larger than that for @xmath3 . in the regime of @xmath69 ,
in particular , we see no splitting for the @xmath3 pairing .
since @xmath70 is estimated to be @xmath71 , @xcite we may be able to distinguish between @xmath3 and @xmath11-like pairings through the presence / absence of zep splitting .
we have also performed similar calculation by letting @xmath72 deviate from unity , which should be the case when @xmath73 deviates from @xmath73 .
the results are plotted in fig .
[ fig:06 ] for various @xmath72 with @xmath67 fixed at @xmath74 . in this case
, we observe an overall shift of the splitted zep , while the magnitude of the splitting remains unchanged .
let us now move on to the model for @xmath58 . in this model ,
as the fermi surface becomes asymmetric with respect to @xmath75 transformation , some injected and reflected quasiparticles feel different signs and the zep appears in the tunneling spectrum @xcite .
when @xmath62 is turned on , a minigap opens at @xmath76 .
this effect again leads to zep splitting , of which the magnitude increases as @xmath62 is increased .
[ fig:07 ] ] although it is by no means easy to pinpoint the origin of the zep splitting analytically , it can be qualitatively explained as follows . for single band models
, the zep appears due to the sign change of the pair potential felt by quasiparticles at the interface . in multiband systems ,
injected and reflected quasiparticles have different band indices , so that an additional phase factor is expected to appear due to interband scattering .
this additional phase factor induces the zep splitting as in the case of pair potentials with broken time reversal symmetry .
to summarize , we have investigated the multiband effect on tunneling spectroscopy . as case studies , we have focused on models for @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 and @xmath29 .
we find that the multiplicity of the fermi surface can lead to a splitting of the zep .
as regards @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 , since @xmath70 is estimated to be @xmath71 , @xcite we can distinguish between @xmath3 and @xmath11-like pairings through the presence / absence of zep splitting .
as mentioned in the introduction , however , a scanning tunneling measurement @xcite actually find no zbcp in the tunneling spectrum of @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 .
since many experiments suggest the existence of nodes in the pair potential in @xmath0-(bedt - ttf)@xmath7 , we believe that the absence of zbcp is not because the pairing symmetry is a simple @xmath77-wave , but because of the roughness of the surface or the random scattering effect by impurities near the interface@xcite , namely , point ( i ) or ( ii ) mentioned in the introduction . as regards the roughness of the surface , we believe it is necessary to study tunneling spectroscopy of @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 in the presence of atomic scale roughness as done by tanuma @xmath15 @xmath16 on a lattice model.@xcite as for the issue of random scattering effect by impurities , asano @xmath15 @xmath16 @xcite have shown , both from analytical and numerical calculation beyond quasiclassical approximations , that impurity scattering near the interface in the high-@xmath17 cuprates can induce a splitting or a disappearance of the zbcp . from this viewpoint , it would also be interesting to study the impurity scattering effect in @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 . in order to clearly determine the pairing symmetry ,
other complementary probes should also be used .
recently , we have shown that magnetotunneling spectroscopy is a promising method to identify the detailed paring symmetry of the unconventional superconductors @xcite .
it would also be interesting to apply this probe to @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 .
it is well known that abs s have serious influence on the josephson current .
there are many works on josephson effect in unconventional superconductors both from theoretical and experimental view point @xcite .
it is also a future problem to study josephson effect in @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 and @xmath29 .
the computations have been performed at the supercomputer center of yukawa institute for theoretical physics , kyoto university .
99 m. sigrist and t.m .
rice , rev .
phys . * 67 * , 505 ( 1995 ) .
van harlingen , rev .
phys . * 67 * , 515 ( 1995 ) .
c.c . tsuei and j.r .
kirtley , rev .
phys . * 72 * , 969 ( 2001 ) .
buchholtz and g. zwicknagl , phys .
b * 23 * , 5788 ( 1981 ) .
hu , phys .
lett . * 72 * , 1526 ( 1994 ) .
y. tanaka and s. kashiwaya , phys .
lett . * 74 * , 3451 ( 1995 ) ; s. kashiwaya , y. tanaka , m. koyanagi , h. takashima , and k. kajimura , phys .
b * 51 * , 1350 ( 1995 ) . m. fogelstrm , d. rainer , and j.a .
sauls , phys .
lett . * 79 * , 281 ( 1997 ) .
s. kashiwaya and y. tanaka , rep .
63 * , 1641 ( 2000 ) ; s. kashiwaya , y. tanaka , m. koyanagi , and k. kajimura , phys . rev .
b * 53 * , 2667 ( 1996 ) .
t. lfwander , v.s .
shumeiko , and g. wendin , supercond .
* 14 * , r53 ( 2001 ) .
m. yamashiro , y. tanaka , and s. kashiwaya , phys . rev .
b * 56 * , 7847 ( 1997 ) ; j. phys .
. jpn . * 67 * , 3364 ( 1998 ) ; m. yamashiro , y. tanaka y. tanuma , and s. kashiwaya , j. phys .
jpn . * 67 * , 3224 ( 1998 ) ; j. phys .
. jpn . * 68 * , 2019 ( 1999 ) ; c. honerkamp and m. sigrist , prog .
. phys . * 100 * , 53 ( 1998 ) ; j. low temp . phys . *
111 * , 895 ( 1998 ) .
y. tanaka , t. asai , n. yoshida , j. inoue , and s. kashiwaya , phys .
b * 61 * , r11902 ( 2000 ) ; y. tanaka , t. hirai , k. kusakabe , and s. kashiwaya , phys . rev .
b * 60 * , 6308 ( 1999 ) ; t. hirai , k. kusakabe , and y. tanaka , physica c * 336*,107 ( 2000 ) ; k. kusakabe and y. tanaka , physica c * 367 * , 123 ( 2002 ) ; j. phys .
solids * 63 * , 1511 ( 2002 ) ; n. stefanakis , phys .
b * 64 * , 224502 ( 2001 ) ; z. c. dong , d. y. xing , and jinming dong , phys . rev .
b * 65 * , 214512 ( 2002 ) ; z. c. dong , d. y. xing , z. d. wang , ziming zheng , and jinming dong , phys .
b * 63 * , 144520 ( 2001 ) ; yu .
s. barash , m. s. kalenkov , and j. kurkijarvi , phys .
b * 62 * , 6665 ( 2000 ) ; m. h. s. amin , a. n. omelyanchouk , and a. m. zagoskin , phys . rev .
b 63 , 212502 ( 2001 ) ; shin - tza wu and chung - yu mou , phys .
b * 66 * , 012512 ( 2002 ) . j .-
x . zhu , b. friedman , and c. s. ting , phys . rev .
b * 59 * , 9558 ( 1999 ) ; s. kashiwaya , y. tanaka , n. yoshida and m.r . beasley , phys .
b * 60 * , 3572 ( 1999 ) ; i. uti and o. t. valls , phys . rev .
b * 60 * , 6320 ( 1999 ) ; n. yoshida , y. tanaka , j. inoue , and s. kashiwaya , j. phys .
jpn . * 68 * , 1071 ( 1999 ) ; t. hirai , n. yoshida , y. tanaka , j. inoue , and s. kashiwaya , j. phys .
. jpn . * 70 * , 1885 ( 2001 ) ; n. yoshida , h. itoh , t. hirai , y. tanaka , j. inoue , and s. kashiwaya , phsica c * 367 * , 135 ( 2002 ) ; t. hirai , y. tanaka , n. yoshida , y. asano , j. inoue , and s. kashiwaya , phys .
b * 67 * 174501 ( 2003 ) .
j. geerk , x.x .
xi , and g. linker , z. phys .
b. * 73 * , 329 ( 1988 ) ; s. kashiwaya , y. tanaka , m. koyanagi , h. takashima , and k. kajimura , phys .
b * 51 * , 1350 ( 1995 ) ; l. alff , h. takashima , s. kashiwaya , n. terada , h. ihara , y. tanaka , m. koyanagi , and k. kajimura , phys . rev .
b * 55 * , 14757 ( 1997 ) ; m. covington , m. aprili , e. paraoanu , l.h .
greene , f. xu , j. zhu , and c.a .
mirkin , phys .
lett . * 79 * , 277 ( 1997 ) ; j. y. t. wei , n .- c .
yeh , d. f. garrigus and m. strasik , phys .
* 81 * , 2542 ( 1998 ) ; h. aubin , l. h. greene , sha jian , and d. g. hinks , phys .
lett . * 89 * , 177001 ( 2002 ) ; a. sharoni , o. millo , a. kohen , y. dagan , r. beck , g. deutscher , and g. koren , phys .
b * 65 * , 134526 ( 2002 ) ; a. biswas , p. fournier , m. m. qazilbash , v. n. smolyaninova , h. balci , and r. l. greene , phys .
* 88 * , 207004 ( 2002 ) ; y. dagan and g. deutscher , phys .
87 * , 177004 ( 2001 ) ; j. w. ekin , y. xu , s. mao , t. venkatesan , d. w. face , m. eddy , and s. a. wolf , phys . rev .
b * 56 * , 13746 ( 1997 ) ; a. kohen , g. leibovitch , and g. deutscher , phys . rev .
lett . * 90 * , 207005 ( 2003 ) .
f. laube , g. goll , h.v .
lhneysen , m. fogelstrm , and f. lichtenberg , phys .
lett . * 84 * , 1595 ( 2000 ) .
mao , k.d .
nelson , r. jin , y. liu , and y. maeno , phys .
lett . * 87 * , 037003 ( 2001 ) .
wlti , h.r .
ott , z. fisk , and j.l .
smith , phys .
* 84 * , 5616 ( 2000 ) .
z.q . mao , m.m .
rosario , k.d .
nelson , k. wu , i.g .
deac , p. schiffer , and y. liu , t. he , k.a .
regan , and r.j .
cava , phys .
b * 67 * , 094502 ( 2003 ) . for a review , see e.g. , t. ishiguro , k. yamaji , and g. saito , _ organic superconductors _ , 2nd ed .
( springer - verlag , heidelberg , 1998 ) .
t. komatsu , n. matsukawa , t. inoue , and g. saito , j. phys .
. jpn . * 65 * , 1340 ( 1996 ) .
y. tanuma , k. kuroki , y. tanaka , and s. kashiwaya , phys .
b * 64 * , 214510 ( 2001 ) ; k. sengupta , i. uti , h .- j .
kwon , v.m .
yakovenko , and s. das sarma , phys .
b * 63 * , 144531 ( 2001 ) .
y. tanuma , y. tanaka , m. yamashiro , and s. kashiwaya , phys .
b * 57 * , 7997 ( 1998 ) .
y. tanuma , y. tanaka , m. ogata , and s. kashiwaya , j. phys .
soc . jpn . *
67 * , 1118 ( 1998 ) ; phys . rev .
b * 60 * , 9817 ( 1999 ) .
k.v . samokhin and m.b .
walker , phys .
b * 64 * , 172506 ( 2001 ) .
i. iguchi , w. wang , m. yamazaki , y. tanaka , and s. kashiwaya , phys .
b * 62 * , r6131 ( 2000 ) ; h. mayaffre , p. wzietek , d. jrome , c. lenoir , and p. batail , phys . rev
. lett . * 75 * , 4122 ( 1995 ) .
de soto , c.p .
slichter , a.m. kini , h.h .
wang , u. geiser , and j.m .
williams , phys .
b * 52 * , 10364 ( 1995 ) .
k. kanoda , k. miyagawa , a. kawamoto , and y. nakazawa , phys .
b * 54 * , 76 ( 1996 ) .
y. nakazawa and k. kanoda , phys .
b * 55 * , r8670 ( 1997 ) .
a. carrington , i.j .
bonalde , r. prozorov , r.w .
giannetta , a.m. kini , j. schlueter , h.h .
wang , u. geiser , and j.m .
williams , phys .
* 83 * , 4172 ( 1999 ) .
m. pinteri , s. tomi , m. prester , d.j .
drobac , o. milat , k. maki , d. schweitzer , i. heinen , and w. strunz , phys .
b * 61 * , 7033 ( 2000 ) .
j. schmalian , phys .
lett . * 81 * , 4232 ( 1998 ) .
h. kino and h. kontani , j. phys .
* 67 * , 3691 ( 1998 ) ; h. kondo and t. moriya , _ ibid .
_ , 3695 ( 1998 ) . k. kuroki and h. aoki , phys .
b * 60 * , 3060 ( 1999 ) .
k. izawa , h. yamaguchi , t. sasaki , and y. matsuda , phys .
lett . * 88 * , 027002 ( 2002 ) .
k. kuroki , t. kimura , r. arita , y. tanaka , and y. matsuda , phys .
b * 65 * , 100516(r ) ( 2002 ) .
t. arai , k. ichimura , k. nomura , s. takasaki , j. yamada , s. nakatsuji , and h. anzai , phys .
b * 63 * , 104518 ( 2001 ) .
y. tanaka , y. tanuma , and s. kashiwaya , phys .
b * 64 * , ( 2001 ) 054510 ; y. tanaka , yu .
v. nazarov and s. kashiwaya , phys .
* 90 * , ( 2003 ) 167003 ; y. tanaka , a. a. golubov and s. kashiwaya , phys .
b * 68 * ( 2003 ) ; n. kitaura , h. itoh , y. asano , y. tanaka , j. inoue , y. tanuma and s. kashiwaya , j. phys . soc
. jpn . * 72 * , ( 2003 ) ; a. a. golubov , m. y. kupriyanov , pisma zh .
fiz * 69 * , 242 ( 1999 ) [ sov .
jetp lett .
* 69 * , 262 ( 1999 ) . ] ; * 67 * , 478 ( 1998 ) [ sov .
jetp lett .
* 67 * , 501 ( 1998 ) . ]
; a. poenicke , yu . s. barash , c. bruder , and v. istyukov , phys . rev .
b * 59 * , 7102 ( 1999 ) ; k. yamada , y. nagato , s. higashitani , and k. nagai , j. phys . soc
. jpn . * 65 * , 1540 ( 1996 ) ; t. lck , u. eckern , and a. shelankov , phys .
b * 63 * , 064510 ( 2002 ) .
y. tanuma , k. kuroki , y. tanaka , r. arita , s. kashiwaya , and h. aoki , phys .
b * 66 * , 094507 ( 2002 ) , k. oshima , t. mori , h. inokuchi , h. urayama , h. yamochi , and g. saito , phys . rev .
b * 38 * , r938 ( 1988 ) .
l. ducasse , m. abderrabba , j. hoarau , m. pesquer , b. gallois , and j. gaultier , j. phys .
c * 19 * , 3805 ( 1986 ) .
p.m. grant , j.phys.(france ) * 44 * , c3 - 847 ( 1983 ) k. yamaji , j. phys .
. jpn . * 55 * , 860 ( 1986 ) .
h. shimahara , phys .
b * 61 * , r14936 ( 2000 ) .
h. yoshino , a. oda , t. sasaki , t. hanajiri , j. yamada , s. nakatsuji , h. anzai , and k. murata , j. phys .
. jpn . * 68 * , 3142 ( 1999 ) .
j. p. pouget , g. shirane , k. bechgaard , and j. m. fabre , phys .
b * 27 * , 5203 ( 1983 ) .
y. tanuma , y. tanaka , k. kuroki , and s. kashiwaya , phys .
b * 66 * , 174502 ( 2002 ) .
m. matsumoto and h. shiba , j. phys .
jpn . * 64 * , 4867 ( 1995 ) ; j. phys .
. jpn . * 64 * , 3384 ( 1995 ) . y. tanuma , y. tanaka , and s. kashiwaya , phys .
b * 64 * , 214519 ( 2001 ) .
s. kashiwya , y. tanaka , m. koyanagi and k. kajimura , j. phys .
solids * 56 * , 1721 ( 1995 ) .
y. tanaka , h. tsuchiura , y. tanuma , and s. kashiwaya , j. phys .
. jpn . * 71 * , 271 ( 2002 ) ; y. tanaka , y. tanuma k. kuroki , and s. kashiwaya , j. phys .
jpn . * 71 * , 2102 ( 2002 ) ; y. tanaka , h. itoh , h. tsuchiura , y. tanuma , j. inoue , and s. kashiwya , j. phys . soc
. jpn . * 71 * , 2005 ( 2002 ) . y. tanaka and s. kashiwaya , phys .
b * 53 * , r11957 ( 1996 ) ; phys . rev .
b * 56 * , 892 ( 1997 ) ; phys . rev .
b * 58 * , 2948 ( 1998 ) ; j. phys .
. jpn . * 68 * , 3485 ( 1999 ) ; j. phys .
. jpn . * 69 * , 1152 ( 2000 ) ; y. s. barash , h. burkhardt , and d. rainer , phys .
lett . * 77 * , 4070 ( 1996 ) ; e. ilichev , v. zakosarenko , r. p. j. ijsselsteijn , v. schultze , h .-
meyer , h. e. hoenig , h. hilgenkamp , and j. mannhart , phys .
lett . * 81 * , 894 ( 1998 ) ; e. ilichev , m. grajcar , r. hlubina , r. p. j. ijsselsteijn , h. e. hoenig , h .-
meyer , a. golubov , m. h. s. amin , a. m. zagoskin , a. n. omelyanchouk , and m. yu .
kupriyanov , phys .
lett . * 86 * , 5369 ( 2001 ) ; h. hilgenkamp , j. mannhart , and b. mayer , phys
b * 53 * , 14586 ( 1996 ) ; f. lombardi , f. tafuri , f. ricci , f. miletto granozio , a. barone , g. testa , e. sarnelli , j. r. kirtley , and c. c. tsuei , phys . rev
. lett . * 89 * , 207001 ( 2002 ) ; h. j. h. smilde , ariando , d. h. a. blank , g. j. gerritsma , h. hilgenkamp , and h. rogalla , phys
lett . * 88 * , 057004 ( 2002 ) ; h. hilgenkamp and j. mannhart , rev .
phys . * 74 * , 485 ( 2002 ) ; t. imaizumi , t. kawai , t. uchiyama , and i. iguchi , phys . rev . lett . * 89 * , 017005 ( 2002 ) ; h. arie , k. yasuda , h. kobayashi , i. iguchi , y. tanaka , and s. kashiwaya : phys . rev .
b * 62 * , 11864 ( 2000 ) ; y. asano , phys . rev .
b * 63 * , 052512 ( 2001 ) ; phys . rev .
b * 64 * , 014511 ( 2001 ) ; phys . rev .
b * 64 * , 224515 ( 2001 ) ; j. phys . soc
. jpn . * 71 * , 905 ( 2002 ) ; y. asano , y. tanaka , m. sigrist , and s. kashiwaya , phys .
b * 67 * , 184505 ( 2003 ) . | we study how the multiplicity of the fermi surface affects the zero - bias peak in conductance spectra of tunneling spectroscopy . as case studies ,
we consider models for organic superconductors @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 and ( tmtsf)@xmath1clo@xmath2 .
we find that multiplicity of the fermi surfaces can lead to a splitting of the zero - bias conductance peak ( zbcp ) .
we propose that the presence / absence of the zbcp splitting is used as a probe to distinguish the pairing symmetry in @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 . |
general relativity teaches us space and time are not independent , but inseparably entangled in a unified _ spacetime_. nevertheless , standard procedure in canonical gravity is to temporarily disregard this lesson , foliating spacetime into spacelike level sets of some time function .
this gives an initial value formulation of general relativity @xcite , with many uses in classical and quantum gravity .
but this approach , depending on an arbitrary , unobservable time function , has strange physical consequences . while the spacetime picture of gravity is described by _
equations , the foliation constrains _ global _ spacetime geometry and topology .
a well - posed initial value formulation demands global hyperbolicity , which in turn implies constant spatial topology @xcite .
this leaves many interesting spacetimes , including even anti - de sitter space , with no ` dynamical ' description . here
we present an alternative , _ fully local _ description of gravitational dynamics , based on the notion of a _ field of observers_. this is useful for a geometric understanding of lorentz symmetry in canonical gravity , for relating geometrodynamics with connection dynamics , for linking canonical and covariant phase spaces , and for various possible extensions of general relativity .
in minkowski space @xmath0 , an observer with velocity @xmath1 in hyperbolic space @xmath2 has a global notion of ` space , ' namely the subspace @xmath3 orthogonal to @xmath1 : @xmath4 an observer thus naturally splits spacetime fields into spatial and temporal parts . in more general spacetimes ,
this picture is valid only ` infinitesimally , ' on each tangent space .
a * field of observers * is a unit future - directed timelike vector field @xmath5something any time - oriented lorentzian manifold has .
such a field suffices to split fields on a _ background _
spacetime into spatial and temporal parts .
but here we are interested in general relativity , where the metric and hence the definition of observer is to be determined by the dynamics
. can we define ` observers ' without using the metric ?
fortunately , the * coframe field * in first - order gravity locally maps spacetime vectors to vectors in @xmath0 , which we view as an * internal spacetime*. this lets us translate between an observer field and a more primitive notion : a field of * internal observers * , assigning an observer @xmath6 to each point @xmath7 of spacetime .
right4 cm ( 0,0 ) * ; ( -20,18 ) * ; ( -20,14 ) * ; ( 20,18 ) * ; ( -35,10)*t_xm ; ( -22,-2)*x ; ( 28,0 ) * y(x ) ; ( 35,11)*r^3,1 ; ( -23,-8)*u=``a '' ; `` a'';(-17,-4 ) * * ; ( 22,-10)*r^3_y=``b '' ; `` b'';(18,-6 ) * * ; _ e@/^.2pc/(-14,-3)*;(12,-6 ) * ; ( 0,-17 ) * ; ( -10,-15);(-2,-8 ) * * ; starting with a smooth spacetime manifold @xmath8 , besides the field @xmath1 , we need : * a nowhere - vanishing 1-form @xmath9 , * an @xmath10-valued 1-form @xmath11 such that + ' '' '' + @xmath12 + is a nondegenerate coframe field . the observer field @xmath5 itself is found by solving @xmath13 .
this implies @xmath14 is dual to @xmath5 , so from @xmath5 s perspective , ` space ' is the kernel of @xmath14 .
@xmath11 annihilates @xmath5 and so is ` purely spatial . '
all other differential forms similarly split into a * spatial part * annihilating @xmath5 , and a * temporal part * of the form @xmath15 . in particular , the spin connection is given by @xmath16 where the spatial 1-form @xmath17 and the scalar @xmath18 both live in @xmath19 .
these constructions are clearly analogous to how spacetime fields are built in adm gravity @xcite . in this language ,
classical field equations split neatly into spatial equations constraining ` initial values ' of the fields and temporal equations corresponding to dynamics .
for example , the spatial part of the vacuum einstein equation @xmath20 is @xmath21 + \xi\ , d^\perp \hat u \right ) = 0\ ] ] where the * spatial differential * @xmath22 depends on the lie derivative @xmath23 along @xmath5 , and @xmath24=d^{\perp}\omega+\omega\wedge\omega$ ] is the * spatial curvature*. the resulting equations also follow from an action given in @xcite with @xmath14 as a background structure analogous to ` @xmath25 ' in foliation - based approaches ; for @xmath26 , where @xmath27 is the lapse , our formulation reduces to adm , rewritten in coframe variables .
but @xmath14 need not be of the form @xmath28 , even locally , in which case the distribution of spatial hyperplanes provided by @xmath14 will not be tangent to any foliation : @xmath29{contact.pdf } } ; ( -50,0)*{\txt{\small \sf spatial \\
\small ( $ \ker(\hat u)$)}}="a " ; " a"+(10,3);(-20,4)**\dir{- } ; " a"+(11,-2);(-25,-5)**\dir{- } ; " a";(-24,-18)**\dir{- } ; ( 50,0)*{\txt{\small \sf attempting a ` loop \\ \small \sf in space ' can result \\
\small \sf in a time translation}}="b " ; ( 15,-2)="c " ; " b";"c"**\dir{- } ; " c";(7.5,1)**\dir{- } ; " c";(7.5,-7)**\dir{- } ; \endxy\ ] ] the frobenius theorem implies the local condition for @xmath14 to determine a foliation is @xmath30 , in which case our equations reduce to a more familiar form .
but why allow configurations with @xmath31 ?
we have already mentioned that doing so avoids making global restrictions other than time - orientability .
more importantly , configurations with @xmath31 are related to foliations by a _
natural symmetry_. lorentz gauge transformations act on internal observers , @xmath32 , but because @xmath33 one can transform either the coframe @xmath34 or the observer field @xmath5 , giving two distinct types of local lorentz symmetry . transforming @xmath5 will in general map a foliation to a configuration where @xmath35 ; invariance under those transformations
is the analog in our framework of independence of foliation in standard hamiltonian approaches .
general relativity _ is _ invariant because @xmath36 and @xmath34 are , but this symmetry is lost in some proposals for going beyond general relativity , as discussed later .
an internal observer @xmath1 is fixed by a subgroup @xmath37 , and lorentz group representations split accordingly : @xmath38_{\pi_y } \ar[dr ] & & { { \mathbb r}}^{3,1 } \ar[dl ] \ar[dr ] & & \scriptstyle \leftarrow \text { \sf adjoint and fundamental reps of $ { { \rm so}}(3,1)$ } \\ { \mathfrak{so}}(3)_y\hspace{-2em } & & { { \mathbb r}}^3_y & & { { \mathbb r}}^1_y & \scriptstyle \leftarrow \text { \sf observer - dependent reps of $ { { \rm so}}(3)_y$ } } \ ] ] a local lorentz transformation changes these splittings , but changes also the rotation group @xmath39 , so all fields transform consistently . as an example , consider the @xmath40 part of the spatial connection , @xmath41 , where @xmath42 projects onto @xmath40 . under a local lorentz transformation of @xmath36
, we get : @xmath43 so that @xmath44 lives in @xmath45 . similarly , the @xmath10-valued ` triad ' @xmath11 transforms to take values in @xmath46 . under transformations living in @xmath47 , @xmath48 and @xmath11 have just the right behavior for a spatial connection and triad
. one can show @xcite that they generalize ashtekar variables @xcite , in the real form due to barbero @xcite . in the ashtekar - barbero formulation , the apparent breaking of lorentz symmetry down to @xmath49 arises by fixing @xmath1 and therefore the subgroup @xmath47 once and for all . for us ,
this breaking occurs ` spontaneously ' : at each spacetime point , @xmath50 selects the subgroup @xmath37 . by transforming @xmath1 along with the dynamical variables ,
the action of the full lorentz group is maintained . breaking @xmath51 symmetry spontaneously has two nice side - effects .
first , it sidesteps second class constraints that must be dealt with in related connection - based approaches .
second , it makes the pair @xmath52 into a ` spatial cartan connection , ' making a precise link between ` geometrodynamics ' and ` connection dynamics . '
see our papers @xcite for details .
in foliation - based approaches , ` time evolution ' is a particular 1-parameter family of spacetime diffeomorphisms : the flow generated by the vector field @xmath53 , moving each spatial slice into the future by intervals of the arbitrary ` time ' function @xmath54 : ( 150,100 ) ( 5,0)(20,20)(10,50)(10,50)(0,80)(10,100 ) ( 130,0)(135,20)(130,50)(130,50)(120,80)(130,100 ) ( 25,0)(1,3)8(50,0)(0,3)10(75,0)(-1,3)4(100,0)(1,3)7 ( 25,25)(0,3)12(50,25)(1,3)6(75,25)(2,3)8(100,25)(2,3)12(125,25)(0,3)17 ( 25,50)(-1,3)4(35,50)(0,1)10(50,50)(1,1)12(75,50)(1,1)17(100,50)(1,3)7 ( 25,75)(1,3)4(50,75)(2,3)12(75,75)(1,3)7(100,75)(2,3)13 ( 135,20)@xmath55(133,35)@xmath56(128,65)@xmath57 ( 14,25)(1,0)118(14,32)(25,39)(56,43)(56,43)(70,43)(83,37)(83,37)(95,34)(112,43)(112,43)(118,47)(125,42)(125,42)(127,40)(132,36 ) ( 13,40)(25,49)(56,55)(56,55)(70,55)(83,59)(83,59)(95,63)(112,62)(112,62)(120,62)(126,67 ) an observer field @xmath5 also generates a flow representing ` time evolution ' .
since @xmath5 is normalized , this flow is parameterized by _ proper time _ of the observer field : ( 150,100 ) ( 5,0)(20,20)(10,50)(10,50)(0,80)(10,100 ) ( 130,0)(135,20)(130,50)(130,50)(120,80)(130,100 ) ( 25,0)(1,3)5(50,0)(0,3)14(75,0)(-1,3)5(100,0)(1,3)5 ( 25,25)(0,3)14(50,25)(1,3)5(75,25)(2,3)13(100,25)(2,3)13(125,25)(0,3)15 ( 25,50)(-1,3)5(35,50)(0,1)14(50,50)(2,3)13(75,50)(2,3)13(100,50)(1,3)5 ( 25,75)(1,3)5(50,75)(2,3)13(75,75)(1,3)5(100,75)(2,3)13 ( 30,0)(40,40)(50,50)(50,50)(55,58)(67,72)(67,72)(70,74)(80,100 ) ( 52,0)(55,30)(72,50)(72,50)(80,60)(95,75)(95,75)(100,81)(108,100 ) ( 85,0)(90,25)(105,50)(105,50)(118,70)(125,100 ) thus , while we have no global notion of space , there is a canonical way to push the _ whole spacetime _
forward by one second of proper time of the observers .
but how do we define _ phase space _ without a foliation ?
the * covariant phase space * of general relativity is its space of solutions , a natural covariant generalization of the ` canonical ' phase space @xcite .
however , not dividing spacetime into space and time , it lacks any obvious link to the conceptual picture of spatial configurations changing in time .
the observer - based formulation could provide this link .
on one hand , if we choose an observer field corresponding to a foliation , we recover canonical gravity . on the other , everything transforms covariantly under change of observer , a local gauge choice .
adjoining the observer field @xmath1 gives us a covariant phase space in which spatial and temporal variables are clearly distinguished , without spoiling local lorentz symmetry .
in general relativity , just as there is no canonical spacelike foliation , there is no canonical choice of observer .
faced with such a situation , rather than making an arbitrary choice , one can _ simultaneously consider all possible choices .
_ individual choices are arbitrary ; the _ space _ of choices is canonical . on the other hand , * observer space * , the space of all possible observers , has manifest physical meaning , and simple topology : it is a 7-dimensional manifold isomorphic to the ` unit future tangent bundle ' of spacetime , locally a product of spacetime with velocity space @xmath2 . in @xcite
we reformulate general relativity directly on observer space , essentially by pulling fields back along the natural projection @xmath58 a connection pulled back to observer space will be flat in the ` velocity ' directions , reflecting the symmetry under changes of observer .
general relativity respects this symmetry , but does _ nature _ ? as we and all our instruments are ` observers , ' we can not probe spacetime geometry directly in any observer - independent way .
the empirical evidence for symmetry under a change of observer could be challenged by future observations .
several modifications of general relativity currently of interest can be studied using observer space .
first , there is a growing interest in models that do not treat spacetime isotropically @xcite . since points in observer space
correspond to directions in spacetime , these anisotropic theories might be described very naturally in these terms .
perhaps more compelling is the question of whether spacetime itself plays any fundamental role in physics .
once we have lifted the theory to observer space , do we still have any need for spacetime ?
in fact , starting with observer space , we can _
reconstruct spacetime_but only when certain ` integrability conditions ' hold @xcite .
the ` relative locality ' proposal @xcite suggests the notion of spacetime itself may be _ observer - dependent_. observer space provides a natural perspective from which to study this possibility , with the potential to move beyond ` special ' and on to ` general ' relative locality .
r. arnowitt , s. deser , and c. w. misner , the dynamics of general relativity , in _ gravitation : an introduction to current research _ , edited by l. witten ( wiley , new york , 1962 ) .
reprint available as http://arxiv.org/abs/gr-qc/0405109[arxiv:gr-qc/0405109 ] . c. crnkovi and e. witten , covariant description of canonical formalism in geometric theories , in _ three hundred years of gravitation _ , edited by s. w. hawking and w. israel ( cambridge university press , cambridge , 1987 ) .
m. j. gotay , j. isenberg , j. e. marsden , and r. montgomery , momentum maps and classical relativistic fields .
part i : covariant field theory , http://arxiv.org/abs/physics/9801019[arxiv:physics/9801019 ] .
j. barbour , shape dynamics .
an introduction , to appear in proceedings of the conference quantum field theory and gravity ( regensburg , 2010 ) , http://arxiv.org/abs/1105.0183/[arxiv:1105.0183 ] ; h. gomes , s. gryb , and t. koslowski , einstein gravity as a 3d conformally invariant theory , _ class .
_ * 28 * , 045005 ( 2011 )
. http://arxiv.org/abs/1010.2481/[arxiv:1010.2481 ]
. g. amelino - camelia , l. freidel , j. kowalski - glikman , and l. smolin , relative locality : a deepening of the relativity principle , second prize in the 2011 essay competition of the gravity research foundation , http://arxiv.org/abs/1106.0313/[arxiv:1106.0313 ] . | hamiltonian gravity , relying on arbitrary choices of ` space , ' can obscure spacetime symmetries .
we present an alternative , manifestly spacetime covariant formulation that nonetheless distinguishes between ` spatial ' and ` temporal ' variables .
the key is viewing dynamical fields from the perspective of a _ field of observers_a unit timelike vector field that also transforms under local lorentz transformations
. on one hand , all fields are spacetime fields , covariant under spacetime symmeties . on the other , when the observer field is normal to a spatial foliation , the fields automatically fall into hamiltonian form , recovering the ashtekar formulation .
we argue this provides a bridge between ashtekar variables and covariant phase space methods .
we also outline a framework where the ` space of observers ' is fundamental , and spacetime geometry itself may be observer - dependent .
essay written for the gravity research foundation + 2012 awards for essays on gravitation . |
let @xmath0 denote the group of orientation - preserving smooth diffeomorphisms of the circle .
it is well known that @xmath0 is a smooth frchet
lie group ( * ? ? ?
let @xmath1 denote the simply connected virasoro group , that is , the central extension by @xmath2 of the simply connected covering of @xmath0 ( see section [ sec - viras ] ) .
unitary representations of @xmath1 have been studied for more than three decades . among unitary representations of @xmath1 , the representations of _ positive energy _
play an important role in physics @xcite @xcite , @xcite .
the positive energy representations are closely related to the unitarizable modules of the virasoro algebra @xmath3 , which is the ( unique ) central extension by @xmath4 of the lie algebra @xmath5 of complex vector fields on the circle with finite fourier series .
the lie bracket of @xmath3 is given by @xmath6 = i(m - n)e^{(m+n)i}\frac{d}{dt}+\delta_{m ,- n}\frac{m(m^2 - 1)}{12}\kappa.\ ] ] the classification of irreducible unitarizable highest weight modules of @xmath3 is by now well known @xcite , @xcite , @xcite .
furthermore , every such module of @xmath3 can be integrated to an irreducible positive energy unitary representation @xmath7 of @xmath1 ( see @xcite and @xcite ) .
here positive energy means that the ( unbounded self - adjoint ) energy operator @xmath8 , where @xmath9 , has a non - negative spectrum ( see definition [ pos - energy - dfeen ] ) . in this paper
, we give a complete classification of the ( in general reducible ) positive energy representations of @xmath1 .
our main result is theorem [ thm - main ] , in which we prove that every positive energy representation of @xmath1 can be expressed in an essentially unique way as a direct integral of irreducible unitary highest weight representations .
let us briefly sketch the main ideas of the proof .
it is well known that the homogeneous manifold @xmath10 has an invariant complex structure .
( this complex manifold is usually called the kirillov - yuriev domain @xcite . )
we show that positive energy representations of @xmath1 can be realized on hilbert spaces of sections of certain holomorphic vector bundles @xmath11 ( defined in section [ sec - hvebun ] ) over the kirillov yuriev domain .
the construction of @xmath11 is canonical .
what is more subtle is to equip @xmath11 with an invariant complex structure .
we will achieve this using the key observation that the @xmath11 are indeed bundles associated to a `` universal '' principal bundle @xmath12 ( defined in section [ ssec - univ ] ) .
therefore an invariant complex stucture on @xmath12 will induce the suitable invariant complex stuctures on the @xmath11 .
the complex structure of the principal bundle @xmath12 is in turn obtained using the one - parameter family of complex structures on @xmath1 that is constructed by lempert @xcite . after realizing positive energy representations on hilbert spaces of holomorphic sections
, we can use the technique of holomorphic induction from @xcite , as well as some new ideas concerning holomorphic reproducing kernels with values in a @xmath13-algebra , to obtain the existence and uniqueness of the desired direct integral decomposition .
we now summarize the content of each of the sections of this paper . in section [ sec - infinitedim ] ,
we briefly review the basic theory of infinite dimensional manifolds and lie groups . in section [ sec - cplxstr ] , we first study invariant complex structures on a general class of principal bundles that arise from lie groups that have a homogeneous space which can be equipped with an invariant complex structure .
next we study the properties of @xmath14-invariant reproducing kernel hilbert spaces of sections of vector bundles associated to the aforementioned principal bundles . for reducible positive energy representations , self - adjoint operators such as the energy operator can have a continuous spectrum and therefore one can not expect to use methods that are typically based on weight vectors . to circumvent this issue , we need a technical tool , called _ arveson spectral subspaces _ , which will be presented in section [ sec - arv ] .
it is worth mentioning that the setup of section 3 is quite general and we expect that it will be applicable to other classes of frchet lie groups . in section [ sec - viras ] ,
we focus our attention on @xmath1 , construct suitable direct integrals of positive energy representations , and prove our main theorem .
+ * acknowledgement . *
the second author thanks professor laszlo lempert for illuminating discussions about his work on the virasoro group . during the completion of this project ,
the second author was supported by an nserc discovery grant and the emerging field project `` quantum geometry '' of fau erlangen nrnberg .
let @xmath15 and @xmath16 be locally convex spaces and @xmath17 be open . given a map @xmath18
, for every @xmath19 and every @xmath20 we define the directional derivative @xmath21 by @xmath22 if the limit exists .
a continuous map @xmath23 is called _ continuously differentiable _ , or @xmath24 , if the maps @xmath25 are defined for all @xmath20 , and the map @xmath26 is continuous .
similarly , the map @xmath27 is called @xmath28 for some positive integer @xmath29 , if the maps @xmath30 are defined and continuous for every @xmath31 .
finally , @xmath27 is called _ smooth _ , or @xmath32 , if it is @xmath28 for every @xmath33 .
based on the above notion of smooth maps we can define smooth locally convex manifolds ( see @xcite and @xcite for detailed expositions ) . from now on , all manifolds are assumed to be smooth . if @xmath34 and @xmath35 are manifolds , then the set of smooth maps from @xmath34 to @xmath35 is denoted by @xmath36 .
let @xmath34 and @xmath35 be manifolds and @xmath37 be a @xmath24 map . if @xmath38 and @xmath39 , then we sometimes write @xmath40 instead of @xmath41 or @xmath42 .
we extend @xmath43 to a map between the complexified tangent bundles , that is , if @xmath44 , then we set @xmath45 . an _ almost complex structure _ on a manifold @xmath34 is a smooth bundle map @xmath46 with the property that @xmath47 is an @xmath2-linear map satisfying @xmath48 .
we can write @xmath49 where @xmath50 and @xmath51 are the @xmath52 eigenspaces of the @xmath4-linear extension of @xmath53 .
the canonical injection @xmath54 results in an isomorphism @xmath55 as complex vector spaces . a smooth map @xmath37 between two almost complex manifolds is called _ holomorphic _ if @xmath56 . the set of holomorphic maps from @xmath34 to @xmath35
is denoted by @xmath57 .
a manifold @xmath34 is called a _ complex manifold _
if it is locally biholomorphic to a complex locally convex space , and the transition maps between the local charts of @xmath34 are holomorphic .
for the theory of holomorphic maps between complex manifolds we refer the reader to @xcite .
an almost complex structure on a real manifold @xmath34 is called _ integrable _ if it is induced by a complex manifold structure on @xmath34 .
the almost complex structure is called _ formally integrable _ if the nijenhuis tensor @xmath58+j[j\vfx,\vfy]+j[\vfx , j\vfy]-[j\vfx , j\vfy]$ ] vanishes for every two locally defined vector fields @xmath59 . if @xmath34 is finite dimensional , then by the newlander nirenberg theorem the notions of integrable and formally integrable coincide .
this is no longer true for infinite dimensional manifolds @xcite .
let @xmath15 be a locally convex space and @xmath17 be an open set .
every smooth vector field @xmath60 on @xmath61 results in a derivation of @xmath62 defined by @xmath63 for every @xmath64 .
the proof of the following lemma is a straightforward induction .
[ uey1jyijn ] let @xmath65 be smooth vector fields on an open set @xmath17 .
then there exists a family of smooth vector fields @xmath66 on @xmath61 , depending on @xmath65 , such that @xmath67 for every @xmath68 and every @xmath64 .
[ x1x2dense ] let @xmath15 be a complex locally convex space , @xmath17 be open , @xmath69 be a family of smooth vector fields on @xmath61 , @xmath70 , and @xmath71 .
let @xmath72 be a positive integer such that @xmath73 if @xmath74 is dense in @xmath75 , then @xmath76 for every @xmath77 and every @xmath78 . if @xmath79 then @xmath80 for every @xmath81 .
since the map @xmath82 is @xmath4-linear and continuous , it follows that @xmath83 for every @xmath84 .
the case @xmath85 is proved by induction on @xmath72 . if @xmath77 , then lemma [ uey1jyijn ] implies that the difference @xmath86 can be written as a sum of directional derivatives of the form @xmath87 where @xmath88 and @xmath89 . by the induction hypothesis , all of the latter terms vanish , and therefore @xmath90 for every @xmath91 .
since @xmath92 is @xmath4-multilinear and continuous , it follows that @xmath76 for every @xmath78 .
we assume that all lie groups are modeled on frchet spaces and have a smooth exponential map .
the exponential map @xmath93 , where @xmath94 , will be denoted by @xmath95 .
lie group homomorphisms are always assumed to be smooth . the universal enveloping algebra of @xmath96 will be denoted by @xmath97 . by a
_ complex lie group _ we mean a lie group that is also a complex manifold with holomorphic multiplication and inversion maps . if @xmath98 and @xmath99 are complex lie groups , by a _ homomorphism of complex lie groups _
@xmath100 , we mean a holomorphic map which is also a group homomorphism . for a lie group @xmath14
we define @xmath101 and @xmath102 by @xmath103 and @xmath104 . if @xmath105 for some @xmath106 , then for simplicity we write @xmath107 and @xmath108 instead of @xmath109 and @xmath110 .
let @xmath14 be a lie group and @xmath34 be a manifold .
if @xmath111 is a smooth map , where @xmath112 is open , then we define @xmath113 if @xmath34 has an almost complex structure @xmath46 , then we define @xmath114 let @xmath15 be a complex locally convex space .
we denote the vector space of continuous linear operators on @xmath15 by @xmath115 .
let @xmath116 denote the group of invertible continuous linear operators on @xmath15 . by a _ strongly continuous _
representation of a lie group @xmath14 on @xmath15 we mean a group homomorphism @xmath117 such that for every @xmath20 , the orbit map @xmath118 defined by @xmath119 is continuous . given a strongly continuous representation @xmath117
, we define @xmath120 if the limit exists .
the group of unitary operators of a complex hilbert space @xmath121 will be denoted by @xmath122 .
a strongly continuous representation @xmath123 is called a _ unitary representation_. a unitary representation @xmath7 is called _ smooth _ if the set @xmath124 is a dense subspace of @xmath121 .
we equip @xmath125 , @xmath126 , and @xmath122 with the operator norm topology .
let @xmath14 be a connected lie group and @xmath127 be a subgroup which is also a lie group .
we assume that @xmath128 is a _ split lie subgroup _ , that is , there exists an open subset @xmath129 of some locally convex vector space @xmath15 and a smooth map @xmath130 such that the map @xmath131 is a diffeomorphism onto an open @xmath132-neighborhood of @xmath14 .
it follows that @xmath14 is a smooth principal @xmath128-bundle over @xmath133 for a suitable manifold structure on @xmath133 .
( in fact the standard argument from the finite dimensional case carries over . )
set @xmath134 , @xmath94 , and @xmath135 .
let @xmath136 be the canonical projection , and set @xmath137 .
let @xmath138 be another connected lie group and set @xmath139 .
let @xmath140 be a lie group homomorphism , and @xmath141 denote the associated bundle corresponding to the smooth action @xmath142 thus , @xmath143 is the quotient of @xmath144 by the right action of @xmath128 defined by @xmath145 the associated bundle @xmath143 is a @xmath14-homogeneous principal @xmath138-bundle over @xmath34 .
we will denote the elements of @xmath143 by equivalence classes @xmath146 $ ] where @xmath147 and @xmath148 .
there is a transitive left action of @xmath144 on @xmath143 given by @xmath149:=[gg',s's^{-1}]\text{\ \ for every } ( g , s)\in g\times s \text { and every } [ g',s']\in \bs.\ ] ] let @xmath150 denote the graph of @xmath151 .
the group @xmath152 is a split lie subgroup of @xmath144 because the map @xmath153 is a diffeomorphism onto an open subset of @xmath144 .
therefore @xmath144 is a smooth principal @xmath152-bundle over @xmath154 .
the lie algebra of @xmath152 is @xmath155 where @xmath156 is the differential of @xmath151 at @xmath157 .
the map @xmath158\mapsto ( g , s^{-1})\gamma(\gamma)\ ] ] is @xmath144-equivariant , and using local charts one can prove that @xmath159 is a diffeomorphism .
we set @xmath160)$ ] . in what follows
, it will be easier to work with @xmath154 than with @xmath143 . from now on
we assume that @xmath138 is a complex lie group , and @xmath34 is a complex manifold with a @xmath14-invariant complex structure that corresponds to a closed lie subalgebra @xmath161 satisfying @xmath162 the latter correspondence is given by the canonical isomorphism @xmath163 .
in fact every @xmath14-invariant formally integrable almost complex structure on @xmath34 corresponds to a lie subalgebra @xmath164 satisfying .
[ galcpx ] the formally integrable @xmath14-invariant almost complex structures on @xmath143 for which the projection map @xmath165 and the right action of @xmath138 on @xmath143 are holomorphic are in one to one correspondence with continuous lie algebra homomorphisms @xmath166 which satisfy @xmath167 and @xmath168 [ prp - cplxggs ] suppose that @xmath143 is equipped with an almost complex structure that satisfies the properties of the proposition . by the diffeomorphism , the almost complex structure of @xmath143 can be transferred to a @xmath144-invariant formally integrable almost complex structure on @xmath154 with the extra property that the projection map @xmath169 and the action map @xmath170 are holomorphic .
let @xmath171 denote the lie subalgebra corresponding to the almost complex structure of @xmath154 by the canonical identification @xmath172 .
thus , @xmath173 satisfies @xmath174 let @xmath175 , where @xmath176 and @xmath177 .
holomorphy of the map implies that @xmath178 and @xmath179 .
but if @xmath180 then @xmath181 , hence @xmath182 .
let @xmath183 be the projection onto the first component .
holomorphy of the projection map @xmath169 implies that @xmath184 .
writing @xmath185 , where @xmath186 , we obtain that @xmath187 and @xmath188 thus there exists a @xmath4-linear map @xmath189 such that @xmath190 . since @xmath191 is a lie subalgebra of @xmath192 , the map @xmath193 is a lie algebra homomorphism .
moreover , the closed graph theorem for frchet spaces ( * ? ? ?
17 ) implies that @xmath193 is continuous .
let @xmath194 be the complex structure of @xmath195 and set @xmath196 , where @xmath197 is the canonical identification @xmath198 . since @xmath199 is a lie algebra homomorphism ,
so is @xmath200 . by the above discussion @xmath201 . setting @xmath202
, we obtain @xmath203 therefore the relation @xmath204 implies that @xmath167 , and follows from @xmath205 .
conversely , suppose we are given a @xmath206 which satisfies the conditions of the proposition .
we define @xmath207 as in .
it is straightforward to check that @xmath207 is a lie subalgebra of @xmath192 which satisfies @xmath208 .
the relation @xmath209 follows from .
next we prove that @xmath210 .
the relation @xmath167 implies that @xmath211 , and from it follows that @xmath212 therefore @xmath213 , which implies that @xmath214 .
we now identify @xmath215 with @xmath216 to obtain a formally integrable @xmath144-invariant almost complex structure on @xmath154 . by the invariance of almost complex structures , holomorphy of the projection
@xmath169 is equivalent to @xmath217 , which is a consequence of .
similarly , holomorphy of the action of @xmath138 on @xmath154 follows from @xmath179 .
let @xmath206 be a continuous lie algebra homomorphism which satisfies the conditions of proposition [ galcpx ] .
suppose that @xmath143 is equipped with the formally integrable almost complex structure that corresponds to @xmath200 .
our next goal is to determine if @xmath143 is indeed a holomorphic principal @xmath138-bundle .
this is achieved in theorem [ thm - vgyekl ] .
let @xmath218 be an open set and @xmath219 be a smooth section .
then we can write @xmath220 as @xmath221\text{\ \ for every } g\in\pgh^{-1}(u),\ ] ] where @xmath136 is the canonical projection , and @xmath222 is a smooth map satisfying @xmath223 for @xmath224 and @xmath225 .
[ lem - schol ] the section @xmath220 is holomorphic if and only if @xmath226 for every @xmath227 and every @xmath228 .
let @xmath229 be defined as in .
the section @xmath230 is holomorphic if and only if @xmath231 let @xmath232 be defined by @xmath233 .
then is equivalent to @xmath234 from and it follows that if @xmath230 is holomorphic , then for every @xmath227 and every @xmath228 , there exists an @xmath235 such that @xmath236 .
the latter equality can be written as @xmath237 .
writing @xmath238 , where @xmath239 , and @xmath240 , where @xmath241 and @xmath242 is the complex structure of @xmath195 , we obtain @xmath243 it follows that @xmath244 and @xmath245 , hence @xmath246 .
the proof of the converse is similar .
[ thm - vgyekl ] suppose that @xmath143 is equipped with the almost complex structure that corresponds to @xmath206 according to proposition [ galcpx ] .
then the following two statements are equivalent .
* @xmath143 is a holomorphic principal @xmath138-bundle over @xmath34 .
* there exists an open @xmath132-neighborhood @xmath247 and a smooth map @xmath248 satisfying @xmath249 moreover , when ( ii ) holds , the complex manifold structure of @xmath143 is uniquely determined by the holomorphic section of @xmath143 given by @xmath250 $ ] in some open neighborhood of the base point @xmath251 .
the implication ( i)@xmath252(ii ) is an immediate consequence of lemma [ lem - schol ] .
next we prove ( ii)@xmath252(i ) .
recall that @xmath128 is a split lie subgroup of @xmath14 , that is , the map @xmath253 given in is a diffeomorphism .
after shrinking @xmath254 we can assume that there exists a connected open subset @xmath255 and a connected open @xmath132-neighborhood @xmath256 in @xmath128 such that the map @xmath257 is a diffeomorphism onto @xmath254 .
let @xmath258 and @xmath259\to h$ ] be a smooth curve such that @xmath260 .
for every @xmath261 $ ] we have @xmath262 , where @xmath263\to \g h$ ] is a smooth map
. then @xmath264 therefore @xmath265 for every @xmath266 and every @xmath267 . as a result
, we obtain an extension of @xmath27 to a smooth map @xmath268 defined by @xmath269 for every @xmath258 and every @xmath224 .
we now define a smooth section @xmath270\text{\ \ for every } g\in \sigma(\sigma')h.\ ] ] if @xmath228 , @xmath266 , and @xmath224 , then @xmath271 therefore lemma [ lem - schol ] implies that @xmath230 is a holomorphic section of @xmath143 . for every @xmath272 we set @xmath273 and define @xmath274 by @xmath275 . note that @xmath276 for every @xmath228 and every @xmath277
. therefore the charts @xmath278\text{\ \ for every } gh\in v_a\text { and every } s\in s,\ ] ] are holomorphic ( lemma [ lem - schol ] ) and turn @xmath143 into a holomorphic principal @xmath138-bundle over @xmath34 .
let @xmath206 be a lie algebra homomorphism that satisfies the conditions of proposition [ galcpx ] and theorem [ thm - vgyekl](ii ) , and suppose that @xmath143 is equipped with the almost complex structure corresponding to @xmath200 .
let @xmath279 be a complex hilbert space , and @xmath280 be a homomorphism of complex lie groups .
let @xmath281 denote the vector bundle over @xmath34 associated to the principal @xmath128-bundle @xmath282 , corresponding to the smooth action @xmath283 that is , @xmath11 is the quotient of @xmath284 by the right action of @xmath128 defined by @xmath285 we define the associated vector bundle @xmath286 similarly .
since the principal @xmath138-bundle @xmath287 and the action of @xmath138 on @xmath279 are holomorphic , @xmath286 is a holomorphic vector bundle over @xmath34 .
the complex manifold structure on @xmath286 can be transferred to @xmath11 by the @xmath14-equivariant diffeomorphism @xmath288,w]\mapsto [ g,\rho(s)w ] \text { \,for } g\in g,\ s\in s,\text { and } w\in\ccw.\ ] ] from now on , we always assume that @xmath11 is equipped with the latter @xmath14-invariant holomorphic vector bundle structure .
let @xmath289 be the projection of the principal @xmath138-bundle @xmath143 over @xmath34 .
then we can write every holomorphic section @xmath290 as @xmath291 $ ] , where @xmath292 is a holomorphic function .
let @xmath293 denote the space of holomorphic sections of the bundle @xmath11 .
we then obtain a linear bijection @xmath294 this bijection intertwines the natural action of @xmath14 on @xmath295 with the action on @xmath296 defined by @xmath297 [ rem : loctriv ] suppose that the bundle @xmath298 has a holomorphic section @xmath299 over the open subset @xmath300 . by lemma [ lem - schol ] , @xmath301 can be written as @xmath302 $ ] , where @xmath303 is a smooth map satisfying @xmath265 for every @xmath147 and every @xmath224 , and @xmath304 for every @xmath147 and every @xmath228 .
the local trivialization of @xmath305 results in a holomorphic trivialization of @xmath306 given by @xmath307.\ ] ] sometimes it is also convenient to realize holomorphic sections of @xmath11 as functions on @xmath14 .
let @xmath308 denote the vector space of smooth maps @xmath309 satisfying @xmath310 the group @xmath14 acts on @xmath308 by left translation , that is , @xmath311 for @xmath312 .
let @xmath313 denote the vector space of maps @xmath314 which satisfy @xmath315 let @xmath316 denote the space of smooth sections of @xmath11 .
then we have a linear isomorphism @xmath317 where @xmath318 $ ] .
this isomorphism restricts to a @xmath14-equivariant isomorphism @xmath319 we now present some results on unitary representations of @xmath14 which can be realized on hilbert spaces that are embedded in @xmath320 , or equivalently , in @xmath313 .
our main tool is the theory of reproducing kernel hilbert spaces . for a detailed exposition of the theory ,
see @xcite .
let @xmath321 be a hilbert space such that , for every @xmath147 , the point evaluation @xmath322 , @xmath323 is continuous . set @xmath324 where @xmath325 denotes the adjoint of @xmath326 .
now let us denote the manifold @xmath143 , equipped with the opposite complex manifold structure , by @xmath327 . by the isomorphism , we can embed @xmath121 in @xmath328 .
the corresponding point evaluations @xmath329 , @xmath330 , are continuous .
set @xmath331 we call the maps and the _ reproducing kernel _ of @xmath121 on @xmath14 , resp . , @xmath143 .
it is easy to check that @xmath332,[g',s'])=\rho(s^{-1})\breve k(g , g')\rho({s'}^{-1})^*.\ ] ] by a standard fact in the theory of reproducing kernel hilbert spaces ( * ? ? ?
i.1.5 ) , the hilbert space @xmath121 is uniquely determined by @xmath333 , resp . , @xmath334 .
a map @xmath335 is called _ @xmath336-invariant _ if it is invariant under the diagonal @xmath14-action on @xmath337 and satisfies @xmath338 we call @xmath334 _ positive definite _ if for every @xmath339 , every @xmath340 , and every @xmath341 , we have @xmath342 .
[ ginvdfn ] a hilbert space @xmath321 with continuous point evaluations @xmath326 , @xmath147 , is called _
@xmath14-invariant _ if it satisfies the following conditions . *
the map @xmath343 is an isometry .
* @xmath121 is an invariant subspace of @xmath313 under the action of @xmath14 by left translation .
the so obtained @xmath14-representation @xmath7 is unitary .
[ ginvr ] let @xmath321 be a hilbert space with continuous point evaluations @xmath326 , @xmath147 and reproducing kernel @xmath344
. then @xmath121 is @xmath14-invariant if and only if @xmath333 satisfies @xmath345 ( cf .
i.4 ) ) . for the sake of completeness
, we recall the argument .
suppose that @xmath121 is @xmath14-invariant
. then @xmath346 .
since the action of @xmath14 on @xmath121 by @xmath347 is unitary , the equality @xmath348 leads to @xmath349 conversely , suppose that holds .
then @xmath350 , so that @xmath351 is an isometry .
next we set @xmath352 .
the relation @xmath353 implies that @xmath354 is a dense subspace of @xmath121 because its orthogonal subspace is trivial .
the invariance of the kernel @xmath333 now implies that @xmath355 which leads to @xmath356 from it follows that @xmath354 is invariant under the action of @xmath14 by left translations .
moreover , also implies that @xmath357 for @xmath358 and @xmath359 .
it follows that the action of every @xmath360 by left translation on @xmath354 extends uniquely to an isometry @xmath361 .
that @xmath362 for every @xmath363 and every @xmath364 follows from @xmath365 finally , from continuity of the matrix coefficients @xmath366 it follows that @xmath7 is strongly continuous .
[ prpevinbe ] for a @xmath14-invariant hilbert space @xmath321 , the following statements are true . *
the unitary @xmath14-representation @xmath367 is smooth .
in fact , the subset @xmath368 is a total subset of @xmath121 consisting of smooth vectors .
* the subspace @xmath369 is closed and @xmath128-invariant . *
set @xmath370 for every @xmath224 .
then @xmath371 is a unitary representation of @xmath128 and @xmath372 is a unitary equivalence between the @xmath128-representations @xmath373 and @xmath374 . *
the reproducing kernel @xmath375 of @xmath121 is holomorphic and @xmath336-invariant and satisfies @xmath376 for @xmath377 $ ] .
\(i ) to prove that @xmath7 is smooth , it is enough to show that @xmath378 for @xmath147 , @xmath379 . since @xmath380 , the map @xmath381 , @xmath382 is smooth
. therefore ( * ? ? ?
* thm 7.2 ) implies that @xmath378 .
\(ii ) clearly @xmath383 is closed because @xmath384 is an isometry . for every @xmath224 , @xmath385 and therefore @xmath386 .
therefore implies that @xmath387 for every @xmath224 . in particular
, @xmath383 is @xmath128-invariant .
\(iii ) since @xmath388 is unitary and @xmath351 is an isometry , from the equality @xmath386 ( proved in ( ii ) above ) it follows that the operator @xmath389 is also unitary . in particular , we obtain @xmath390 .
the relation @xmath391 now implies that @xmath373 and @xmath374 are unitarily equivalent .
\(iv ) for every @xmath147 , @xmath392 .
it follows that @xmath393 for every @xmath394 .
since @xmath7 is unitary , we obtain @xmath395 , and consequently @xmath396,[g',s'])\|\leq \|\rho(s^{-1})\|\cdot \|\rho({s'}^{-1})^*\|.\ ] ] this means that the map @xmath397 is locally bounded . moreover , by and we can embed @xmath121 in @xmath398 , so that the general theory of reproducing kernels ( * ? ? ?
a.iii.9(iii ) ) implies that the map @xmath397 is holomorphic .
the remaining statements follow from the relation between @xmath334 and @xmath333 , and @xmath399 .
[ adprp - gi - ii - iii ] let @xmath321 be a @xmath14-invariant hilbert space and @xmath7 be the unitary representation of @xmath14 by left translation , defined in definition [ ginvdfn ] .
set @xmath400 .
then , * @xmath401 is a dense subspace of @xmath121 .
* @xmath402 is a dense subspace of @xmath121 .
suppose , in addition , that there exists an @xmath403-invariant decomposition @xmath404 such that @xmath405 , where @xmath406 is the lie algebra defined in , and @xmath407 .
then , * @xmath408 is a dense subspace of @xmath121 .
* @xmath409 , where @xmath410 .
* assume that every closed @xmath411-invariant subspace of @xmath279 is also @xmath412-invariant .
set @xmath370 for every @xmath224 .
let @xmath413 , resp . ,
@xmath414 , denote the commutants of @xmath415 , resp .
, @xmath416 .
then the map @xmath417 is an isomorphism of von neumann algebras .
\(i ) from proposition [ prpevinbe](i ) and we have @xmath418 for @xmath147 and @xmath379
. therefore @xmath419 , and proposition [ prpevinbe](i ) implies that @xmath420 is a dense subset of @xmath121 .
( ii ) given a smooth map @xmath421 , where @xmath422 is a complex hilbert space , we define @xmath423 we extend @xmath424 to @xmath425 as usual , by setting @xmath426 and @xmath427 . set @xmath428 fix @xmath429 and @xmath430 , and set @xmath431 . by proposition [ prpevinbe](i ) , the orbit map @xmath432 , @xmath119 is smooth , and @xmath433 by differentiating both sides of the latter equality , we obtain @xmath434 next we prove that @xmath435 where @xmath436 is the principal anti - automorphism of @xmath437 , defined by @xmath438 for @xmath439 .
it suffices to prove the latter statement for @xmath440 where @xmath441 . for @xmath442 ,
the statement is obvious . for @xmath443 ,
we prove the statement by induction . for every @xmath444 , @xmath445 since @xmath446 is a dense subspace of @xmath121 , we obtain @xmath447 .
setting @xmath448 in , we now obtain @xmath449 .
it follows immediately that @xmath450 to complete the proof of ( ii ) , we need to show that @xmath451 .
fix @xmath364 such that @xmath452 for every @xmath425 . as in remark
[ rem : loctriv ] , let @xmath453 $ ] be a holomorphic section of @xmath305 , where @xmath61 is an open connected neighborhood of @xmath454
. then the map @xmath455 , @xmath456 , is holomorphic , and a computation using the leibniz rule shows that @xmath457 for every @xmath425
. set @xmath458 for every @xmath459 and every @xmath147 .
then @xmath460 is a smooth vector field on @xmath34 , and @xmath461 for every @xmath462 and every @xmath459 .
lemma [ x1x2dense ] implies that , in every local chart , we have @xmath463 for every @xmath464 and every @xmath465 . since @xmath466 is holomorphic it vanishes on @xmath61 .
it follows that @xmath467 .
\(iii ) first we prove that @xmath468 .
let @xmath469 , @xmath470 . applying with @xmath471
, we obtain for @xmath472 and @xmath473 the relation @xmath474 since @xmath472 was arbitrary , we obtain @xmath475 . from proposition [ prpevinbe](ii )
it now follows that @xmath476 and @xmath477 for every @xmath478 .
consequently , @xmath479 , and from ( ii ) we derive that @xmath408 is dense in @xmath121 .
\(iv ) we have shown in the proof of ( iii ) that @xmath480 .
next we prove that @xmath481 .
let @xmath482 .
then @xmath483 for every @xmath484 and every @xmath485 .
in particular , @xmath486 . from ( iii ) it follows that @xmath487 is dense in @xmath121 .
moreover , @xmath488 because @xmath489 this shows that @xmath490 .
\(v ) first note that ( iv ) implies that @xmath383 is invariant under @xmath413 .
thus , the map is well - defined .
now let @xmath491 such that @xmath492 .
then @xmath493 , so that ( i ) implies that @xmath494 . therefore the map is an injection .
next we prove surjectivity of . since every von neumann algebra is generated by orthogonal projections ( * ?
* chap . 1 , 1.2 ) , and
images of von neumann algebras under restriction maps are von neumann algebras ( * ? ? ?
* chap . 1 , 2.1 , prop . 1 )
, it is enough to show that every orthogonal projection in @xmath414 is contained in the image of the map .
let @xmath495 .
setting @xmath496 and @xmath497 yields an @xmath128-invariant orthogonal decomposition @xmath498 now set @xmath499 for @xmath500 .
surjectivity of follows if we can prove that @xmath501 , because then @xmath502 is a @xmath14-invariant orthogonal direct sum , and the orthogonal projection @xmath503 onto @xmath504 satisfies @xmath505 .
we now prove that @xmath506 . because of the relation @xmath391 ( proposition [ prpevinbe](iii ) )
, the decomposition corresponds to an orthogonal decomposition @xmath507 , where each @xmath508 , @xmath509 , is a closed and @xmath411-invariant subspace .
thus , each @xmath508 is @xmath412-invariant , and since @xmath373 is unitary , it is also @xmath510-invariant .
fix @xmath511 for @xmath509 , and let @xmath512 be the reproducing kernel of @xmath121 , as defined in . by proposition [ prpevinbe](iv ) , the map @xmath513 is holomorphic .
but if @xmath514 $ ] , then @xmath515 , and therefore the map vanishes on the diagonal subset of @xmath516 . since every holomorphic map on @xmath517 is uniquely determined by its diagonal values ( * ? ? ?
a.iii.7 ) , the map should also vanish for every @xmath518 .
consequently , for every @xmath511 , @xmath509 , we have @xmath519 , [ \yek,\yek])w_2\rag=0.\ ] ] that is , @xmath520 .
this implies that @xmath501 .
[ pexgi ] let @xmath375 be a holomorphic , @xmath336-invariant , positive definite map which satisfies @xmath521 for @xmath522 $ ] .
then there exists a @xmath14-invariant hilbert space @xmath523 whose reproducing kernel , as defined in , is equal to @xmath334 .
this follows from standard facts in the theory of reproducing kernels .
therefore we will only sketch the argument . by the isomorphisms
and it suffices to construct @xmath524 .
set @xmath525 where @xmath526 .
we can define a pre - hilbert structure on @xmath354 by the relation @xmath527 .
let @xmath121 denote the completion of @xmath354 . from (
i.1.9(iii ) ) it follows that @xmath528 .
continuity of point evaluations of @xmath121 follows from the equality @xmath529 for @xmath364 , @xmath330 , and @xmath430 ( see also ( * ? ? ?
* thm i.1.4 ) ) .
the latter equality also implies that @xmath530 , from which it follows that @xmath531 .
next note that @xmath532 , that is , @xmath533 is an isometry .
finally , the condition of definition [ ginvdfn](ii ) follows from @xmath336-invariance of @xmath334 , which implies the @xmath14-invariance of the kernel @xmath333 on @xmath534 .
[ patmston ] for every holomorphic representation @xmath535 , there exists at most one @xmath14-invariant hilbert space @xmath321 .
the hilbert space @xmath121 is uniquely determined by its reproducing kernel ( * ? ? ?
i.1.5 ) . by and we can assume @xmath524 , and it suffices to show that the corresponding @xmath336-invariant kernel @xmath375 is uniquely determined by @xmath536 . since @xmath537 is an isometry , we have @xmath538 . from @xmath336-invariance
we obtain @xmath332 , [ g , s ] ) = \rho(s^{-1 } ) k(p_\circ , p_\circ ) \rho(s^{-1})^ * = \rho(s^{-1 } ) \rho(s^{-1})^*\ ] ] for every @xmath539 .
therefore @xmath540 is uniquely determined by @xmath536 for every @xmath541\in \bs$ ] . from proposition [ prpevinbe](iv ) , and the fact that every holomorphic map @xmath542 is uniquely determined by its diagonal values ( * ? ? ?
a.iii.7 ) , it follows that @xmath543 is uniquely determined by @xmath536 for every @xmath544 .
we now recall some results from ( * ? ? ?
* appendix a ) on arveson spectral subspaces .
these spectral subspaces were originally introduced in @xcite .
let @xmath254 be a complete complex locally convex space .
a strongly continuous representation @xmath545 is called _ equicontinuous _ if , for every @xmath546-neighborhood @xmath547 , there exists a @xmath546-neighborhood @xmath548 such that @xmath549 for every @xmath550 .
an equicontinuous representation @xmath551 yields a representation @xmath552 , given by @xmath553 . for every @xmath554 , we set @xmath555 where @xmath556 .
similarly , we set @xmath557 for every @xmath558 , we define the _ arveson spectral subspace _
@xmath559 where @xmath560 denotes the closure of @xmath561 . from (
a.6 ) it follows that the vector space @xmath562 is a closed subspace of @xmath254 that is invariant under @xmath563 for every @xmath550 ( see also @xcite ) .
[ fl1rcpid ] let @xmath7 be a smooth unitary representation of @xmath14 .
let @xmath459 , and set @xmath564 for every @xmath550 .
then @xmath565 for every @xmath566 .
let @xmath567 denote the @xmath568-algebra of borel subsets of @xmath2 , and @xmath569 denote the spectral measure associated to the self - adjoint operator @xmath570 .
since @xmath571 , the operator @xmath572 is bounded .
fix @xmath573 and set @xmath574 for every @xmath575 .
then , @xmath576 [ p(i)-(iv ) ] let @xmath7 be a smooth unitary representation of @xmath14 and @xmath459 .
set @xmath564 and @xmath577 for every @xmath550 .
assume that @xmath551 is equicontinuous .
then the following statements hold .
* @xmath578 for every @xmath566 .
moreover , @xmath579 * set @xmath580 for every @xmath581
. then , @xmath582 * let @xmath569 denote the spectral measure associated to the self - adjoint operator @xmath570 , where @xmath567 is the @xmath568-algebra of borel subsets of @xmath2 .
let @xmath583 be an open interval .
then @xmath584 is dense in @xmath585 . *
assume that @xmath586 , and that @xmath7 is obtained , according to proposition [ prpevinbe ] , from an embedding @xmath587 as a @xmath14-invariant hilbert space , where @xmath280 is a homomorphism of complex lie groups .
furthermore , suppose that there exists an @xmath403-invariant decomposition @xmath404 such that @xmath405 , where @xmath406 is the lie algebra defined in , and @xmath588 .
set @xmath589 .
then , @xmath590 where @xmath591 , and the right hand side denotes the closure of @xmath592 .
\(i ) this is a special case of ( * ? ? ?
* thm 2.3 ) .
\(ii ) this is a special case of ( * ? ? ?
* thm 3.1 ) .
\(iii ) fix @xmath593 and @xmath594 .
choose @xmath595 such that @xmath596 .
choose compact intervals @xmath597 such that @xmath598 , and a sequence @xmath599 in @xmath600 such that @xmath601 , @xmath602 for all @xmath603 , and @xmath604 .
let @xmath605 be schwartz functions such that @xmath606 for every @xmath607 .
set @xmath608 . from ( i ) and lemma [ fl1rcpid ]
it follows that @xmath609 .
let @xmath610 denote the characteristic function of @xmath611 , so that @xmath612
. then @xmath613 , and therefore @xmath614
. moreover , @xmath615 ( see ( * ? ? ?
* thm xii.2.6(c ) ) ) , and consequently , @xmath616 finally , since @xmath617 , we obtain @xmath618 .
\(iv ) set @xmath619 and @xmath620 .
then @xmath621 , so that ( ii ) implies that @xmath622 , where @xmath623 . by proposition [ adprp - gi - ii - iii](iii )
, @xmath408 is a dense subspace of @xmath121 .
therefore @xmath624 is also a dense subspace of @xmath121 .
since @xmath624 is also closed in @xmath121 , we obtain @xmath625 .
in this section we use the general results of section [ sec - cplxstr ] in the special case of the virasoro group . let @xmath0 denote the group of orientation - preserving smooth diffeomorphisms of the circle . as shown in ( * ? ? ?
4.4.6 ) , @xmath0 is a smooth frchet
lie group .
note that @xmath626 , and the simply connected covering of @xmath0 is @xmath627 from now on @xmath14 will denote the simply connected virasoro group @xmath1 , that is , the central extension of @xmath628 by @xmath2 whose multiplication is defined by @xmath629 where @xmath630 is the bott cocycle @xmath631 .
the center of @xmath14 is @xmath632 where @xmath633 for every @xmath634 .
let @xmath94 .
we can represent every element of @xmath439 by an infinite series @xmath635 , where @xmath636 and @xmath637 here @xmath638 , and @xmath639 is chosen such that the lie bracket of @xmath640 is given by @xmath641=(n - m)\bd_{m+n}+2\pi im^3 \delta^{}_{m+n,0}\vecc\text { \,for every } m , n\in\z.\ ] ] note that in the above formula , the choice of the cocycle is different from which is more commonly used in the literature , but this will not have a significant effect on our presentation .
observe that @xmath642 , where the complex conjugation @xmath643 is uniquely specified by @xmath644 and @xmath645 .
moreover , @xmath646 for every @xmath647 . from now on , let @xmath128 denote the connected lie subgroup of @xmath14 which corresponds to the lie subalgebra @xmath648 . by a well - known result of kirillov and yuriev @xcite ,
the homogeneous manifold @xmath134 is a complex manifold ( diffeomorphic to a complex frchet domain ) with a @xmath14-invariant complex structure which corresponds to the lie subalgebra @xmath649 , where @xmath650 following @xcite , we define a one - parameter family of ( integrable ) left - invariant almost complex structures on @xmath14 .
namely , for every @xmath651 we write @xmath652 where @xmath653 the canonical identification @xmath654 yields a left - invariant almost complex structure on @xmath14 .
this almost complex structure induces an almost complex structure on @xmath128 corresponding to the direct sum decomposition @xmath655 , where @xmath656 and @xmath657 . from now on , to specify the almost complex structures chosen on @xmath14 and @xmath128 , we write @xmath658 and @xmath659 .
note that @xmath660 as complex lie groups .
[ kiriyur ] for every @xmath661 , the almost complex structure of @xmath658 is integrable . the complex manifold @xmath658 is a holomorphic principal @xmath659-bundle over @xmath34 with a global trivialization .
this is proved in ( * ? ? ?
* thm 4.2 ) .
actually lempert gives the proof for @xmath662 where @xmath663 is the discrete subgroup of @xmath14 generated by @xmath664 .
since the pullback of an integrable complex structure to a covering manifold is also integrable , it follows that lempert s result also holds for the covering group @xmath14 .
let @xmath665 denote the complexification of @xmath128 and @xmath666 denote the canonical embedding , so that @xmath667 is the canonical injection .
set @xmath668 , where the right hand side is the quotient of the action defined in .
we equip @xmath12 with the almost complex structure that corresponds , according to proposition [ galcpx ] , to the canonical projection @xmath669 .
fix @xmath671 and let @xmath672 be defined by @xmath673 , where @xmath674 is the canonical projection
. set @xmath675 , where @xmath676 is the identity map .
we equip @xmath677 with the almost complex structure that corresponds , according to proposition [ galcpx ] , to @xmath678 .
let @xmath679 be a holomorphic global section for the trivial bundle @xmath680 , and set @xmath681 the map @xmath682 given by @xmath683 $ ] is a holomorphic diffeomorphism .
therefore lemma [ lem - schol ] implies that @xmath684 for every @xmath671 , let @xmath685 denote the lie subgroup of @xmath670 corresponding to the lie subalgebra @xmath686 .
now fix @xmath687 such that @xmath688 .
the map @xmath689 , @xmath690 is an isomorphism of complex lie groups .
the map @xmath691 is smooth , and from it follows that @xmath692 for every @xmath228 and every @xmath147 .
therefore the proposition follows from theorem [ thm - vgyekl ] .
we now need the following important observation .
let @xmath693 be a lie group homomorphism . from (
* thm iii.1.5 ) it follows that @xmath536 can be extended uniquely to a homomorphism of complex lie groups @xmath694 .
therefore the construction of section [ sec - hvebun ] makes @xmath695 a holomorphic vector bundle over @xmath34 . from now on
, we assume that @xmath11 is equipped with the latter holomorphic vector bundle structure .
the @xmath128-invariance of @xmath699 follows from the relation @xmath701 and @xmath403-invariance of @xmath702 .
next we prove that @xmath703 is dense .
set @xmath704 .
then @xmath705 is a closed and @xmath14-invariant subspace of @xmath121 , and the representation of @xmath14 on @xmath705 is of positive energy .
now set @xmath706 let @xmath707 be the spectral measure associated to the self - adjoint operator @xmath708 . by proposition [ p(i)-(iv)](iii ) , there exists a nonzero @xmath709 .
now set @xmath710 , and define @xmath577 and @xmath564 for every @xmath550 .
it is straightforward to check that @xmath711 and that @xmath551 is equicontinuous . in view of @xmath712
we obtain with proposition [ p(i)-(iv)](ii ) that @xmath713)= \mathrm e_{[a - n-\frac12,a - n+\frac12]}\cch=\{0\ } \text{\,\ for every $ n>0$.}\ ] ] consequently , @xmath714 , which is a contradiction .
let @xmath7 be a positive energy representation of @xmath14 .
we fix some notation which will appear in the rest of this article .
set @xmath715 .
note that @xmath699 is @xmath128-invariant .
now let @xmath716 denote the unitary representation of @xmath128 that is obtained from the restriction of @xmath717 .
that is , @xmath718 for every @xmath224 .
let @xmath719 denote the spectral measure corresponding to @xmath716 , in the sense of stone s theorem ( * ? ? ?
* thm x.2.1 ) . here
@xmath720 denotes the dual of @xmath128 , and @xmath721 denotes the @xmath568-algebra of borel subsets of @xmath720 .
* @xmath693 is continuous with respect to the operator norm .
* @xmath693 is a lie group homomorphism . *
the spectral measure corresponding to @xmath536 has compact support .
* @xmath536 has a unique extension to a homomorphism of complex lie groups @xmath694 .
[ rglvgqendv ] let @xmath7 be a positive energy representation of @xmath14 , @xmath724 be a closed @xmath128-invariant subspace , and @xmath725 be the orthogonal projection on @xmath279 .
set @xmath726 for every @xmath224 , so that @xmath723 is a unitary representation of @xmath128 .
assume the following statements .
let @xmath738 denote the space of continuous sections of @xmath11 , equipped with the compact - open topology . by ( * ?
iii.12 ) , the space @xmath293 of holomorphic sections of @xmath11 is a closed subspace of @xmath738 . for every @xmath739
, we set @xmath740\text { for every } g\in g.\ ] ] then @xmath741 , and the map @xmath742 defined by @xmath743 is continuous .
we have shown above that @xmath737 for every @xmath733 . therefore from the isomorphism it follows that @xmath744 .
since @xmath446 is a dense subspace of @xmath121 , we obtain that @xmath745 , that is , @xmath737 for every @xmath739 . if @xmath746 for some @xmath747 , then @xmath748 , that is , @xmath749 .
therefore the map @xmath750 is an injection .
checking the @xmath14-equivariance relation @xmath751 , and continuity of point evaluations @xmath752 for @xmath147 , are straightforward .
the reproducing kernel of @xmath121 , as defined in , is given by @xmath753 , and therefore it satisfies .
remark [ ginvr ] now implies that @xmath121 is @xmath14-invariant .
the equality @xmath754 follows from proposition [ adprp - gi - ii - iii](iv ) .
[ pipij ] let @xmath7 be a positive energy representation of @xmath14 .
then there exists a sequence @xmath755 , where @xmath756 , of positive energy representations of bounded type such that @xmath757 , and the spectral measures @xmath758 , as defined in , are mutually singular .
let @xmath759 denote the one - parameter subgroup @xmath760 , and let @xmath761 denote the spectral measure of the self - adjoint operator @xmath762 .
then @xmath763 , where @xmath764 .
since @xmath765 , every @xmath766 is @xmath14-invariant .
therefore , from now on , we can assume that @xmath767 for some @xmath647 .
now let @xmath768 be the one - parameter subgroup @xmath769 and let @xmath770 denote the spectral measure of the self - adjoint operator @xmath696 .
set @xmath771 .
note that @xmath772 , because @xmath7 is a positive energy representation .
set @xmath773 and @xmath774 .
then @xmath775 is @xmath128-invariant , and from proposition [ p(i)-(iv)](iii ) it follows that @xmath776 is a dense subspace of @xmath775 . on the other hand , setting @xmath710 in proposition [ p(i)-(iv)](ii ) implies that @xmath777 ( cf .
the proof of proposition [ prpposcyq ] ) . consequently , @xmath778 is a dense subspace of @xmath775 .
now let @xmath779 denote the representation of @xmath128 defined by @xmath780 for @xmath224 .
the above arguments imply that the spectral measure of @xmath781 is compactly supported , and therefore @xmath782 is a smooth lie group homomorphism . from (
* ? ? ? * thm
iii.1.5 ) it follows that there exists a unique extension of @xmath781 to a homomorphism of complex lie groups @xmath783 .
set @xmath784 where @xmath785 , and let @xmath786 be the representation of @xmath14 defined by @xmath787 for @xmath147 . by proposition [ rglvgqendv ] , the representation @xmath786 can be realized as a @xmath14-invariant hilbert space in @xmath788 , and @xmath789 .
consequently , @xmath786 is of bounded type .
we now repeat the above process with the representation of @xmath14 on @xmath790 .
that is , we set @xmath791 , @xmath792 , @xmath793 , @xmath794 , and so on . note that it is possible that @xmath795 , but the spectral measures on @xmath775 and @xmath796 will be mutually singular , because @xmath797 .
let @xmath798 be an irreducible positive energy representation of @xmath14 .
then from ( * ? ? ?
4.11 ) it follows that @xmath798 is a _
highest weight _ representation , that is , @xmath799 is discrete and there is a unique eigenvector of @xmath800 in @xmath801 corresponding to its smallest eigenvalue .
note that proposition [ rglvgqendv ] and proposition [ patmston ] imply that @xmath798 is uniquely determined by its highest weight .
let @xmath802 be the set of unitary characters of @xmath128 which appear as highest weights of irreducible positive energy representations of @xmath14 . by the results of @xcite and @xcite ,
@xmath803 can be identified with the set of highest weights of unitarizable modules of the virasoro algebra , whose description is by now well known @xcite , ( * ? ? ?
3.3 ) , @xcite .
( smoothness of the irreducible positive energy representations is addressed in @xcite . )
it follows that @xmath804 is a closed subset of @xmath720 . in this paper
we do not need the explicit description of @xmath804 . by proposition [ pipij ]
we can assume that @xmath7 is of bounded type .
let @xmath807 denote the @xmath13-algebra generated by @xmath808 . by remark [ rmk : i - iv ]
, @xmath536 has a unique extension to a homomorphism of complex lie groups @xmath809 .
note that @xmath810 . * step 1 .
* by proposition [ rglvgqendv ] we can assume that @xmath7 is embedded as a @xmath14-invariant hilbert space in @xmath811 .
let @xmath812 be the corresponding reproducing kernel .
next we prove that @xmath813 for every @xmath814 .
if @xmath514 $ ] , then @xmath815 . now let @xmath816 denote the canonical projection .
the quotient @xmath817 is a complex banach space .
the map @xmath818 vanishes on the diagonal , and proposition [ prpevinbe](iv ) implies that @xmath819 is holomorphic .
therefore @xmath820 vanishes for every @xmath814 .
consequently , @xmath821 for every @xmath814 .
* the @xmath13-algebra of @xmath128 is isomorphic to @xmath822 , the algebra of complex - valued continuous functions on @xmath720 which vanish at infinity .
let @xmath823 denote the kernel of the canonical map @xmath824 then @xmath825 .
since @xmath826 is compact , it follows that the image of is a dense subspace of @xmath827 , and @xmath828 , where @xmath829 is the space of continuous complex - valued functions on @xmath830 . by elementary @xmath13-theory (
1.8 ) the map induces an isomorphism @xmath831 of @xmath13-algebras
. * step 3 . * fix @xmath832 and let @xmath833 denote the character of @xmath827 defined by the relation @xmath834 for every @xmath835 . note that @xmath836 for every @xmath224 , and since both sides are holomorphic functions of @xmath837 , equality holds for every @xmath837 as well .
now set @xmath838 .
it is clear that @xmath839 is holomorphic and @xmath840-invariant .
moreover , @xmath841 is positive definite , because every character @xmath842 is completely positive in the sense of ( * ? ? ?
iv.3.3 ) . consequently , by proposition [ pexgi ] , there exists a @xmath14-invariant hilbert space @xmath843 with reproducing kernel @xmath841 . * step 4 .
* let @xmath844 denote the smooth unitary representation of @xmath14 on @xmath845 that is defined in proposition [ prpevinbe](i ) . by the isomorphism we can assume @xmath846 .
set @xmath847 , so that @xmath848 . note that @xmath849 , because @xmath850 .
now s. kobayashi s criterion for irreducibility of holomorphic induction ( * ? ? ?
* thm iv.1.10 ) implies that @xmath844 is irreducible ( see also proposition [ adprp - gi - ii - iii ] ) .
it is straightforward to check that @xmath851 for every @xmath224 .
therefore proposition [ adprp - gi - ii - iii](iv ) implies that @xmath852 , and proposition [ p(i)-(iv)](iv ) implies that @xmath844 is a highest weight representation with highest weight @xmath853 .
thus from the definition of @xmath854 it follows that @xmath855 .
now recall that in step 3 , @xmath853 is an arbitrary element of @xmath856 .
consequently , @xmath857 .
[ adprpexgin ] let @xmath723 be a unitary representation of @xmath128 , and @xmath858 denote the spectral measure corresponding to @xmath536 .
assume that @xmath859 is a compact subset of @xmath854 .
then there exists a @xmath14-invariant hilbert space @xmath860 .
let @xmath861 denote the @xmath13-algebra generated by @xmath862 . by an argument similar to step 2 in the proof of proposition [ p - sseqh+ ]
, we obtain a canonical isomorphism of @xmath13-algebras @xmath863 . for every @xmath864 ,
let @xmath865 be the character of @xmath827 defined by @xmath866 for every @xmath867 , and let @xmath844 denote the irreducible highest weight representation corresponding to @xmath853 .
let @xmath868 be the one - dimensional @xmath853-weight space of @xmath869 .
next we use a construction similar to the universal representation of a @xmath13-algebra . set @xmath870 so that @xmath871 is a representation of @xmath14 , and @xmath872 is a representation of @xmath128 .
set @xmath873 from the definition of @xmath842 it follows that @xmath874 for every @xmath224 , so that @xmath875 for every @xmath224 .
moreover , @xmath876 is faithful because @xmath877 .
consequently , elementary @xmath13-algebra theory implies that @xmath878 is an isomorphism of @xmath13-algebras between @xmath827 and the @xmath13-algebra @xmath879 that is generated by @xmath880 .
since @xmath881 is a bounded unitary representation of @xmath128 , it has a unique holomorphic extension to @xmath670 and by proposition [ rglvgqendv ] we can embed @xmath871 into @xmath882 as a @xmath14-invariant hilbert space .
let @xmath883 be the @xmath884-invariant reproducing kernel of @xmath885 .
an argument similar to step 1 in the proof of proposition [ p - sseqh+ ] shows that @xmath886 for every @xmath814 .
the map @xmath887 is completely positive in the sense of ( * ? ? ?
iv.3.3 ) , and therefore the kernel @xmath888 , defined by @xmath889 , is positive definite . furthermore , the kernel @xmath334 is @xmath336-invariant , holomorphic , and satisfies @xmath538 .
consequently , proposition [ pexgi ] implies that there exists a @xmath14-invariant hilbert space @xmath860 corresponding to @xmath334 . for every @xmath855 ,
let @xmath844 denote the irreducible positive energy representation of @xmath14 with highest weight @xmath853 .
we will denote the inner product of @xmath845 by @xmath891 .
let @xmath892 be a finite positive borel measure on @xmath854 . in this section we will construct a direct integral unitary representation @xmath893 throughout this section
, @xmath894 will denote the cyclic unitary representation of @xmath128 corresponding to @xmath892 , that is , @xmath895 and the action of @xmath128 is defined by @xmath896 by proposition [ rglvgqendv ] , the representation @xmath844 can be realized as a @xmath14-invariant hilbert space in @xmath897 , and in fact proposition [ patmston ] implies that there exists a unique @xmath898-invariant holomorphic kernel @xmath899 satisfying @xmath900 .
set @xmath901 , so that @xmath902 .
it suffices to prove the statement for every subset of the form @xmath905 , where @xmath906 is open and relatively compact .
choose a finite positive borel measure @xmath892 on @xmath854 such that @xmath907 .
by proposition [ adprpexgin ] there exists a @xmath14-invariant hilbert space @xmath908 .
let @xmath909 be the reproducing kernel of @xmath910 , as defined in . by proposition [ prpevinbe](iv )
, @xmath911 is holomorphic .
let @xmath912 denote the @xmath13-algebra generated by @xmath913 .
an argument similar to step 1 in the proof of proposition [ p - sseqh+ ] shows that @xmath914 for every @xmath915 . by an argument similar to step 2 in the proof of proposition [ p - sseqh+ ]
, we obtain a canonical isomorphism of @xmath13-algebras @xmath916 . for every @xmath917 ,
let @xmath918 be the character of @xmath919 that is defined by the relation @xmath920 for every @xmath921 .
an argument similar to step 3 in the proof of proposition [ p - sseqh+ ] shows that @xmath922 is holomorphic , @xmath898-invariant , and positive definite .
moreover , @xmath923 .
uniqueness of the reproducing kernel of @xmath845 implies that @xmath924 the statement of the lemma now follows from continuity of the canonical map @xmath925 . to construct the direct integral @xmath926
, we consider the family @xmath927 as a field of hilbert spaces over @xmath854 ( in the sense of ( * ? ? ?
ii ) ) . fix a countable dense subset @xmath928 of @xmath143 , and consider the sections @xmath929 , defined by @xmath930 for every @xmath855 and every @xmath339 .
* for every @xmath855 , the set @xmath932 is a total subset of @xmath845 .
* for every @xmath933 , the map @xmath934 , @xmath935 , is measurable .
* for every @xmath147 , the operator field @xmath936 is measurable .
\(i ) recall that @xmath937 .
thus , for every @xmath938 , we have @xmath939 now let @xmath940 be a subsequence of @xmath941 that converges to a point @xmath330 . since @xmath841 is continuous , from it follows that @xmath942 in @xmath845 .
furthermore , proposition [ prpevinbe](i ) states that @xmath943 is a total subset of @xmath845 .
therefore @xmath944 is a total subset of @xmath845 .
\(iii ) by @xmath898-invariance of @xmath841 , we can write @xmath946 . therefore , by lemma [ lwhmmc ] , the map @xmath947 is continuous .
the statement is now proved using a standard pointwise convergence argument based on ( * ? ? ? * chap . 1 ,
1 , lem . 1 ) . we are now in a position to define @xmath890 .
a section @xmath948 is called _ measurable _ if , for every @xmath339 , the map @xmath949 is measurable . by standard facts in the theory of direct integral hilbert spaces
( see @xcite or ( * ? ? ?
* appendix ii ) ) , lemma [ add - lm - msrbity ] implies that the set @xmath950 , consisting of all measurable sections @xmath951 which satisfy @xmath952 , is a separable hilbert space with inner product @xmath953 . setting @xmath954 we obtain a representation of @xmath14 on @xmath950 by unitary operators ( but we still need to prove that @xmath955 is a smooth unitary representation of positive energy ) .
[ rmkcktnk ] let @xmath956 be a partition of @xmath854 into measurable and relatively compact subsets , and set @xmath957 for every @xmath958 and every @xmath339 .
then @xmath959 , and the @xmath960 are mutually singular measures with compact support . moreover , @xmath961 .
because of remark [ rmkcktnk ] , we can assume that @xmath962 is compact .
let @xmath963 be the one - parameter subgroup @xmath769 . using ( * ? ? ?
* thm x.6.16 ) for the restriction @xmath964 , we obtain @xmath965 .
next we prove that @xmath890 is a smooth unitary representation of @xmath14 .
given @xmath966 and @xmath967 , we set @xmath968 lemma [ lwhmmc ] implies that @xmath969 , and ( * ? ? ?
ii.1.7 ) implies that the fields @xmath970 generate a dense subspace of @xmath950 . for every @xmath360
, we have @xmath971 the map @xmath972 , @xmath973 is smooth . from
we obtain @xmath974 where @xmath975 is defined as in .
from linearity of @xmath842 it follows that an inequality similar to holds for directional derivatives of @xmath976 . by differentiating under the integral sign ( ( * ? ? ?
2.1 ) ) it follows that the map @xmath977 is smooth .
finally , from ( * ? ? ?
* thm 7.2 ) it follows that @xmath978 is a smooth vector for @xmath890 . by remark [ rmkcktnk ]
we can assume that @xmath962 is compact . by the proof of proposition [ pimupe ] ,
the sections @xmath970 , defined in , belong to @xmath982 . applying ( * ? ? ?
2 ) to one - parameter subgroups of @xmath14 implies that @xmath983 .
therefore , to complete the proof of the proposition , it suffices to show that if a section @xmath984 satisfies @xmath985 , then @xmath986 for almost every @xmath855 . to prove the latter claim ,
again we note that ( * ? ? ?
2 ) implies that there exists a measurable set @xmath987 with @xmath988 such that @xmath989 for every @xmath990 and every @xmath991 .
( here we use the fact that for every @xmath992 , the map @xmath993 , @xmath994 is continuous . ) finally , proposition [ rglvgqendv ] implies that @xmath995 .
[ thm - main ] let @xmath7 be a positive energy representation of @xmath14 .
then @xmath996 where @xmath997 are mutually singular finite positive measures on @xmath854 , and the @xmath998 are constructed as in _ section [ sec - dirint]_. moreover , the measures @xmath999 , are unique up to equivalence
. * step 1 . *
assume that @xmath716 is a cyclic representation of @xmath128 , and the spectral measure @xmath805 , defined in , has compact support .
then there exists a finite positive borel measure with compact support @xmath892 on @xmath854 such that @xmath1001 .
proposition [ prpposcyq ] and proposition [ rglvgqendv ] imply that @xmath7 can be realized as a @xmath14-invariant hilbert space in @xmath1002 . for a similar reason , @xmath890
can also be realized as a @xmath14-invariant hilbert space in @xmath1002 .
therefore by proposition [ patmston ] we obtain @xmath1003 .
* step 2 . *
assume that @xmath716 is not necessarily cyclic , but the support of @xmath805 is compact .
then @xmath1004 where the @xmath960 are mutually singular finite positive borel measures on @xmath854 . by proposition [ adprp - gi - ii - iii](v ) , the above decomposition of @xmath1005 corresponds to a decomposition @xmath1006 where @xmath1007 for @xmath1008 ( here @xmath1009 is the hilbert space of @xmath1010 , and @xmath1011 is the hilbert space of @xmath1012 ) .
since the representations @xmath1013 are cyclic , step 1 implies that @xmath1014 for every @xmath1008 .
this results in the decomposition for @xmath7 . + * step 3 . * finally ,
note that proposition [ pipij ] together with step 2 imply the existence of the direct sum decomposition without any assumptions on @xmath716 . next we prove uniqueness of the decomposition .
suppose that a decomposition of @xmath7 as in is given . by considering @xmath702-invariants , and using proposition [ adwolueh ] , we obtain a corresponding direct sum decomposition @xmath1015 where the @xmath960 are mutually singular finite positive borel measures . since @xmath128 is a locally compact abelian group ,
the decomposition is unique up to equivalence of measures .
therefore the decomposition of @xmath7 is also unique .
r. goodman , n. wallach , _ projective unitary positive - energy representations of @xmath1016 _ , j. funct .
63 * ( 1985 ) , 299321 .
r. hamilton , _ the inverse function theorem of nash and moser _ , bull .
* 7 * ( 1982 ) , 65222 .
e. hewitt , k. a. ross , `` abstract harmonic analysis .
ii : structure and analysis for compact groups .
analysis on locally compact abelian groups , '' grundlehren der mathematischen wissenschaften * 152 * , springer , new york
berlin , 1970 .
v. g. kac , _ some problems on infinite - dimensional lie algebras and their representations _ , in `` lie algebras and related topics , '' lecture notes in mathematics , vol .
933 , springer , berlin / heidelberg / new york , 1982 .
v. g. kac , a. k. raina , `` bombay lectures on highest weight representations of infinite dimensional lie algebras , '' advanced series in mathematical physics * 2 * , world scientific publishing co. , inc . ,
teaneck , nj , 1987 .
neeb , _ holomorphic realization of unitary representations of banach lie groups _ , in `` lie groups : structure , actions and representations , '' huckleberry , a. , penkov , i. , zuckerman , g. ( eds . ) , progress in mathematics * 306 * , 2013 ; 185223 .
rudin , w. , `` functional analysis , '' mcgraw hill , 1973 m. takesaki , _ theory of operator algebras i _ , encyclopaedia of mathematical sciences , 124 .
operator algebras and non - commutative geometry , 5 .
springer - verlag , berlin , 2002 . | we give a complete classification of all positive energy unitary representations of the virasoro group .
more precisely , we prove that every such representation can be expressed in an essentially unique way as a direct integral of irreducible highest weight representations . |
the bulk viscosity ( in the following denoted by @xmath5 ) describes the response of a fluid to a compression or expansion .
compared to other transport coefficients such as the shear viscosity or the self diffusion constant , it is the least studied transport coefficient .
this is surprising since @xmath5 is for instance a central quantity in the description of the damping of longitudinal sound .
it is also an important quantity to probe slow dynamic processes such as the critical slowing down near the critical point of a liquid
gas transition or the liquid liquid unmixing transition in a binary fluid .
we will briefly discuss these issues below .
a microscopic expression for @xmath5 is given by a green kubo formula ( boon and yip , 1980 ) , @xmath7 with @xmath8 denoting cartesian components ( @xmath9 ) .
@xmath10 , @xmath0 and @xmath11 are volume , temperature and boltzmann s constant , respectively . in the microcanonical ensemble ,
@xmath12 is equal to the difference between the pressure at time @xmath13 and that at @xmath14 , @xmath15 , where @xmath16 is equal to the diagonal elements of the pressure tensor @xmath17 defined as follows : @xmath18 \ .
\label{eq_i_2}\ ] ] herein @xmath19 and @xmath20 are respectively the @xmath8th component of momentum and position of particle @xmath21 , and @xmath22 is the @xmath8th component of the force acting on particle @xmath21 .
note that in order to calculate the shear viscosity @xmath4 one has to use the non diagonal elements of the pressure tensor in the green kubo integral ( @xmath23 ) ( boon and yip , 1980 ) : @xmath24 eqs .
( [ eq_i_1 ] ) , ( [ eq_i_2 ] ) , and ( [ eq_i_3 ] ) can be used to calculate @xmath5 and @xmath4 from equilibrium fluctuations in a molecular dynamics ( md ) computer simulation . indeed , in one of the pioneering md studies of a lennard jones liquid near its triple point by levesque _
et al . _
( levesque et al . , 1973 )
the viscosities were determined by green kubo formulas .
note that a recently proposed `` new '' formula by okumura and yonezawa ( okumura and yonezawa , 2002 ) just expresses the pressure fluctuations in eq .
( [ eq_i_1 ] ) in terms of the pair correlation function and the interatomic potentials .
alternative methods to determine @xmath5 are based on non equilibrium molecular dynamics ( nemd ) simulations .
heyes ( heyes , 1984 ; heyes , 1986 ) proposed a nemd scheme where the volume of the system is changed from @xmath10 to @xmath25 at @xmath14 which leads to a change of the pressure .
then one follows the relaxation of the pressure to its equilibrium value @xmath26 and measures @xmath27 where @xmath16 is the instantaneous pressure at time @xmath13 .
@xmath5 is then given by @xmath28 dt \ .
\label{eq_i_4}\ ] ] certainly , eq . ( [ eq_i_4 ] ) is only valid if @xmath29 is small enough to allow the application of linear response theory .
another nemd approach was proposed by hoover _
et al . _
( hoover et al . , 1980 ) .
the latter authors impose a frequency dependent small perturbation by changing the volume of the system by a periodic compression and expansion with a frequency @xmath30 . as a result @xmath31
is obtained for several values of @xmath30 and then an extrapolation to zero frequency may be possible .
the method of hoover _ et al .
_ might be especially useful for liquid states where @xmath5 exhibits a long time tail .
however , some knowledge of the frequency dependence of @xmath31 is required to extrapolate it accurately from finite frequencies to zero .
the aforementioned methods have been mainly used in feasibility studies where the bulk viscosity was determined , e.g. , for a lennard jones fluid at a single state near its triple point ( see refs .
( levesque et al . , 1973 ; hoover et al . , 1980 ; heyes , 1984 ; okumura and yonezawa , 2002 ) ) .
only in a small number of simulations , @xmath5 has been investigated systematically .
one of these rare studies is the md simulation of symmetrical lennard
jones mixtures by vogelsang and hoheisel ( vogelsang and hoheisel , 1988 ; hoheisel , 1993 ) who considered systems of 256 particles at moderate densities ( i.e. far from the triple point ) . in this
work @xmath5 as well as @xmath4 were calculated by means of the green kubo formulas , eqs .
( [ eq_i_1 ] ) and ( [ eq_i_3 ] ) .
an interesting result of this study was that the ratio @xmath32 is ( much ) larger than one if the fluid mixture has a ( strongly ) associating character or a ( strongly ) demixing character . in both of the latter cases the bulk viscosity increases quickly whereas the shear viscosity remains essentially constant . as a consequence
it is expected that @xmath5 shows a strong increase near the coexistence line of a fluid fluid unmixing transition .
in contrast to the small number of simulations , there are many theoretical investigations of the bulk viscosity in the context of the dynamics near the liquid gas critical point ( kawasaki , 1976 ; kadanoff and swift , 1968 ; swift , 1968 ; hohenberg and halperin , 1977 ; folk and moser , 1995 ; onuki , 1997 ; onuki , 2002 ) .
these works predict that the bulk viscosity exhibits a strong divergence near the critical point of a gas liquid liquid transition .
in contrast to that , the shear viscosity is expected to show a very weak divergence ( logarithmic divergence ) at the critical point ( if there is at all a divergence in this quantity ) .
the latter predictions have been confirmed experimentally .
an example is @xmath33he in the vicinity of the critical temperature @xmath34 : at @xmath35 on the critical isochore , @xmath5 is about 50 poise whereas @xmath4 is equal to @xmath36 poise ( kogan and meyer , 1998 ; onuki , 2002 ) . in the present work we consider a simple model of a _ dense _
liquid mixture near and at a liquid
liquid unmixing transition and , apart from static susceptibilities , we calculate the shear and the bulk viscosity . although we are not able to determine these quantities very close to the critical point , we find a behavior which agrees qualitatively with the aforementioned theoretical predictions for the critical dynamics : @xmath5 shows a stronger increase than @xmath4 when approaching a state on the coexistence line about 15% below the critical point and , furthermore , at the latter point , @xmath5 is significantly larger than @xmath4 , i.e. @xmath37 .
the rest of the paper is organized as follows : in the next section we briefly comment on the details of the simulation as well as the lennard
jones model and its phase diagram .
the static properties and the transport coefficients ( shear and bulk viscosity ) as obtained from the simulation are then presented in sec .
finally we summarize the results in sec .
the model that we consider in this work is a binary lennard
jones mixture .
thus , the interaction potential between a particle of type @xmath8 and a particle of type @xmath38 ( @xmath39 ) is given by @xmath40 \ , \label{eq1}\ ] ] @xmath41 being the distance between the two particles .
for the lennard jones parameters @xmath42 and @xmath43 we choose @xmath44 , @xmath45 and @xmath46 .
lengths , energies , and temperatures are measured respectively in units of @xmath47 , @xmath48 , and @xmath49 . in the molecular dynamics ( md )
part equal masses are chosen for a and b particles , i.e. @xmath50 .
the potential is truncated and shifted at @xmath51 .
the model mixture that we have defined so far is obviously completely symmetrical . whether it has the tendency towards association or demixing
is controlled by the parameter @xmath52 .
we use @xmath53 which implies the possibility of a fluid fluid unmixing transition .
since we are interested in the dense liquid state we have chosen a density @xmath54 , which provides the absence of crystallization in the temperature range of interest , @xmath55 .
note that for densities @xmath56 the phase behavior of symmetrical lj mixtures have been extensively investigated by wilding ( wilding , 1997 ) .
the simulations were done as follows : we started with a random mixture with an equal number of a and b particles . by using standard monte carlo ( mc ) in the canonical ensemble with trial displacements of particles in the range @xmath57 $ ] , we equilibrated the system for @xmath58 monte carlo steps ( mcs ) per particle .
then , we switched on a mc simulation in the semigrand canonical ensemble , i.e. , at the end of each displacement sweep an identity switch of @xmath59 randomly chosen particles was attempted , @xmath60 or @xmath61 ( @xmath62 being the total number of particles ) . note that in the metropolis criterion of the semigrand canonical moves , the chemical potential energy difference @xmath63 ( @xmath64 : chemical potential of species @xmath65 ) has to be taken into account in addition to the energy change in the boltzmann factor . in order to localize the coexistence curve of the liquid
liquid unmixing transition in the present case , one has to just set @xmath66 which is simply due to the symmetry of our model . in order to determine the phase diagram we have performed five independent runs with a length of 400000 mcs per particle where we started the averaging after 100000 mcs in each run ( for more details of this calculation see ref .
( das et al . , 2003 ) ) .
[ fig1 ] shows the phase diagram in the @xmath0@xmath67 plane for the system sizes @xmath68 , and 3200 ( @xmath69 is the concentration of @xmath70 particles ) . due to the symmetry of the model we know _ a priori _ that the critical point is located at @xmath71 . as we can infer from fig .
[ fig1 ] , the finite size effects near the critical point are small for @xmath72 , and for @xmath73 the data agree within the statistical errors .
we have estimated the critical temperature @xmath34 from power law fits according to the three dimensional ising universality class ( binder and ciccotti , 1996 ; binder and heermann , 2002 ) , @xmath74 where @xmath75 is a critical amplitude which is used , as well as @xmath34 , as a fitting parameter . from the fits with eq .
( [ eq2 ] ) we obtain @xmath76 for @xmath77 . for a more accurate estimate of @xmath34 , we would have to perform a finite scaling analysis ( binder and ciccotti , 1996 ; binder and heermann , 2002 ) .
apart from the phase diagram the mc in the semigrand
canonical ensemble yields also equilibrated configurations exactly along the coexistence line .
we used them as starting configurations for molecular dynamics ( md ) simulations to determine the static quantities and the transport coefficients that are presented in the next section . in the md
, the equations of motion were integrated by means of the velocity verlet algorithm ( binder and ciccotti , 1996 ) with a time step @xmath78 [ in units of the time @xmath79 .
starting point for the md were the configurations with 1600 particles at @xmath80 that correspond to the concentration @xmath81 at the coexistence line .
configurations in the one phase region at the latter value of @xmath67 were obtained by heating up the system and equilibrating it for @xmath58 time steps at constant temperature with the use of an andersen thermostat ( das et al .
, 2003 ) . then , microcanonical runs were added from which we computed the static and dynamic quantities that are shown in the next section .
the path along which we determined the latter quantities is indicated in fig .
[ fig1 ] by crosses : apart from the coexistence state at @xmath80 , which is about 15% below the critical point with respect to temperature , the temperatures @xmath82 , 1.7 , 3.0 , and 6.0 were analyzed ( note that also other paths around the coexistence line are studied in ref .
( das et al . , 2003 ) ) .
one may wonder why we have not studied states that are much closer to the critical point .
but due to the diverging correlation length that is accompanied by the approach of the critical point , we would have to consider systems that contain much more than 1600 particles as in our work .
furthermore , the critical slowing down would require very long runs to equilibrate the system and to determine the transport coefficients with reasonable error bars .
the latter point is especially a severe problem for transport coefficients such as the shear or the bulk viscosity .
these are collective quantities and require many independent runs and/or a long time averaging since they are not subject to a self averaging as one particle quantities such as the self diffusion constant .
however , compared to many previous works , our choice of @xmath62 is relatively large .
even the very recent computation of the bulk viscosity by okumura and yonezawa ( okumura and yonezawa , 2002 ) was only done for a small system of @xmath83 particles .
in this section we present the results for the static and dynamic properties of the symmetrical lj system along the path in the phase diagram which is indicated in fig .
as described in the previous section , we have generated first five independent configurations at each temperature .
all these configurations were used as initial configurations for microcanonical md runs over 4.8 million time steps ( the time step was 0.01 @xmath84 , see previous section ) .
thus , at each temperature 24 million time steps were done to determine the quantities of interest .
as we shall see in the following , this effort was large enough to get a reasonable estimate of bulk and shear viscosities . as we see in fig .
[ fig1 ] the states at @xmath80 on the coexistence line are about 15% below the critical point with respect to temperature .
although these points are not very close to the critical point one may expect that the approach of the critical point is reflected in thermodynamic quantities such as the isothermal compressibility @xmath2 and the concentration susceptibility @xmath3 ( defined below ) .
@xmath2 can be calculated from the static number number density structure factor @xmath85 in the limit of wavenumber @xmath86 ( hansen and mcdonald , 1986 ) , @xmath87 with @xmath88 being the total density of the system ( in our case @xmath88 as well as the boltzmann constant @xmath11 are equal to one ) .
the structure factor @xmath85 for a binary ab mixture is defined by ( hansen and mcdonald , 1986 ) @xmath89 where @xmath90 ( @xmath91 $ ] ) are the partial structure factors , @xmath92 \right > \label{eqsec3_2}\ ] ] with @xmath93 for @xmath23 and @xmath94 for @xmath95 . in eq .
( [ eqsec3_2 ] ) the indices @xmath96 run over the number of particles of species @xmath8 and @xmath38 , respectively , and @xmath97 is the position of the @xmath21th particle of species @xmath8 . fig .
[ fig2 ] shows @xmath85 for @xmath80 , 1.7 , 3.0 , and 6.0 . for wavenumbers @xmath98 that correspond to distances smaller or equal the typical nearest neighbor distance , say
@xmath99 , the typical behavior of this quantity for simple dense liquid can be identified : upon decreasing the temperature the amplitude especially of the first peak increases and the peaks become narrower .
the small values of @xmath85 for @xmath86 reflect the fact that the considered dense liquid state is hardly compressible . it might be surprising that even at coexistence @xmath100 does not show any tendency to increase significantly for @xmath86 .
the amplitude of @xmath85 at small @xmath98 appears to be even a monotonic decreasing function with decreasing temperature .
however , the relevant thermodynamic quantity in our context is @xmath2 , that we have extracted from @xmath85 by extrapolating this function to @xmath101 . as we see in the inset of fig .
[ fig2 ] , @xmath2 increases significantly with decreasing temperature which shows that for states around @xmath80 , long ranged static correlations , i.e. the presence of the critical point , still affect the behavior of @xmath2 .
the `` vicinity '' of the critical point is more pronounced in the structure factor of the concentration densities , @xmath102 , than in @xmath103 .
@xmath102 can be also expressed by a linear combination of the partial structure factors ( hansen and mcdonald , 1986 ) , i.e. @xmath104 in the limit @xmath105 the structure factor @xmath102 is related to the static concentration susceptibility @xmath3 by @xmath106 note that we have determined @xmath3 directly via a fluctuation relation by semigrand
canonical mc runs ( see ref .
( das et al . , 2003 ) ) .
so , it was not necessary to extrapolate @xmath102 to @xmath101 . as we see in fig .
[ fig3 ] this would be a difficult task because , in contrast to @xmath85 , @xmath102 steeply increases for @xmath105 .
as we can infer from the inset of fig .
[ fig3 ] , @xmath3 is about a factor of 2 larger at the coexistence state at @xmath80 than at @xmath107 .
it is remarkable that @xmath102 exhibits almost no temperature dependence for @xmath99 in the broad temperature range @xmath108 . for the computation of the bulk and shear viscosities we have used the green kubo ( gk ) formulas , eqs .
( [ eq_i_1 ] ) and ( [ eq_i_3 ] ) .
the alternative methods that are based on nemd require essentially the same computational effort .
furthermore , in the heyes method , see eq .
( [ eq_i_4 ] ) , one has to choose the perturbation @xmath29 small enough to ensure that this perturbation justifies the application of linear response theory .
thus , one has to study the dependence of the measured bulk viscosity @xmath5 on @xmath29 ( of course , in the linear response regime the apparent @xmath5 is independent of @xmath29 ) .
the hoover method has in addition the drawback that one has to extrapolate the frequency dependent viscosity to zero frequency .
however , a comparative study of the different nemd and gk methods to measure transport coefficients in a simulation is an interesting future project since the nemd methods may give additional physical insight into the microscopic mechanism of different transport processes .
[ fig4 ] shows @xmath109 and @xmath110 for four temperatures .
these quantities are defined by eqs .
( [ eq_i_1 ] ) and ( [ eq_i_3 ] ) where one has to replace @xmath111 in the integral by @xmath13 .
we see that @xmath109 and @xmath110 approach indeed plateaus at long times the values of which correspond to the hydrodynamic shear and bulk viscosities , respectively . at low temperatures
, there is a qualitative difference in @xmath110 as compared to @xmath112 : e.g. at @xmath80 , @xmath109 is essentially constant for @xmath113 .
in contrast to that , @xmath110 exhibits a second strong increase and it reaches the plateau value for @xmath114 .
this is due to a long time tail in the autocorrelation function of the pressure fluctuations .
note that the decrease of @xmath110 for @xmath115 is due to the fact that the statistics is much worse at long times .
@xmath5 and @xmath4 are plotted in fig .
[ fig5]a as a function of inverse temperature .
whereas @xmath4 exhibits only a very weak temperature dependence , @xmath5 increases significantly in the vicinity of the coexistence state at @xmath80 . as we can see in fig .
[ fig5]b the ratio @xmath116 is in the whole considered temperature range @xmath117 above one , and it reaches a value of about 3.3 at @xmath80 .
one expects such a behavior from theories of the critical dynamics of the liquid gas transition ( onuki , 2002 ) . according to these theories the long ranged critical fluctuations cause a slowing down of the system s response to a compression or expansion ( described by @xmath5 ) . on the other hand , the response to the shearing of the system
is hardly affected by the critical fluctuations ( and thus @xmath4 ) . in our case , at a state about 15% below the critical point , there is already a significant increase of static correlations which makes the behavior of @xmath32 plausible .
since the data presented in this paper are taken at an off - critical concentration , one could also attempt to interpret them in terms of a singular behavior at the `` spinodal temperature '' @xmath118 ( limit of metastability ) ( binder , 1987 ) . according to the mean field theory of symmetric binary mixtures
, one should expect that the static concentration susceptibility @xmath3 for @xmath119 behaves as @xmath120^{-1 } \label{eqn12}\ ] ] where near the critical temperature the spinodal temperature @xmath121 is the inverse function of the concentration @xmath122 along the spinodal curve , given by the equation @xmath123 ( binder , 1987 ) . further away from @xmath34 , a simple expression for @xmath121 exists for the lattice ( ising ) model of symmetric binary mixtures , namely @xmath124/2 . \label{eqn13}\ ] ] here we have emphasized by this notation that the mean field estimate @xmath125 of the critical temperature for systems with short range forces normally exceeds the actual critical temperature distinctly ( also eq .
( [ eqn12 ] ) does then not hold for @xmath67 near @xmath126 and @xmath0 near the actual critical temperature , since @xmath127 , where the actual susceptibility exponent @xmath128 ( binder and ciccotti , 1996 ; binder and heermann , 2002 ) ) .
although we do not really expect that eq .
( [ eqn12 ] ) and a related mean - field divergence for the bulk viscosity @xmath5 is a good approximation for our lennard - jones system , we present a plot of @xmath129 vs. @xmath0 and @xmath130 vs. @xmath0 in fig [ fig6 ] .
mean field theory would predict that the data fall on straight lines and both straight lines should hit the abscissa in the same point which then is the estimate of @xmath131 .
indeed the data points close to the coexistence curve are compatible with such analysis , with @xmath132 .
of course , one should not put too much weight on this analysis , since the temperature range over which we need to extrapolate is larger than the temperature range where actual data are fitted .
also the estimate from eq .
( [ eqn13 ] ) would be much lesser , namely @xmath133 , if the distinction between the actual @xmath34 and @xmath125 is ignored .
we caution the reader that anyway the concept of a spinodal is of doubtful validity outside of mean field theory ( binder,1987 ) , although in the experimental literature on binary mixtures ( both in metallic alloys and in polymer blends , for instance ) it is widely used .
we have used computer simulations to investigate transport coefficients of a dense symmetrical lennard
jones mixture that were calculated along a path towards a liquid liquid miscibility gap ending at a coexistence state about 15% below the critical point .
the main result of our study is that the bulk viscosity @xmath5 increases significantly near the coexistence state whereas the shear viscosity @xmath4 does not show any change near coexistence .
@xmath4 remains to exhibit a very weak temperature dependence also when it passes the coexistence line .
the behavior of @xmath5 and @xmath4 can be qualitatively understood by theories of critical dynamics ( see ref .
( onuki , 2002 ) ) . in future studies we plan to compute the transport properties also closer to the critical point .
of course , in such studies much larger system sizes than those used in this work have to be considered .
moreover , the emergence of critical slowing down requires simulations on much longer time scales . + the present research was supported by the deutsche forschungsgemeinschaft ( dfg ) under grant no .
bi314/18 ( spp 1120 ) .
one of the authors ( j. h. ) acknowledges the support of the dfg under grants no .
ho 2231/2 - 1 and ho 2231/2 - 2 .
: : binder , k. ( 1987 ) .
theory of first order phase transitions .
phys . , 50 , 783 .
: : binder , k. and ciccotti , g. , eds .
monte carlo and molecular dynamics of condensed matter systems .
italian physical society , bologna .
: : binder , k. and heermann , d.w .
monte carlo simulations in statistical physics .
an introduction , 4th edition .
springer , berlin .
: : boon , j.p . and
yip , s. ( 1980 ) .
molecular hydrodynamics .
mcgraw hill , new york .
: : das , s.k . ,
horbach , j. and binder , k. ( 2003 ) .
transport phenomena and microscopic structure in partially miscible binary fluids : a simulation study of the symmetrical lennard
jones mixture .
j. chem .
119 , 1547 .
: : folk , r. and moser , g. ( 1995 ) .
nonuniversal dynamical crossover in pure and binary fluids near a critical point .
75 , 2706 .
: : hansen , j.p . and mcdonald , i.r .
theory of simple liquids .
academic , london . : : heyes , d.m .
( 1984 ) . j.
, faraday trans .
80 , 1363 .
: : heyes , d.m .
: : hoheisel , c. ( 1993 ) .
theoretical treatment of liquids and liquid mixtures .
elsevier , amsterdam , pp .
: : hohenberg , p.c . and
halperin , b.i .
theory of dynamic critical phenomena .
mod , phys .
: : hoover , w.g . , evans , d.j . ,
hickman , r.b . ,
ladd , a.j.c .
, ashurst , w.t . , and moran , b. ( 1980 ) .
jones triple
point bulk and shear viscosities .
green kubo theory , hamiltonian mechanics , and nonequilibrium molecular dynamics .
a 22 , 1690 .
: : kadanoff , l.p . and swift , j. ( 1968 ) .
transport coefficients near the liquid gas critical point .
phys . rev .
: : kawasaki , k. ( 1976 ) .
mode coupling and critical dynamics . in : domb , c. and green , m.s .
( eds . ) , _ phase transitions and critical phenomena _ , vol .
5a , pp . 166405 . academic press , london . : : kogan , a.b . and meyer , h. ( 1998 ) .
sound propagation in @xmath33he and @xmath134he above the liquid
vapor critical point .
j. low temp .
110 , 899 .
: : levesque , d. , verlet , l. , and krkijarvi , j. ( 1973 ) .
computer `` experiments '' on classical fluids .
iv . transport properties and time correlation functions of the lennard
jones liquid near its triple point .
a 7 , 1690 .
: : okumura , h. and yonezawa , f. ( 2002 ) .
new formula for the bulk viscosity constructed from the interatomic potential and the pair distribution function . j.
chem . phys .
116 , 7400 .
: : onuki , a. ( 1997 ) .
dynamic equations and bulk viscosity near the gas liquid critical point .
phys . rev .
e 55 , 403 .
: : onuki , a. ( 2002 ) .
phase transition dynamics .
cambridge university press , cambridge , p. 247 .
: : swift , j. ( 1968 ) .
transport coefficients near the consolute temperature of a binary liquid mixture .
173 , 257 .
: : vogelsang , r. and hoheisel , c. ( 1988 ) .
thermal transport coefficients including the soret coefficient for various liquid lennard
jones mixtures .
a 38 , 6296 .
: : wilding , n.b . ( 1997 ) .
critical end point behavior in a binary fluid mixture .
e 55 , 6624 . as indicated .
the crosses at @xmath81 mark the states for which the structure and dynamics was investigated ( note that also @xmath135 and @xmath136 were studied ) .
the solid lines are fits with eq .
( [ eq2 ] ) and the dashed lines are just guides to the eye . , width=529 ] for the four indicated temperatures .
the inset shows the isothermal compressibility @xmath2 as a function of temperature .
@xmath2 is estimated from the extrapolated value @xmath137 [ see eq .
( [ eqsec3_0 ] ) ] .
, width=529 ] | a lennard jones model of a binary dense liquid ( a , b ) with a symmetrical miscibility gap is investigated by means of computer simulation methods . semigrand
canonical monte carlo simulations yield the phase diagram in the @xmath0@xmath1 plane ( @xmath0 : temperature , @xmath1 : concentration of a or b particles ) as well as equilibrated configurations at coexistence
. then molecular dynamics simulations use these configurations to determine static properties ( isothermal compressibility @xmath2 and concentration susceptibility @xmath3 ) as well as the shear and the bulk viscosity @xmath4 and @xmath5 , respectively .
the latter quantities are calculated along a path approaching the coexistence line from high temperatures in the one phase region and ending at a state at the coexistence line about 15% below the critical point .
we find that @xmath2 and @xmath3 increase significantly near the coexistence line reflecting the vicinity of the critical point .
whereas @xmath6 exhibits a weak temperature dependence , @xmath5 increases significantly near the coexistence curve . |
in this paper , we study properties of solutions of differential equations that are defined on various real foliations . in section 2 , we investigate @xmath2 foliations on compact sets @xmath4 defined by vector fields @xmath5 .
we study the differential equation @xmath6 which is defined on the foliation .
we show : if @xmath3 is @xmath2 on a neighborhood of @xmath4 , then there is a @xmath2 function @xmath1 on @xmath4 so that @xmath6 we prove theorem 1 by mathematical induction on the number of derivatives .
we also construct some particular examples to examine this theorem and to motivate the choice of equation . in section 3 , we study foliations with singularities . on the real line , we find a solvable differential equation and we prove that @xmath7 is continuous and bounded if @xmath8 is continuous and bounded .
however , on a circle , we show that there is no such solvable differential equation . in section 4 ,
we construct certain line bundles on @xmath2 foliations .
we show that in this case we can solve @xmath9 with values in the line bundle .
this is the main result in this paper : if @xmath3 is @xmath2 on a foliated set @xmath4 , then there is a bounded and continuous function @xmath1 such that @xmath10 .
consider a function @xmath8 on a real line which is continuous and bounded .
we want to find a bounded function @xmath7 so that @xmath11 .
if we let @xmath12 , then @xmath13 where @xmath14 is a constant .
obviously @xmath7 is not a bounded function . to solve this problem
, we modify the differential equation by adding @xmath7 to the left part of the equation and get @xmath15 .
then , we solve this new equation .
first , we multiply the equation by @xmath16 and get @xmath17 which is the same as @xmath18 and hence @xmath19 .
we get a general formula for @xmath7 : @xmath20 now we will prove the continuity of the function @xmath7 by showing that @xmath21 is continuous if @xmath8 is continuous .
let @xmath22 be an arbitrary small positive number and @xmath23 .
we will show that there exist a real number @xmath24 such that @xmath25 holds for every pair of @xmath26 if @xmath27 .
@xmath28 since @xmath29 is bounded and continuous , there exist a @xmath24 such that @xmath25
if if @xmath27 . this proves the continuity of @xmath7 .
[ figure 1 . ]
now we expand this problem to a square on a plane with lines with an irrational slope @xmath30 .
as shown in figure 1 ( a ) , whenever the line hits the upper ( or right ) boundary of the square , it goes straight down ( or left ) to the same x ( or y ) value on the lower ( or left ) boundary with the same slope .
then we have a continuous and bounded function @xmath31 and a bounded function @xmath32 which is differentiable in the direction@xmath33such that @xmath34 .
we found that the same problem exists here as in the real line case condition .
let @xmath35 , then we have @xmath36 and @xmath37 .
setting@xmath38 , we get @xmath39 which is an unbounded function .
again we multiply both sides of the equation by @xmath16 and get @xmath40 and @xmath41 . here
if we set @xmath42 , we get @xmath43 which is a bounded function .
however , we notice that by multiplying @xmath16 to the equation , we created a difference between the y values of the line when it reaches the right side of the square and moves to the left side . at the right side of the square , the line has the expression of @xmath44 while at the left side , the expression becomes @xmath45 .
the y values of the two ends are not equal to each other any more .
we use a line bundle to fix this difference .
( we will discuss line bundle in more detail in section 4 . )
we connect the upper and lower sides together to form a hollow cylinder .
then , we bend the cylinder along its axis and make it to be a ring as shown in figure 2 . [ fig : figure2 ] let @xmath46 be a real number closed to 0 and @xmath47 , u be a set that contains all the points on the square with @xmath48 , and v be a set contains the points with @xmath49 .
we define @xmath50 as a set of points lies on @xmath51 and @xmath52 as the set of points lies on @xmath53 . both @xmath50 and @xmath54
are intersection of sets @xmath55 and @xmath56 .
based on the assumptions , we have @xmath57 @xmath58 @xmath59 @xmath60 with @xmath61 , we have @xmath62 . for @xmath52 , @xmath63 .
+ now , we are going to find the general form of @xmath7 . @xmath64
fix a point on the line with coordinates @xmath65 , it is easy to know that @xmath66 .
according to the formulae we derived above , we have @xmath67 since the function @xmath31 is defined on the interval [ 0,1 ] , in order to do the integration from negative infinity to @xmath68 , we need to expand its definition .
we assume that @xmath31 is a periodic function with the period of 1,i.e .
@xmath69 . what we want to know is that whether the function @xmath32 has the same property or not .
+ > from equation ( 5 ) , @xmath70 let @xmath71 , @xmath72 so @xmath32 is a periodic function with period of 1 .
+ next , we will prove the continuity of function @xmath73 .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath24 such that @xmath74 and @xmath75 holds for every pair of @xmath26 and @xmath76 if @xmath77 and @xmath78 . because of the continuity of the function @xmath79 , there exists a @xmath80 such that @xmath74 .
+ without loss of generality , we assume @xmath81 .
@xmath82dh\\ & & + \int_x^{x'}e^hv(\sqrt{2}h+y'-\sqrt{2}x')dh |}\\ & < & { | \int_{-\infty}^xe^h \frac{\epsilon}{2e^x } dh+\int_x^{x'}e^hv(h , \sqrt{2}h+y'-\sqrt{2}x')dh |}\\ & = & { | \frac{\epsilon}{2e^x } e^x + \int_x^{x'}e^hv(h , \sqrt{2}h+y'-\sqrt{2}x')dh |}\\ & \leq & \frac{\epsilon}{2 } + { | \int_x^{x'}e^hv(h , \sqrt{2}h+y'-\sqrt{2}x')dh |}\end{aligned}\ ] ] to make the condition @xmath75 holds , we have @xmath83 .
let @xmath84 .
again because of the continuity of the function @xmath31 , there exist a @xmath85 such that @xmath86 @xmath87 let @xmath88 . if @xmath77 and @xmath78 , then all conditions hold and this proves the continuity of the function @xmath89 .
+ consider two concentric circles with radius 1 and 2 respectively . between them , there is a spiral which approximates both of the circles but
never touches them ( shown as figure 3 ) .
[ fig : figure3 ] a continuous and bounded function @xmath90 is defined on the two circles and the spiral .
we want to find another continuous , bounded and differentiable function @xmath91 such that @xmath92 .
+ on circles , we have @xmath93 @xmath94 @xmath95 @xmath96 on the spiral , the radius is a function of the angel which is @xmath97 the function @xmath98 has following property on the spiral : @xmath99 @xmath100 we want to find that function @xmath101 has the same property : @xmath102 @xmath103 if we set @xmath104 , then @xmath105 and hence @xmath106 which is not a bounded function .
this indicates that the differential equation @xmath107 is not solvable . to solve the problem
, we add @xmath101 to the left side of the equation and get a new equation @xmath108 .
we found that we can solve this new equation by multiplying @xmath109 to both sides of the equation and get , @xmath110 which is the same as @xmath111 now we are going to find the general form for @xmath101 .
+ on the circles , we have @xmath112 @xmath113 fix a point on a circle with coordinates @xmath114 , according to the formulae we derived above , we have @xmath115 then , we need to make sure that the function @xmath116 is periodic with period of @xmath117 , i.e. we need to verify that the equation @xmath118 holds . + from equation ( 16 ) , @xmath119 let @xmath120 , @xmath121 this proves that @xmath122 is a periodic function with period of @xmath117 .
+ let @xmath123 @xmath124 on the spiral , we have , @xmath125 @xmath126 substitute ( 17 ) , ( 18 ) back to the solution , we get @xmath127 fix a point on the spiral with coordinates @xmath128 , according to the formulae we derived above , we have @xmath129 now we show the continuity of the solution we got .
we are going to examine the continuity of @xmath130 , @xmath131 , and the continuity between them .
we will fist show that the solution on the circles is continuous .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath80 such that @xmath132 holds for every pair of @xmath133 if @xmath134 .
without loss of generality , we assume @xmath135 .
@xmath136 because of the continuity of @xmath137 , there exists a @xmath80 such that @xmath138 then we show that the solution on the spiral is continuous .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath85 such that @xmath139 holds for every pair of @xmath133 if @xmath140 . without loss of generality , we assume @xmath135 .
@xmath141 because of the continuity of @xmath142 , there exists a @xmath85 such that @xmath143 then we show that the solution between the circles and the spiral is continuous .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath144 such that @xmath145 | } < \epsilon$ ] and @xmath146 holds for every pair of @xmath133 and @xmath147 if @xmath148 and @xmath149.@xmath150 firstly , we show the continuity between @xmath101 and @xmath151 when @xmath152 . without loss of generality ,
we assume @xmath153 .
@xmath154dh -u(1 , \theta')+u(1 , \theta ) |}\end{aligned}\ ] ] because of the continuity of @xmath151 that we have just shown , there exists a @xmath155 such that @xmath156 according to our assumption ( equation ( 12 ) ) , when @xmath157 goes to negative infinity , we have @xmath158 therefore , there exists a @xmath159 such that @xmath160dh | } < \frac{\epsilon}{2},\ ] ] if @xmath152 which is the same with @xmath161 .
let @xmath162 .
if @xmath163 and @xmath164 , then we have @xmath165dh -u(1 , \theta')+u(1 , \theta ) | } < \epsilon\ ] ] this proves the continuity between @xmath101 and @xmath151 when @xmath152 .
secondly , we show the continuity between @xmath101 and @xmath166 when @xmath167 .
@xmath168dh -u(2 , \theta')+u(2 , \theta ) |}\end{aligned}\ ] ] because of the continuity of @xmath166 that we have just shown , there exists a @xmath155 such that @xmath169 according to our assumption ( equation ( 13 ) ) , when @xmath157 goes to positive infinity , we have @xmath170 therefore , there exists a @xmath159 such that @xmath171dh | } < \frac{\epsilon}{2},\ ] ] if @xmath172 which is the same with @xmath173 .
let @xmath174 .
if @xmath175 and @xmath176 , then we have @xmath177dh -u(2 , \theta')+u(2 , \theta ) | } < \epsilon\ ] ] this proves the continuity between @xmath101 and @xmath166 when @xmath167 .
now we change the problem setting again .
we still have two concentric circles with radius 1 and 2 respectively , and infinitely many spirals between them .
the spirals go though every point between the two circles and each of them goes to approximate both of the circles ( shown as figure 4 ) .
[ fig : figure4 ] the spirals will neither touch each other nor the two circles so that if we pick an arbitrary small part of the foliation , we can stretch the curves to be straight parallel lines . a continuous and bounded function @xmath178 is defined on the two circles and the spiral , i.e. it is defined on the closed area between the two circles .
we want to find another continuous and bounded function @xmath179 which is differentiable along the spirals and the circles such that @xmath180 where @xmath181 is the gradient of function @xmath179 .
@xmath181 has the form of @xmath182 where @xmath183 and @xmath184 are continuous functions of @xmath185 and @xmath186 . since the definition of the functions are on circles and spirals , it would be more convenient if we transform the coordinates into polar systems .
let @xmath187 , @xmath157 be two variables in the corresponding polar system .
then we have @xmath188 notice that the radius @xmath187 is a function of the angle @xmath157 , we have the expression of @xmath187 on spiral s : @xmath189 where @xmath190 .
we also need the expression for @xmath187 and @xmath157 in terms of @xmath185 and @xmath186 so that we know which spiral is point @xmath191 on . from equation
( 22 ) , @xmath192\ ] ] from ( 20 ) ( 21 ) , @xmath193 therefore , we can solve for the expressions of @xmath157 and @xmath194 in terms of @xmath185 and @xmath186 .
@xmath195-\arctan \frac{y}{x}\end{aligned}\ ] ] notice that the expressions for both @xmath157 and @xmath194 are uniformly continuous .
that indicates that if @xmath178 is continuous , then @xmath90 is continuous and if @xmath179 is continuous , then @xmath196 is continuous and vice versa .
so we know the transformation from cartesian to polar coordinates is : @xmath197 the transformation from polar to cartesian coordinates is : @xmath198 then , we do the transformation of gradient .
the gradient of function @xmath199 is @xmath200 fix @xmath194 , from ( 21 ) ( 22 ) we get @xmath201 @xmath202 \tilde{\nabla}\\ & = & \left [ \begin{array}{cc } \cos \theta & -r \sin \theta \\
\sin \theta & r \cos \theta \end{array } \right ] \left [ \begin{array}{c } \frac{dr}{d\theta } \\ 1 \end{array } \right ] \\ & = & \left [ \begin{array}{c } \cos \theta \frac{dr}{d\theta } -r \sin \theta\\ \sin \theta \frac{dr}{d\theta } + r \cos \theta \end{array } \right ] \\ & = & \left [ \begin{array}{c } f(x , y)\\ g(x , y ) \end{array } \right ] \end{aligned}\ ] ] hence we know @xmath203 then , we have to make sure that the transformed variables @xmath185 and @xmath186 are eligable under our conditions . by assumption , @xmath204 and @xmath205 .
it is easy to see that @xmath206 goes from @xmath207 to @xmath208 .
however , the value of @xmath206 gives infinite many choices for @xmath157 .
so we look at equation ( 26 ) . given @xmath204 , @xmath157 can only be in an interval with a length smaller than @xmath117 . according to the properties of @xmath209 , we have @xmath210 where @xmath46 is a integer .
thus we only have a unique value for @xmath157 and this transformation is eligable .
now we start to investigate the properties of all the curves . on the circles
, we have @xmath93 @xmath94 @xmath95 @xmath96 on spiral s , function @xmath211 has following property on the spiral : @xmath212 @xmath213 we want to show that function @xmath199 has the same property : @xmath214 @xmath215 if we set @xmath216 , then @xmath217 and hence @xmath218 which is not a bounded function .
this indicates that the differential equation @xmath219 is not solvable .
therefore , @xmath180 is not solvable either . to solve the problem
, we add @xmath101 to the left side of the equation and get a new equation @xmath220 .
we found that we can solve this new equation by multiplying @xmath109 to both sides of the equation and get , @xmath221 which is the same as @xmath222 so , our original equation based on the cartesian coordinates becomes @xmath223 .
now we are going to find the general form for @xmath179 through finding solution for @xmath199 . on the circles
, we have @xmath224 @xmath225 fix a point on a circle with coordinates @xmath114 , according to the formulae we derived above , we have @xmath115 then , we need to make sure that the function @xmath226 is periodic with period of @xmath117 , i.e. we need to verify that the equation @xmath118 holds . + from equation ( 23 ) , @xmath119 let @xmath120 , @xmath227 this proves that @xmath228 is a periodic function with period of @xmath117 .
+ now we will get the solution formulae on the spirals .
let @xmath229 @xmath230 on the spiral , we have , @xmath231 @xmath232 substitute ( 30 ) , ( 31 ) back to the solution , we get @xmath233 fix a point on the spiral with coordinates @xmath128 , according to the formulae we derived above , we have @xmath234 next step is to show the continuity property of the solution .
we are going to prove the following theorem : the function @xmath179 is continuous if the function @xmath178 is continuous .
now we show the continuity of the solution we got .
we are going to examine the continuity of @xmath130 , @xmath235 , between the spirals and between the spirals and the circles .
we know @xmath90 is continuous since @xmath178 is continuous .
before we show the continuity of all the things listed above , we first examine the continuity of the function @xmath236 in case we need it to know the properties of the foliation .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath237 such that @xmath238 holds for every pair of @xmath133 and @xmath239 if @xmath240 and @xmath241 .
we have @xmath242 because of the continuity of the @xmath243 function , there exists a @xmath237 such that @xmath244 if @xmath240 and @xmath241 .
* step 1 . * to show the continuity of the whole solution , we start with showing that the solution on the circles is continuous .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath80 such that @xmath132 holds for every pair of @xmath133 if @xmath134 . without loss of generality , we assume @xmath135 .
@xmath136 because of the continuity of @xmath137 , there exists a @xmath80 such that @xmath245 * step 2 . * showing the solution is continuous on each single spiral .
we randomly choose a spiral s with the radius function @xmath246 .
since @xmath157 is the only variable here , we can write the radius number of spiral s to be @xmath247 which is obviously a continuous function .
since both @xmath248 and @xmath249 are continuous , @xmath250 is again a continuous function .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath85 such that @xmath251 holds for every pair of @xmath133 if @xmath140 . without loss of generality , we assume @xmath135 .
@xmath252 because of the continuity of @xmath250 , there exists a @xmath85 such that @xmath253 * step 3 . * showing the solution is continuous between different spirals .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath144 such that @xmath254 holds for every pair of @xmath133 and @xmath239 if @xmath148 and @xmath255 . without loss of generality , we assume @xmath135 .
let @xmath256 , @xmath257 , @xmath258 and @xmath259 .
@xmath260 dh-\int_{\theta}^{\theta ' } e^hv(r(h , s ' ) , h)dh | } \\ & = & { | \int_{-\infty}^{-l } e^h [ v(r(h , s ) , h)-v(r(h , s ' ) , h ) ] dh+ \int_{-l}^{\theta } e^h [ v(r(h , s ) , h)-v(r(h , s ' ) , h ) ] dh\\ & & -\int_{\theta}^{\theta ' } e^hv(r(h , s ' ) , h)dh |}\end{aligned}\ ] ] where l is a large positive number to be chosen . because of the continuity of @xmath261 , there exists a @xmath262 such that @xmath263 if @xmath163 .
+ if @xmath264 and @xmath265 , because of the continuity of @xmath266 , @xmath267 can be arbitrary small . then , because of the continuity of @xmath261 , there exists a small @xmath268 such that @xmath269 if @xmath270 which is the same with @xmath264 and @xmath265 .
after that , we evaluate l. let @xmath271 where @xmath272 is a positive real number . then we have @xmath273 therefore , we have @xmath274 dh | } < \frac{\epsilon}{4},\ ] ] if @xmath270 . since @xmath275 is a continuous and bounded function
, we have @xmath276 if @xmath277 which is the same with @xmath264 and @xmath265 therefore , @xmath278
dh\\ & < & \int_{-\ln \frac{4m}{\epsilon}}^{\theta } e^\theta \frac{\epsilon}{4(\theta - ( -\ln \frac{4m}{\epsilon}))e^\theta))e^\theta } dh \\ & = & \frac{\epsilon}{4 } , \end{aligned}\ ] ] if @xmath277 .
then , let @xmath279 , we have @xmath280 which gives , @xmath281 if @xmath148 and @xmath255 which is the same with @xmath264 and @xmath265 * step 4 . * showing the solution is continuous between each spiral and the inner circle .
we know that @xmath275 is continuous function for all s. let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath282 such that @xmath283 | } < \epsilon$ ] holds for every pair of @xmath133 and @xmath284 if @xmath285 and @xmath286 . without loss of generality , we assume @xmath153
. we will start with the cartesian coordinates .
@xmath287dh -u(1 , \theta')+u(1 , \theta ) |}\end{aligned}\ ] ] because of the continuity of @xmath151 that we have just shown , there exists a @xmath288 such that @xmath289 according to our assumption ( equation ( 21 ) ) , when @xmath152 , we have @xmath290 therefore , there exists a @xmath291 such that @xmath292dh | } < \frac{\epsilon}{2},\ ] ] if @xmath152 which is the same with @xmath293 .
let @xmath294 .
if @xmath295 and @xmath296 , then we have @xmath297dh -u(1 , \theta')+u(1 , \theta ) | } < \epsilon\ ] ] this proves the continuity between @xmath199 and @xmath151 for all s when @xmath152 .
* step 5 . * showing the solution is continuous between each spiral and the outer circle . we know that @xmath298 is continuous function for all s. let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath299 such that @xmath300 | } < \epsilon$ ] holds for every pair of @xmath133 and @xmath301 when @xmath167 if @xmath302 and @xmath303 . without loss of generality , we assume @xmath153 .
@xmath304dh -u(2 , \theta')+u(2 , \theta ) |}\end{aligned}\ ] ] because of the continuity of @xmath166 that we have just shown , there exists a @xmath305 such that @xmath306 according to our assumption ( equation ( 22 ) ) , when @xmath167 , we have @xmath307 therefore , there exists a @xmath308 such that @xmath309dh | } < \frac{\epsilon}{2},\ ] ] if @xmath167 which is the same with @xmath310 .
let @xmath311 .
if @xmath312 and @xmath313 , then we have @xmath314dh -u(2 , \theta')+u(2 , \theta ) | } < \epsilon\ ] ] this proves the continuity between @xmath199 and @xmath166 for all s when @xmath167 . in conlusion ,
. then we have @xmath316 if @xmath317 and @xmath318 . therefore , @xmath122 is continuous and hence @xmath179 is also continuous .
consider a vector field @xmath5 in @xmath319 , we have a compact set which is a union of integral curves of vector field @xmath5 .
so , in particular we assume the the vector field @xmath5 does not vanish at any point in @xmath4 .
let @xmath56 be a continuous function on a compact set @xmath4 .
based on previous experience , we want to find a continuous function @xmath55 which is differentiable along the integral curves so that on each curve , @xmath320 imagine there is a particle moving along the curves , @xmath5 can be viewed as the velocity of the particle and @xmath321 is the speed .
we could set @xmath322 by changing the coordinates so that the particles move at speed of 1 in all directions on all curves . here
is another way to state this problem . consider a compact set @xmath4 in @xmath319 .
there is only one curve through every point in @xmath4 .
the curves vary continuously , have similar slopes and do not intersect with each other .
let @xmath56 be a continuous function on @xmath4 , we will find a continuous function @xmath55 which is differentiable along the curves so that on each curve , @xmath323 where @xmath324 is the distance that a point on a curve go through and @xmath325 is a coefficient function which is continuous and positive .
notice that since function @xmath325 is defined on a compact set @xmath4 , @xmath325 is bounded automatically .
to relate the two case , let @xmath326 , then @xmath327 . from ( 34 ) , we have @xmath328 which is the same with ( 35 ) .
thus we know the two conditions are identical and we can solve the general foliations through solving @xmath329 .
we first focus on the real line case . on a real line ,
our differential equation becomes @xmath330 .
based on previous experience , we multiply @xmath16 to the equation and get : @xmath331 let @xmath332 , we have @xmath333 which is obviously a bounded function .
hence the equation @xmath330 is solvable .
so we have the general formulae for u(x ) : @xmath334 if @xmath335 is periodic functions with period of 1 , we want to know whether @xmath336 is also periodic or not .
let @xmath337 .
@xmath338 so @xmath336 is a periodic function with period of 1 .
we also consider another solvable equation @xmath339 .
@xmath340 set v(x)=1 , we have @xmath341 which is obviously a bounded function . then we know our modified equation @xmath339 which is the same with @xmath342 is solvable .
so we have the general formulae for u(x ) : @xmath343 going back to the original case , our equation is : @xmath344 if @xmath335 and @xmath345 are periodic functions with periods of 1 , we want to know whether @xmath336 is also periodic or not .
let @xmath337 .
@xmath346 so @xmath336 is a periodic function with period of 1 .
but we are not going to discuss more about this equation in this paper .
we have the following assumptions : let vector field @xmath347 , and @xmath348 .
let @xmath4 be a compact union of integral curves .
then there exists a map @xmath349
@xmath350 let @xmath351 be an initial point on the compact set @xmath4 .
then , by assumption , there is only one curve @xmath352 goes thought @xmath185 parameterized by time @xmath324 . when @xmath353 , @xmath354 since @xmath185 is the initial point . then , we map the initial point to another point after time @xmath324 by @xmath349 .
we have @xmath355 since @xmath347 , the map @xmath356 . for each @xmath357 ,
the map @xmath358 has a @xmath2 inverse map @xmath359 in a neighborhood of @xmath4 .
@xmath360 on a compact set @xmath4 , if @xmath56 is continuous along an integral curve , then @xmath55 is continuous along the curve .
pick any integral curve @xmath352 , we have a function @xmath29 which is defined as @xmath361 then we have a differential equation @xmath362 where @xmath363 is a bounded function which is differentiable along the curve . we know that @xmath364 so what we need to show is the continuity of @xmath363 given that @xmath29 is continuous
we have the general formulae of @xmath363 which is : @xmath365 we will prove the continuity of @xmath363 by showing that @xmath366 is continuous .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath24 such that @xmath367 holds for every pair of @xmath368 if @xmath369 . without loss of generality , we assume @xmath370 .
@xmath371 let @xmath372 , because of the continuity of @xmath29 , there exist a @xmath373 such that @xmath374 @xmath375 if @xmath376 , then @xmath367 and hence @xmath363 is continuous .
therefore , @xmath377 is continuous . on a compact set @xmath4 ,
if @xmath56 is continuous function , then @xmath55 is continuous .
proof . according to what we have done above , we have @xmath378 when @xmath353 , we know that @xmath354 since @xmath185 is the initial point
. then we have @xmath379 by assumption , we have @xmath380 so our formulae for @xmath55 when @xmath353 is @xmath381 let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath24 such that @xmath382 holds for every pair of @xmath26 if @xmath383 .
let @xmath384 be a large number to be chosen .
@xmath385dw+\int_{-l}^0 [ e^wv(\phi ( x , w))- e^wv(\phi ( x ' ,
w))]dw |}\end{aligned}\ ] ] let @xmath272 be the maximum of function @xmath56 , then @xmath386 @xmath387dw\\ & < & \int_{-\infty}^{-l } e^w 2mdw\\ & = & 2me^{-l}<\frac{\epsilon}{2}\end{aligned}\ ] ] solve the inequation above , we get @xmath388 on the other hand , because of the continuity of @xmath56 and @xmath349 , there exists a @xmath389 such that @xmath390dw<\frac{\epsilon}{2}\ ] ] if @xmath383 .
therefore , @xmath55 is continuous on @xmath4 .
then we are going to find that if @xmath56 is better than continuous , what will happen to @xmath55 . on a compact set @xmath4 , if @xmath391 , then @xmath392 . proof .
if @xmath392 , from ( 47 ) , we will have @xmath393 we will show that equation above holds .
let @xmath394 notice that @xmath395 .
then the equation that we are going to prove becomes @xmath396 by definition of derivative , we have @xmath397 by the mean - value theorem , @xmath398 where @xmath399 .
@xmath400 @xmath401 where @xmath384 is a large number to be chosen .
let @xmath272 be the maximum value of @xmath402 and @xmath403 be the maximum value of @xmath404 .
let @xmath22 be an arbitrary small number .
@xmath405 solve the inequation above , we get @xmath406 similarly , we can get @xmath407 by setting @xmath408 then we make @xmath409 and get the following equations : @xmath410 @xmath411 when @xmath412 , @xmath413 .
hence we know that @xmath414 therefore , @xmath392 . on a compact set @xmath4 , if @xmath415 , then @xmath416
. proof .
we will show this theorem by mathematical induction . initial step . from theorem 4 ,
we know when @xmath417 , if @xmath391 , then @xmath392 . induction step .
assume that for @xmath418 , if @xmath419 , then @xmath420 .
that is to say that we have @xmath421 we will show that for @xmath422 , if @xmath423 , then @xmath424 .
that is to say , we will show that the following equation holds : @xmath425 let @xmath426 since @xmath427 , it is obvious that @xmath428
. let @xmath429 we will show that @xmath430 by definition of derivative , we have @xmath431 by the mean - value theorem , @xmath432 where @xmath399 .
@xmath433 @xmath434 where @xmath384 is a large number to be chosen .
let @xmath272 be the maximum value of @xmath435 and @xmath403 be the maximum value of @xmath436 .
let @xmath22 be an arbitrary small number .
@xmath437 solve the inequation above , we get @xmath406 similarly , we can get @xmath407 by setting @xmath438 then we make @xmath409 and get the following equations : @xmath439 @xmath440 when @xmath412 , @xmath413 .
hence we know that @xmath441 therefore , @xmath442 and @xmath424 .
in the previous section , we only considered nice vector field @xmath348 everywhere .
do we really need that condition ? in this section , we are going to check with vector field @xmath443 on a compact set @xmath4 with singularities at the origin and see whether that works well .
since we still need a bounded vector field , @xmath444 would be a good choice .
however , it would be a huge mass when we do the calculations with @xmath445 , so we modify function @xmath446 into : @xmath447 we will first solve the problem on a real line .
let @xmath8 be a bounded and continuous function .
we will solve the differential equation @xmath448 where @xmath1 is a bounded function .
let @xmath12 .
when @xmath449 , we have @xmath450 obviously , @xmath451 is not bounded around the origin .
so this differential equation is not solvable
. then we multiply @xmath185 to both sides of equation ( 59 ) and get @xmath452 let @xmath12 . when @xmath449 , we have @xmath453 when @xmath454 , we have @xmath455 since @xmath456 is not bounded , this differential equation is not solvable either
. then we multiply @xmath16 to both sides of equation ( 60 ) and get @xmath457 let @xmath12 .
when @xmath449 , we have @xmath458 since @xmath459 then when @xmath460 , @xmath461 .
so @xmath7 is bounded between -1 and 1 . when @xmath454 , we have @xmath462 since @xmath463 , we know @xmath7 is bounded when @xmath454 . when @xmath464 , we have @xmath465 since @xmath466 is not bounded when @xmath467 , @xmath7 is not bounded when @xmath468
. then we modify equation ( 61 ) into @xmath469 when @xmath470 , we have the same bounded function @xmath7 as what we got from equation ( 60 ) .
when @xmath471 , we have @xmath472 since @xmath473 then when @xmath460 , @xmath461 .
so @xmath7 is bounded between -1 and 0 . when @xmath464 , we have @xmath474 since @xmath463 , we know @xmath7 is bounded when @xmath464 .
therefore , @xmath7 is bounded through the real line and differential equation ( 62 ) is solvable
. now we will derive the general formulae for @xmath7 in equation ( 62 ) . when @xmath475 , we have @xmath476 when @xmath454 , we have @xmath477\end{aligned}\ ] ] when @xmath478 , we have @xmath479 when @xmath464 , we have @xmath480\end{aligned}\ ] ] then we are going to show the continuity of @xmath7 if @xmath8 is continuous by show that @xmath7 is continuous in each of the four parts and between neighboring parts .
let @xmath22 be an arbitrary small number .
when @xmath475 , we will show that there exist a real number @xmath80 such that @xmath481 holds for every pair of @xmath26 if @xmath482 .
@xmath483 since @xmath8 is bounded and @xmath484 between 0 and 1 , we can find a maximum value @xmath272 for @xmath485 .
therefore , there exists @xmath80 such that @xmath486 if @xmath482 .
when @xmath454 , we will show that there exist a real number @xmath85 such that @xmath487 holds for every pair of @xmath26 if @xmath488 .
@xmath489 we fix @xmath185 at a random point , we have a finite value of @xmath490 .
therefore , we have @xmath491 if @xmath85 is small enough and @xmath488 .
when @xmath478 , we will show that there exist a real number @xmath144 such that @xmath492 holds for every pair of @xmath26 if @xmath493 .
@xmath494 since @xmath8 is bounded and @xmath495 between 0 and 1 , we can find a maximum value @xmath272 for @xmath496 .
therefore , there exists @xmath144 such that @xmath497 if @xmath493 .
when @xmath464 , we will show that there exist a real number @xmath282 such that @xmath498 holds for every pair of @xmath26 if @xmath499 .
@xmath500 we fix @xmath185 at a random point , we have a finite value of @xmath501 .
therefore , we have @xmath502 if @xmath282 is small enough and @xmath499 .
to show the continuity between different parts of @xmath7 , we need to have the following equations : @xmath503 @xmath504 since we have shown @xmath7 is continuous on each part and equations ( 63 ) ( 64 ) and ( 65 ) hold , we know that @xmath7 is continuous .
consider a vector field @xmath505 on circle with singularities on it .
notice that there are infinitely many singularities on the circle since we encounter one or more singularities every @xmath117 distance .
so the function @xmath506 has to be a periodic function with period of @xmath117 .
let @xmath507 based on previous experience , we have our differential equation in the form of @xmath508=f(\theta)v(\theta)\ ] ] where @xmath509 is a continuous and periodic function , @xmath510 is continuous and @xmath511 is continuous , periodic and differentiable along @xmath157 .
obviously , there are singularities at @xmath512 and @xmath513 .
now we generate the general formulae for @xmath511 which is @xmath514 since @xmath509 periodic , let @xmath515 , we have @xmath516 so to make @xmath511 to be periodic , @xmath510 has to satisfy @xmath517 where @xmath14 is a constant .
we found that if the function @xmath518 cancels the singularity of @xmath519 at 0 , i.e. @xmath520 , then there would be a problem at @xmath521 since there is a @xmath522 outside the integral .
then , to avoid the point @xmath53 , we change our integral interval to @xmath523 and get @xmath524 since we can always find a @xmath8 such that @xmath525 there is no way for @xmath511 to be periodic .
we partition a circle into upper part and lower part with two overlaps at the left and right ends respectively .
let @xmath526 be the set that contains the upper part of the circle and @xmath527 contains the lower part .
we define @xmath384 as a set of contains left overlap area and @xmath528 as the set contains the right overlap area .
both @xmath384 and @xmath528 are intersection of sets @xmath526 and @xmath527 .
base on the assumptions , we have @xmath529 @xmath530 then we separate the differential equation @xmath531 into the following two equations : @xmath532 thus we separate functions @xmath1 and @xmath3 into four functions : @xmath533 and @xmath534 . on the set @xmath527 which is the lower part of the circle , we have @xmath535 on the set @xmath526 which is the upper part of the circle , we have @xmath536 on the set @xmath384 which is the left overlap part of the circle , we have @xmath537 on the set @xmath528 which is the right overlap part of the circle , we have @xmath538 let @xmath539 notice that in set @xmath384 , @xmath540 and in set @xmath528 , @xmath541 .
we have to make sure that @xmath542 and @xmath543 and @xmath544 match each other on the overlap part .
thus our differential equation becomes @xmath548 based on previous experience , @xmath549 is solvable and the general formulae for @xmath549 is @xmath550 since functions @xmath549 and @xmath551 are defined on the circle , they have to be both periodic with period of @xmath117 .
so we have @xmath552 is periodic and we will show that @xmath549 is periodic too .
let @xmath553 , we have @xmath554 hence we know that @xmath549 is a periodic function .
let @xmath22 be an arbitrary small number .
we will show that there exist a real number @xmath24 such that @xmath556 holds for every pair of @xmath133 if @xmath557 .
@xmath558 we fix @xmath157 at a random point , we have a finite value of @xmath559 . therefore , we have @xmath560 if @xmath24 is small enough and @xmath557 . in this section , we have two concentric circles with radius 1 and 2 respectively , and infinite many spirals between them just as what we have in section 4 .
the spirals goes though every point between the two circles and each of them goes to approximate both of the circles .
the spirals will neither touch each other nor the two circles so that if we pick an arbitrary small part of the foliation , we can stretch the curves to be straight parallel lines .
a continuous and bounded function @xmath178 is defined on the two circles and the spiral , i.e. it is defined on the closed area between the two circles .
we want to find another continuous and bounded function @xmath179 which is differentiable along the spirals such that @xmath561 where @xmath5 is the vector field .
@xmath5 has the form of @xmath562 where @xmath183 and @xmath184 are continuous functions of @xmath185 and @xmath186 . since the definition of the functions are on circles and spirals , it would be more convenient if we transform the coordinates into polar systems .
let @xmath187 , @xmath157 be two variables in the corresponding polar system .
then we have @xmath188 notice that the radius @xmath187 is a function of the angle @xmath157 , we have the expression of @xmath187 on spiral s : @xmath189 where @xmath190 . on the circles
, we have @xmath93 @xmath94 @xmath95 @xmath96 on spiral s , function @xmath211 has following property on the spiral : @xmath212 @xmath213 we want to show that function @xmath199 has the same property : @xmath214 @xmath215 we partition the annulus into upper part and lower part with two overlaps at the left and right ends respectively .
let @xmath526 be the set that contains the upper part of the annulus and @xmath527 contains the lower part .
we define @xmath384 as a set contains left overlap area and @xmath528 as the set contains the right overlap area .
both @xmath384 and @xmath528 are intersection of sets @xmath526 and @xmath527 .
base on the assumptions , we have @xmath564 @xmath530 then we separate the differential equation @xmath531 into the following two equations : @xmath532 thus we separate functions @xmath1 and @xmath3 into four functions : @xmath533 and @xmath534 . on the set @xmath527 which is the lower part of the annulus , we have @xmath535 on the set @xmath526 which is the upper part of the annulus , we have @xmath536 on the set @xmath384 which is the left overlap part of the annulus , we have @xmath537 on the set @xmath528 which is the right overlap part of the annulus , we have @xmath538 let @xmath539 notice that in set @xmath384 , @xmath540 and in set @xmath528 , @xmath541 .
we have to make sure that @xmath542 and @xmath543 and @xmath544 match each other on the overlap part .
thus our differential equation becomes @xmath565 based on previous experience , @xmath566 is solvable and the general formulae for @xmath566 is @xmath567 we have shown the continuity of @xmath566 in section 2.4 . consider a laminated set @xmath4 on @xmath568 which is a union of integral curves .
we cover the laminated set @xmath4 by finitely many boxes .
we have an orientation of the set @xmath4 that define the direction that the arc length increases along the curves .
imagine there is a particle moving along the curves at a velocity of 1 .
assume along the direction of the curves , there are drops in the overlap area .
we define the drop between boxes as following : in each box @xmath569 , we have a function @xmath570 .
we assume that at the left boundary of the box , @xmath571 for all curves .
that is to say , when @xmath353 , we have @xmath572 where @xmath324 is distance that a particle go through in the box since the velocity of the particle is always 1 .
suppose we have two boxes @xmath573 and @xmath574 and the overlap of the two boxes @xmath575 .
there are some curves go from @xmath573 to @xmath574 . on curve @xmath384
, we have a dropping constant given by @xmath576 based on the assumptions , on leaf @xmath384 we have @xmath577 @xmath578 assume we have three boxes @xmath573 @xmath574 and @xmath579 and they all have overlaps with each other .
then we have @xmath580 this proves the consistency of our assumption
. then we show our main results .
we separate the differential equation @xmath581 into ( infinite ) many equations : @xmath582 in box @xmath583 , we have @xmath584 on overlap @xmath585 , we have @xmath586 let @xmath587 @xmath588 notice that on overlap @xmath585 , @xmath589 .
@xmath590 similarly , we have @xmath591 on the overlap @xmath585 .
therefore , our differential equation becomes @xmath592 which is the equation we start with this paper .
the general formulae of @xmath593 is @xmath594 we have shown the continuity of @xmath593 in section 2 .
john e. fornaess + mathematics department + the university of michigan + ann arbor , michigan 48109 + usa + [email protected] + xiaoai chai + mathematics department + the university of michigan + ann arbor , michigan 48109 + usa + [email protected] + | in this paper , we discuss foliations by real curves . we investigate differential equations which are modifications of @xmath0 along leaves .
our focus is on having a solution operator so that @xmath1 is continuous ( or @xmath2 ) if @xmath3 is continuous ( or @xmath2 ) .
xiaoai chai john e. fornss |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.