首页 | 本学科首页   官方微博 | 高级检索  
相似文献
 共查询到20条相似文献,搜索用时 31 毫秒
1.
This paper investigates the properties of (0) optimal policies in the model of [2]. It is shown that, if * = ( 0 * , 1 * ,..., n * , n +1/* , ...) is a-discounted optimal policy, then ( 0 * , 1 * , ..., n * ) for alln0 is also a-discounted optimal policy. Under some condition we prove that stochastic stationary policy n * corresponding to the decision rule n * is also optimal for the same discounting factor. We have also shown that for each-optimal stochastic stationary policy 0 * , 0 * can be decomposed into several decision rules to which the corresponding stationary policies are also-optimal separately; and conversely, a proper convex combination of these decision rules is identified with the former 0 * . We have further proved that for any (,)-optimal policy, say *=( 0 * , 1 * , ..., n * , n +1/* , ...), n–1 * ) is ((1– n )–1 e, ) optimal forn>0. At the end of this paper we mention that the results about convex combinations and decompositions of optimal policies of § 4 in [1] can be extended to our case.Project supported by the Science Fund of the Chinese Academy of Sciences.  相似文献   

2.
In this paper, we will use the Birkhoff's ergodic theorem to do some finer analysis on the spectral properties of slant Toeplitz operators. For example, we will show that if is an invertibleL function on the unit circle, then almost every point in (A * ) is not an eigenvalue ofA * . More specifically, we will show that the point spectrum ofA * is contained in a circle with positive radius.  相似文献   

3.
C. Hightower found two infinite sequences of gaps in the Markov spectrum, ( n , n ) and ( n , n ) with n and n both Markov elements, converging to . This paper exhibits Markov elements n * and n * such that, for alln 1, ( n * , n ) and ( n n * ) are gaps in the Markov spectrum. Other results include showing that, for alln 1, n is completely isolated, while the other endpoints of the gaps are limit points in the Markov spectrum.  相似文献   

4.
Summary We consider a (possibly) vector-valued function u: RN, Rn, minimizing the integral , 2-2/(n*1)<p<2, whereD i u=u/x i or some more general functional retaining the same behaviour, we prove higher integrability for Du: D1 u,..., Dn–1 u Lp/(p-1) and Dnu L2; this result allows us to get existence of second weak derivatives: D(D1 u),...,D(Dn–1u)L2 and D(Dn u) L p.This work has been supported by MURST and GNAFA-CNR.  相似文献   

5.
Let T~=i=1 n irvi:V V=[v1,. . . .,vn] X, where i V* and X is a Banach space. Let T= i=1 nuivi: X V be an extension of T~ to all of X (i.e., ui X*) such that T has minimal (operator) norm. (E.g., if T~=I, T is a minimal projection from X onto V.) Then it is necessary and sufficient that u:=u_1,. . . ,un is given by (v:=v1,. . . ,vn)extv(u) Vn,where the notion of a v-extremal (extv) of u is properly defined.The condition above leads in many important cases to a simple geometric interpretation of minimal projections. Furthermore, by applying this formula to the case X=Lp, we obtain a linear n-dimensional analog of the Hölder equality condition (M is given by extv(u)=Mv)1/p u · Mv = 1/q u · Mv,wherever v is differentiable.We point out several applications, including the determination of the absolute projection constant of n p   相似文献   

6.
Summary We consider the problem of comparing large finite and infinite systems with locally interacting components, and present a general comparison scheme for the case when the infinite system is nonergodic. We show that this scheme holds for some specific models. One of these is critical branching random walk onZ d . Let t denote this system, and let t N denote a finite version of t defined on the torus [–N,N] d Z d . Ford3 we prove that for stationary, shift ergodic initial measures with density , that ifT(N) andT(N)/(2N+1)d s[0,] asN, then {v }, 0 is the set of extremal invariant measures for the infinite system t andQ s is the transition function of Feller's branching diffusion. We prove several extensions and refinements of this result. The other systems we consider are the voter model and the contact process.Work supported in part by the National Science Foundation under Grant DMS-8802055, by the U.S. Army Research Office through the Mathematical Sciences Institute at Cornell University and by the Deutsche Forschungsgemeinschaft through the SFB 123 at the Universität Heidelberg  相似文献   

7.
Let * be the equilateral triangulation of the plane and let 1 * be the equilateral triangle formed by four triangles of *. We study the space of piecewise polynomial functions in C k (R 2) with support 1 *, having a sufficiently high degree n and which are invariant with respect to the group of symmetries of 1 *. Such splines are called 1 *-splines. We first compute the dimension of this space in function of n and k. Then, for any fixed k0, we prove the existence of 1 *-splines of class C k and minimal degree, but these splines are not unique. Finally, we describe an algorithm computing the Bernstein–Bézier coefficients of these splines.  相似文献   

8.
We give efficiency estimates for proximal bundle methods for finding f*minXf, where f and X are convex. We show that, for any accuracy <0, these methods find a point xkX such that f(xk)–f* after at most k=O(1/3) objective and subgradient evaluations.  相似文献   

9.
We look at an extension of the steady state delay probability inM/M/s/s + c systems to nonintegral number of serverss and queue capacityc, which we call GED function. We show that this function is increasing and concave in the queue capacity. We find that if c 1, the reciprocal of the GED function is convex in the traffic intensity and the GED function is increasing in the traffic intensity if is below some * s,c, and decreasing if is greater than * s,c, Moreover, * s,c is increasing in the number of servers and, fors 1, * i,c=1 p* s,c < 2.Research supported by Grant BD/645/90-RM from Junta Nacional de Investigação Científica e Tecnológica.On leave from: Departamento de Matemätica, Instituto Superior Técnico, Av. Rovisco Pais, 1096 Lisboa Codex, Portugal.  相似文献   

10.
11.
We study the limiting behavior of the weighted central paths{(x(), s())} > 0 in linear programming at both = 0 and = . We establish the existence of a partition (B ,N ) of the index set { 1, ,n } such thatx i() ands j () as fori B , andj N , andx N (),s B () converge to weighted analytic centers of certain polytopes. For allk 1, we show that thekth order derivativesx (k) () ands (k) () converge when 0 and . Consequently, the derivatives of each order are bounded in the interval (0, ). We calculate the limiting derivatives explicitly, and establish the surprising result that all higher order derivatives (k 2) converge to zero when .  相似文献   

12.
For any locally compact groupG, we show that any locally tight homomorphism from a real directed semigroup intoM 1 (G) (semigroup of probability measures onG) has a shift which extends to a continuous one-parameter semigroup. IfG is ap-adic algebraic group then the above holds even iff is not locally tight. These results are applied to give sufficient conditions for embeddability of some translate of limits of sequences of the form {v n kn } and M 1 (G) such that ()= M , for somek>1 and AutG (cf. Theorems 2.1, 2.4, 3.7).  相似文献   

13.
Zeta functions v(z; q)= n=1 [jvn(q)]–z and partition functions Zv(t; q)=n exp[–tj vn 2 (q)] related to the zeros jvn(q) of the Bessel q-functions Jv(x; q) and J v (2) (x; q) are studied and explicit formulas for v(2n; q) at n=±1, ±2, ... are obtained. The poles of v(z; q) in the complex plane and the corresponding residues are found. Asymptotics of the partition functions Zv(t; q) at t 0 are investigated.Translated from Teoreticheskaya i Matematicheskaya Fizika, Vol. 107, No. 3, pp. 397–414, June, 1996.  相似文献   

14.
The application of the ML method in linear regression requires a parametric form for the error density. When this is not available, the density may be parameterized by its cumulants ( i ) and the ML then applied. Results are obtained when the standardized cumulants ( i ) satisfy i = i+2/ 2 (i+2)/2 =O(v i ) asv 0 fori>0.Research financed in part by the Research Center of the Athens University of Economics and Business.  相似文献   

15.
Let XoR n be an irreducible analytic germ and the order space of its field of meromorphicfunetion germs. A formal half-branch in Xo is a kind of C-map germ c[0,)Xo; an ordering is centered at c if it contains the functions which are positive on c. We obtain a partition 1,...,d, d=dim Xo, of the set * of central (i.e.: centered at some half-branch) orderings, according to the dimension of half-branches. Then we show that all e, e= 1,.,d, as well as the set \* of noncentral orderings, are dense in . Finally, we solve the 17th Hubert Problem for analytic germs.  相似文献   

16.
a k f k , f k L 2, w-, (2), w(n) — . a k f k N {a k }l 2, {a k }l 2 ( 1, 2, 1a, 2a). ( 2) [8]. , {a k } w-.  相似文献   

17.
LetE be an ample vector bundle of rankr on a compact complex manifoldX of dimension 3 with detE=–K x, andi(X) the index ofX. Then it is proved in this note thati(X)r unless (X,E)(1 × 2,p*O(1) q*), wherep,q are the projections and is isomorphic toO(2) O(1) or the tangent bundleT of 2. This result gives a counterexample to the conjecture formed by T. Peternell.  相似文献   

18.
LetH be a germ of holomorphic diffeomorphism at 0 . Using the existence theorem for quasi-conformal mappings, it is possible to prove that there exists a multivalued germS at 0, such thatS(ze 2i )=HS(z) (1). IfH is an unfolding of diffeomorphisms depending on (,0), withH 0=Id, one introduces its ideal . It is the ideal generated by the germs of coefficients (a i (), 0) at 0 k , whereH (z)–z=a i ()z i . Then one can find a parameter solutionS (z) of (1) which has at each pointz 0 belonging to the domain of definition ofS 0, an expansion in seriesS (z)=z+b i ()(z–z 0) i with , for alli.This result may be applied to the bifurcation theory of vector fields of the plane. LetX be an unfolding of analytic vector fields at 0 2 such that this point is a hyperbolic saddle point for each . LetH (z) be the holonomy map ofX at the saddle point and its associated ideal of coefficients. A consequence of the above result is that one can find analytic intervals , , transversal to the separatrices of the saddle point, such that the difference between the transition mapD (z) and the identity is divisible in the ideal . Finally, suppose thatX is an unfolding of a saddle connection for a vector fieldX 0, with a return map equal to identity. It follows from the above result that the Bautin ideal of the unfolding, defined as the ideal of coefficients of the difference between the return map and the identity at any regular pointz, can also be computed at the singular pointz=0. From this last observation it follows easily that the cyclicity of the unfoldingX , is finite and can be computed explicity in terms of the Bautin ideal.Dedicated to the memory of R. Mañé  相似文献   

19.
Let T be the domain in N defined by the inequalities O < 1 < ... < N < +. Put N+k = /2(1+k) (k=1,2,3), N+4=1+2, and denote byF() the set of functions x=x(u,v)=(x1(u,v),...,xp(u,v)), (p2) of class , where B is the unit disk u2+v2<1, which maps the circular arcs k={w=ei:k<<K+1} (k=1,..., N+3) into the straight lines containing the edges ak, ak+1 (aN+4=a1) of a polygon IRp. Then we show that the function ()= inf xF() D(x) is analytic in T. This generalizes and sharpens an unproved result of I. Marx and M. Shiffman (see [4]).

Hans Lewy und Charles B. Morrey gewidmet  相似文献   

20.
For a collection C of convex bodies let h(C) be the minimum number m with the property that every element K of C can be covered by m or fewer smaller homothetic copies of K. Denote by C d * the collection of all duals of cyclic d-polytopes, d 2. We show that h(C 2k * )=(k +1)2 for any k 2. We also prove the inequalities (d+1) (d+3)/4 h(C d * ) (d+1) 2/2$ for any d 2.  相似文献   

设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号