文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > ENTROPY-CONVERGENCE, INSTABILITY IN STIELTJES AND HAMBURGER MOMENT PROBLEMS
侵权投诉

ENTROPY-CONVERGENCE, INSTABILITY IN STIELTJES AND HAMBURGER MOMENT PROBLEMS

The recovering of a positive density, of which a finite number of moments is assigned, is considered (in the Stieltjes and Hamburger moment problems). In the choice of the approximant the Maximum Entropy approach is adopted. Two main problems are taken int

ENTROPY-CONVERGENCE, INSTABILITY IN STIELTJES AND HAMBURGER MOMENT PROBLEMS M. Frontini - A. Tagliani The recovering of a positive density, of which a nite number of moments is assigned, is considered (in the Stieltjes and Hamburger moment problems). In the choice of the approximant the Maximum Entropy approach is adopted. Two main problems are taken into account. 1. A review of di erent criteria concerning the determinacy and indeterminacy of the in nite moment problem is presented. This review includes criteria based on the moments themselves or on the coe cients of the three-terms recurrence relation characterizing the orthogonal polynomials arising from the moment sequence. A particular criterion, guaranteeing the determinacy of the moment problem, is endowed with a geometrical interpretation, which allows the proof of entropy convergence. More precisely, it is proved that, whenever the sequence of assigned moments arises a determinate moment problem, then the approximants converge in entropy to the recovering function. 2. The entropy convergence allows us to employ a large number of moments. Unstability of the recovering is proved when an increasing number of moments is used. Subsequently the condition number, the ratio between the relative error of the function and the relative error of the modi ed moment, is analyzed. We also prove that the condition number is related to the orthogonal polynomials generated by the sequence of moments.

Abstract

1 Introduction A typical case in the Physics framework consists in the knowledge of the rst few expectation values of a positive function f (x) and in trying to obtain information on the values of such a function. We are concerned with an indeterminate moment problem, for which di erent solutions can be used. The Maximum Entropy (ME) method, on Information Theory, chooses among all the admissible approximants, the least biased of them which is compatible with the partial information represented by few known moments. The analytical form of the approximant is obtained by maximizing the information Shannon-entropy functional under the constraints provided by the M+ 1 moments 0;:::; M

S (f )=?

Z

R

f (x) ln f (x)dx? ( 0? 1)( f (x)dx? 0)? R

Z

M X j=1

j(

Z

R

xj f (x)dx? j ): (1:1)

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email