Berry-Esseen bound for high dimensional asymptotic approximation of Wilks' Lambda distribution

被引：9

作者：

Ulyanov, Vladimir V.

Wakaki, Hirofumi ^{[1
]}

Fujikoshi, Yasunori

机构：

[1] Hiroshima Univ, Grad Sch Sci, Dept Math, Higashihiroshima 724, Japan

[2] Moscow MV Lomonosov State Univ, Fac Computat Math & Cybernet, Dept Math Stat, Moscow, Russia

[3] Chuo Univ, Grad Sch Sci & Engn, Dept Math, Bunkyo Ku, Tokyo 112, Japan

来源：

STATISTICS & PROBABILITY LETTERS | 2006年 / 76卷 / 12期

基金：

俄罗斯基础研究基金会;

关键词：

asymptotic distribution; Berry-Esseen bound; high dimensional approximation; Wilks' lambda distribution;

D O I：

10.1016/j.spl.2005.12.027

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Let Lambda = \S-e\/\S-e + S-h\, where S-h and S-e are independently distributed as Wishart distributions W-p(q, Sigma) and W-p(n, Sigma), respectively. Then Lambda is distributed as Wilks' lambda distribution Lambda(p,q,n) which appears as the distributions of various multivariate likelihood ratio tests. In this paper, we derive a Berry-Essen bound for a high dimensional asymptotic approximation of the distribution of T = -n log Lambda when p/n --> c is an element of (0, 1). (C) 2006 Elsevier B.V. All rights reserved.

引用

页码：1191 / 1200

页数：10