首页 | 本学科首页   官方微博 | 高级检索  
     


A Donsker-Varadhan type of invariance principle
Authors:Naresh C. Jain
Affiliation:(1) Dept. of Mathematics, University of Minnesota, 55455 Minneapolis, MN, USA
Abstract:Summary LetX1,X2,h. be i.i.d. random variables in the domain of attraction of a stable lawG, and denote Sn = X1 + Xn, Ln(ohgr, A) =n–1
$$L_n (omega ,A) = n^{ - 1} sumlimits_{j = 0}^{n - 1} {chi _A ({{S_j } mathord{left/ {vphantom {{S_j } {a(n)}}} right. kern-nulldelimiterspace} {a(n)}})}$$
satisfiesa(n)–1SnrarrG. Large deviation probability estimates of Donsker-Varadhan type are obtained forLn(ohgr, ·), and these are then used to study the behavior of ldquosmallrdquo values of (Sn/a(n). These latter results are analogues of Strassen's results which described the behavior of ldquolargerdquo values of (Sn/a(n)) when the limit law was Gaussian. The limiting constants are seen to depend only on the limit lawG and not on the distribution ofX1. The techniques used are those developed by Donsker and Varadhan in their theory of large deviations.This work was partially supported by the National Science Foundation.
Keywords:
本文献已被 SpringerLink 等数据库收录!
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