Optimal Hoeffding bounds for discrete reversible Markov chains
成果类型：
Article
署名作者：
León, CA; Perron, F
署名单位：
Universidad de Concepcion; Universite de Montreal
刊物名称：
ANNALS OF APPLIED PROBABILITY
ISSN/ISSBN：
1050-5164
DOI：
10.1214/105051604000000170
发表日期：
2004
页码：
958-970
关键词：
摘要：
We build optimal exponential bounds for the probabilities of large deviations of sums Sigma(k)(=1)(n) f (X(k)) where (X(k)) is a finite reversible Markov chain and f is an arbitrary bounded function. These bounds depend only on the stationary mean Sigmapi f the end-points of the support of f, the sample size n and the second largest eigenvalue lambda of the transition matrix.