期刊
SIAM JOURNAL ON CONTROL AND OPTIMIZATION
卷 50, 期 5, 页码 2743-2762出版社
SIAM PUBLICATIONS
DOI: 10.1137/110832288
关键词
Markov chain; Markov chain Monte Carlo; asymptotic variance; average-case analysis; worst-case analysis; rate of convergence; reversibility; nonreversibility
资金
- National Science Council [98-2115-M-001-007-MY3, 99-2115-M-305-001, 100-2115-M-001-006]
Let chi be a finite space and let pi be an underlying probability on chi. For any real-valued function f defined on chi, we are interested in calculating the expectation of f under pi. Let X-0, X-1,..., X-n,... be a Markov chain generated by some transition matrix P with invariant distribution pi. The time average, 1/n Sigma(n-1)(k=0) f(X-k), is a reasonable approximation to the expectation, E-pi[f(X)]. Which matrix P minimizes the asymptotic variance of 1/n Sigma(n-1)(k=0) f(X-k)? The answer depends on f. Rather than a worst-case analysis, we will identify the set of P's that minimize the average asymptotic variance, averaged with respect to a uniform distribution on f.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据