Rigorous confidence bounds for MCMC under a geometric drift condition
详细信息 查看全文
- 作者：Krzysztof Ł ; atuszyń ; ski ; Wojciech Niemiro
- 关键词：MCMC estimation ; Confidence intervals ; Mean square error ; Markov chain ; Drift condition ; Simulation cost
- 刊名：Journal of Complexity
- 出版年：2011
- 出版时间：February 2011
- 年：2011
- 卷：27
- 期：1
- 页码：23-38
- 全文大小：295 K
文摘
We assume a drift condition towards a small set and bound the mean square error of estimators obtained by taking averages along a single trajectory of a Markov chain Monte Carlo algorithm. We use these bounds to construct fixed-width nonasymptotic confidence intervals. For a possibly unbounded function , let be the value of interest and its MCMC estimate. Precisely, we derive lower bounds for the length of the trajectory n and burn-in time t which ensure that The bounds depend only and explicitly on drift parameters, on the V-norm of f, where V is the drift function and on precision and confidence parameters . Next we analyze an MCMC estimator based on the median of multiple shorter runs that allows for sharper bounds for the required total simulation cost. In particular the methodology can be applied for computing posterior quantities in practically relevant models. We illustrate our bounds numerically in a simple example.