期刊
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
卷 45, 期 17, 页码 5098-5111出版社
TAYLOR & FRANCIS INC
DOI: 10.1080/03610926.2014.936562
关键词
Markov chain Monte Carlo; Optimal scaling; random-walk Metropolis-Hastings; Robbins-Monro
资金
- Medical Research Council, UK
- Australian Research Council [DP0877432]
- MRC [MR/J013838/1] Funding Source: UKRI
- Medical Research Council [MR/J013838/1] Funding Source: researchfish
We present an adaptive method for the automatic scaling of random-walk Metropolis-Hastings algorithms, which quickly and robustly identifies the scaling factor that yields a specified overall sampler acceptance probability. Our method relies on the use of the Robbins-Monro search process, whose performance is determined by an unknown steplength constant. Based on theoretical considerations we give a simple estimator of this constant for Gaussian proposal distributions. The effectiveness of our method is demonstrated with both simulated and real data examples.
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