Comment extracting more diagnostic information from a single run using the cusum path plot.pdf

Comment: Extracting more diagnostic informationfrom a single run using Cusum Path PlotBin YuThe article by Besag, Green, Higdon, and Mengersen adds to a series of recent papers (Be-sag and Green (1993), Geyer and Thompson (1993), and Gelman and Rubin (1992b)) in makingMarkov chain Monte Carlo (MCMC) methods accessible to more statisticians, especially appliedstatisticians. I am glad to see that dierent algorithms are reviewed in a unied way and manyexamples are given. Although the article gives general recommendations as to which algorithmsand sampling scans to choose, there is not much discussion on the empirical monitoring of conver-gence of the Markov chains. Since the convergence issue is very critical to the success of MCMCmethods, and something close to my heart, I will make this issue my topic here. In particular,using the prostate cancer example in the Besag et al paper and the Ising model example in Gelmanand Rubin (1992a), I illustrate that the cusum path plot in Yu and Mykland (1994) can eectivelybring out the local mixing property of the Markov chain.It had been believed by many MCMC researchers (including this author) that information solelyfrom a single run of a Markov chain can be misleading since, for example, it can get trapped at alocal mode of the target density. Consequently, additional information beyond that from a singlerun has been introduced to the convergence diagnostics. Gelman and Rubin (1992b) proposed amultiple chain approach in the MCMC context, followed by Liu, Liu, and Rubin (1992) and Roberts(1992). Yu (1994) introduced additional information to a single run by taking advantage of theunnormalized target density. In the context of Gibbs samplers, Ritter and Tanner (1992) and Cui,Bin Yu is Assistant Professor, Department of Statistics, University of California, Berkeley, CA 94720-3860.Research supported in part by Grant DMS-9322817 from the National Science Foundation and Grant DAAH04-94-G-0232 from the Army Research Oce. 1 Tann