Solving the Poisson equation using coupled Markov chains

This article shows how coupled Markov chains that meet exactly after a random number of iterations can be used to generate unbiased estimators of the solutions of the Poisson equation. We establish connections to recently-proposed unbiased estimators of Markov chain equilibrium expectations. We further employ the proposed estimators of solutions of the Poisson equation to construct unbiased estimators of the asymptotic variance of Markov chain ergodic averages, involving a random but finite computing time. We formally study the proposed methods under realistic assumptions on the meeting times of the coupled chains and on the existence of moments of test functions under the target distribution. We describe experiments in toy examples such as the autoregressive model, and in more challenging settings.
View on arXiv