Estimating MCMC convergence rates using common random number simulation

Abstract

This paper presents how to use common random number (CRN) simulation to evaluate Markov chain Monte Carlo (MCMC) convergence to stationarity. We provide an upper bound on the Wasserstein distance of a Markov chain to its stationary distribution after $N$ steps in terms of averages over CRN simulations. We apply our bound to Gibbs samplers on a model related to James-Stein estimators, a variance component model, and a Bayesian linear regression model. Using our examples, we show that the CRN-based simulation combined with a coalescing condition to generate a total variation bound converges to zero much more faster than the available drift and minorization bounds, while also converging at the same rate as the one-shot coupling bound.

References in corpus (9)

• General state space Markov chains and MCMC algorithms
• Sufficient burn-in for Gibbs samplers for a hierarchical random effects model
• Coupling and Convergence for Hamiltonian Monte Carlo
• Quantitative bounds for Markov chain convergence: Wasserstein and total variation distances
• Convergence analysis of the Gibbs sampler for Bayesian general linear mixed models with improper priors
• Mixing time of the adjacent walk on the simplex
• Convergence analysis of block Gibbs samplers for Bayesian linear mixed models with $p>N$
• Adapted Wasserstein distance between the laws of SDEs
• Coupling-based Convergence Diagnostic and Stepsize Scheme for Stochastic Gradient Descent