Bounding and estimating MCMC convergence rates using common random number simulations
Authors:
- Sabrina SixtaSSSabrina Sixta
- Jeffrey S. RosenthalJRJeffrey S. Rosenthal
Preprint
Abstract
This paper explores how and when to use common random number (CRN) simulation to evaluate MCMC convergence rates. We discuss how CRN simulation is closely related to theoretical convergence rate techniques such as one-shot coupling and coupling from the past. We present conditions under which the CRN technique generates an unbiased estimate of the Wasserstein distance between two random variables. We also discuss how unbiasedness of the Wasserstein distance between two Markov chains over a single iteration does not extend to unbiasedness over multiple iterations. 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 result to a Bayesian regression Gibbs sampler.
Keywords
All Related Versions
This publication has 39 references indexed in Scilit:
- Convergence rate bounds for iterative random functions using one-shot couplingStatistics and Computing, 2022
- Dimension free convergence rates for Gibbs samplers for Bayesian linear mixed modelsStochastic Processes and their Applications, 2022
- Optimal Transport for Applied Mathematicians2015
- Bayesian Methods in Finance2012
- A First Course in Bayesian Statistical Methods2009
- Preface2008
- One-shot coupling for certain stochastic recursive sequencesStochastic Processes and their Applications, 2002
- A simulation approach to convergence rates for Markov chain Monte Carlo algorithmsStatistics and Computing, 1998
- Strong uniform times and finite random walksAdvances in Applied Mathematics, 1987
- Inequalities for E k(X, Y) when the marginals are fixedProbability Theory and Related Fields, 1976