Scalable couplings for the random walk Metropolis algorithm

TL;DR

This paper introduces scalable coupling methods for the random walk Metropolis algorithm that improve high-dimensional performance, addressing limitations of existing couplings and providing a framework for asymptotic optimality.

Contribution

It proposes a low-rank modification of synchronous coupling and a mitigation of reflection coupling's shortcomings, advancing coupling design for high-dimensional MCMC.

Findings

The new coupling is provably optimally contractive in high dimensions.

The modified reflection coupling reduces the shortcomings of the original.

Numerical experiments demonstrate improved performance in high-dimensional settings.

Abstract

There has been a recent surge of interest in coupling methods for Markov chain Monte Carlo algorithms: they facilitate convergence quantification and unbiased estimation, while exploiting embarrassingly parallel computing capabilities. Motivated by these, we consider the design and analysis of couplings of the random walk Metropolis algorithm which scale well with the dimension of the target measure. Methodologically, we introduce a low-rank modification of the synchronous coupling that is provably optimally contractive in standard high-dimensional asymptotic regimes. We expose a shortcoming of the reflection coupling, the state of the art at the time of writing, and we propose a modification which mitigates the issue. Our analysis bridges the gap to the optimal scaling literature and builds a framework of asymptotic optimality which may be of independent interest. We illustrate the applicability of our proposed couplings, and the potential for extending our ideas, with various numerical experiments.