Construction of weakly CUD sequences for MCMC sampling

TL;DR

This paper develops methods to construct and verify weakly CUD sequences for MCMC, demonstrating their effectiveness in reducing variance in high-dimensional Bayesian sampling.

Contribution

It introduces general techniques for proving sequences are WCUD, identifies specific WCUD sequences, and shows how operations on WCUD sequences preserve their properties.

Findings

WCUD sequences can reduce variance in high-dimensional MCMC

New methods for verifying WCUD sequences are proposed

Numerical experiments show significant variance reduction

Abstract

In Markov chain Monte Carlo (MCMC) sampling considerable thought goes into constructing random transitions. But those transitions are almost always driven by a simulated IID sequence. Recently it has been shown that replacing an IID sequence by a weakly completely uniformly distributed (WCUD) sequence leads to consistent estimation in finite state spaces. Unfortunately, few WCUD sequences are known. This paper gives general methods for proving that a sequence is WCUD, shows that some specific sequences are WCUD, and shows that certain operations on WCUD sequences yield new WCUD sequences. A numerical example on a 42 dimensional continuous Gibbs sampler found that some WCUD inputs sequences produced variance reductions ranging from tens to hundreds for posterior means of the parameters, compared to IID inputs.