Adaptation of the Independent Metropolis-Hastings Sampler with Normalizing Flow Proposals

TL;DR

This paper introduces an adaptive independent Metropolis-Hastings sampler using normalizing flows as proposals, extending convergence theory and demonstrating improved performance on synthetic and physical system examples.

Contribution

It extends adaptive MCMC convergence theory to normalizing flow proposals and develops an update scheme using stochastic gradient descent.

Findings

Improved sampling efficiency with normalizing flow proposals.

Successful application to synthetic and physical systems.

Theoretical guarantees for convergence of the adaptive method.

Abstract

Markov Chain Monte Carlo (MCMC) methods are a powerful tool for computation with complex probability distributions. However the performance of such methods is critically dependant on properly tuned parameters, most of which are difficult if not impossible to know a priori for a given target distribution. Adaptive MCMC methods aim to address this by allowing the parameters to be updated during sampling based on previous samples from the chain at the expense of requiring a new theoretical analysis to ensure convergence. In this work we extend the convergence theory of adaptive MCMC methods to a new class of methods built on a powerful class of parametric density estimators known as normalizing flows. In particular, we consider an independent Metropolis-Hastings sampler where the proposal distribution is represented by a normalizing flow whose parameters are updated using stochastic gradient descent. We explore the practical performance of this procedure on both synthetic settings and in the analysis of a physical field system and compare it against both adaptive and non-adaptive MCMC methods.