Delayed rejection Hamiltonian Monte Carlo for sampling multiscale distributions

Abstract

The efficiency of Hamiltonian Monte Carlo (HMC) can suffer when sampling a distribution with a wide range of length scales, because the small step sizes needed for stability in high-curvature regions are inefficient elsewhere. To address this we present a delayed rejection variant: if an initial HMC trajectory is rejected, we make one or more subsequent proposals each using a step size geometrically smaller than the last. We extend the standard delayed rejection framework by allowing the probability of a retry to depend on the probability of accepting the previous proposal. We test the scheme in several sampling tasks, including multiscale model distributions such as Neal's funnel, and statistical applications. Delayed rejection enables up to five-fold performance gains over optimally-tuned HMC, as measured by effective sample size per gradient evaluation. Even for simpler distributions, delayed rejection provides increased robustness to step size misspecification. Along the way, we provide an accessible but rigorous review of detailed balance for HMC.

30 pages, 10 figures

References in corpus (14)

• The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo
• A Conceptual Introduction to Hamiltonian Monte Carlo
• A General Framework for the Parametrization of Hierarchical Models
• Piecewise-Deterministic Markov Chain Monte Carlo
• A General Metric for Riemannian Manifold Hamiltonian Monte Carlo
• Hamiltonian Monte Carlo Without Detailed Balance
• Extra Chance Generalized Hybrid Monte Carlo
• Generalizing Hamiltonian Monte Carlo with Neural Networks
• Faster Hamiltonian Monte Carlo by Learning Leapfrog Scale: an offline randomized solution
• A general perspective on the Metropolis-Hastings kernel
• Manifold lifting: scaling MCMC to the vanishing noise regime
• Evaluating the Implicit Midpoint Integrator for Riemannian Manifold Hamiltonian Monte Carlo
• Implicit Hamiltonian Monte Carlo for Sampling Multiscale Distributions
• Ensemble preconditioning for Markov chain Monte Carlo simulation

Cited by in corpus (2)

• A Differentiable, End-to-End Forward Model for 21 cm Cosmology: Estimating the Foreground, Instrument, and Signal Joint Posterior
• Incorporating Local Step-Size Adaptivity into the No-U-Turn Sampler using Gibbs Self Tuning