Discontinuous Hamiltonian Monte Carlo for discrete parameters and discontinuous likelihoods

Abstract

Hamiltonian Monte Carlo has emerged as a standard tool for posterior computation. In this article, we present an extension that can efficiently explore target distributions with discontinuous densities. Our extension in particular enables efficient sampling from ordinal parameters though embedding of probability mass functions into continuous spaces. We motivate our approach through a theory of discontinuous Hamiltonian dynamics and develop a corresponding numerical solver. The proposed solver is the first of its kind, with a remarkable ability to exactly preserve the Hamiltonian. We apply our algorithm to challenging posterior inference problems to demonstrate its wide applicability and competitive performance.

15 pages (4 figures) + 18 page (3 figures) supplement. Accepted by Biometrika. Code available at https://github.com/aki-nishimura/discontinuous-hmc

References in corpus (6)

• Piecewise Deterministic Markov Processes for Scalable Monte Carlo on Restricted Domains
• Leave Pima Indians alone: binary regression as a benchmark for Bayesian computation
• Optimizing The Integrator Step Size for Hamiltonian Monte Carlo
• On the convergence of Hamiltonian Monte Carlo
• Probabilistic Path Hamiltonian Monte Carlo
• Hamiltonian Monte Carlo for Probabilistic Programs with Discontinuities

Cited by in corpus (22)

• Bayesian inference of scaled versus fractional Brownian motion
• Conditional independence by typing
• Densities of Almost Surely Terminating Probabilistic Programs are Differentiable Almost Everywhere
• On Probabilistic Termination of Functional Programs with Continuous Distributions
• Oops I Took A Gradient: Scalable Sampling for Discrete Distributions
• Zig-zag sampling for discrete structures and non-reversible phylogenetic MCMC
• Quantum-Inspired Hamiltonian Monte Carlo for Bayesian Sampling
• Sampling algorithms in statistical physics: a guide for statistics and machine learning
• Mixed Hamiltonian Monte Carlo for Mixed Discrete and Continuous Variables
• Stein Variational Inference for Discrete Distributions
• Hamiltonian Monte Carlo for Probabilistic Programs with Discontinuities
• Deterministic Gibbs Sampling via Ordinary Differential Equations
• Supermartingales, Ranking Functions and Probabilistic Lambda Calculus
• Bayesian Inference with the l1-ball Prior: Solving Combinatorial Problems with Exact Zeros
• Scalable Approximate Inference and Some Applications
• Super-Efficient Exact Hamiltonian Monte Carlo for the von Mises Distribution
• Sampling in Combinatorial Spaces with SurVAE Flow Augmented MCMC
• Reliable Categorical Variational Inference with Mixture of Discrete Normalizing Flows
• Zigzag path connects two Monte Carlo samplers: Hamiltonian counterpart to a piecewise deterministic Markov process
• Controlling for sparsity in sparse factor analysis models: adaptive latent feature sharing for piecewise linear dimensionality reduction
• LSB: Local Self-Balancing MCMC in Discrete Spaces
• Direct Sampling with a Step Function