On the convergence of Hamiltonian Monte Carlo

Abstract

This paper discusses the irreducibility and geometric ergodicity of the Hamiltonian Monte Carlo (HMC) algorithm. We consider cases where the number of steps of the symplectic integrator is either fixed or random. Under mild conditions on the potential $\F$ associated with target distribution $\pi$, we first show that the Markov kernel associated to the HMC algorithm is irreducible and recurrent. Under more stringent conditions, we then establish that the Markov kernel is Harris recurrent. Finally, we provide verifiable conditions on $\F$ under which the HMC sampler is geometrically ergodic.

References in corpus (5)

• On the Geometric Ergodicity of Hamiltonian Monte Carlo
• Randomized Hamiltonian Monte Carlo
• Learning Generative Models with Visual Attention
• Geometric integrators and the Hamiltonian Monte Carlo method
• Compressible Generalized Hybrid Monte Carlo

Cited by in corpus (23)

• On the Geometric Ergodicity of Hamiltonian Monte Carlo
• Rapid Mixing of Hamiltonian Monte Carlo on Strongly Log-Concave Distributions
• Coupling and Convergence for Hamiltonian Monte Carlo
• Discontinuous Hamiltonian Monte Carlo for discrete parameters and discontinuous likelihoods
• Sampling as optimization in the space of measures: The Langevin dynamics as a composite optimization problem
• Convergence of unadjusted Hamiltonian Monte Carlo for mean-field models
• Fast mixing of Metropolized Hamiltonian Monte Carlo: Benefits of multi-step gradients
• Couplings for Andersen Dynamics
• Diffusion approximations and control variates for MCMC
• A general perspective on the Metropolis-Hastings kernel
• Geometric Methods for Sampling, Optimisation, Inference and Adaptive Agents
• Faster Convergence of Stochastic Gradient Langevin Dynamics for Non-Log-Concave Sampling
• Geometry-Aware Hamiltonian Variational Auto-Encoder
• Nonreversible MCMC from conditional invertible transforms: a complete recipe with convergence guarantees
• Online Sampling from Log-Concave Distributions
• A Unifying and Canonical Description of Measure-Preserving Diffusions
• An Introduction to Hamiltonian Monte Carlo Method for Sampling
• Markov Chain Monte Carlo Methods for Estimating Systemic Risk Allocations
• Sparse Bayesian mass-mapping using trans-dimensional MCMC
• Fast Convergence of Langevin Dynamics on Manifold: Geodesics meet Log-Sobolev
• Generative Particle Variational Inference via Estimation of Functional Gradients
• Entropy-based adaptive Hamiltonian Monte Carlo
• Stochastic Approximation Hamiltonian Monte Carlo