Deterministic Gibbs Sampling via Ordinary Differential Equations
arXiv:2106.10188
Abstract
Deterministic dynamics is an essential part of many MCMC algorithms, e.g. Hybrid Monte Carlo or samplers utilizing normalizing flows. This paper presents a general construction of deterministic measure-preserving dynamics using autonomous ODEs and tools from differential geometry. We show how Hybrid Monte Carlo and other deterministic samplers follow as special cases of our theory. We then demonstrate the utility of our approach by constructing a continuous non-sequential version of Gibbs sampling in terms of an ODE flow and extending it to discrete state spaces. We find that our deterministic samplers are more sample efficient than stochastic counterparts, even if the latter generate independent samples.
References in corpus (5)
- Irreversible Monte Carlo Algorithms for Efficient Sampling
- Discontinuous Hamiltonian Monte Carlo for discrete parameters and discontinuous likelihoods
- Monte Carlo simulation of classical spin models with chaotic billiards
- Driving Markov chain Monte Carlo with a dependent random stream
- How to view an MCMC simulation as a permutation, with applications to parallel simulation and improved importance sampling