Extra Chance Generalized Hybrid Monte Carlo
arXiv:1407.8107 · doi:10.1016/j.jcp.2014.09.037
Abstract
We study a method, Extra Chance Generalized Hybrid Monte Carlo, to avoid rejections in the Hybrid Monte Carlo method and related algorithms. In the spirit of delayed rejection, whenever a rejection would occur, extra work is done to find a fresh proposal that, hopefully, may be accepted. We present experiments that clearly indicate that the additional work per sample carried out in the extra chance approach clearly pays in terms of the quality of the samples generated.
arXiv comment13 pages, 8 figures (including 1 diagram, 3 pseudocodes), 1 table, updated (corrections, typos and new appendix)
References in corpus (6)
- MCMC using Hamiltonian dynamics
- Numerical integrators for the Hybrid Monte Carlo method
- Cost of the Generalised Hybrid Monte Carlo Algorithm for Free Field Theory
- Compressible Generalized Hybrid Monte Carlo
- Reducing the effect of Metropolization on mixing times in molecular dynamics simulations
- Hamiltonian Monte Carlo with Reduced Momentum Flips
Cited by in corpus (16)
- On the Geometric Ergodicity of Hamiltonian Monte Carlo
- Geometric integrators and the Hamiltonian Monte Carlo method
- Modified Hamiltonian Monte Carlo for Bayesian inference
- Hamiltonian Monte Carlo Without Detailed Balance
- Identifying the Optimal Integration Time in Hamiltonian Monte Carlo
- Convergence of unadjusted Hamiltonian Monte Carlo for mean-field models
- Palindromic 3-stage splitting integrators, a roadmap
- A general perspective on the Metropolis-Hastings kernel
- Testing algorithms for critical slowing down
- Geometric Methods for Sampling, Optimisation, Inference and Adaptive Agents
- Peskun-Tierney ordering for Markov chain and process Monte Carlo: beyond the reversible scenario
- Sampling algorithms in statistical physics: a guide for statistics and machine learning
- HMC: avoiding rejections by not using leapfrog and some results on the acceptance rate
- Delayed rejection Hamiltonian Monte Carlo for sampling multiscale distributions
- A Markov Jump Process for More Efficient Hamiltonian Monte Carlo
- Improving the convergence of Markov chains via permutations and projections