Fast Markov Chain Monte Carlo Algorithms via Lie Groups
arXiv:1901.08606
Abstract
From basic considerations of the Lie group that preserves a target probability measure, we derive the Barker, Metropolis, and ensemble Markov chain Monte Carlo (MCMC) algorithms, as well as variants of waste-recycling Metropolis-Hastings and an altogether new MCMC algorithm. We illustrate these constructions with explicit numerical computations, and we empirically demonstrate on a spin glass that the new algorithm converges more quickly than its siblings.
arXiv commentAccepted to AISTATS 2020; proofs included here as an appendix (but relegated to supplementary info in conference version)
References in corpus (12)
- Group Equivariant Convolutional Networks
- Markov Chain Monte Carlo Method without Detailed Balance
- Spherical CNNs
- Efficient computation of the Zassenhaus formula
- A Review of Multiple Try MCMC algorithms for Signal Processing
- The Sherrington-Kirkpatrick model: an overview
- Optimal Monte Carlo Updating
- Automorphism Groups of Graphical Models and Lifted Variational Inference
- MCMC Using Ensembles of States for Problems with Fast and Slow Variables such as Gaussian Process Regression
- Optimization of Markov process violates detailed balance condition
- Contextual Symmetries in Probabilistic Graphical Models
- On the stochastic Lie algebra