Galilean and Hamiltonian Monte Carlo
Author:
Open Access
- 5 December 2019
- conference paper
- Published by MDPI AG in Proceedings
- Vol. 33 (1) , 19
Abstract
Galilean Monte Carlo (GMC) allows exploration in a big space along systematic trajectories, thus evading the square-root inefficiency of independent steps. Galilean Monte Carlo has greater generality and power than its historical precursor Hamiltonian Monte Carlo because it discards second-order propagation under forces in favour of elementary force-free motion. Nested sampling (for which GMC was originally designed) has similar dominance over simulated annealing, which loses power by imposing an unnecessary thermal blurring over energy.Keywords
This publication has 4 references indexed in Scilit:
- Bayesian computation in big spaces-nested sampling and Galilean Monte CarloAIP Conference Proceedings, 2012
- Nested sampling for general Bayesian computationBayesian Analysis, 2006
- Hybrid Monte CarloPhysics Letters B, 1987
- Optimization by Simulated AnnealingScience, 1983