Understanding the Hastings algorithm

DOI10.1080/03610918.2013.777455MaRDI QIDQ5252830zbMATH OpenOpenAlexFDO

Authors David D. L. Minh, Do Le Minh (Paul)

Publication date 3 June 2015

Published in Communications in Statistics. Simulation and Computation (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1408.4438

zbMATH Keywords

Hastings algorithmMarkov chain Monte Carlo methodMetropolis-Hastings algorithmsimulation

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Monte Carlo methods (65C05) Numerical analysis or methods applied to Markov chains (65C40)

Abstract: The Hastings algorithm is a key tool in computational science. While mathematically justified by detailed balance, it can be conceptually difficult to grasp. Here, we present two complementary and intuitive ways to derive and understand the algorithm. In our framework, it is straightforward to see that the celebrated Metropolis-Hastings algorithm has the highest acceptance probability of all Hastings algorithms.

Recommendations

• The Hastings algorithm at fifty
• Hastings-Metropolis algorithms and reference measures
• Sampling unnormalized probabilities: an alternative to the Metropolis-Hastings algorithm
• Un algorithme de Hastings-Metropolis avec apprentissage séquentiel
• What do we know about the Metropolis algorithm?

Cites work

• A geometric interpretation of the Metropolis-Hastings algorithm.
• Markov chains for exploring posterior distributions. (With discussion)
• Monte Carlo sampling methods using Markov chains and their applications
• Monte Carlo strategies in scientific computing
• Optimum Monte-Carlo sampling using Markov chains
• Regenerative Markov chain Monte Carlo for any distribution

Cited in

(9)

• A method of generating random vectors with a given probability density function
• Layer sampling
• Regenerative Markov chain importance sampling
• Bayesian networks: regenerative Gibbs samplings
• Learning an efficient constructive sampler for graphs
• Solving inverse problems using data-driven models
• Fast updating multipole coulombic potential calculation
• Bayesian networks: generating independent samples
• Handling polynomial and transcendental functions in SMT via unconstrained optimisation and topological degree test

This page was built for publication: Understanding the Hastings algorithm

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5252830)