next up previous
Next: The Metropolis-Hastings algorithm: principle Up: An introduction to Markov Previous: Monte Carlo integration

Sampling algorithms

  The reference sampling algorithm in Bayesian statistics is the Metropolis-Hastings algorithm. The principle is to use an auxiliary transition kernel and to split the sampling into two steps: a proposal step and an acceptance step, so as to guarantee the reversibility condition.





Bob Fisher
Fri Jul 26 09:56:32 BST 2002