Résumé : This is a short introduction to the many directions of current research in Bayesian computational statistics, from accelerating MCMC algorithms, to using partly deterministic Markov processes like the bouncy particle and the zigzag samplers, to approximating the target or the proposal distributions in such methods. The main illustration focuses on the evaluation of normalising constants and ratios of normalising constants.

Mots-Clés : Bayesian computation; Bayesian statistics; Markov chain Monte Carlo; MCMC simulation; computational methods