A0534
Title: Ergodicity of some reversible proposal MCMC and its application to Bayesian inference for stochastic processes
Authors: Kengo Kamatani - ISM (Japan) [presenting]
Abstract: Ergodicity of some Markov chain Monte Carlo (MCMC) methods with reversible proposal kernel is studied. This class of MCMC is useful for Bayesian inference problems since we can adjust MCMC algorithm directly to the invariant probability measure of the proposal kernel. Ergodicity theorem provides some useful information to choose appropriate MCMC for Bayesian inference problem. We illustrate the effect by the simulation study for discretely observed stochastic processes.