Chapter 9

Markov Chain Monte Carlo Methods

9.1 Introduction

Markov Chain Monte Carlo (MCMC) methods encompass a general framework of methods introduced by Metropolis et al. [197] and Hastings [138] for Monte Carlo integration. Recall (see Section 5.2) that Monte Carlo integration estimates the integral

Ag(t)dt

with a sample mean, by restating the integration problem as an expectation with respect to some density function f(·). The integration problem then is reduced to finding a way to generate samples from the target density f(·).

The MCMC approach to sampling from f(·) is to construct a Markov chain with stationary distribution f(·), and run the chain for a sufficiently long time until the chain converges (approximately) to its stationary ...

Get Statistical Computing with R now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.