Perfectly random sampling of truncated multinormal distributions

Abstract

The target measure mu is the distribution of a random vector in a box B, a Cartesian product of bounded intervals. The Gibbs sampler is a Markov chain with invariant measure mu. A 'coupling from the past' construction of the Gibbs sampler is used to show ergodicity of the dynarnics and to perfectly simulate mu. An algorithm to sample vectors with multinormal distribution truncated to B is then implemented.