Paul Damien

Overview

Paul Damien is a Bayesian statistician (University of Michigan, Ann Arbor at time of DWW 1999; later University of Texas at Austin). He co-developed the auxiliary variable Gibbs sampling framework for non-conjugate and hierarchical models with Wakefield and Walker.

Key Contributions / Features

• Damien-Wakefield-Walker (1999): Co-developer of the auxiliary variable Gibbs sampler; Theorem 1 establishes that introducing $U_i \sim \text{Uniform}(0, \ell_i(x))$ converts $f(x) \propto \pi(x)\prod_i \ell_i(x)$ into a Gibbs sampler with uniform conditionals plus one truncated-$\pi$ draw. See Auxiliary Variable Sampler.

Related

• Auxiliary Variable Sampler
• Jon Wakefield
• Stephen Walker
• Gibbs Sampler
• Data Augmentation
• Markov Chain Monte Carlo