Overview
Walter (Wally) R. Gilks is a British statistician (MRC Biostatistics Unit, Cambridge; later University of Leeds) central to the practical Markov chain Monte Carlo revolution — a developer of adaptive rejection sampling, an architect of the BUGS software, and editor of the standard reference Markov Chain Monte Carlo in Practice.
Key Contributions / Features
- Adaptive rejection sampling (Gilks-Wild 1992): a tuning-free method for sampling log-concave full conditionals, a key enabler of generic Gibbs sampling (Gibbs Sampler).
- Applications of Gibbs sampling in medicine (Gilks et al. 1993): lead author of the survey that made the applied case for MCMC across biostatistics.
- Bayesian mixed models for longitudinal data (Gilks et al. 1993): generalized the Laird-Ware model to multiple non-nested random effects with Gibbs-sampling estimation and censoring (Mixed-Effects Regression).
- Co-architect of BUGS/WinBUGS and editor of Markov Chain Monte Carlo in Practice (Gilks, Richardson, Spiegelhalter 1996).
Related