School Seminars and Colloquia

The Generalized Gibbs Sampler

Statistics Seminar

by Jonathan Keith


Institution: Monash University
Date: Tue 24th July 2012
Time: 1:00 PM
Location: Room 213, Richard Berry Building

Abstract: The Generalised Gibbs Sampler (GGS) is a recently proposed
Markov chain Monte Carlo technique that is trans-dimensional, that is,
it can sample from distributions defined on spaces in which the
dimension varies from point to point or in which points are not easily
defined in terms of co-ordinates. Such spaces arise in problems
involving model selection and model averaging and in a number of
interesting problems in Computational Biology and Ecology. Such
problems have hitherto been the domain of the Reversible-jump Sampler,
but the method described here provides an alternative that is easy to
implement and highly efficient. The GGS provides a very general
framework for Markov chain Monte Carlo simulation. Not only the
conventional Gibbs Sampler, but also a variety of well known and
lesser known samplers emerge as special cases. A new construction,
which I refer to as the Neighbourhood Sampler, also emerges as a
special case. I illustrate the new samplers with a number of examples
drawn from Computational Biology and Ecology.

For More Information: contact Farshid Jamshidi. E-mail: farshid.jamshidi@unimelb.edu.au