PLACE: 321 Snedecor
SPEAKER:
Krishna Athreya
Iowa State University
TITLE:
Two MCMC issues: Improper prior and Perfect
sampling.
ABSTRACT:
Issue #1
In Bayesian inference often improper priors are used provided
they yield proper
conditionals for the posterior so that the Gibbs sampler could
be used. Is the
inference based on such runs reliable or justified?
Issue # 2.
When is perfect sampling from the target distribution possible?
The talk will
discuss these two issues and their resolution.
COFFEE: 3:45 p.m., 104 Snedecor