Bayesian nonparametric analysis of reversible Markov chains

Bayesian nonparametric analysis of reversible Markov chains

Neyman Seminar
Jan 27, 2014, 01:00 PM - 02:00 PM | 1011 Evans Hall | Happening As Scheduled
Sergio Bacallado, Stanford University
We discuss the construction of a conjugate prior for reversible Markov chains with a countably infinite number of states through a random walk with reinforcement. The talk with touch on different properties of the prior, including conjugacy, a sufficientness characterization, and an exact algorithm for posterior sampling [1]. The prior finds applications in the analysis of molecular dynamics...