Sticky Manifold Piecewise Deterministic Monte Carlo for Bayesian Model Averaging

We design a framework for combining PDMP dynamics in different dimensions to sample from mixed-dimensional targets, in a rejection-free and non-reversible manner, without requiring additional likelihood evaluations for transitions between dimensions. Examples include Bayesian PCA with unknown rank and survival extrapolation over different parametric families.

arXiv

1 Collaboration

It has been a long journey; the beginning was in front of a blackboard at the Isaac Newton Institute. It was my first time visiting the UK, or indeed anywhere outside East Asia. At that time, I was curious about a latent variable model called the item response model, or the ideal point model in political science. The Overleaf project still bears its name, though the final version of the manuscript no longer pertains to it.