A0155
Title: Particle rolling MCMC with double-block sampling
Authors: Naoki Awaya - Stanford University (United States)
Yasuhiro Omori - University of Tokyo (Japan) [presenting]
Abstract: An efficient particle Markov chain Monte Carlo methodology is proposed for the rolling-window estimation of state space models. The particles are updated to approximate the long sequence of posterior distributions as the estimation window is moved. To overcome the well-known weight degeneracy problem that causes the poor approximation, a practical double-block sampler with the conditional sequential Monte Carlo update is introduced, where one lineage from multiple candidates is chosen for the set of current state variables. The proposed sampler is justified in the augmented space through theoretical discussions. In the illustrative examples, it is shown to be successful in accurately estimating the posterior distributions of the model parameters.
