90007545

open access

Version of Record

Replica Exchange Particle-Gibbs Method with Ancestor Sampling

English (英語)

Journal of the Physical Society of Japan

89(10)

104801-104801

The Physical Society of Japan

2020-10-15

2020-10-21

Particle-Gibbs (PG) method is a Markov chain Monte Carlo (MCMC) method for sampling from the joint posterior distribution of latent variables and parameters in a nonlinear state space model, which is a probabilistic time-series model. Recently, particle-Gibbs with ancestor sampling (PGAS) was proposed as a method with better sampling efficiency of PG. However, even with PG and PGAS, it is often difficult to realize sampling from the joint posterior distribution with a finite number of samples. In this paper, we propose replica exchange particle-Gibbs with ancestor sampling (REPGAS) as a method to overcome this problem by combining PGAS with the replica exchange method. We also demonstrate the effectiveness of the proposed method by using simulated data obtained from a benchmark nonlinear state space model and the Izhikevich model, which is a computational model of the membrane potential of a neuron.

学術雑誌論文

This article is published by the Physical Society of Japan under the terms of the Creative Commons Attribution 4.0 License. Any further distribution of this work must maintain attribution to the author(s) and the title of the article, journal citation, and DOI.