Spatiotemporal blocking of the bouncy particle sampler for efficient inference in state space models

01/08/2021
by   Jacob Vorstrup Goldman, et al.
0

We propose a novel blocked version of the continuous-time bouncy particle sampler of Bouchard- Côté et al. [2018] applicable to any differentiable probability density. Motivated by Singh et al. [2017], we also introduce an alternative implementation that leads to significant improvement in terms of effective sample size per second, and furthermore allows for parallelization at the cost of an extra logarithmic factor. The new algorithms are particularly efficient for latent state inference in high-dimensional state space models, where blocking in both space and time is necessary to avoid degeneracy of the proposal kernel. The efficiency of the blocked bouncy particle sampler, in comparison with both the standard implementation of the bouncy particle sampler and the particle Gibbs algorithm of Andrieu et al. [2010], is illustrated numerically for both simulated data and a challenging real-world financial dataset.

READ FULL TEXT

page 14

page 17

research
02/16/2015

Particle Gibbs for Bayesian Additive Regression Trees

Additive regression trees are flexible non-parametric models and popular...
research
07/31/2020

A Note on Particle Gibbs Method and its Extensions and Variants

High-dimensional state trajectories of state-space models pose challenge...
research
10/13/2022

Marginalized particle Gibbs for multiple state-space models coupled through shared parameters

We consider Bayesian inference from multiple time series described by a ...
research
02/04/2014

Particle Metropolis adjusted Langevin algorithms for state space models

Particle MCMC is a class of algorithms that can be used to analyse state...
research
09/22/2020

The computational cost of blocking for sampling discretely observed diffusions

Many approaches for conducting Bayesian inference on discretely observed...
research
09/29/2021

An efficient pseudo marginal method for state space models

Pseudo Marginal Metropolis-Hastings (PMMH) is a general approach to carr...

Please sign up or login with your details

Forgot password? Click here to reset