Parallelizing particle filters with butterfly interactions

DOI10.1111/SJOS.12408MaRDI QIDQ5118463zbMATH OpenOpenAlexWikidataFDO

Authors Kari Heine, Nick Whiteley, A. Taylan Cemgil

Publication date 8 September 2020

Published in Scandinavian Journal of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1812.01502

zbMATH Keywords

bootstrap particle filter (BPF)hidden Markov modelparallelismparticle filterparticle interactionsequential Monte Carlo

Mathematics Subject Classification ID

Sequential estimation (62L12) Markov processes: estimation; hidden Markov models (62M05) Inference from stochastic processes and prediction (62M20)

Abstract: Bootstrap particle filter (BPF) is the corner stone of many popular algorithms used for solving inference problems involving time series that are observed through noisy measurements in a non-linear and non-Gaussian context. The long term stability of BPF arises from particle interactions which in the context of modern parallel computing systems typically means that particle information needs to be communicated between processing elements, which makes parallel implementation of BPF nontrivial. In this paper we show that it is possible to constrain the interactions in a way which, under some assumptions, enables the reduction of the cost of communicating the particle information while still preserving the consistency and the long term stability of the BPF. Numerical experiments demonstrate that although the imposed constraints introduce additional error, the proposed method shows potential to be the method of choice in certain settings.

Recommendations

• Limit theorems for sequential MCMC methods
• Nested particle filters for online parameter estimation in discrete-time state-space Markov models
• An introduction to sequential Monte Carlo
• Particle-kernel estimation of the filter density in state-space models
• A tutorial on particle filters
• Following a moving target -- Monte Carlo inference for dynamic Bayesian models
• \(\mathrm{SMC}^2\): an efficient algorithm for sequential analysis of state space models
• Analysis of a sequential Monte Carlo method for optimization in dynamical systems
• A uniformly convergent adaptive particle filter

Cited in

(7)

• Multilevel bootstrap particle filter
• Subsampling sequential Monte Carlo for static Bayesian models
• Vectorized and parallel particle filter SMC parameter estimation for stiff ODEs
• scientific article; zbMATH DE number 5925014 (Why is no real title available?)
• scientific article; zbMATH DE number 2044131 (Why is no real title available?)
• On parallel implementation of sequential Monte Carlo methods: the island particle model
• Parallel particle filtering

This page was built for publication: Parallelizing particle filters with butterfly interactions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5118463)