Multilevel particle filters: normalizing constant estimation

DOI10.1007/S11222-016-9715-5MaRDI QIDQ1702280zbMATH OpenOpenAlexFDO

Authors Ajay Jasra, Kengo Kamatani, Prince Peprah Osei, Yan Zhou

Publication date 28 February 2018

Published in Statistics and Computing (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1605.04963

zbMATH Keywords

particle filterfilteringmultilevel Monte Carlodiffusions

Mathematics Subject Classification ID

Monte Carlo methods (65C05) Signal detection and filtering (aspects of stochastic processes) (60G35) Interacting random processes; statistical mechanics type models; percolation theory (60K35)

Abstract: In this article we introduce two new estimates of the normalizing constant (or marginal likelihood) for partially observed diffusion (POD) processes, with discrete observations. One estimate is biased but non-negative and the other is unbiased but not almost surely non-negative. Our method uses the multilevel particle filter of Jasra et al (2015). We show that, under assumptions, for Euler discretized PODs and a given $varepsilon>0$. in order to obtain a mean square error (MSE) of $mathcalO(varepsilo{n}^2)$ one requires a work of $mathcalO(varepsilo{n}^{-2.5})$ for our new estimates versus a standard particle filter that requires a work of $mathcalO(varepsilo{n}^{-3})$. Our theoretical results are supported by numerical simulations.

Recommendations

• Multilevel Particle Filters
• An algorithm for approximating the second moment of the normalizing constant estimate from a particle filter
• Multilevel particle filters for the non-linear filtering problem in continuous time
• Unbiased filtering of a class of partially observed diffusions
• Multilevel estimation of normalization constants using ensemble Kalman-Bucy filters

Cites work

• scientific article; zbMATH DE number 5919872 (Why is no real title available?)
• scientific article; zbMATH DE number 54145 (Why is no real title available?)
• scientific article; zbMATH DE number 2106098 (Why is no real title available?)
• A nonasymptotic theorem for unnormalized Feynman-Kac particle models
• Inference in hidden Markov models.
• Mean field simulation for Monte Carlo integration
• Multilevel Monte Carlo Path Simulation
• Multilevel Particle Filters
• Multilevel sequential Monte Carlo samplers
• On coupling particle filter trajectories
• On particle Gibbs sampling
• Particle Filters for Partially Observed Diffusions
• The Monte-Carlo method for filtering with discrete-time observations

Cited in

(18)

• Bayesian parameter inference for partially observed stochastic differential equations driven by fractional Brownian motion
• A Wasserstein coupled particle filter for multilevel estimation
• Normalizing constants of log-concave densities
• Multilevel particle filters for a class of partially observed piecewise deterministic Markov processes
• Calculating Bayesian model evidence for porous-media flow using a multilevel estimator
• Multilevel particle filters for Lévy-driven stochastic differential equations
• Multilevel estimation of normalization constants using ensemble Kalman-Bucy filters
• Multilevel particle filters for the non-linear filtering problem in continuous time
• Central limit theorems for coupled particle filters
• Unbiased estimation of the solution to Zakai's equation
• Unbiased and multilevel methods for a class of diffusions partially observed via marked point processes
• Bayesian static parameter estimation for partially observed diffusions via multilevel Monte Carlo
• Multilevel Particle Filters
• On coupling particle filter trajectories
• Advanced Multilevel Monte Carlo Methods
• An algorithm for approximating the second moment of the normalizing constant estimate from a particle filter
• Multilevel Monte Carlo for smoothing via transport methods
• Constructing unbiased gradient estimators with finite variance for conditional stochastic optimization

This page was built for publication: Multilevel particle filters: normalizing constant estimation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1702280)