Central limit theorem for sequential Monte Carlo methods and its application to Bayesian inference

Abstract

The term ``sequential Monte Carlo methods'' or, equivalently, ``particle filters,'' refers to a general class of iterative algorithms that performs Monte Carlo approximations of a given sequence of distributions of interest (π_t). We establish in this paper a central limit theorem for the Monte Carlo estimates produced by these computational methods. This result holds under minimal assumptions on the distributions π_t, and applies in a general framework which encompasses most of the sequential Monte Carlo methods that have been considered in the literature, including the resample-move algorithm of Gilks and Berzuini [J. R. Stat. Soc. Ser. B Stat. Methodol. 63 (2001) 127-146] and the residual resampling scheme. The corresponding asymptotic variances provide a convenient measurement of the precision of a given particle filter. We study, in particular, in some typical examples of Bayesian applications, whether and at which rate these asymptotic variances diverge in time, in order to assess the long term reliability of the considered algorithm.

Published at http://dx.doi.org/10.1214/009053604000000698 in the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (2)

• Recursive Monte Carlo filters: Algorithms and theoretical analysis
• Sequential monte carlo samplers

Cited by in corpus (104)

• On Particle Methods for Parameter Estimation in State-Space Models
• Recursive Monte Carlo filters: Algorithms and theoretical analysis
• Sequential Monte Carlo smoothing for general state space hidden Markov models
• Comparison of Resampling Schemes for Particle Filtering
• Three discussions of the paper "sequential quasi-Monte Carlo sampling", by M. Gerber and N. Chopin
• Iterated filtering
• Multilevel Sequential Monte Carlo Samplers
• Convergence of adaptive mixtures of importance sampling schemes
• Monte Carlo Confidence Sets for Identified Sets
• Particle Filters and Data Assimilation
• A multi-resolution, non-parametric, Bayesian framework for identification of spatially-varying model parameters
• A general theory of particle filters in hidden Markov models and some applications
• Lookahead Strategies for Sequential Monte Carlo
• Particle Filtering for Large Dimensional State Spaces with Multimodal Observation Likelihoods
• Stability properties of some particle filters
• Generalized fiducial inference for normal linear mixed models
• Concentration inequalities for mean field particle models
• Rao-Blackwellized particle smoothers for conditionally linear Gaussian models
• Sequential Monte Carlo Methods for System Identification
• Sequential Monte Carlo Samplers for capital allocation under copula-dependent risk models
• Controlled Sequential Monte Carlo
• Particle Smoothing for Hidden Diffusion Processes: Adaptive Path Integral Smoother
• Nested Sequential Monte Carlo Methods
• On off-line and on-line Bayesian filtering for uncertainty quantification of structural deterioration
• Consistency of the Adaptive Multiple Importance Sampling
• Independent Resampling Sequential Monte Carlo Algorithms
• Twisted particle filters
• Non-asymptotic Error Bounds for Sequential MCMC and Stability of Feynman-Kac Propagators
• Marginalizing Gaussian Process Hyperparameters using Sequential Monte Carlo
• On the Mathematical Theory of Ensemble (Linear-Gaussian) Kalman-Bucy Filtering
• Dynamic filtering of static dipoles in magnetoencephalography
• Limit theorems for sequential MCMC methods
• On Particle Learning
• On the Forward Filtering Backward Smoothing particle approximations of the smoothing distribution in general state spaces models
• On sequential Monte Carlo, partial rejection control and approximate Bayesian computation
• Stochastic filtering for multiscale stochastic reaction networks based on hybrid approximations
• Sequential Monte Carlo with transformations
• Learning of state-space models with highly informative observations: a tempered Sequential Monte Carlo solution
• Asymptotic optimality of adaptive importance sampling
• Streaming Variational Monte Carlo
• A class of fast exact Bayesian filters in dynamical models with jumps
• Asymptotic genealogies of interacting particle systems with an application to sequential Monte Carlo
• Particle approximation improvement of the joint smoothing distribution with on-the-fly variance estimation
• Nonasymptotic analysis of adaptive and annealed Feynman-Kac particle models
• Infinite-dimensional gradient-based descent for alpha-divergence minimisation
• Central Limit Theorems for Coupled Particle Filters
• Importance Sampling: Intrinsic Dimension and Computational Cost
• Non-asymptotic Error Bounds for Sequential MCMC Methods in Multimodal Settings
• Annealed Important Sampling for Models with Latent Variables
• On the auxiliary particle filter
• Advanced Interacting Sequential Monte Carlo Sampling for Inverse Scattering
• Particle Filters for Partially Observed Diffusions
• Perfect simulation using atomic regeneration with application to Sequential Monte Carlo
• Optimal schedules for annealing algorithms
• Correctness of Sequential Monte Carlo Inference for Probabilistic Programming Languages
• An ergodic theorem for the weighted ensemble method
• On the Convergence of Adaptive Sequential Monte Carlo Methods
• An Ensemble Kushner-Stratonovich (EnKS) Nonlinear Filter: Additive Particle Updates in Non-Iterative and Iterative Forms
• Butterfly resampling: asymptotics for particle filters with constrained interactions
• Variance Estimation in Adaptive Sequential Monte Carlo
• Unifying Sequential Monte Carlo with Resampling Matrices
• Theory of Parallel Particle Filters for Hidden Markov Models
• The divide-and-conquer sequential Monte Carlo algorithm: theoretical properties and limit theorems
• Safe and adaptive importance sampling: a mixture approach
• Optimal input potential functions in the interacting particle system method
• Annealed Flow Transport Monte Carlo
• The Practical Scope of the Central Limit Theorem
• Sequential Quasi-Monte Carlo
• Fully Parallel Particle Learning for GPGPUs and Other Parallel Devices
• Generalized Bayesian Filtering via Sequential Monte Carlo
• Scalable Bayesian reduced-order models for high-dimensional multiscale dynamical systems
• Bayesian Inference for Logistic Regression Models Using Sequential Posterior Simulation
• Quantitative approximations of evolving probability measures and sequential Markov Chain Monte Carlo methods
• Adaptive semiparametric Bayesian differential equations via sequential Monte Carlo
• A Dynamic Bi-orthogonality based Approach for Uncertainty Quantification of Stochastic Systems with Discontinuities
• A Unification of Weighted and Unweighted Particle Filters
• Online inference in Markov modulated nonlinear dynamic systems: a Rao-Blackwellized particle filtering approach
• Nudging the particle filter
• An Adaptive Resample-Move Algorithm for Estimating Normalizing Constants
• Particle Gibbs Sampling for Bayesian Phylogenetic inference
• Approximating Posterior Predictive Distributions by Averaging Output From Many Particle Filters
• Filtering with State-Observation Examples via Kernel Monte Carlo Filter
• A particle filter approach to approximate posterior Cramér-Rao lower bound
• State-Observation Sampling and the Econometrics of Learning Models
• A New Evolutionary Bayesian Approach Incorporating Additive Path Correction for Nonlinear Inverse Problems
• Adaptive Importance Sampling meets Mirror Descent: a Bias-variance tradeoff
• Free energy computations by minimization of Kullback-Leibler divergence: an efficient adaptive biasing potential method for sparse representations
• Uncertainty quantification in complex systems using approximate solvers
• Sequential Monte Carlo Methods for Option Pricing
• Diffusion Monte Carlo method: numerical analysis in a simple case
• An ensemble Kushner-Stratonovich-Poisson filter for recursive estimation in nonlinear dynamical systems
• A central limit type theorem for Gaussian mixture approximations to the nonlinear filtering problem
• Robust Particle Density Tempering for State Space Models
• Noise Robust Online Inference for Linear Dynamic Systems
• Fluctuations, stability and instability of a distributed particle filter with local exchange
• A Stochastically Evolving Non-local Search and Solutions to Inverse Problems with Sparse Data
• Bayesian Conditional Monte Carlo Algorithms for Sequential Single and Multi-Object filtering
• Parallelising Particle Filters with Butterfly Interactions
• Variance reduction for particle filters of systems with time-scale separation
• Adaptive Metropolis-Hastings Sampling using Reversible Dependent Mixture Proposals
• Shape Modeling with Spline Partitions
• A Particle Filter based Multi-Objective Optimization Algorithm: PFOPS
• A critical analysis of resampling strategies for the regularized particle filter
• Numerically stable online estimation of variance in particle filters