Multilevel Sequential Monte Carlo Samplers for Normalizing Constants

Abstract : This article considers the Sequential Monte Carlo (SMC) approximation of ratios of normalizing constants associated to posterior distributions which in principle rely on continuum models. Therefore, the Monte Carlo estimation error and the discrete approximation error must be balanced. A multilevel strategy is utilized to substantially reduce the cost to obtain a given error level in the approximation as compared to standard estimators. Two estimators are considered and relative variance bounds are given. The theoretical results are numerically illustrated for two Bayesian inverse problems arising from elliptic Partial Differential Equations (PDEs). The examples involve the inversion of observations of the solution of (i) a one-dimensional Poisson equation to infer the diffusion coefficient, and (ii) a two-dimensional Poisson equation to infer the external forcing.
Type de document :
Article dans une revue
ACM Transactions on Modeling and Computer Simulation, Association for Computing Machinery, 2017, 27 (3), pp.1 - 22. 〈10.1145/3092841〉
Liste complète des métadonnées

https://hal.inria.fr/hal-01669115
Contributeur : Pierre Del Moral <>
Soumis le : mercredi 20 décembre 2017 - 15:36:01
Dernière modification le : jeudi 11 janvier 2018 - 06:22:12

Identifiants

Collections

Citation

Pierre Del Moral, Ajay Jasra, Kody J. H. Law, Yan Zhou. Multilevel Sequential Monte Carlo Samplers for Normalizing Constants. ACM Transactions on Modeling and Computer Simulation, Association for Computing Machinery, 2017, 27 (3), pp.1 - 22. 〈10.1145/3092841〉. 〈hal-01669115〉

Partager

Métriques

Consultations de la notice

173