Delayed Acceptance Elliptical Slice Sampling

ABSTRACT Slice sampling is a well‐established Markov chain Monte Carlo method for approximate sampling of target distributions which are only known up to a normalizing constant. The method is based on choosing a new state on a slice, i.e., a (super‐)level set of the target density function. However,...

Full description

Saved in:
Bibliographic Details
Published inProceedings in applied mathematics and mechanics Vol. 25; no. 1
Main Authors Bitterlich, Kevin, Rudolf, Daniel, Sprungk, Björn
Format Journal Article
LanguageEnglish
Published 01.03.2025
Online AccessGet full text

Cover

More Information
Summary:ABSTRACT Slice sampling is a well‐established Markov chain Monte Carlo method for approximate sampling of target distributions which are only known up to a normalizing constant. The method is based on choosing a new state on a slice, i.e., a (super‐)level set of the target density function. However, this process may require several evaluations of the target density and, thus, become computationally demanding, particularly, for Bayesian inference with costly likelihoods appearing, e.g., in the Bayesian approach to inverse problems. In this paper we exploit cheap (deterministic) approximations of the density and propose the delayed acceptance version of the elliptical slice sampler. Numerical experiments illustrate the potential benefits in cost reduction of the novel approach.
ISSN:1617-7061
1617-7061
DOI:10.1002/pamm.70005