Variable transformation to obtain geometric ergodicity in the random-walk Metropolis algorithm

Abstract

A random-walk Metropolis sampler is geometrically ergodic if its equilibrium density is super-exponentially light and satisfies a curvature condition [Stochastic Process. Appl. 85 (2000) 341-361]. Many applications, including Bayesian analysis with conjugate priors of logistic and Poisson regression and of log-linear models for categorical data result in posterior distributions that are not super-exponentially light. We show how to apply the change-of-variable formula for diffeomorphisms to obtain new densities that do satisfy the conditions for geometric ergodicity. Sampling the new variable and mapping the results back to the old gives a geometrically ergodic sampler for the original variable. This method of obtaining geometric ergodicity has very wide applicability.

Published in at http://dx.doi.org/10.1214/12-AOS1048 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org). With Corrections

References in corpus (7)

• MCMC using Hamiltonian dynamics
• General state space Markov chains and MCMC algorithms
• A General Framework for the Parametrization of Hierarchical Models
• Batch means and spectral variance estimators in Markov chain Monte Carlo
• Sufficient burn-in for Gibbs samplers for a hierarchical random effects model
• Nonasymptotic bounds on the estimation error of MCMC algorithms
• Rigorous confidence bounds for MCMC under a geometric drift condition

Cited by in corpus (9)

• Ergodicity of the zigzag process
• Component-Wise Markov Chain Monte Carlo: Uniform and Geometric Ergodicity under Mixing and Composition
• Stacking for Non-mixing Bayesian Computations: The Curse and Blessing of Multimodal Posteriors
• Neuronized Priors for Bayesian Sparse Linear Regression
• Strong Consistency of Multivariate Spectral Variance Estimators
• Deterministic Bayesian Information Fusion and the Analysis of its Performance
• Bayesian inverse problems with $l_1$ priors: a Randomize-then-Optimize approach
• IID Sampling from Intractable Multimodal and Variable-Dimensional Distributions
• Entropy-based adaptive Hamiltonian Monte Carlo