Nonasymptotic bounds on the mean square error for MCMC estimates via renewal techniques

DOI10.1007/978-3-642-27440-4_31zbMATH Open1271.65007arXiv1101.5837OpenAlexW2113896743WikidataQ96748975 ScholiaQ96748975MaRDI QIDQ5326129FDOQ5326129

Authors: Krzysztof Łatuszyński, Błażej Miasojedow, Wojciech Niemiro 

Publication date: 31 July 2013

Published in: Springer Proceedings in Mathematics & Statistics (Search for Journal in Brave)

Abstract: The Nummellin's split chain construction allows to decompose a Markov chain Monte Carlo (MCMC) trajectory into i.i.d. "excursions". RegenerativeMCMC algorithms based on this technique use a random number of samples. They have been proposed as a promising alternative to usual fixed length simulation [25, 33, 14]. In this note we derive nonasymptotic bounds on the mean square error (MSE) of regenerative MCMC estimates via techniques of renewal theory and sequential statistics. These results are applied to costruct confidence intervals. We then focus on two cases of particular interest: chains satisfying the Doeblin condition and a geometric drift condition. Available explicit nonasymptotic results are compared for different schemes of MCMC simulation.

Full work available at URL: https://arxiv.org/abs/1101.5837

Recommendations

• Nonasymptotic bounds on the estimation error of MCMC algorithms
• On the applicability of regenerative simulation in Markov chain Monte Carlo
• Exponential inequalities for unbounded functions of geometrically ergodic Markov chains: applications to quantitative error bounds for regenerative Metropolis algorithms
• Rigorous confidence bounds for MCMC under a geometric drift condition
• Regenerative Markov chain Monte Carlo for any distribution

zbMATH Keywords

confidence intervalsalgorithmmean square errorrenewal theorynonasymptotic boundssequential statisticsMarkov chain Monte Carlo trajectorysplit chain construction

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Monte Carlo methods (65C05) Numerical analysis or methods applied to Markov chains (65C40) Parametric tolerance and confidence regions (62F25)

Cites Work

• Title not available (Why is that?)
• On the geometric ergodicity of hybrid samplers
• Title not available (Why is that?)
• MC's for MCMC'ists
• Markov chains and stochastic stability
• General state space Markov chains and MCMC algorithms
• Geometric ergodicity of Gibbs and block Gibbs samplers for a hierarchical random effects model
• Geometric ergodicity and hybrid Markov chains
• Sufficient burn-in for Gibbs samplers for a hierarchical random effects model.
• Rates of convergence of the Hastings and Metropolis algorithms
• Gibbs sampling for a Bayesian hierarchical general linear model
• Title not available (Why is that?)
• Minorization Conditions and Convergence Rates for Markov Chain Monte Carlo
• Rigorous confidence bounds for MCMC under a geometric drift condition
• Random generation of combinatorial structures from a uniform distribution
• On the geometric ergodicity of Metropolis-Hastings algorithms
• On variance conditions for Markov chain CLTs
• Renewal theory and computable convergence rates for geometrically erdgodic Markov chains
• Nonasymptotic bounds on the estimation error of MCMC algorithms
• A splitting technique for Harris recurrent Markov chains
• Fixed Precision MCMC Estimation by Median of Products of Averages
• On Monte Carlo methods for Bayesian multivariate regression models with heavy-tailed errors
• Explicit error bounds for lazy reversible Markov chain Monte Carlo
• A mixture representation of \(\pi\) with applications in Markov chain Monte Carlo and perfect sampling.
• A New Approach to the Limit Theory of Recurrent Markov Chains
• Regeneration in Markov Chain Samplers
• On Excess Over the Boundary
• Hoeffding's inequality for uniformly ergodic Markov chains
• On the applicability of regenerative simulation in Markov chain Monte Carlo
• \(V\)-subgeometric ergodicity for a Hastings-Metropolis algorithm
• A regeneration proof of the central limit theorem for uniformly ergodic Markov chains
• How to couple from the past using a read-once source of randomness
• Catalytic perfect simulation

Cited In (5)

• Hit-and-run for numerical integration
• Error bounds of MCMC for functions with unbounded stationary variance
• Orlicz Integrability of Additive Functionals of Harris Ergodic Markov Chains
• Variable transformation to obtain geometric ergodicity in the random-walk Metropolis algorithm
• Nonasymptotic bounds on the estimation error of MCMC algorithms

This page was built for publication: Nonasymptotic bounds on the mean square error for MCMC estimates via renewal techniques

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5326129)