Convergence rates for a hierarchical Gibbs sampler

DOI10.3150/15-BEJ758MaRDI QIDQ502894zbMATH OpenFDO

Authors Oliver Jovanovski, Neal Madras

Publication date 11 January 2017

Published in Bernoulli (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1402.4733, https://projecteuclid.org/euclid.bj/1475001368

zbMATH Keywords

convergence ratecouplinggamma distributionhierarchical Gibbs samplerMarkov chainstochastic monotonicity

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Bayesian inference (62F15)

Abstract: We establish some results for the rate of convergence in total variation of a Gibbs sampler to its equilibrium distribution. This sampler is motivated by a hierarchical Bayesian inference construction for a gamma random variable. Our results apply to a wide range of parameter values in the case that the hierarchical depth is 3 or 4, and are more restrictive for depth greater than 4. Our method involves showing a relationship between the total variation of two ordered copies of our chain and the maximum of the ratios of their respective co-ordinates. We construct auxiliary stochastic processes to show that this ratio does converge to 1 at a geometric rate.

Review text

Specific results are given for the rate of convergence of a Markov chain resulting from the application of Gibbs sampling to a hierarchical gamma model. In particular, observations are assumed to come from conditional iid sampling from a gamma distribution with fixed shape parameter and random scale parameter. In turn, the inverse scale parameter is assumed to be sampled from a higher level gamma distribution, and so on 3 or 4 times. The results are rather technical and the proofs rely on repeated applications of coupling and of properties of Markov processes. Explicit formulas for the convergence rates of Gibbs samplers are not common; here, the convergence considered is in the total variation metric.

Recommendations

• Stability of the Gibbs sampler for Bayesian hierarchical models
• scientific article; zbMATH DE number 1191611
• scientific article; zbMATH DE number 597902
• Rates of convergence for Gibbs sampling for variance component models
• Convergence rate of Gibbs sampler and its application

Cited in

(14)

• Rate of convergence of the Gibbs sampler in the Gaussian case
• Sufficient burn-in for Gibbs samplers for a hierarchical random effects model.
• Rates of convergence for Gibbs sampling for variance component models
• Convergence rate of Markov chain methods for genomic motif discovery
• On the convergence complexity of Gibbs samplers for a family of simple Bayesian random effects models
• Convergence rate bounds for iterative random functions using one-shot coupling
• On the convergence rate of the ``out-of-order block Gibbs sampler
• Convergence rates of symmetric scan Gibbs sampler
• Stability of the Gibbs sampler for Bayesian hierarchical models
• scientific article; zbMATH DE number 597902 (Why is no real title available?)
• Convergence rate and concentration inequalities for Gibbs sampling in high dimension
• The Efficiency of Next-Generation Gibbs-Type Samplers: An Illustration Using a Hierarchical Model in Cosmology
• Spectral telescope: convergence rate bounds for random-scan Gibbs samplers based on a hierarchical structure
• Convergence rate of Gibbs sampler and its application

This page was built for publication: Convergence rates for a hierarchical Gibbs sampler

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q502894)