We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Dimension free convergence rates for Gibbs samplers for Bayesian linear mixed models

Abstract: The emergence of big data has led to a growing interest in so-called convergence complexity analysis, which is the study of how the convergence rate of a Monte Carlo Markov chain (for an intractable Bayesian posterior distribution) scales as the underlying data set grows in size. Convergence complexity analysis of practical Monte Carlo Markov chains on continuous state spaces is quite challenging, and there have been very few successful analyses of such chains. One fruitful analysis was recently presented by Qin and Hobert (2021b), who studied a Gibbs sampler for a simple Bayesian random effects model. These authors showed that, under regularity conditions, the geometric convergence rate of this Gibbs sampler converges to zero as the data set grows in size. It is shown herein that similar behavior is exhibited by Gibbs samplers for more general Bayesian models that possess both random effects and traditional continuous covariates, the so-called mixed models. The analysis employs the Wasserstein-based techniques introduced by Qin and Hobert (2021b).
Comments: 48 pages
Subjects: Statistics Theory (math.ST); Computation (stat.CO)
Cite as: arXiv:2103.06324 [math.ST]
  (or arXiv:2103.06324v2 [math.ST] for this version)

Submission history

From: Zhumengmeng Jin [view email]
[v1] Wed, 10 Mar 2021 20:06:26 GMT (33kb)
[v2] Wed, 11 Aug 2021 16:35:11 GMT (39kb)

Link back to: arXiv, form interface, contact.