Between Dec 19, 2024 and Jan 2, 2025, datasets can be submitted to DRUM but will not be processed until after the break. Staff will not be available to answer email during this period, and will not be able to provide DOIs until after Jan 2. If you are in need of a DOI during this period, consider Dryad or OpenICPSR. Submission responses to the UDC may also be delayed during this time.
 

Output Analysis for Markov Chain Monte Carlo

Loading...
Thumbnail Image

Persistent link to this item

Statistics
View Statistics

Journal Title

Journal ISSN

Volume Title

Title

Output Analysis for Markov Chain Monte Carlo

Published Date

2017-02

Publisher

Type

Thesis or Dissertation

Abstract

Markov chain Monte Carlo (MCMC) is a sampling method used to estimate expectations with respect to a target distribution. An important question is when should sampling stop so that we have good estimates of these expectations? The key to answering this question lies in assessing the Monte Carlo error through a multivariate Markov chain central limit theorem (CLT). The multivariate nature of this Monte Carlo error largely has been ignored in the MCMC literature. This dissertation discusses the drawbacks of the current univariate methods of terminating simulation and introduces a multivariate framework for terminating simulation. Theoretical properties of the procedures are established. A multivariate effective sample size is defined and estimated using strongly consistent estimators of the covariance matrix in the Markov chain CLT, a property that is shown for the multivariate batch means estimator and the multivariate spectral variance estimator. A critical aspect of this procedure is that a lower bound on the number of effective samples required for a pre-specified level of precision can be determined a priori. This lower bound depends on the problem only in the dimension of the expectation being estimated, and not on the underlying stochastic process. This result is obtained by drawing a connection between terminating the simulation via effective sample size and terminating it using a relative standard deviation fixed-volume sequential stopping rule. The finite sample properties of the proposed methods are demonstrated in a variety of examples. The proposed method requires the existence of a Markov chain CLT, establishing which requires bounding the rate of convergence of the Markov chains. We establish a geometric rate of convergence for a class of Bayesian penalized regression models. We also present examples showing how it may be easier to establish rates of convergence for linchpin variable samplers than for its competitors.

Description

University of Minnesota Ph.D. dissertation. February 2017. Major: Statistics. Advisor: Galin Jones. 1 computer file (PDF); ix, 200 pages.

Related to

Replaces

License

Collections

Series/Report Number

Funding information

Isbn identifier

Doi identifier

Previously Published Citation

Other identifiers

Suggested citation

Vats, Dootika. (2017). Output Analysis for Markov Chain Monte Carlo. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/185631.

Content distributed via the University Digital Conservancy may be subject to additional license and use restrictions applied by the depositor. By using these files, users agree to the Terms of Use. Materials in the UDC may contain content that is disturbing and/or harmful. For more information, please see our statement on harmful content in digital repositories.