Communications in Applied Mathematics and Computational Science Vol. 5, No. 1, 2010 (original) (raw)

Abstract
We propose a family of Markov chain Monte Carlo methods whose performance is unaffected by affine tranformations of space. These algorithms are easy to construct and require little or no additional computational overhead. They should be particularly useful for sampling badly scaled distributions. Computational tests show that the affine invariant methods can be significantly faster than standard MCMC methods on highly skewed distributions.

Abstract

We propose a family of Markov chain Monte Carlo methods whose performance is unaffected by affine tranformations of space. These algorithms are easy to construct and require little or no additional computational overhead. They should be particularly useful for sampling badly scaled distributions. Computational tests show that the affine invariant methods can be significantly faster than standard MCMC methods on highly skewed distributions.

Keywords

Markov chain Monte Carlo, affine invariance, ensemble samplers

Mathematical Subject Classification 2000

Primary: 65C05

Milestones

Received: 6 November 2009

Accepted: 29 November 2009

Published: 31 January 2010

Authors

Courant Institute New York University 251 Mercer St. New York, NY 10012 United States
Jonathan Weare
Courant Institute New York University 251 Mercer St. New York, NY 10012 United States