SED navigation bar go to SED home page go to SED seminars page go to NIST home page SED Home Page SED Contacts SED Projects SED Products and Publications Search SED Pages

Statistical Engineering Division Seminar

Using Sequential Importance Sampling to Speed up MCMC

Dr. Isabel Beichl
Mathematical and Computational Sciences Division
Information Technology Laboratory
Wednesday, May 21, 2008, 10:30-11:30 AM
Building 222, Room A330

Abstract

The Monte Carlo Markov Chain Method (MCMC) has been written about extensively and there is an extensive theory developed. But getting answers to problems is still difficult because MCMC can be slow. We present two methods for speeding up MCMC for the monomer-dimer problem based on the method of sequential importance sampling (SIS). Our method computes optimal fugacities. The other computes the mixing rate efficiently.

NIST Contact: Dr. Charles Hagwood, (301) 975-2846.

Date created: 5/13/2008
Last updated: 5/13/2008
Please email comments on this WWW page to sedwww@cam.nist.gov.