This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.
Abstract Details
Activity Number:
|
347
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
|
Sponsor:
|
IMS
|
Abstract - #308359 |
Title:
|
Regenerative Simulation for Variable-at-a-Time Metropolis-Hastings Algorithms
|
Author(s):
|
Ronald Charles Neath*+ and Galin L. Jones
|
Companies:
|
Baruch College, CUNY and University of Minnesota
|
Address:
|
One Bernard Baruch Way, New York, NY, 10010,
|
Keywords:
|
Markov chain Monte Carlo ;
Monte Carlo standard error ;
Regenerative simulation ;
Monte Carlo EM algorithm
|
Abstract:
|
Regenerative simulation (RS) is a technique for identifying times at which a Markov chain "probabilistically restarts itself." The paths taken between regeneration times are independent, and this result forms the basis for one popular approach to Markov chain Monte Carlo (MCMC) standard error estimation. Mykland, Tierney, and Yu (1995) showed how to simulate regeneration times for several MCMC algorithms, including the Metropolis-Hastings independence sampler (MHIS). In practice, it is common for MHIS updates to be conducted one variable at a time (e.g., the Monte Carlo EM algorithm of McCulloch, 1997), and thus Mykland et al's formula does not apply. In this talk, we derive regeneration probabilities for the variable-at-a-time independence sampler. We apply our results to a simple problem in Bayesian inference, and compare the RS approach with the method of batch means.
|
The address information is for the authors that have a + after their name.
Authors who are presenting talks have a * after their name.
Back to the full JSM 2010 program
|
2010 JSM Online Program Home
For information, contact jsm@amstat.org or phone (888) 231-3473.
If you have questions about the Continuing Education program, please contact the Education Department.