Book chapter 1250 views
How rates of convergence for gibbs fields depend on the interaction and the kind of scanning used
Markov Process and Controlled Markov Chains
Swansea University Author: Yuzhi Cai
Abstract
In this paper we describe recent empirical work using perfect simulation toinvestigate how rates of convergence for Gibbs fields might depend onthe interaction between sites andthe kind of scanning used. We also give some experiment results onKendall's (1998) perfect simulation method for area-...
Published in: | Markov Process and Controlled Markov Chains |
---|---|
Published: |
2002
|
URI: | https://cronfa.swan.ac.uk/Record/cronfa15301 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Abstract: |
In this paper we describe recent empirical work using perfect simulation toinvestigate how rates of convergence for Gibbs fields might depend onthe interaction between sites andthe kind of scanning used. We also give some experiment results onKendall's (1998) perfect simulation method for area-interactionprocess, which show that the repulsive case could be quicker orslower than the attractive case for different choices ofthe parameters. |
---|---|
Keywords: |
McMC, Gibbs sampler, Perfect simulation, coalescence time, attractive, repulsive, area-interaction |
College: |
Faculty of Humanities and Social Sciences |
End Page: |
498 |