Activity Number:
|
303
|
Type:
|
Topic Contributed
|
Date/Time:
|
Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Computing*
|
Abstract - #300670 |
Title:
|
Multi-resolution Genetic Algorithms and Markov Chain Monte Carlo
|
Author(s):
|
Christopher Holloman*+ and Herbert Lee and David Higdon
|
Affiliation(s):
|
Duke University and Duke University and Los Alamos National Laboratory
|
Address:
|
Box 90251, Durham, North Carolina, 27708-0251, USA
|
Keywords:
|
Bayesian statistics ; parallel processing ; simulated tempering
|
Abstract:
|
Due to modern advances in computing power, the use of increasingly complex models has become practical. One class of large models that often relies on numerical techniques for parameter estimation is multi-resolution models. Unfortunately, numerical maximization and sampling techniques used to estimate parameters in such complex models often explore the parameter space slowly, resulting in unreliable or unstable estimates. We propose a multi-resolution genetic algorithm that incorporates elements of simulated tempering to allow efficient estimation of parameters in multi-scale models. This algorithm can also be adapted to perform Markov chain Monte Carlo sampling from a posterior distribution in a Bayesian setting, which can greatly improve mixing and exploration of the posterior. Parallel implementation is addressed. These methods are demonstrated on an example from groundwater hydrology.
|