Combining Multiple Computer Models for Uncertainty Quantification in Posterior Predictive Inference

• Activity Number: 456
• Type: Topic Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Matthew T. Pratola*+
• Companies: Los Alamos National Laboratory
• Keywords: computer experiments ; bayesian model averaging ; climate models ; latent variable ; uncertainty quantification

Abstract

Computer models enable scientists to investigate real-world phenomena in a virtual laboratory using computer experiments. Frequently, multiple computer models are available which make different assumptions in attempting to simulate the physical process of interest. Combining multiple computer models has not been well-addressed in the computer experiments literature. We develop a Bayesian multi-model statistical framework to address this problem. In particular, our approach describes a latent model-space, which may be integrated out to fully incorporate model uncertainty in the posterior inference.