|
Activity Number:
|
2
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 3, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #300138 |
|
Title:
|
Calibration and Ensemble Prediction for Multiple Computer Models and Data Sources
|
|
Author(s):
|
Brian Williams*+ and Thomas J. Santner and Max Morris
|
|
Companies:
|
Los Alamos National Laboratory and The Ohio State University and Iowa State University
|
|
Address:
|
Statistical Sciences Group , Los Alamos National Lab, NM, 87545,
|
|
Keywords:
|
Computer experiments ; Gaussian process ; Multivariate ; Bayesian methods ; Model averaging
|
|
Abstract:
|
With the proliferation of mathematical models and numerical methods for their solution, we are often faced with several computer models that can be used to predict the same outcome. While these models have common engineering inputs, their different physics or biological simplifications often lead to model-specific calibration parameters. In addition, some applications allow multiple data sources which allow calibration of subsets of the common parameters in one or more of the computational models. This talk will describe methodology that uses simultaneously the outputs from all available computer models and data sources to calibrate unknown model parameters and to provide ensemble predictions of the code output. An example will be given to illustrate the proposed methods.
|
