|
Activity Number:
|
250
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #308282 |
|
Title:
|
Simultaneous Calibration and Tuning of Computer Experiments
|
|
Author(s):
|
Gang Han*+ and Thomas Santner
|
|
Companies:
|
The Ohio State University and The Ohio State University
|
|
Address:
|
2517 Burlawn Court, Columbus, OH, 43235,
|
|
Keywords:
|
Hierarchical Bayesian Model ; product power exponential correlation ; kriging ; Metropolis Hastings algorithm ; optimization ; root mean squared prediction error
|
|
Abstract:
|
Calibrating complex computer codes to field data and setting tuning parameters in computer outputs are both problems of considerable interests to researchers. We propose a methodology that optimizes the tuning parameters, estimates the distribution of calibration variables, and predicts the untried physical experiment responses. A Bayesian model is constructed based on conditional Gaussian stochastic processes and diffuse prior distributions. The model is implemented by Markov Chain Monte Carlo methodology. The program is illustrated with an example and applied in a biomechanics engineering problem.
|
