skip to main content

DOE PAGESDOE PAGES

Title: Efficient calibration for imperfect computer models

Many computer models contain unknown parameters which need to be estimated using physical observations. Furthermore, the calibration method based on Gaussian process models may lead to unreasonable estimate for imperfect computer models. In this work, we extend their study to calibration problems with stochastic physical data. We propose a novel method, called the L 2 calibration, and show its semiparametric efficiency. The conventional method of the ordinary least squares is also studied. Theoretical analysis shows that it is consistent but not efficient. Here, numerical examples show that the proposed method outperforms the existing ones.
Authors:
 [1] ;  [2]
  1. Chinese Academy of Sciences (CAS), Beijing (China)
  2. Georgia Inst. of Technology, Atlanta, GA (United States)
Publication Date:
Report Number(s):
DOE-GT-0010548-4
Journal ID: ISSN 0090-5364; FG02-13ER26159; TRN: US1702896
Grant/Contract Number:
SC0010548
Type:
Accepted Manuscript
Journal Name:
Annals of Statistics
Additional Journal Information:
Journal Volume: 43; Journal Issue: 6; Journal ID: ISSN 0090-5364
Publisher:
Institute of Mathematical Statistics
Research Org:
Georgia Tech Research Corp., Atlanta, GA (United States)
Sponsoring Org:
USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR) (SC-21)
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING
OSTI Identifier:
1405172

Tuo, Rui, and Wu, C. F. Jeff. Efficient calibration for imperfect computer models. United States: N. p., Web. doi:10.1214/15-AOS1314.
Tuo, Rui, & Wu, C. F. Jeff. Efficient calibration for imperfect computer models. United States. doi:10.1214/15-AOS1314.
Tuo, Rui, and Wu, C. F. Jeff. 2015. "Efficient calibration for imperfect computer models". United States. doi:10.1214/15-AOS1314. https://www.osti.gov/servlets/purl/1405172.
@article{osti_1405172,
title = {Efficient calibration for imperfect computer models},
author = {Tuo, Rui and Wu, C. F. Jeff},
abstractNote = {Many computer models contain unknown parameters which need to be estimated using physical observations. Furthermore, the calibration method based on Gaussian process models may lead to unreasonable estimate for imperfect computer models. In this work, we extend their study to calibration problems with stochastic physical data. We propose a novel method, called the L2 calibration, and show its semiparametric efficiency. The conventional method of the ordinary least squares is also studied. Theoretical analysis shows that it is consistent but not efficient. Here, numerical examples show that the proposed method outperforms the existing ones.},
doi = {10.1214/15-AOS1314},
journal = {Annals of Statistics},
number = 6,
volume = 43,
place = {United States},
year = {2015},
month = {12}
}