Gaussian process emulation of dynamic computer codes

被引：133

作者：

Conti, S. ^{[1
]}

Gosling, J. P. ^{[2
]}

Oakley, J. E. ^{[3
]}

O'Hagan, A. ^{[3
]}

机构：

[1] Hlth Protect Agcy, Ctr Infect, London NW9 5EQ, England

[2] Dept Environm Food & Rural Affairs, Cent Sci Lab, Sand Hutton YO41 1LZ, Yorks, England

[3] Univ Sheffield, Dept Probabil & Stat, Sheffield S3 7RH, S Yorkshire, England

来源：

BIOMETRIKA | 2009年 / 96卷 / 03期

基金：

英国工程与自然科学研究理事会;

关键词：

Bayesian inference; Computer experiment; Dynamic simulator; Emulation; Gaussian process; Iterative modelling; UNCERTAINTY; MODELS; ERROR;

D O I：

10.1093/biomet/asp028

中图分类号：

Q [生物科学];

学科分类号：

07 ; 0710 ; 09 ;

摘要：

Computer codes are used in scientific research to study and predict the behaviour of complex systems. Their run times often make uncertainty and sensitivity analyses impractical because of the thousands of runs that are conventionally required, so efficient techniques have been developed based on a statistical representation of the code. The approach is less straightforward for dynamic codes, which represent time-evolving systems. We develop a novel iterative system to build a statistical model of dynamic computer codes, which is demonstrated on a rainfall-runoff simulator.

引用

页码：663 / 676

页数：14

共 15 条

[1] Estimating the prediction mean squared error in Gaussian stochastic processes with exponential correlation structure [J].

Abt, M .

SCANDINAVIAN JOURNAL OF STATISTICS, 1999, 26 (04) :563-578

[2]

[Anonymous], 2000, Probability and Statistics series

[3] A Simulation Approach to Bayesian Emulation of Complex Dynamic Computer Models [J].

Bhattacharya, Sourabh .

BAYESIAN ANALYSIS, 2007, 2 (04) :783-815

[4]

Cressie NAC, 1993, STAT SPATIAL DATA, DOI [10.1002/9781119115151, DOI 10.1002/9781119115151]

[5]

Kennedy M, 2008, J ROY STAT SOC A, V171, P109

[6] Bayesian calibration of computer models [J].

Kennedy, MC ;

O'Hagan, A .

JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 2001, 63 :425-450

[7] Towards a Bayesian total error analysis of conceptual rainfall-runoff models: Characterising model error using storm-dependent parameters [J].

Kuczera, George ;

Kavetski, Dmitri ;

Franks, Stewart ;

Thyer, Mark .

JOURNAL OF HYDROLOGY, 2006, 331 (1-2) :161-177

[8] EXPLORATORY DESIGNS FOR COMPUTATIONAL EXPERIMENTS [J].

MORRIS, MD ;

MITCHELL, TJ .

JOURNAL OF STATISTICAL PLANNING AND INFERENCE, 1995, 43 (03) :381-402

[9] Bayesian analysis of computer code outputs: A tutorial [J].

O'Hagan, A. .

RELIABILITY ENGINEERING & SYSTEM SAFETY, 2006, 91 (10-11) :1290-1300

[10] Bayesian inference for the uncertainty distribution of computer model outputs [J].

Oakley, J ;

O'Hagan, A .

BIOMETRIKA, 2002, 89 (04) :769-784

← 1 2 →