Robust Gaussian stochastic process emulation

DOI10.1214/17-AOS1648zbMATH Open1408.62155arXiv1708.04738OpenAlexW2963446226MaRDI QIDQ1991689FDOQ1991689

Authors: Mengyang Gu, Xiaojing Wang, James O. Berger 

Publication date: 30 October 2018

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: We consider estimation of the parameters of a Gaussian Stochastic Process (GaSP), in the context of emulation (approximation) of computer models for which the outcomes are real-valued scalars. The main focus is on estimation of the GaSP parameters through various generalized maximum likelihood methods, mostly involving finding posterior modes; this is because full Bayesian analysis in computer model emulation is typically prohibitively expensive. The posterior modes that are studied arise from objective priors, such as the reference prior. These priors have been studied in the literature for the situation of an isotropic covariance function or under the assumption of separability in the design of inputs for model runs used in the GaSP construction. In this paper, we consider more general designs (e.g., a Latin Hypercube Design) with a class of commonly used anisotropic correlation functions, which can be written as a product of isotropic correlation functions, each having an unknown range parameter and a fixed roughness parameter. We discuss properties of the objective priors and marginal likelihoods for the parameters of the GaSP and establish the posterior propriety of the GaSP parameters, but our main focus is to demonstrate that certain parameterizations result in more robust estimation of the GaSP parameters than others, and that some parameterizations that are in common use should clearly be avoided. These results are applicable to many frequently used covariance functions, e.g., power exponential, Mat{'e}rn, rational quadratic and spherical covariance. We also generalize the results to the GaSP model with a nugget parameter. Both theoretical and numerical evidence is presented concerning the performance of the studied procedures.

Full work available at URL: https://arxiv.org/abs/1708.04738

Recommendations

• Jointly robust prior for Gaussian stochastic process in emulation, calibration and variable selection
• Computer emulation with nonstationary Gaussian processes
• Default priors for Gaussioan processes
• The effect of the nugget on Gaussian process emulators of computer models
• Flexible Correlation Structure for Accurate Prediction and Uncertainty Quantification in Bayesian Gaussian Process Emulation of a Computer Model

zbMATH Keywords

emulationobjective priorsLatin hypercube designGaussian stochastic processposterior proprietyrobust parameter estimationanisotropic covariance

Mathematics Subject Classification ID

Inference from spatial processes (62M30) Statistical block designs (62K10) Robustness and adaptive procedures (parametric inference) (62F35)

Cites Work

• Quasi-regression
• The design and analysis of computer experiments.
• Design and analysis of computer experiments. With comments and a rejoinder by the authors
• Bayesian calibration of computer models. (With discussion)
• Bayesian Design and Analysis of Computer Experiments: Use of Derivatives in Surface Prediction
• Towards reconciling two asymptotic frameworks in spatial statistics
• Objective Bayesian Analysis of Spatially Correlated Data
• Inconsistent Estimation and Asymptotically Equal Interpolations in Model-Based Geostatistics
• Model-based geostatistics.
• Default priors for Gaussioan processes
• Parallel partial Gaussian process emulation for computer models with massive output
• Handbook of spatial statistics.
• Objective Bayesian analysis of spatial data with measurement error
• Space and space-time modeling using process convolutions
• The effect of the nugget on Gaussian process emulators of computer models
• Computer model validation with functional output
• An Approach to Statistical Spatial-Temporal Modeling of Meteorological Fields
• Objective Bayesian analysis for a spatial model with nugget effects
• Objective Bayesian analysis of spatial data with uncertain nugget and range parameters
• Objective Bayesian analysis of geometrically anisotropic spatial data
• Automating emulator construction for geophysical hazard maps
• Objective Bayesian analysis of spatial models with separable correlation functions

Cited In (31)

• Adaptive Bayesian nonparametric regression using a kernel mixture of polynomials with application to partial linear models
• Prediction of non-stationary response functions using a Bayesian composite Gaussian process
• Emulating satellite drag from large simulation experiments
• Heavy-Tailed Density Estimation
• Robust additive Gaussian process models using reference priors and cut-off-designs
• Jointly robust prior for Gaussian stochastic process in emulation, calibration and variable selection
• Default priors for Gaussioan processes
• Gaussian process hyper-parameter estimation using parallel asymptotically independent Markov sampling
• Comparing emulation methods for a high-resolution storm surge model
• The Zero Problem: Gaussian Process Emulators for Range-Constrained Computer Models
• Objective Bayesian Analysis of a Cokriging Model for Hierarchical Multifidelity Codes
• Bayesian fixed-domain asymptotics for covariance parameters in a Gaussian process model
• A higher-order singular value decomposition tensor emulator for spatiotemporal simulators
• Beyond Matérn: On A Class of Interpretable Confluent Hypergeometric Covariance Functions
• Optimal compromise between incompatible conditional probability distributions, with application to objective Bayesian kriging
• Coupling computer models through linking their statistical emulators
• Propriety of the reference posterior distribution in Gaussian process modeling
• Gaussian Process Emulation for High-Dimensional Coupled Systems
• Screening the Discrepancy Function of a Computer Model
• Novel statistical emulator construction for volcanic ash transport model Ash3d with physically motivated measures
• Modeling Nonstationary and Asymmetric Multivariate Spatial Covariances via Deformations
• Ice model calibration using semicontinuous spatial data
• Efficient kriging using interleaved lattice-based designs with low fill and high separation distance properties
• A Bayesian meta-modeling approach for Gaussian stochastic process models using a non informative prior
• Fast Nonseparable Gaussian Stochastic Process With Application to Methylation Level Interpolation
• Diagnostics-driven nonstationary emulators using kernel mixtures
• Generalized probabilistic principal component analysis of correlated data
• Fitting Matérn smoothness parameters using automatic differentiation
• Bayesian Projected Calibration of Computer Models
• Computer emulation with nonstationary Gaussian processes
• Scaled Gaussian stochastic process for computer model calibration and prediction

Uses Software

• DiceKriging
• DiceOptim
• Virtual library

This page was built for publication: Robust Gaussian stochastic process emulation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1991689)