Accurate emulators for large-scale computer experiments

Abstract

Large-scale computer experiments are becoming increasingly important in science. A multi-step procedure is introduced to statisticians for modeling such experiments, which builds an accurate interpolator in multiple steps. In practice, the procedure shows substantial improvements in overall accuracy, but its theoretical properties are not well established. We introduce the terms nominal and numeric error and decompose the overall error of an interpolator into nominal and numeric portions. Bounds on the numeric and nominal error are developed to show theoretically that substantial gains in overall accuracy can be attained with the multi-step approach.

Published in at http://dx.doi.org/10.1214/11-AOS929 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (2)

• A new and flexible method for constructing designs for computer experiments
• Construction of nested space-filling designs

Cited by in corpus (17)

• Practical heteroskedastic Gaussian process modeling for large simulation experiments
• Composite Gaussian process models for emulating expensive functions
• Calibrating a large computer experiment simulating radiative shock hydrodynamics
• On the construction of nested space-filling designs
• Cross-validation based adaptive sampling for Gaussian process models
• Variable selection and sensitivity analysis using dynamic trees, with an application to computer code performance tuning
• Potentially Predictive Variance Reducing Subsample Locations in Local Gaussian Process Regression
• A theoretical framework for calibration in computer models: parametrization, estimation and convergence properties
• Local Gaussian process approximation for large computer experiments
• Four-Point, 2D, Free-Ranging, IMSPE-Optimal, Twin-Point Designs
• Parallel Bayesian Additive Regression Trees
• Predicting the Output From a Stochastic Computer Model When a Deterministic Approximation is Available
• Transitional annealed adaptive slice sampling for Gaussian process hyper-parameter estimation
• Controlling Sources of Inaccuracy in Stochastic Kriging
• Speeding up neighborhood search in local Gaussian process prediction
• Massively parallel approximate Gaussian process regression
• Fast prediction of deterministic functions using sparse grid experimental designs