Designs for Computer Experiments

TLDR

A computer experiment generates observations by running a computer model at chosen inputs and recording the output, and prediction of the response at untried inputs is achieved by modeling the systematic departure from a linear model as a stochastic process. The study aims to select input settings that enable efficient prediction of the response in computer experiments. The authors address the choice of stochastic‑process model and the computation of efficient designs, applying the approach to chemical kinetics problems. Key words include computer‑aided design, experimental design, prediction, response surface, spatial statistics, and supercomputing.

Abstract

Abstract A computer experiment generates observations by running a computer model at inputs x and recording the output (response) Y. Prediction of the response Y to an untried input is treated by modeling the systematic departure of Y from a linear model as a realization of a stochastic process. For given data (selected inputs and the computed responses), best linear prediction is used. The design problem is to select the inputs to predict efficiently. The issues of choice of stochastic-process model and computation of efficient designs are addressed, and applications are made to some chemical kinetics problems. KEY WORDS: Computer-aided designExperimental designPredictionResponse surfaceSpatial statisticsSupercomputing