Designing Computer Experiments

Thomas J. Santner
Department of Statistics
Ohio State University

This talk will describe the two fundamental approaches used to construct experimental designs for running computer experiments. The first class of methods are geometric, space-filling techniques and the second are stochastic process-based methods. The former are useful in the initial stages of investigating the input- output relationship of a code and for verification of computer codes. The latter methods are typically used to satisfy quantitative criterion such as improving the overall predicted input-output surface or optimizing a computer code.

Return to Program