|
Abstract:
|
This is the first of a series of two talks that will discuss some statistical approaches to analyzing data from computer experiments. We will begin by introducing the basic components of a computer experiment, emphasizing how computer experiments differ from physical experiments. We will then describe some common goals of computer experiments. Following this overview, we will discuss strategies for selecting an experimental design. We will present several popular space-filling designs including Latin hypercube designs, uniform designs, minimax designs, maximin designs, Sobol sequences, and designs based on multiple criteria. We will discuss software that is available for generating designs, with examples. We will also briefly describe other strategies based on statistical criteria, including sequential strategies.
|
