|
Abstract:
|
Jerry Sacks is well known for his pioneering work on computer experiments starting in the late 1980s. Prior to that, Jerry's work included a remarkable series of papers over a span of 20 years on robust design and analysis for regression, largely in collaboration with Don Ylvisaker. By reviewing that literature, we will argue that the necessary statistical tools---nonparametric departures from the assumed model---were primed and ready for the computer-experiment revolution. The nonparametric departures from a regression model played a much larger, indeed starring, role in modeling the output function of a computer model by a Gaussian stochastic process, but many of the basics are common to both areas. Also, people often ask how statistical design and analysis of computer experiments emerged in the late 1980s. The talk will recount some of that story: how Jerry, along with Don Ylvisaker, Toby Mitchell, Henry Wynn and others, saw the connections between their statistical tools and the revolution in the way that scientists were starting to conduct experiments.
|
