Multiple response surface methods in computer simulation

This paper reviews the application of multiple re sponse surfaces to multiple-variable optimization problems and describes how these techniques may be used in analyzing computer simulation experiments. An example with four response surfaces illustrates the method. In it a simulation of a tank duel is analyzed to determine the values of two independent (input) variables that will optimize four dependent (output) variables simultaneously. The problem examined is that of training battle-tank crews, and the optimization procedure used is based on the Geoffrion-Dyer interactive vector maximal algorithm.