Optimization of a thin-film multilayer design by use of the generalized simulated-annealing method

Abstract

The generalized simulated-annealing method is a statistical optimization method for finding the global minimum of a multivariable function. It was initially used in complicated lens system design. We report success in applying this method to thin-film multilayer design. The local minima trapping problem mentioned previously [W. J. Wild and H. Buhay, Opt. Lett. 11, 745 (1986)] with the simulated-annealing method is not found here because of the use of a proper evaluation function [I. O. Bohachevsky, M. E. Johnson, and M. L. Stein, Technometrics 28, 209 (1986)]. The effects of the initial configuration and the random step size on the simulation process are discussed systematically.