Markov-chain Monte Carlo approach to the design of multilayer thin-film optical coatings

We reconsider the problem of locating the globally optimal solution of a multilayer-optical-coating design problem, within some predetermined space of parameters, with the aim of obtaining a robust technique that requires a minimum of user intervention. The approach we adopt centers on exploring the space of the parameters of interest by using a Markov-chain Monte Carlo sampling algorithm. This technique enables one to locate the global optimum automatically with high confidence and without the need for a good starting design. It also allows the trivial inclusion of prior constraints on the variables and provides a natural means for investigating the robustness of the optimal solution.