Bayesian approach to global optimization and application to multiobjective and constrained problems (Q1176841)

The authors recall some earlier results which they obtained with the Bayesian approach to the problem of global optimization. They present new proofs about the asymptotic behavior of the density of observations around the point of global minimum of the objective function and of the parameters of the methods. Algorithms are presented integrating this technique together with a set of linear or nonlinear constraints. Even the case of multiobjective optimization is treated and shown to be sometimes reducible to global optimization. Finally the advantages are shown of treating the Bayesian approach to global optimization in an interactive mode. The authors state that this procedure is furthermore almost inevitable in multiobjective optimization, the efficiency of which depends on the reduction of multidimensional data. The article ends with the presentation of the corresponding software either in FORTRAN or in \(C\)-language. In this last version, which is intended for use on PCs, large facilities of graphical representation are also provided.