Integral global optimization method for nonlinear games (Q805506)

A new, derivative-free method is proposed for the solution of non-convex continuous games defined over compact robust sets that may be disconnected and with nonsmooth boundary. The method is based on the integral global optimization algorithm for mathematical programming and it does not employ gradient-based techniques nor the notion of convexity. On this basis, certain iterative algorithms are developed for which convergence theorems are proved. The algorithms provide complete exact solutions of the game irrespective of the existence of saddle points. Finally, games with uncertainties are introduced and the applicability of the method and of the algorithms to such games is demonstrated.