A New Look at Solving Minimax Problems with Coevolution

In recent papers, [1, 2, 3], coevolutionary genetic algorithms have been used to solve so-called minimax problems from mechanical structure optimisation, constrained optimisation and scheduling domains. The applications have been quite successful, but the algorithms used require the minimax problems to have a certain property; the problem has to be symmetric in the two search-spaces. In the present article it is argued that the proposed algorithms will fail to converge if the problem is not symmetric. A new approach solving the problem will be demonstrated.