On the effective coupling of optimization algorithms to solve inverse problems of electromagnetism

The hybrid optimization tool presented in this paper combines generational genetic algorithms (GA) and variable metric (VM) optimizers. Both GA and VM may deal with the same, parameterized description of the inverse problem and a switch from GA to Vh? is being made automatically. GA is used to locate a subdomain in the design variables space containing the global minimum of the objective function. This global minimum may be found then, quickly and precisely, with the deterministic optimizer. The crucial concern of the hybrid algorithm design is to switch from stochastic to deterministic algorithm is such a way as to ensure that the global solution will be found in the fastest way. The article is focused on the algorithm which is able to determine whether GA has already found the desired subdomain described above. This algorithm is based on the cluster analysis using seed points and density-determined hyperspheres.