Global optimization by an improved differential evolutionary algorithm

A hybrid differential evolutionary (DE) algorithm for global optimization is proposed. In the new algorithm, the stochastic properties of chaotic systems are used to spread the individuals in search spaces as much as possible, the pattern search method is employed to speed up the local exploiting and the DE operators are used to jump to a better point. The global convergence is proved. Three typical chaotic systems are investigated in detail. Numerical experiments on benchmark examples including 13 high dimensional functions demonstrate that the new method achieved an improved success rate and final solution with less computational effort.