Abstract
A global optimization strategy based on Differential Evolution (DE) is proposed. The new optimization strategy divides the total population into two swarms, assimilative and evolutionary swarms. The different operations to generate offspring vector are implemented in different swarms. In each generation, the best member of evolutionary swarm will be exalted to join in the assimilative swarm to replace the worst performance member. For verifying the efficiency of the new optimization strategy, some Yagi-Uda antennas are optimized as examples. The results show that the new global optimizer is robust for multi-modal objective function optimization.