Distributed evolutionary algorithm for optimization in electromagnetics

This paper presents a distributed universal evolutionary optimization environment designed for optimal design in electromagnetics. The optimizer can be tuned to act as a genetic algorithm or as an evolutionary strategy. Basic principles of optimizer design are discussed. The system is coded in Java and uses a remote method invocation technique to distribute computational tasks between local and/or remote servers. T.E.A.M Workshop problem 25 was used to estimate parallel performance of the system. An example of application is the optimal location of electrodes for electroconvulsive stimulation is presented