Hybrid Genetic Programming With Accelerating Genetic Algorithm Optimizer for 3-D Metamaterial Design

Abstract

The computational efficiency of existing genetic programming (GP) software is improved through the addition of parallelization and hybridization with a low-level optimizer. The low-level optimizer is implemented using genetic algorithm, and two different versions of the developed hybrid GP program are presented. Timing and efficiency performance are discussed, which includes a method of making the original optimizer more than eight times faster. Efficiency results indicate that through the use of the optimizer, with sufficient variables included, significantly fewer (about ¼ to ½) GP generations are required to achieve working metamaterial designs. Two low-frequency broadband (225–450 MHz) ground plane design examples, which are 25.3% thinner than a previously published design covering the same frequency band, are included.