Hybrid genetic programming with accelerating optimizer for 3D metamaterial design

Abstract

In this paper, the hybridization of previously developed genetic programming (GP) software with a low-level optimizer, which uses genetic algorithm (GA), is presented. Key variables are selected for low-level optimization of each GP topology. Two low frequency design examples are included, with each covering at least 225–450 MHz. The first design (from the 142nd generation of the original GP software) has 73.4% bandwidth (±90° reflection phase) while being 17.0% thinner than a previously presented design covering this band, whereas the second design (from the 6th generation of the hybrid software) has 68.9% bandwidth and is 25.3% thinner.