Antenna Array Synthesis Using Derivative, Non-Derivative and Random Search Optimization

The literature on antenna array optimization has recently focused on direct search methods that use random decision making, e.g., genetic/evolutionary algorithms (GA) and particle swarm optimization (PSO). In this paper we compare results for test problems from the literature using direct search methods that do not use random decision making, such as Nelder-Mead, with those that do, such as PSO. We also revisit gradient based search methods, specifically Newton-like optimization methods that also do not rely upon randomness. Search efficiency is measured in terms of the total number of function evaluations required to converge to an acceptable pattern. For these test problems it is found that direct search without randomness is far more efficient than direct search with randomness; and that non-random gradient based search is most efficient of all.