A comparison of multi-objective optimizers using a microstrip patch antenna test problem

This study compares the convergence efficiency of three different multi-objective optimizers in the analysis of microstrip patch antennas. Each function evaluation is computationally expensive, making the convergence rate a critical factor in whether these algorithms are feasible for real-world antenna design and other electromagnetic applications. Each optimizer has a unique design philosophy to solve the problem. The optimizers chosen include a genetic algorithm (BORG), an evolutionary strategy (MOCMA-ES) and a scalarization algorithm (MOEAD). Tradeoffs between S11 and axial ratio bandwidth, front-to-back ratio and overall size are studied for a series of designs with increasing complexity.