Nature-inspired metaheuristic optimization algorithms for FDTD dispersion modeling

Abstract

Optimization algorithms have been employed for a variety of applications such as engineering design optimization, machine learning, control systems, computer science and software engineering. Among various optimization approaches, nature-inspired metaheuristic optimization algorithms excel in addressing complex optimization problems by considering various constraints and optimizing a wide array of variables and target functions. In finite-difference time-domain (FDTD) methods for complex dispersive media, it is crucial to derive accurate dispersion model parameters that satisfy the numerical stability conditions by applying an optimization algorithm. In this work, we apply five representative nature-inspired metaheuristic optimization algorithms to extract accurate and numerically stable dispersion modeling parameters: continuous genetic algorithm, particle swarm optimization (PSO), artificial bee colony, grey wolf optimization, and coyote optimization algorithm. To achieve a comprehensive analysis, this study examines the FDTD dispersion modeling for various materials across different frequency ranges. The numerical examples illustrate that PSO excels at extracting numerically stable and highly accurate parameters for the FDTD dispersion model.