進化的アルゴリズムによる波長選択集光マイクロレンズの逆設計解析

抄録

This paper demonstrates an inverse design to explore the possible best binary diffractive microlens which functions as a wavelength selective concentrator. EA-FDTD method in which evolutionary algorithm (EA) was incorporated with finite difference time domain (FDTD) computation was used for optimization of the lens structure. The present method successfully generated the selective concentrator structures within practical computation time. This implies that the present method has a potential to design complicated but desirable photonic microstructures which the conventional methods hardly design efficiently. In addition, effects of EA parameters such as mutation rate and crossover rate on convergence speed were tested. Transformation of the lens structure was monitored generation by generation. It was found that there was a suitable set of EA parameters for each generation steps to accelerate convergence. If EA parameters can be controlled properly at each generation steps, convergence speed becomes faster. It was also found that the best structure was obtained efficiently if the degree of freedom for the part of the lens structure was controlled properly during the generations. Those results will contribute to further development of EA-FDTD method.