[11a-PA2-1~17] 3.2 Equipment optics and materials

[11a-PA2-6] Design of multilevel anisotropic diffraction gratings based on deep reinforcement learning

〇Hiroyuki Okamoto¹, Furukawa Hirotomo¹, Noda Kohei², Sakamoto Moritsugu², Sasaki Tomoyuki², Ono Hiroshi² (1.NIT, Anan Col., 2.Nagaoka Univ. of Technol.)

Keywords:multilevel anisotropic diffraction gratings, deep reinforcement learning

Time-cost and human-resources are needed to design new multilevel anisotropic diffraction gratings. To reduct these costs, we contrived to use computers. Furthermore, as an advantage of using a computer for structural design , computer can design a structure beyond human ideas. We developed the system that designes multilevel anisotropic diffraction gratings based on deep reinforcement learning.