Global and Multi-objective Optimization for Lens Design by Real-coded Genetic Algorithms
Isao Ono, Shigenobu Kobayashi, Kōji Yoshida · International Optical Design Conference · 1998
Local search methods have been traditionally used in lens design. They have a serious problem that the quality of lenses found by them heavily depends on starting points. Recently, several global optimization methods for lens design have been proposed(1-5),(8). Of these studies, Betensky(1) and Chen, et al.(2) have proposed methods based on bit-string genetic algorithms and the DLS method. However, all conventional local and global methods for lens design do not explicitly handle plural criteria. Therefore, a lens designer must usually find an appropriate starting point and a proper tradeoff-ratio among plural criteria by trail and error.