Complex Inverse Design of Meta-optics by Segmented Hierarchical Evolutionary Algorithm.

With the recent burgeoning advances in nano-optics, ultracompact, miniaturized photonic devices with high-quality and spectacular functionalities are highly desired. Such devices' design paradigms often call for the solution of a complex inverse nonanalytical/semianalytical problem. However, currently reported strategies dealing with amplitude-controlled meta-optics devices achieved limited functionalities mainly due to restricted search space and demanding computational schemes. Here, we established a segmented hierarchical evolutionary algorithm, aiming to solve large-pixelated, complex inverse meta-optics design and fully demonstrate the targeted performance. This paradigm allows significantly extended search space at a rapid converging speed. As typical complex proof-of-concept examples, large-pixelated meta-holograms are chosen to demonstrate the validity of our design paradigm. An improved fitness function is proposed to reinforce the performance balance among image pixels, so that the image quality is improved and computing speed is further accelerated. Broadband and full-color meta-holograms with high image fidelities using binary amplitude control are demonstrated experimentally. Our work may find important applications in the advanced design of future nanoscale high-quality optical devices.