Adaptive Genetic Algorithm for Optical Metasurfaces Design.

Area of Science:

• Optics and Photonics
• Materials Science
• Computational Science

Background:

• Conventional optical metasurfaces face limitations in structural complexity and phase engineering for emerging applications.
• Existing design methods struggle to meet the increasing demands for tailorable and multifunctional metasurface platforms.

Purpose of the Study:

• To introduce digitized-binary elements as high-dimensional building blocks for complex metasurface design.
• To demonstrate the adaptive genetic algorithm (AGA) as an effective optimizer for these demanding applications.
• To showcase the versatility of binary metasurfaces and AGA through four challenging optical problems.

Main Methods:

• Development and application of digitized-binary elements for metasurface construction.
• Utilizing an adaptive genetic algorithm (AGA) for optimizing complex metasurface designs.
• Solving four distinct optical problems: beam-steering reflectarray, dual-beam leaky-wave antenna, birefringent metasurface, and visible-transparent infrared metasurface.

Main Results:

• Successfully designed and demonstrated four novel metasurface applications overcoming conventional limitations.
• Achieved high performance in beam-steering, dual-beam diffraction, compact birefringence, and thermal management.
• Generated large datasets from AGA, suitable for machine learning and deep learning algorithms.

Conclusions:

• Digitized-binary elements combined with AGA offer a powerful approach for designing complex, multifunctional optical metasurfaces.
• The AGA technique effectively addresses computational and fabrication challenges in metasurface design.
• Generated datasets pave the way for rapid, data-driven solutions in metasurface research using AI.