Multifunctional Metasurface Design with a Generative Adversarial Network (Advanced Optical Materials 5/2021). Issue 5 (4th March 2021)

• Record Type: Journal Article
• Authors: An, Sensong; Zheng, Bowen; Tang, Hong; Shalaginov, Mikhail Y.; Zhou, Li; Li, Hang; Kang, Myungkoo; Richardson, Kathleen A.; Gu, Tian; Hu, Juejun; Fowler, Clayton; Zhang, Hualiang
• Is Part Of: Advanced optical materials. Volume 9:Issue 5(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-03-04
• Subjects: deep learning -- inverse design -- metasurfaces -- neural networks -- photonics; Optical materials -- Periodicals; Photonics -- Periodicals; 620.11295
• DOI: 10.1002/adom.202170019
• Languages: English
• ISSNs: 2195-1071

Abstract

Abstract : This cover picture illustrates the deep‐learning approach for designing multifunctional all‐dielectric meta‐devices as proposed by Clayton Fowler, Hualiang Zhang, and co‐workers in article number 2001433 . With the help of a fully‐trained Generative Adversarial Network (GAN), various meta‐atom designs can be generated based on different design goals, which can be used to achieve multifunctional meta‐optics devices/systems (e.g. frequency multiplexed metalens as shown in the cover image). Several metasurface devices including multifunctional beam deflectors and lenses are designed and verified to showcase the efficacy of the proposed deep‐learning framework. This approach offers new opportunities to design and advance meta‐optic devices with enhanced time efficiency and performance.