2020/05/26 by Abhishek Mall, Abhijeet Patil, Mall, Abhishek +5
Materials Science · Biochemistry, Genetics and Molecular Biology · #Metamaterials and Metasurfaces Applications #Animal Vocal Communication and Behavior
paper · pdf · doi:10.48550/arxiv.2005.12796
The conventional approach to nanophotonic metasurface design and optimization\nfor a targeted electromagnetic response involves exploring large geometry and\nmaterial spaces, which is computationally costly, time consuming and a highly\niterative process based on trial and error. Moreover, the non-uniqueness of\nstructural designs and high non-linearity between electromagnetic response and\ndesign makes this problem challenging. To model this non-intuitive relationship\nbetween electromagnetic response and metasurface structural design as a\nprobability distribution in the design space, we introduce a cyclical deep\nlearning (DL) based framework for inverse design of nanophotonic metasurfaces.\nThe proposed framework performs inverse design and optimization mechanism for\nthe generation of meta-atoms and meta-molecules as metasurface units based on\nDL models and genetic algorithm. The framework includes consecutive DL models\nthat emulate both numerical electromagnetic simulation and iterative processes\nof optimization, and generate optimized structural designs while simultaneously\nperforming forward and inverse design tasks. A selection and evaluation of\ngenerated structural designs is performed by the genetic algorithm to construct\na desired optical response and design space that mimics real world responses.\nImportantly, our cyclical generation framework also explores the space of new\nmetasurface topologies. As an example application of utility of our proposed\narchitecture, we demonstrate the inverse design of gap-plasmon based half-wave\nplate metasurface for user-defined optical response. Our proposed technique can\nbe easily generalized for designing nanophtonic metasurfaces for a wide range\nof targeted optical response.\n