Paper
20 January 2022 Inverse design nanorod hyperbolic metamaterial by transformer
Author Affiliations +
Proceedings Volume 12154, 13th International Photonics and OptoElectronics Meetings (POEM 2021); 1215416 (2022) https://doi.org/10.1117/12.2625336
Event: 13th International Photonics and OptoElectronics Meetings (POEM 2021), 2021, Wuhan, China
Abstract
Recently, deep learning methods have revolutionized the design of nanophotonic devices, which provides a new way to efficiently design nanophotonic devices. Here, we demonstrated a deep learning method using attention mechanisms to inverse design nanophotonic devices, the mean relative error of the predicted value can be as low as 4.1% or less. Using the encoder part of Transformer, the long sequence of spectral data can be mapped to the structural parameters of the nanorod hyperbolic metamaterial. The inverse design model based on the attention mechanisms is good at processing sequence data and can be calculated in parallel, which is an effective way to design nanophotonic devices.
© (2022) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Ruoqin Yan and Tao Wang "Inverse design nanorod hyperbolic metamaterial by transformer", Proc. SPIE 12154, 13th International Photonics and OptoElectronics Meetings (POEM 2021), 1215416 (20 January 2022); https://doi.org/10.1117/12.2625336
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Nanophotonics

Metamaterials

Nanorods

Transformers

Computer programming

Data modeling

Finite-difference time-domain method

Back to Top