Deep learning enables large-scale inverse design of free-form metasurfaces
Abstract
Metasurfaces are ultrathin optical elements that control light through carefully engineered structures smaller than the wavelength of light, enabling compact devices with functionalities that are difficult to achieve with conventional optics. However, exploiting their full design freedom over large areas has been prohibited by the large computational cost of repeated electromagnetic simulations required for inverse design. Here we introduce a physics-tailored deep-learning surrogate model that exploits the locality and symmetries of electromagnetic interactions to generalize from small-scale simulations to metasurfaces orders of magnitude larger. The model captures nonlocal interactions while predicting optical fields more than four orders of magnitude faster than the conventional electromagnetic solver it is trained on, enabling gradient-based topology optimization under arbitrary illumination. We validate the approach through the inverse design and experimental realization of free-form metagratings with diffraction angles up to 75 degrees, as well as a millimeter-scale free-form holographic metasurface more than 1,000 wavelengths across and comprising 1.6 billion degrees of freedom. By decoupling the scale of electromagnetic simulation from the scale of inverse design, our approach enables system-scale free-form photonic devices that retain nanoscale design freedom.
Related
- Inverse design of poles and zeros: tailoring phase control in metasurfaces
- An Agent-Driven, Cross-Validated Inverse-Design Pipeline for Dielectric Metasurfaces
- Diffusion-Based Inverse Design of Dielectric Resonator Metasurfaces for Shaping Smart Electromagnetic Environments
- Inverse design of large-scale freeform meta-optics by breaking the memory wall of full-wave simulation
- Interpretable Inverse Design of Metal-Organic Frameworks with Large Language Model Agents
BibTeX
@misc{lilja2026deep,
title = {{Deep learning enables large-scale inverse design of free-form metasurfaces}},
author = {Viktor A. Lilja and Khosro Zangeneh Kamali and Mikael Käll and Philippe Tassin},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.05021},
eprint = {2610.05021},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.05021},
}
RIS
TY - UNPB
TI - Deep learning enables large-scale inverse design of free-form metasurfaces
AU - Viktor A. Lilja
AU - Khosro Zangeneh Kamali
AU - Mikael Käll
AU - Philippe Tassin
PY - 2026
DA - 2026/10/04
PB - arXiv
DO - 10.48550/arxiv.2610.05021
UR - https://arxiv.org/abs/2610.05021
AB - Metasurfaces are ultrathin optical elements that control light through carefully engineered structures smaller than the wavelength of light, enabling compact devices with functionalities that are difficult to achieve with conventional optics. However, exploiting their full design freedom over large areas has been prohibited by the large computational cost of repeated electromagnetic simulations required for inverse design. Here we introduce a physics-tailored deep-learning surrogate model that exploits the locality and symmetries of electromagnetic interactions to generalize from small-scale simulations to metasurfaces orders of magnitude larger. The model captures nonlocal interactions while predicting optical fields more than four orders of magnitude faster than the conventional electromagnetic solver it is trained on, enabling gradient-based topology optimization under arbitrary illumination. We validate the approach through the inverse design and experimental realization of free-form metagratings with diffraction angles up to 75 degrees, as well as a millimeter-scale free-form holographic metasurface more than 1,000 wavelengths across and comprising 1.6 billion degrees of freedom. By decoupling the scale of electromagnetic simulation from the scale of inverse design, our approach enables system-scale free-form photonic devices that retain nanoscale design freedom.
ER -