Deep inverse models for the real-time design of optical metasurfaces from near-field responses
Deep inverse models (DIMs) can circumvent the simulation bottlenecks of traditional optical metasurface design, but they have predominantly been applied to low-dimensional targets and isolated meta-atoms characterized under the local phase approximation (LPA). Their efficacy on high-dimensional near-field targets produced by strongly coupled, heterogeneous structures remains largely uncharacterized. We benchmark modern forward emulators and DIMs from three model families on full-wave simulations of a heterogeneous nanopillar array that captures inter-element near-field coupling directly. The preferred model depends on the candidate budget: conditional invertible neural networks (cINNs) give lower design error at one to two orders of magnitude lower cost, while the neural adjoint attains the lowest field error once ∼10 2 or more candidates are searched, an advantage that holds under full-wave verification. We then introduce a multi-wavelength framework that conditions a single emulator on a continuous wavelength descriptor, enabling inversion under varying degrees of spectral specification, and show that additional bands separate near-degenerate solutions that a single band cannot distinguish.
Authors
- Jordan M. Malof (ORCID: https://orcid.org/0000-0002-7851-4920)
- Saad Lahrichi (ORCID: https://orcid.org/0000-0003-2249-9255)
- Marshall Lindsay
- Scott Kovaleski
- Matthew Maschmann
- Steven price
- Derek Anderson
- Ethan Mick
- Stanton Price
Institutions
- United States Army (US)
- University of Missouri (US)
- U.S. Army Engineer Research and Development Center (US)
Publication Details
- Journal
- Optics Continuum
- Published
- 2026-11-19
- DOI
- https://doi.org/10.1364/optcon.616299
- Primary Topic
- Metamaterials and Metasurfaces Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00