SUPR
Machine Learning for Nanophotonics
Dnr:

NAISS 2025/5-182

Type:

NAISS Medium Compute

Principal Investigator:

Philippe Tassin

Affiliation:

Chalmers tekniska högskola

Start Date:

2025-04-01

End Date:

2026-03-01

Primary Classification:

10304: Condensed Matter Physics

Webpage:

Allocation

Abstract

In this project, we are using machine learning for inverse design in photonics. Data sets for different types of devices have been obtained (elsewhere) or are in the process of being constructed. This compute project is then used to train neural networks with data from full-wave simulations to predict the optical properties of new nanophotonic devices and also to perform free-form inverse design (topology optimization), i.e., obtain a design for an nanophotonic device with a desired optical response.