Abstract
Nanophotonic inverse design uses gradient-based optimization to design the topologies of devices such as waveguides and lenses. Since the optimization process is sensitive to initial conditions yet computationally expensive, it is advantageous to balance the exploration-exploitation tradeoff by allocating optimization resources only to promising designs. Inspired by hyperparameter tuning in machine learning, this disclosure proposes Asynchronous Topology Optimization with Early Stopping (ATOES), where nanophotonic devices are optimized in parallel and resources are allocated to more promising designs via an early stopping algorithm.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Wee, Zi Jian; Lim, Hock Soon; Hachmioune, Sabrine; Chaudhary, Nidhi; and Kim, Heonjung, "Parallel nanophotonic inverse design with early stopping", Technical Disclosure Commons, (November 27, 2025)
https://www.tdcommons.org/dpubs_series/8941