Physics-informed recurrent neural network for time dynamics in optical resonances
Abstract
Resonance structures and features are ubiquitous in optical science. However, capturing their time dynamics in real-world scenarios suffers from long data acquisition time and low analysis accuracy due to slow convergence and limited time windows. Here we report a physics-informed recurrent neural network to forecast the time-domain response of optical resonances and infer corresponding resonance frequencies by acquiring a fraction of the sequence as input. The model is trained in a two-step multi-fidelity framework for high-accuracy forecast, using first a large amount of low-fidelity physical-model-generated synthetic data and then a small set of high-fidelity application-specific data. Through simulations and experiments, we demonstrate that the model is applicable to a wide range of resonances, including dielectric metasurfaces, graphene plasmonics and ultra-strongly coupled Landau polaritons, where our model captures small signal features and learns physical quantities. The demonstrated machine-learning algorithm can help to accelerate the exploration of physical phenomena and device design under resonance-enhanced light–matter interaction.
Additional Information
© 2022 Nature Publishing Group. Received 06 September 2021; Accepted 16 February 2022; Published 28 March 2022. J.F. and W.G. thank the University of Utah start-up fund for support. X.L. acknowledges support from the Caltech Postdoctoral Prize Fellowship and the IQIM. C.Y. acknowledges support from grants NSF-2047176 and NSF-2008144. Data availability: The source data of all figures in both main text and Supplementary Information are available at https://github.com/GaoUtahLab/Cascaded_GRU_Networks. The Zenodo version is available at ref. 45. Source data are provided with this paper. Code availability: The code for the models in all three demonstrations of optical resonances and that support the plots within this paper and other findings of this study is available at https://github.com/GaoUtahLab/Cascaded_GRU_Networks. The Zenodo version is available at ref. 45. Contributions: C.Y. and W.G. conceived the idea and designed the project. Y.T. performed the modeling and calculations with the help of J.F., J.M., M.Q., C.Y. and W.G. X.L. helped with the analysis of the Landau polariton data. Y.T. and W.G. wrote the manuscript. All authors discussed the manuscript and provided feedback. The authors declare no competing interests. Peer review information: Nature Computational Science thanks Andrey Baydin, Bowen Zheng and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Kaitlin McCardle, in collaboration with the Nature Computational Science team.Attached Files
Submitted - 2109.09837.pdf
Supplemental Material - 43588_2022_215_MOESM1_ESM.pdf
Supplemental Material - 43588_2022_215_MOESM2_ESM.zip
Supplemental Material - 43588_2022_215_MOESM3_ESM.zip
Supplemental Material - 43588_2022_215_MOESM4_ESM.zip
Files
Name | Size | Download all |
---|---|---|
md5:cbee53053f79ecea6555588c612f1296
|
74.8 kB | Preview Download |
md5:ebf005e0c5cf69f386708c2c21b94107
|
4.0 MB | Preview Download |
md5:c65ca1491853b883b590adfd8ae749a1
|
18.0 kB | Preview Download |
md5:dff0d513800c03568299a6d6886bf521
|
4.0 MB | Preview Download |
md5:53df00197a96fb29b5c611b549b8b5c8
|
4.9 MB | Preview Download |
Additional details
- Eprint ID
- 113435
- DOI
- 10.1038/s43588-022-00215-2
- Resolver ID
- CaltechAUTHORS:20220214-170107696
- University of Utah
- Caltech
- Institute for Quantum Information and Matter (IQIM)
- CCF-2047176
- NSF
- CCF-2008144
- NSF
- Created
-
2022-02-14Created from EPrint's datestamp field
- Updated
-
2022-04-05Created from EPrint's last_modified field