Predictive and generative machine learning models for photonic crystals
© 2020 Thomas Christensen et al., published by De Gruyter, Berlin/Boston 2020. The prediction and design of photonic features have traditionally been guided by theory-driven computational methods, spanning a wide range of direct solvers and optimization techniques. Motivated by enormous advances in...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Walter de Gruyter GmbH
2021
|
| Online Access: | https://hdl.handle.net/1721.1/132457 |
| _version_ | 1826208659729809408 |
|---|---|
| author | Christensen, Thomas Loh, Charlotte Picek, Stjepan Jakobović, Domagoj Jing, Li Fisher, Sophie Ceperic, Vladimir Joannopoulos, John D Soljačić, Marin |
| author_facet | Christensen, Thomas Loh, Charlotte Picek, Stjepan Jakobović, Domagoj Jing, Li Fisher, Sophie Ceperic, Vladimir Joannopoulos, John D Soljačić, Marin |
| author_sort | Christensen, Thomas |
| collection | MIT |
| description | © 2020 Thomas Christensen et al., published by De Gruyter, Berlin/Boston 2020. The prediction and design of photonic features have traditionally been guided by theory-driven computational methods, spanning a wide range of direct solvers and optimization techniques. Motivated by enormous advances in the field of machine learning, there has recently been a growing interest in developing complementary data-driven methods for photonics. Here, we demonstrate several predictive and generative data-driven approaches for the characterization and inverse design of photonic crystals. Concretely, we built a data set of 20,000 two-dimensional photonic crystal unit cells and their associated band structures, enabling the training of supervised learning models. Using these data set, we demonstrate a high-accuracy convolutional neural network for band structure prediction, with orders-of-magnitude speedup compared to conventional theory-driven solvers. Separately, we demonstrate an approach to high-throughput inverse design of photonic crystals via generative adversarial networks, with the design goal of substantial transverse-magnetic band gaps. Our work highlights photonic crystals as a natural application domain and test bed for the development of data-driven tools in photonics and the natural sciences. |
| first_indexed | 2024-09-23T14:09:31Z |
| format | Article |
| id | mit-1721.1/132457 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T14:09:31Z |
| publishDate | 2021 |
| publisher | Walter de Gruyter GmbH |
| record_format | dspace |
| spelling | mit-1721.1/1324572021-09-21T03:23:27Z Predictive and generative machine learning models for photonic crystals Christensen, Thomas Loh, Charlotte Picek, Stjepan Jakobović, Domagoj Jing, Li Fisher, Sophie Ceperic, Vladimir Joannopoulos, John D Soljačić, Marin © 2020 Thomas Christensen et al., published by De Gruyter, Berlin/Boston 2020. The prediction and design of photonic features have traditionally been guided by theory-driven computational methods, spanning a wide range of direct solvers and optimization techniques. Motivated by enormous advances in the field of machine learning, there has recently been a growing interest in developing complementary data-driven methods for photonics. Here, we demonstrate several predictive and generative data-driven approaches for the characterization and inverse design of photonic crystals. Concretely, we built a data set of 20,000 two-dimensional photonic crystal unit cells and their associated band structures, enabling the training of supervised learning models. Using these data set, we demonstrate a high-accuracy convolutional neural network for band structure prediction, with orders-of-magnitude speedup compared to conventional theory-driven solvers. Separately, we demonstrate an approach to high-throughput inverse design of photonic crystals via generative adversarial networks, with the design goal of substantial transverse-magnetic band gaps. Our work highlights photonic crystals as a natural application domain and test bed for the development of data-driven tools in photonics and the natural sciences. 2021-09-20T18:22:31Z 2021-09-20T18:22:31Z 2020 2020-10-30T18:44:26Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/132457 en 10.1515/NANOPH-2020-0197 Nanophotonics Creative Commons Attribution 4.0 International license https://creativecommons.org/licenses/by/4.0/ application/pdf Walter de Gruyter GmbH De Gruyter |
| spellingShingle | Christensen, Thomas Loh, Charlotte Picek, Stjepan Jakobović, Domagoj Jing, Li Fisher, Sophie Ceperic, Vladimir Joannopoulos, John D Soljačić, Marin Predictive and generative machine learning models for photonic crystals |
| title | Predictive and generative machine learning models for photonic crystals |
| title_full | Predictive and generative machine learning models for photonic crystals |
| title_fullStr | Predictive and generative machine learning models for photonic crystals |
| title_full_unstemmed | Predictive and generative machine learning models for photonic crystals |
| title_short | Predictive and generative machine learning models for photonic crystals |
| title_sort | predictive and generative machine learning models for photonic crystals |
| url | https://hdl.handle.net/1721.1/132457 |
| work_keys_str_mv | AT christensenthomas predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT lohcharlotte predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT picekstjepan predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT jakobovicdomagoj predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT jingli predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT fishersophie predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT cepericvladimir predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT joannopoulosjohnd predictiveandgenerativemachinelearningmodelsforphotoniccrystals AT soljacicmarin predictiveandgenerativemachinelearningmodelsforphotoniccrystals |