Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity
Graph neural networks (GNNs) have recently been used to learn the representations of crystal structures through an end-to-end data-driven approach. However, a systematic top-down approach to evaluate and understand the limitations of GNNs in accurately capturing crystal structures has yet to be esta...
| Main Authors: | , , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
American Association for the Advancement of Science
2024
|
| Online Access: | https://hdl.handle.net/1721.1/154281 |
| _version_ | 1824458180547051520 |
|---|---|
| author | Gong, Sheng Yan, Keqiang Xie, Tian Shao-Horn, Yang Gomez-Bombarelli, Rafael Ji, Shuiwang Grossman, Jeffrey C. |
| author2 | Massachusetts Institute of Technology. Department of Materials Science and Engineering |
| author_facet | Massachusetts Institute of Technology. Department of Materials Science and Engineering Gong, Sheng Yan, Keqiang Xie, Tian Shao-Horn, Yang Gomez-Bombarelli, Rafael Ji, Shuiwang Grossman, Jeffrey C. |
| author_sort | Gong, Sheng |
| collection | MIT |
| description | Graph neural networks (GNNs) have recently been used to learn the representations of crystal structures through an end-to-end data-driven approach. However, a systematic top-down approach to evaluate and understand the limitations of GNNs in accurately capturing crystal structures has yet to be established. In this study, we introduce an approach using human-designed descriptors as a compendium of human knowledge to investigate the extent to which GNNs can comprehend crystal structures. Our findings reveal that current state-of-the-art GNNs fall short in accurately capturing the periodicity of crystal structures. We analyze this failure by exploring three aspects: local expressive power, long-range information processing, and readout function. To address these identified limitations, we propose a straightforward and general solution: the hybridization of descriptors with GNNs, which directly supplements the missing information to GNNs. The hybridization enhances the predictive accuracy of GNNs for specific material properties, most notably phonon internal energy and heat capacity, which heavily rely on the periodicity of materials. |
| first_indexed | 2024-09-23T12:47:59Z |
| format | Article |
| id | mit-1721.1/154281 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2025-02-19T04:21:48Z |
| publishDate | 2024 |
| publisher | American Association for the Advancement of Science |
| record_format | dspace |
| spelling | mit-1721.1/1542812025-01-07T04:42:48Z Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity Gong, Sheng Yan, Keqiang Xie, Tian Shao-Horn, Yang Gomez-Bombarelli, Rafael Ji, Shuiwang Grossman, Jeffrey C. Massachusetts Institute of Technology. Department of Materials Science and Engineering Massachusetts Institute of Technology. Department of Mechanical Engineering Graph neural networks (GNNs) have recently been used to learn the representations of crystal structures through an end-to-end data-driven approach. However, a systematic top-down approach to evaluate and understand the limitations of GNNs in accurately capturing crystal structures has yet to be established. In this study, we introduce an approach using human-designed descriptors as a compendium of human knowledge to investigate the extent to which GNNs can comprehend crystal structures. Our findings reveal that current state-of-the-art GNNs fall short in accurately capturing the periodicity of crystal structures. We analyze this failure by exploring three aspects: local expressive power, long-range information processing, and readout function. To address these identified limitations, we propose a straightforward and general solution: the hybridization of descriptors with GNNs, which directly supplements the missing information to GNNs. The hybridization enhances the predictive accuracy of GNNs for specific material properties, most notably phonon internal energy and heat capacity, which heavily rely on the periodicity of materials. 2024-04-25T13:49:25Z 2024-04-25T13:49:25Z 2023-11-10 2024-04-25T13:43:44Z Article http://purl.org/eprint/type/JournalArticle 2375-2548 https://hdl.handle.net/1721.1/154281 Sheng Gong et al. ,Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity.Sci. Adv.9, eadi3245 (2023). en 10.1126/sciadv.adi3245 Science Advances Creative Commons Attribution-Noncommercial http://creativecommons.org/licenses/by-nc/4.0/ application/pdf American Association for the Advancement of Science American Association for the Advancement of Science |
| spellingShingle | Gong, Sheng Yan, Keqiang Xie, Tian Shao-Horn, Yang Gomez-Bombarelli, Rafael Ji, Shuiwang Grossman, Jeffrey C. Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title | Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title_full | Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title_fullStr | Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title_full_unstemmed | Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title_short | Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity |
| title_sort | examining graph neural networks for crystal structures limitations and opportunities for capturing periodicity |
| url | https://hdl.handle.net/1721.1/154281 |
| work_keys_str_mv | AT gongsheng examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT yankeqiang examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT xietian examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT shaohornyang examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT gomezbombarellirafael examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT jishuiwang examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity AT grossmanjeffreyc examininggraphneuralnetworksforcrystalstructureslimitationsandopportunitiesforcapturingperiodicity |