Machine learning potentials for metal-organic frameworks using an incremental learning approach
Abstract Computational modeling of physical processes in metal-organic frameworks (MOFs) is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior. Density functional theory (DFT) may describe interatomic interactions at the qua...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-02-01
|
| Series: | npj Computational Materials |
| Online Access: | https://doi.org/10.1038/s41524-023-00969-x |
| _version_ | 1811165755948924928 |
|---|---|
| author | Sander Vandenhaute Maarten Cools-Ceuppens Simon DeKeyser Toon Verstraelen Veronique Van Speybroeck |
| author_facet | Sander Vandenhaute Maarten Cools-Ceuppens Simon DeKeyser Toon Verstraelen Veronique Van Speybroeck |
| author_sort | Sander Vandenhaute |
| collection | DOAJ |
| description | Abstract Computational modeling of physical processes in metal-organic frameworks (MOFs) is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior. Density functional theory (DFT) may describe interatomic interactions at the quantum mechanical level, but is computationally too expensive for systems beyond the nanometer and picosecond range. Herein, we propose an incremental learning scheme to construct accurate and data-efficient machine learning potentials for MOFs. The scheme builds on the power of equivariant neural network potentials in combination with parallelized enhanced sampling and on-the-fly training to simultaneously explore and learn the phase space in an iterative manner. With only a few hundred single-point DFT evaluations per material, accurate and transferable potentials are obtained, even for flexible frameworks with multiple structurally different phases. The incremental learning scheme is universally applicable and may pave the way to model framework materials in larger spatiotemporal windows with higher accuracy. |
| first_indexed | 2024-04-10T15:42:42Z |
| format | Article |
| id | doaj.art-262edf4f921e4cc1a18f2bc43d156179 |
| institution | Directory Open Access Journal |
| issn | 2057-3960 |
| language | English |
| last_indexed | 2024-04-10T15:42:42Z |
| publishDate | 2023-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Computational Materials |
| spelling | doaj.art-262edf4f921e4cc1a18f2bc43d1561792023-02-12T12:17:53ZengNature Portfolionpj Computational Materials2057-39602023-02-01911810.1038/s41524-023-00969-xMachine learning potentials for metal-organic frameworks using an incremental learning approachSander Vandenhaute0Maarten Cools-Ceuppens1Simon DeKeyser2Toon Verstraelen3Veronique Van Speybroeck4Center for Molecular Modeling, Ghent UniversityCenter for Molecular Modeling, Ghent UniversityCenter for Molecular Modeling, Ghent UniversityCenter for Molecular Modeling, Ghent UniversityCenter for Molecular Modeling, Ghent UniversityAbstract Computational modeling of physical processes in metal-organic frameworks (MOFs) is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior. Density functional theory (DFT) may describe interatomic interactions at the quantum mechanical level, but is computationally too expensive for systems beyond the nanometer and picosecond range. Herein, we propose an incremental learning scheme to construct accurate and data-efficient machine learning potentials for MOFs. The scheme builds on the power of equivariant neural network potentials in combination with parallelized enhanced sampling and on-the-fly training to simultaneously explore and learn the phase space in an iterative manner. With only a few hundred single-point DFT evaluations per material, accurate and transferable potentials are obtained, even for flexible frameworks with multiple structurally different phases. The incremental learning scheme is universally applicable and may pave the way to model framework materials in larger spatiotemporal windows with higher accuracy.https://doi.org/10.1038/s41524-023-00969-x |
| spellingShingle | Sander Vandenhaute Maarten Cools-Ceuppens Simon DeKeyser Toon Verstraelen Veronique Van Speybroeck Machine learning potentials for metal-organic frameworks using an incremental learning approach npj Computational Materials |
| title | Machine learning potentials for metal-organic frameworks using an incremental learning approach |
| title_full | Machine learning potentials for metal-organic frameworks using an incremental learning approach |
| title_fullStr | Machine learning potentials for metal-organic frameworks using an incremental learning approach |
| title_full_unstemmed | Machine learning potentials for metal-organic frameworks using an incremental learning approach |
| title_short | Machine learning potentials for metal-organic frameworks using an incremental learning approach |
| title_sort | machine learning potentials for metal organic frameworks using an incremental learning approach |
| url | https://doi.org/10.1038/s41524-023-00969-x |
| work_keys_str_mv | AT sandervandenhaute machinelearningpotentialsformetalorganicframeworksusinganincrementallearningapproach AT maartencoolsceuppens machinelearningpotentialsformetalorganicframeworksusinganincrementallearningapproach AT simondekeyser machinelearningpotentialsformetalorganicframeworksusinganincrementallearningapproach AT toonverstraelen machinelearningpotentialsformetalorganicframeworksusinganincrementallearningapproach AT veroniquevanspeybroeck machinelearningpotentialsformetalorganicframeworksusinganincrementallearningapproach |