Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction
Machine learning has grown in use for constructing potential energy surfaces for their ability to theoretically recreate any function given enough training as well as their fast predictive powers after being trained. When trained on ab initio data, this enables simulation of a large number of ab-ini...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-06-01
|
| Series: | Carbon Trends |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2667056923000123 |
| _version_ | 1797809079924555776 |
|---|---|
| author | Kazuumi Fujioka Eric Lam Brandon Loi Rui Sun |
| author_facet | Kazuumi Fujioka Eric Lam Brandon Loi Rui Sun |
| author_sort | Kazuumi Fujioka |
| collection | DOAJ |
| description | Machine learning has grown in use for constructing potential energy surfaces for their ability to theoretically recreate any function given enough training as well as their fast predictive powers after being trained. When trained on ab initio data, this enables simulation of a large number of ab-initio-quality trajectories. Here, rigorous benchmarking of these machine-learned potential energy surfaces—both in terms of their static errors and dynamics errors—is carried out for the HBr+ + HCl system. In a novel comparison, both neural networks and a kernel regression method are compared for a global potential energy surface, covering multiple dissociation channels. Further, comparison with ab initio molecular dynamics simulations enables one of the first direct comparisons of dynamic, ensemble-average properties of the system. Finally, comparison with experimental results reveals remarkable agreement for the sGDML method for training sets of thousands to tens of thousands of molecular configurations. |
| first_indexed | 2024-03-13T06:47:13Z |
| format | Article |
| id | doaj.art-44c236f6de214cf7b4eaea86c82cf7a6 |
| institution | Directory Open Access Journal |
| issn | 2667-0569 |
| language | English |
| last_indexed | 2024-03-13T06:47:13Z |
| publishDate | 2023-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Carbon Trends |
| spelling | doaj.art-44c236f6de214cf7b4eaea86c82cf7a62023-06-08T04:20:01ZengElsevierCarbon Trends2667-05692023-06-0111100257Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reactionKazuumi Fujioka0Eric Lam1Brandon Loi2Rui Sun3University of Hawaii at Manoa, Department of Chemistry, 2545 McCarthy Mall, Honolulu, HI 96822-2275, United StatesUniversity of Hawaii at Manoa, Department of Chemistry, 2545 McCarthy Mall, Honolulu, HI 96822-2275, United StatesUniversity of Hawaii at Manoa, Department of Chemistry, 2545 McCarthy Mall, Honolulu, HI 96822-2275, United StatesCorresponding author.; University of Hawaii at Manoa, Department of Chemistry, 2545 McCarthy Mall, Honolulu, HI 96822-2275, United StatesMachine learning has grown in use for constructing potential energy surfaces for their ability to theoretically recreate any function given enough training as well as their fast predictive powers after being trained. When trained on ab initio data, this enables simulation of a large number of ab-initio-quality trajectories. Here, rigorous benchmarking of these machine-learned potential energy surfaces—both in terms of their static errors and dynamics errors—is carried out for the HBr+ + HCl system. In a novel comparison, both neural networks and a kernel regression method are compared for a global potential energy surface, covering multiple dissociation channels. Further, comparison with ab initio molecular dynamics simulations enables one of the first direct comparisons of dynamic, ensemble-average properties of the system. Finally, comparison with experimental results reveals remarkable agreement for the sGDML method for training sets of thousands to tens of thousands of molecular configurations.http://www.sciencedirect.com/science/article/pii/S2667056923000123Molecular dynamicsAb initio molecular dynamicsMachine learningNeural networkKernel regressionBimolecular reaction |
| spellingShingle | Kazuumi Fujioka Eric Lam Brandon Loi Rui Sun Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction Carbon Trends Molecular dynamics Ab initio molecular dynamics Machine learning Neural network Kernel regression Bimolecular reaction |
| title | Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction |
| title_full | Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction |
| title_fullStr | Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction |
| title_full_unstemmed | Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction |
| title_short | Ab initio molecular dynamics benchmarking study of machine-learned potential energy surfaces for the HBr+ + HCl reaction |
| title_sort | ab initio molecular dynamics benchmarking study of machine learned potential energy surfaces for the hbr hcl reaction |
| topic | Molecular dynamics Ab initio molecular dynamics Machine learning Neural network Kernel regression Bimolecular reaction |
| url | http://www.sciencedirect.com/science/article/pii/S2667056923000123 |
| work_keys_str_mv | AT kazuumifujioka abinitiomoleculardynamicsbenchmarkingstudyofmachinelearnedpotentialenergysurfacesforthehbrhclreaction AT ericlam abinitiomoleculardynamicsbenchmarkingstudyofmachinelearnedpotentialenergysurfacesforthehbrhclreaction AT brandonloi abinitiomoleculardynamicsbenchmarkingstudyofmachinelearnedpotentialenergysurfacesforthehbrhclreaction AT ruisun abinitiomoleculardynamicsbenchmarkingstudyofmachinelearnedpotentialenergysurfacesforthehbrhclreaction |