A biasing potential is derived from the uncertainty of a neural network ensemble and used to modify the potential energy surface in molecular dynamics simulations and facilitate the determination of underrepresented structural regions.
This is a preview of subscription content, log in via an institution to check access.
References
Behler, J. & Parrinello, M. Phys. Rev. Lett. 98, 146401 (2007).
Bartók, A. P., Kondor, R. & Csányi, G. Phys. Rev. B 87, 184115 (2013).
Schütt, K. et al. SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. In 31st Conference on Neural Information Processing Systems 991–1002 (NIPS, 2017).
Smith, J. S., Isayev, O. & Roitberg, A. E. Chem. Sci. 8, 3192–3203 (2017).
Batzner, S. et al. Nat. Commun. 13, 2453 (2022).
Musaelian, A. et al. Nat. Commun. 14, 579 (2023).
Vandermause, J. et al. npj Comput. Mater. 6, 20 (2020).
Kulichenko, M. et al. Nat. Comput. Sci. https://doi.org/10.1038/s43588-023-00406-5 (2023).
Bussi, G. & Laio, A. Nat. Rev. Phys. 2, 200–212 (2020).
Behler, J. Int. J. Quantum Chem. 115, 1032–1050 (2015).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The author declares no competing interests.
Rights and permissions
About this article
Cite this article
Batzner, S. Biasing energy surfaces towards the unknown. Nat Comput Sci 3, 190–191 (2023). https://doi.org/10.1038/s43588-023-00420-7
Published:
Issue Date:
DOI: https://doi.org/10.1038/s43588-023-00420-7
- Springer Nature America, Inc.