Advancing molecular simulation with equivariant interatomic potentials

被引：15

作者：

Batzner, Simon ^{[1
,2
]}

Musaelian, Albert ^{[1
]}

Kozinsky, Boris ^{[1
,3
]}

机构：

[1] Harvard Univ, Cambridge, MA 02138 USA

[2] Google DeepMind, Mountain View, CA 94043 USA

[3] Robert Bosch LLC Res & Technol Ctr, Watertown, MA USA

来源：

NATURE REVIEWS PHYSICS | 2023年 / 5卷 / 08期

关键词：

Compendex;

D O I：

10.1038/s42254-023-00615-x

中图分类号：

O59 [应用物理学];

学科分类号：

摘要：

Deep learning has the potential to accelerate atomistic simulations, but existing models suffer from a lack of robustness, sample efficiency, and accuracy. Simon Batzner, Albert Musaelian, and Boris Kozinsky outline how exploiting the symmetry of Euclidean space offers a new way to address these challenges. Simon Batzner, Albert Musaelian and Boris Kozinsky discuss how exploiting the symmetry of Euclidean space can help tackle challenges in molecular simulations.

引用

页码：437 / 438

页数：2

共 10 条

[1]

Batatia I., 2022, PREPRINT, DOI 10.48550/arXiv.2206.07697

[2] E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials [J].