CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling
Lawrence Berkeley National Laboratory · University of California, Berkeley · +2 more institutions
Abstract
Abstract Large-scale simulations with complex electron interactions remain one of the greatest challenges for atomistic modelling. Although classical force fields often fail to describe the coupling between electronic states and ionic rearrangements, the more accurate ab initio molecular dynamics suffers from computational complexity that prevents long-time and large-scale simulations, which are essential to study technologically relevant phenomena. Here we present the Crystal Hamiltonian Graph Neural Network (CHGNet), a graph neural network-based machine-learning interatomic potential (MLIP) that models the universal potential energy surface. CHGNet is pretrained on the energies, forces, stresses and magnetic…
Citation impact
- FWCI
- 66.49
- Percentile
- 100%
- References
- 67
Authors
7- BDBowen DengCorresponding
Lawrence Berkeley National Laboratory, University of California, Berkeley
- PZPeichen Zhong
Lawrence Berkeley National Laboratory, University of California, Berkeley
- KJKyuJung Jun
Lawrence Berkeley National Laboratory, University of California, Berkeley
- JRJanosh Riebesell
Lawrence Berkeley National Laboratory, University of Cambridge
- KHKevin Han
Lawrence Berkeley National Laboratory
Topics & keywords
- Degrees of freedom (physics and chemistry)
- Hamiltonian (control theory)
- Statistical physics
- Ionic bonding
- Electron
- Ab initio
- Molecular dynamics
- Computer science
Funding
- NSNational Science Foundation
- UDU.S. Department of EnergyAwards: KC23MP, DE-AC0205CH11231
- NENational Energy Research Scientific Computing CenterAward: DE-AC0205CH11231
- OOOffice of ScienceAward: DE-AC0205CH11231
- BEBasic Energy SciencesAwards: KC23MP, DE-AC0205CH11231
- LBLawrence Berkeley National LaboratoryAward: DE-AC0205CH11231