Machine-learning Interatomic Potentials

An Atomic Cluster Expansion Potential for Twisted Multilayer Graphene

We have developed an Atomic Cluster Expansion (ACE) potential for simulating twisted multilayer graphene and tested it on a range of simulation tasks. We proposed an approach to construct training and test datasets that incorporates all possible twist angles and local stacking, including incommensurate ones. To achieve this, we generated configurations with periodic boundary conditions suitable for DFT calculations, and then introduced an internal twist and shift within those supercell structures. We further refined the dataset through active learning filtering, guided by Bayesian uncertainty quantification. We validated our model for accuracy and robustness through a wide range of numerical tests.

Y. Wang, D. Clark, S. Das, Z. Zhu, D. Massatt, V. Gavini, M. Luskin, and C. Ortner. An atomic cluster expansion potential for twisted multilayer graphene. arXiv2506.15061, 2025.