Abstract
The atomic cluster expansion is a general polynomial expansion of the atomic energy in multi-atom basis functions. Here we implement the atomic cluster expansion in the performant C++ code PACE that is suitable for use in large-scale atomistic simulations. We briefly review the atomic cluster expansion and give detailed expressions for energies and forces as well as efficient algorithms for their evaluation. We demonstrate that the atomic cluster expansion as implemented in PACE shifts a previously established Pareto front for machine learning interatomic potentials toward faster and more accurate calculations. Moreover, general purpose parameterizations are presented for copper and silicon and evaluated in detail. We show that the Cu and Si potentials significantly improve on the best available potentials for highly accurate large-scale atomistic simulations.
Highlights
Atomistic modeling and simulation requires efficient computation of energies and forces
We investigated in detail the 2D hcp lattice; both embedded atom method (EAM) and spectral neighbor analysis potential (SNAP) potentials show dynamic instabilities related to out-of-plane atomic displacements for the 3 × 3 × 1 supercell that we used in our calculations (Supplementary Fig. 2)
We present a performant implementation of atomic cluster expansion (ACE) in the form of the PACE code
Summary
Atomistic modeling and simulation requires efficient computation of energies and forces. The ML models construct representations of atomic structure that are used in various regression algorithms to predict energies and forces. The recently developed atomic cluster expansion (ACE)[1] provides a complete and efficient representation of atomic properties as a function of the local atomic environment in terms of many-body correlation functions. The details of how these were constructed are provided in the Supplementary Methods While these benchmarks establish advanced computational performance, we demonstrate the capacity of the ACE framework to develop highly accurate parameterizations: we present two parameterizations of interatomic potentials for Cu and Si that outperform available ML-based potentials in terms of performance, accuracy, and generalizability
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
