REVIEW 3 cited by
A Performance and Cost Assessment of Machine Learning Interatomic Potentials
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors --- Behler-Parrinello symmetry functions, smooth overlap of atomic positions (SOAP), the Spectral Neighbor Analysis Potential (SNAP) bispectrum components, and moment tensors --- using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model, and consequently computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.
Forward citations
Cited by 3 Pith papers
-
Machine-learned interatomic potential for titanium carbide MXenes: Application to ion irradiation simulations
A machine-learned atomic model for Ti-carbide MXenes reproduces DFT accuracy and yields the first large-scale molecular-dynamics statistics for ion irradiation damage and implantation.
-
A Robust Machine Learned Interatomic Potential for Nb: Collision Cascade Simulations with accurate Defect Configurations
A SNAP machine-learned potential for niobium reproduces the DFT-predicted ⟨111⟩ self-interstitial ground state and yields that orientation in 5 keV collision cascades, while EAM and FS potentials favor ⟨110⟩.
-
Machine-learning interatomic potential for radiation damage and defects in tungsten
A new Gaussian Approximation Potential for tungsten reproduces defect, surface, liquid, and short-range repulsion energetics near DFT accuracy, making it suitable for radiation damage molecular dynamics.
Discussion (0). Continue with ORCID to comment.