A Performance and Cost Assessment of Machine Learning Interatomic Potentials

2019/06/30 by Yunxing Zuo, Chi Chen, Xiangguo Li +10 · 57 citations
Chemistry · Materials Science · Physics and Astronomy · #Crystallography and molecular interactions #Machine Learning in Materials Science #X-ray Diffraction in Crystallography #cond-mat.mtrl-sci #physics.comp-ph

paper · pdf · doi:10.1021/acs.jpca.9b08723

arxiv created 2019/07/24 · openalex publication_date 2020/01/09 · arxiv updated 2020/02/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31

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.

Citations

• Projector augmented-wave method
• Fast Parallel Algorithms for Short-Range Molecular Dynamics
• Constructing high‐dimensional neural network potentials: A tutorial review
• Efficient iterative schemes forab initiototal-energy calculations using a plane-wave basis set
• Self-Consistent Equations Including Exchange and Correlation Effects
• An Electrostatic Spectral Neighbor Analysis Potential (eSNAP) for Lithium Nitride
• DScribe: Library of descriptors for machine learning in materials science

Cited by

• Universally Accurate or Specifically Inadequate? Stress‐Testing General Purpose Machine Learning Interatomic Potentials
• Electrostatic embedding machine learning for ground and excited state molecular dynamics of solvated molecules
• Four Generations of High-Dimensional Neural Network Potentials
• Accuracy of Machine Learning Potential for Predictions of Multiple-Target Physical Properties*
• Site Disorder Drives Cyanide Dynamics and Fast Ion Transport in Li6PS5CN
• Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems
• Gaussian Process Regression for Materials and Molecules
• Machine learning interatomic potentials in biomolecular modeling: principles, architectures, and applications
• Benchmarking Universal Machine Learning Interatomic Potentials for Elastic Property Prediction
• Benchmarking 34 OpenKIM Nickel Potentials with an Emphasis on Surfaces and Extended Defects
• Challenges and opportunities for machine learning potentials in transition path sampling: alanine dipeptide and azobenzene studies
• Active learning and explicit electrostatics enable accurate modeling of electrolytes
• Macro-Dipole-Constrainted Learning of Atomic Charges for Accurate Electrostatic Potentials at Electrochemical Interfaces
• Improved capabilities of the TurboGAP code for radiation induced cascade simulations: an illustration with silicon
• Extrapolation of Machine-Learning Interatomic Potentials for Organic and Polymeric Systems
• SMC-X: A Distributed Scalable Monte Carlo Simulation Method for Chemically Complex Alloys
• Bayesian Neural Networks versus deep ensembles for uncertainty quantification in machine learning interatomic potentials
• Machine-Learning Potentials for Efficient Simulations of Anisotropic Colloids
• Effects of training machine-learning potentials for radiation damage simulations using different pseudopotentials
• Achieving DFT accuracy in short range ordering and stacking fault energy using moment tensor potential for CoCrFeNi and CoCrNi
• Suitability of ReaxFF potential for MD modelling of lithium across low and high temperatures
• Machine learning force fields and coarse-grained variables in molecular dynamics: application to materials and biological systems
• Migration as a Probe: A Generalizable Benchmark Framework for Specialist vs. Generalist Machine-Learned Force Fields
• Evaluating Moment Tensor Potential in Ag-Cu Alloy: Accuracy, Transferability, and Phase Diagram Fidelity
• Atomistic mechanisms of phase transitions in all-temperature barocaloric material KPF₆
• Advancing Material Modeling in Hydrocodes Beyond Equations of State
• Operationalizing Serendipity: Multi-Agent AI Workflows for Enhanced Materials Characterization with Theory-in-the-Loop
• Model Accuracy and Data Heterogeneity Shape Uncertainty Quantification in Machine Learning Interatomic Potentials
• Expert-Guided LLM Reasoning for Battery Discovery: From AI-Driven Hypothesis to Synthesis and Characterization
• Molecular dynamics of liquid–electrode interface by integrating Coulomb interaction into universal neural network potential
• Machine learning interatomic potential for predicting the thermal properties of uranium nitride
• Training Data Set Refinement for the Machine Learning Potential of Li-Si Alloys via Structural Similarity Analysis
• Classical and reactive molecular dynamics: Principles and applications in combustion and energy systems
• A Study on the Fine-Tuning Performance of Universal Machine-Learned Interatomic Potentials (U-MLIPs)
• Long-term atomistic finite-temperature substitutional diffusion
• A Universal Graph Deep Learning Interatomic Potential for the Periodic Table
• Quasicrystal bulk and surface energies from density functional theory
• Machine-learning interatomic potentials from a users perspective: A comparison of accuracy, speed and data efficiency
• Machine-Learned Potentials for Solvation Modeling
• Scaling deep learning for materials discovery
• Path-integral molecular dynamics with actively-trained and universal machine learning force fields
• From oligomers to entangled polymers: How to train a transferable machine learning interatomic potential
• Exploration of amorphous V₂O₅ as cathode for magnesium batteries
• The structure and migration of twin boundaries in tetragonal β-Sn: an application of machine learning based interatomic potentials
• The evolution of machine learning potentials for molecules, reactions and materials
• Machine-Learned Interatomic Potentials for Predicting Physicochemical Properties of Molten Metal-Salt Systems for Calcium Electrolysis
• Interpretable, Physics-Informed Learning Reveals Sulfur Adsorption and Poisoning Mechanisms in 13-Atom Icosahedra Nanoclusters
• AccelNet: Exact backward-compatible acceleration of polynomial angular descriptors through Cartesian moment factorization
• Scalable Data-Driven Basis Selection for Linear Machine Learning Interatomic Potentials
• High thermal conductivity in semiconducting Janus and non-Janus diamanes
• Moment Tensor Potential and Equivariant Tensor Network Potential with explicit dispersion interactions
• Machine learning potentials for multicomponent systems: The Ti-Al binary system
• Bayesian force fields from active learning for simulation of inter-dimensional transformation of stanene
• Learning properties of ordered and disordered materials from multi-fidelity data
• A Physically-informed Graph-based Order Parameter for the Universal Characterization of Atomic Structures
• Dynamically polarisable force-fields for surface simulations via\n multi-output classification Neural Networks
• Neural network potential from bispectrum components: A case study on crystalline silicon

Related

• Machine-learning interatomic potentials from a users perspective: A comparison of accuracy, speed and data efficiency
• Optimal data generation for machine learned interatomic potentials
• Universally Accurate or Specifically Inadequate? Stress‐Testing General Purpose Machine Learning Interatomic Potentials
• Application-specific machine-learned interatomic potentials: exploring the trade-off between DFT convergence, MLIP expressivity, and computational cost
• Beyond potentials: integrated machine-learning models for materials