Skip to Content
MaterialsNeural Network Potentials

Neural Network Potentials

Machine learning interatomic potentials — training, evaluation, and deployment for atomistic simulations.

Tutorials Tools Datasets Awesome Lists

Tutorials

ML for Materials Course Jupyter
Neural network potentials Uncertainty
Transformers for Materials Jupyter
LLMs for materials science

Tools

Universal Potentials

MACE Equivariant architecture with MACE-MP foundation models CHGNet Universal potential for atomistic modeling SevenNet Scalable equivariant interatomic network

Training Frameworks

NequIP E(3)-equivariant interatomic potentials SchNetPack Deep learning for molecules and materials FLARE Fast and accurate interatomic potentials with active learning FitSNAP Training SNAP interatomic potentials NeuralForceField PyTorch-based force field

Simulation & Analysis

ASE Atomic Simulation Environment — interfaces to many codes GPAW Density-functional theory Python code QUIP Software tools for molecular dynamics simulations phonopy Phonon calculations at harmonic levels pymatgen Python Materials Genomics — analysis and manipulation

Cloud Platforms

Matlantis Accelerated materials discovery platform Mat3ra Materials modeling cloud engine

Datasets

Large-Scale Computational Datasets

OMat24 DFT for inorganic crystals (Meta) 110M entries LeMat-Bulk Inorganic material structures 6.7M structures LeMat-Traj Inorganic material trajectories 113M trajectories MatPES Structures from 300K MD simulations ~400K structures MP-ALOE r2SCAN DFT for universal MLIPs ~1M calculations Open Catalyst 2020 Surface relaxations for catalysis 1.2M relaxations Carbon Data Carbon trajectories 22.9M atoms

Awesome Lists

510+ atomistic ML projects  Materials informatics software  Major atomistic simulation engines 
Materials DiscoveryProperty Prediction