Equivariant message passing for the prediction of tensorial properties and molecular spectra

Abstract

Message passing neural networks have become a method of choice for learning on graphs, in particular the prediction of chemical properties and the acceleration of molecular dynamics studies. While they readily scale to large training data sets, previous approaches have proven to be less data efficient than kernel methods. We identify limitations of invariant representations as a major reason and extend the message passing formulation to rotationally equivariant representations. On this basis, we propose the polarizable atom interaction neural network (PaiNN) and improve on common molecule benchmarks over previous networks, while reducing model size and inference time. We leverage the equivariant atomwise representations obtained by PaiNN for the prediction of tensorial properties. Finally, we apply this to the simulation of molecular spectra, achieving speedups of 4-5 orders of magnitude compared to the electronic structure reference.

Accepted at ICML 2021

Cited by in corpus (43)

• E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials
• DeePMD-kit v2: A software package for Deep Potential models
• SpookyNet: Learning Force Fields with Electronic Degrees of Freedom and Nonlocal Effects
• A Comprehensive Survey on Deep Graph Representation Learning
• Inverse design of 3d molecular structures with conditional generative neural networks
• Perspective on integrating machine learning into computational chemistry and materials science
• Evaluation of the MACE Force Field Architecture: from Medicinal Chemistry to Materials Science
• AdsorbML: A Leap in Efficiency for Adsorption Energy Calculations using Generalizable Machine Learning Potentials
• Advances of Machine Learning in Materials Science: Ideas and Techniques
• Informing Geometric Deep Learning with Electronic Interactions to Accelerate Quantum Chemistry
• Excited state, non-adiabatic dynamics of large photoswitchable molecules using a chemically transferable machine learning potential
• SchNetPack 2.0: A neural network toolbox for atomistic machine learning
• Equivariant graph neural networks for fast electron density estimation of molecules, liquids, and solids
• Equivariant Graph Neural Networks for 3D Macromolecular Structure
• GeoMol: Torsional Geometric Generation of Molecular 3D Conformer Ensembles
• Predicting Properties of Periodic Systems from Cluster Data: A Case Study of Liquid Water
• Fast and Sample-Efficient Interatomic Neural Network Potentials for Molecules and Materials Based on Gaussian Moments
• On the design space between molecular mechanics and machine learning force fields
• Machine Learning and Data-Driven Methods in Computational Surface and Interface Science
• Modeling electronic response properties with an explicit-electron machine learning potential
• Accelerating the prediction of inorganic surfaces with machine learning interatomic potentials
• Data efficiency and extrapolation trends in neural network interatomic potentials
• SPA${}^{\mathrm{H}}$M: the Spectrum of Approximated Hamiltonian Matrices representations
• Reducing Training Data Needs with Minimal Multilevel Machine Learning (M3L)
• GPU-Accelerated Approximate Kernel Method for Quantum Machine Learning
• Optimized Multifidelity Machine Learning for Quantum Chemistry
• Similarity Equivariant Graph Neural Networks for Homogenization of Metamaterials
• Geometric and Physical Quantities Improve E(3) Equivariant Message Passing
• Molecular-orbital-based Machine Learning for Open-shell and Multi-reference Systems with Kernel Addition Gaussian Process Regression
• Beyond Independent Error Assumptions in Large GNN Atomistic Models
• Towards Symbolic XAI -- Explanation Through Human Understandable Logical Relationships Between Features
• Coherent energy and force uncertainty in deep learning force fields
• Molecular Dipole Moment Learning via Rotationally Equivariant Gaussian Process Regression with Derivatives in Molecular-orbital-based Machine Learning
• PEMP: Leveraging Physics Properties to Enhance Molecular Property Prediction
• Molecular relaxation by reverse diffusion with time step prediction
• Scaling Graph Neural Networks to Large Proteins
• Molecular Dynamics and Machine Learning Unlock Possibilities in Beauty Design -- A Perspective
• A simple equivariant machine learning method for dynamics based on scalars
• Do Graph Neural Networks Work for High Entropy Alloys?
• FastCHGNet: Training one Universal Interatomic Potential to 1.5 Hours with 32 GPUs
• Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model
• Learning Atomic Multipoles: Prediction of the Electrostatic Potential with Equivariant Graph Neural Networks
• Detect the Interactions that Matter in Matter: Geometric Attention for Many-Body Systems