Unified Representation of Molecules and Crystals for Machine Learning

Abstract

Accurate simulations of atomistic systems from first principles are limited by computational cost. In high-throughput settings, machine learning can reduce these costs significantly by accurately interpolating between reference calculations. For this, kernel learning approaches crucially require a representation that accommodates arbitrary atomistic systems. We introduce a many-body tensor representation that is invariant to translations, rotations, and nuclear permutations of same elements, unique, differentiable, can represent molecules and crystals, and is fast to compute. Empirical evidence for competitive energy and force prediction errors is presented for changes in molecular structure, crystal chemistry, and molecular dynamics using kernel regression and symmetric gradient-domain machine learning as models. Applicability is demonstrated for phase diagrams of Pt-group/transition-metal binary systems.

Final revised version, major update

References in corpus (20)

• Machine Learning Force Fields
• Big Data of Materials Science - Critical Role of the Descriptor
• DScribe: Library of Descriptors for Machine Learning in Materials Science
• Towards Exact Molecular Dynamics Simulations with Machine-Learned Force Fields
• Machine learning prediction errors better than DFT accuracy
• On-the-fly machine learning force field generation: Application to melting points
• Identifying structural flow defects in disordered solids using machine learning methods
• FCHL revisited: faster and more accurate quantum machine learning
• Alchemical and structural distribution based representation for improved QML
• Equivariant message passing for the prediction of tensorial properties and molecular spectra
• Understanding molecular representations in machine learning: The role of uniqueness and target similarity
• On the Completeness of Atomic Structure Representations
• Gaussian Process Regression for Geometry Optimization
• Representations of molecules and materials for interpolation of quantum-mechanical simulations via machine learning
• Comment on "Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning"
• Atomistic Structure Learning Algorithm with surrogate energy model relaxation
• Sensitivity and Dimensionality of Atomic Environment Representations used for Machine Learning Interatomic Potentials
• Predicting Gas-Particle Partitioning Coefficients of Atmospheric Molecules with Machine Learning
• Machine Learning for compositional disorder: A Comparison Between Different Descriptors and Machine Learning Frameworks
• Conformational analysis of tannic acid: environment effects in electronic and reactivity properties

Cited by in corpus (44)

• SchNet - a deep learning architecture for molecules and materials
• DScribe: Library of Descriptors for Machine Learning in Materials Science
• Machine learning prediction errors better than DFT accuracy
• SchNetPack: A Deep Learning Toolbox For Atomistic Systems
• Atom-Density Representations for Machine Learning
• Machine-learned multi-system surrogate models for materials prediction
• Orbital Graph Convolutional Neural Network for Material Property Prediction
• Optimizing many-body atomic descriptors for enhanced computational performance of machine learning based interatomic potentials
• Updates to the DScribe Library: New Descriptors and Derivatives
• Representing individual electronic states for machine learning GW band structures of 2D materials
• Tutorial: How to Train a Neural Network Potential
• Detection of Topological Materials with Machine Learning
• Convolutional neural networks for atomistic systems
• Prediction of Atomization Energy Using Graph Kernel and Active Learning
• ET-AL: Entropy-Targeted Active Learning for Bias Mitigation in Materials Data
• Decoding the Competing Effects of Dynamic Solvation Structures on Nuclear Magnetic Resonance Chemical Shifts of Battery Electrolytes via Machine Learning
• Lightweight and Effective Tensor Sensitivity for Atomistic Neural Networks
• Linear Jacobi-Legendre expansion of the charge density for machine learning-accelerated electronic structure calculations
• Kernel based quantum machine learning at record rate : Many-body distribution functionals as compact representations
• Size and Temperature Transferability of Direct and Local Deep Neural Networks for Atomic Forces
• Machine Learning for compositional disorder: A Comparison Between Different Descriptors and Machine Learning Frameworks
• Towards Structural Reconstruction from X-Ray Spectra
• Machine learning for phase ordering dynamics of charge density waves
• Machine learning predictions for local electronic properties of disordered correlated electron systems
• Prediction of surface reconstructions using MAGUS
• Inorganic Crystal Structure Prototype Database based on Unsupervised Learning of Local Atomic Environments
• Machine Learning Framework for Modeling Exciton-Polaritons in Molecular Materials
• Efficient ensemble uncertainty estimation in Gaussian Processes Regression
• Electronic Descriptors for Supervised Spectroscopic Predictions
• Efficient dataset generation for machine learning perovskite alloys
• Information Bottleneck in Peptide Conformation Determination by X-ray Absorption Spectroscopy
• Molecular dynamics-driven global tetra-atomic potential energy surfaces: Application to the AlF dimer
• Molecular Hessian matrices from a machine learning random forest regression algorithm
• Kinetics of orbital ordering in cooperative Jahn-Teller models: Machine-learning enabled large-scale simulations
• High-performance descriptor for magnetic materials: Accurate discrimination of magnetic structure
• A Fuzzy Classification Framework to Identify Equivalent Atoms in Complex Materials and Molecules
• Structural Descriptors and Information Extraction from X-ray Emission Spectra: Aqueous Sulfuric Acid
• MADAS -- A Python framework for assessing similarity in materials-science data
• Machine-Learned Potentials for Solvation Modeling
• Implementation of the Emulator-based Component Analysis
• Machine-learning force-field models for dynamical simulations of metallic magnets
• Efficient interpolation of molecular properties across chemical compound space with low-dimensional descriptors
• Machine learning approach for vibronically renormalized electronic band structures
• Integer linear programming for unsupervised training set selection in molecular machine learning