Sensitivity and Dimensionality of Atomic Environment Representations used for Machine Learning Interatomic Potentials

Abstract

Faithfully representing chemical environments is essential for describing materials and molecules with machine learning approaches. Here, we present a systematic classification of these representations and then investigate: (i) the sensitivity to perturbations and (ii) the effective dimensionality of a variety of atomic environment representations, and over a range of material datasets. Representations investigated include Atom Centred Symmetry Functions, Chebyshev Polynomial Symmetry Functions (CHSF), Smooth Overlap of Atomic Positions, Many-body Tensor Representation and Atomic Cluster Expansion. In area (i), we show that none of the atomic environment representations are linearly stable under tangential perturbations, and that for CHSF there are instabilities for particular choices of perturbation, which we show can be removed with a slight redefinition of the representation. In area (ii), we find that most representations can be compressed significantly without loss of precision, and further that selecting optimal subsets of a representation method improves the accuracy of regression models built for a given dataset.

References in corpus (23)

• Gaussian Approximation Potentials: the accuracy of quantum mechanics, without the electrons
• Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties
• SchNet - a deep learning architecture for molecules and materials
• AFLOW: An automatic framework for high-throughput materials discovery
• Quantum-Chemical Insights from Deep Tensor Neural Networks
• A Spectral Analysis Method for Automated Generation of Quantum-Accurate Interatomic Potentials
• A Performance and Cost Assessment of Machine Learning Interatomic Potentials
• DScribe: Library of Descriptors for Machine Learning in Materials Science
• SISSO: a compressed-sensing method for identifying the best low-dimensional descriptor in an immensity of offered candidates
• Machine Learning Unifies the Modelling of Materials and Molecules
• Universal Fragment Descriptors for Predicting Electronic Properties of Inorganic Crystals
• Machine learning a general purpose interatomic potential for silicon
• Insightful classification of crystal structures using deep learning
• Efficient and Accurate Machine-Learning Interpolation of Atomic Energies in Compositions with Many Species
• Automatic Selection of Atomic Fingerprints and Reference Configurations for Machine-Learning Potentials
• Achieving DFT accuracy with a machine-learning interatomic potential: thermomechanics and defects in bcc ferromagnetic iron
• A neural network interatomic potential for the phase change material GeTe
• Extending the Accuracy of the SNAP Interatomic Potential Form
• Atom2Vec: learning atoms for materials discovery
• Atom-Density Representations for Machine Learning
• Pattern Learning Electronic Density of States
• NOMAD 2018 Kaggle Competition: Solving Materials Science Challenges Through Crowd Sourcing
• Quantifying Chemical Structure and Atomic Energies in Amorphous Silicon Networks

Cited by in corpus (12)

• Physics-inspired structural representations for molecules and materials
• Machine-learning interatomic potentials for materials science
• Unified Representation of Molecules and Crystals for Machine Learning
• Representations of molecules and materials for interpolation of quantum-mechanical simulations via machine learning
• Efficient implementation of atom-density representations
• A Physically-informed Graph-based Order Parameter for the Universal Characterization of Atomic Structures
• Optimal radial basis for density-based atomic representations
• Mean-Field Density Matrix Decompositions
• A smooth basis for atomistic machine learning
• Structural Descriptors and Information Extraction from X-ray Emission Spectra: Aqueous Sulfuric Acid
• Structure-Driven Prediction of Magnetic Order in Uranium Compounds
• Compressing local atomic neighbourhood descriptors