Abstract:In this work we apply methods for describing three-dimensional images to the problem of encoding atomic environments in a way that is invariant to rotations, translations, and permutations of the atoms and, crucially, can be decoded back into the original environment modulo global orientation without the need for training a model. From the point of view of decoding, the descriptor is optimally complete and can be extended to arbitrary order, allowing for a systematic convergence of the fidelity of the description. In experiments on molecules ranging from 3 to 29 atoms in size, we demonstrate that positions can be decoded with a 97% success rate and positions plus species with a 70% rate of success, rising to 95% if a second fingerprint is used. In all cases, consistent recovery is observed for molecules with 17 or fewer atoms. Additionally, we evaluate the descriptor's performance in predicting the energies and forces of bulk Ni, Cu, Li, Mo, Si, and Ge by means of a neural network model trained on DFT data. When comparing to six machine learning interaction potential methods that use various descriptors and regression schemes, our descriptor is found to be competitive, in several cases outperforming well established methods. The combined ability to both decode and make property predictions from a representation that does not need to be learned lays the foundations for a novel way of building generative models that are tasked with solving the inverse problem of predicting atomic arrangements that are statistically likely to have certain desired properties. https://doi.org/10.1103/PhysRevB.104.144110 ©2021 American Physical Society

On the Completeness of Atomic Structure Representations

Completeness of Atomic Structure Representations

Through the eyes of a descriptor: Constructing complete, invertible descriptions of atomic environments

Atom-Density Representations for Machine Learning

Unified theory of atom-centered representations and message-passing machine-learning schemes

Complete and Efficient Covariants for 3D Point Configurations with Application to Learning Molecular Quantum Properties

Physically Motivated Recursively Embedded Atom Neural Networks: Incorporating Local Completeness and Nonlocality

Continuous and optimally complete description of chemical environments using Spherical Bessel descriptors

Electronic structure based descriptor for characterizing local atomic environments

Discovering the Building Blocks of Atomic Systems using Machine Learning

Equivariant representations for molecular Hamiltonians and N-center atomic-scale properties

Physics-inspired Equivariant Descriptors of Non-bonded Interactions

Unveiling Intrinsic Many-Body Complexity by Compressing Single-Body Triviality

Local invertibility and sensitivity of atomic structure-feature mappings

Designing hybrid descriptors for improved machine learning models in atomistic materials science simulations

Investigating 3D Atomic Environments for Enhanced QSAR

Decomposing Chemical Space: Applications to the Machine Learning of Atomic Energies

Persistent homology-based descriptor for machine-learning potential

Travelling fronts in an array of coupled symmetric bistable units

Three-dimensional theory for interaction between atomic ensembles and free-space light

Understanding the Structure of QM7b and QM9 Quantum Mechanical Datasets Using Unsupervised Learning