Abstract:Having access to accurate electron densities in chemical systems, especially for dynamical systems involving chemical reactions, ion transport, and other charge transfer processes, is crucial for numerous applications in materials chemistry. Traditional methods for computationally predicting electron density data for such systems include quantum mechanical (QM) techniques, such as density functional theory. However, poor scaling of these QM methods restricts their use to relatively small system sizes and short dynamic time scales. To overcome this limitation, we have developed a deep neural network machine learning formalism, which we call deep charge density prediction (DeepCDP), for predicting charge densities by only using atomic positions for molecules and condensed phase (periodic) systems. Our method uses the weighted smooth overlap of atomic positions to fingerprint environments on a grid-point basis and map it to electron density data generated from QM simulations. We trained models for bulk systems of copper, LiF, and silicon; for a molecular system, water; and for two-dimensional charged and uncharged systems, hydroxyl-functionalized graphane, with and without an added proton. We showed that DeepCDP achieves prediction R2 values greater than 0.99 and mean squared error values on the order of 10−5e2 Å−6 for most systems. DeepCDP scales linearly with system size, is highly parallelizable, and is capable of accurately predicting the excess charge in protonated hydroxyl-functionalized graphane. We demonstrate how DeepCDP can be used to accurately track the location of charges (protons) by computing electron densities at a few selected grid points in the materials, thus significantly reducing the computational cost. We also show that our models can be transferable, allowing prediction of electron densities for systems on which it has not been trained but that contain a subset of atomic species on which it has been trained. Our approach can be used to develop models that span different chemical systems and train them for the study of large-scale charge transport and chemical reactions.

Electronic Structure Prediction of Multi-million Atom Systems Through Uncertainty Quantification Enabled Transfer Learning

Uncertainty Quantification and Propagation in Atomistic Machine Learning

Multi-task learning for molecular electronic structure approaching coupled-cluster accuracy

Electronic excited states from physically-constrained machine learning

Electronic Excited States from Physically Constrained Machine Learning

Uncertainty-Quantified Hybrid Machine Learning/Density Functional Theory High Throughput Screening Method for Crystals

Uncertainty quantification by direct propagation of shallow ensembles

Coarse-Grained Density Functional Theory Predictions via Deep Kernel Learning

Evaluating uncertainty-based active learning for accelerating the generalization of molecular property prediction

Machine Learning Electron Density Prediction Using Weighted Smooth Overlap of Atomic Positions

Transferability in Machine Learning for Electronic Structure via the Molecular Orbital Basis

Bypassing the computational bottleneck of quantum-embedding theories for strong electron correlations with machine learning

Predicting The One-Particle Density Matrix With Machine Learning

Fast and Accurate Uncertainty Estimation in Chemical Machine Learning

Quantum Hardware-Enabled Molecular Dynamics via Transfer Learning

Linear Jacobi-Legendre expansion of the charge density for machine learning-accelerated electronic structure calculations

Hierarchical Transfer Learning: An Agile and Equitable Strategy for Machine-Learning Interatomic Models

Electronic structure prediction of medium and high entropy alloys across composition space

MLTB: Enhancing Transferability and Extensibility of Density Functional Tight Binding Theory with Multi-body Interaction Corrections

Constructing Accurate and Efficient General-Purpose Atomistic Machine Learning Model with Transferable Accuracy for Quantum Chemistry