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.

Coarse-Grained Density Functional Theory Predictions via Deep Kernel Learning

Deepks: A Comprehensive Data-Driven Approach Toward Chemically Accurate Density Functional Theory

Accelerating finite-temperature Kohn-Sham density functional theory with deep neural networks

A deep learning framework to emulate density functional theory

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

Ensemble learning of coarse-grained molecular dynamics force fields with a kernel approach.

Deep Neural Network Computes Electron Densities and Energies of a Large Set of Organic Molecules Faster than Density Functional Theory (DFT)

Density Functionals with Quantum Chemical Accuracy: From Machine Learning to Molecular Dynamics

Learning Equivariant Non-Local Electron Density Functionals

Prediction of Electronic Properties of Radical-Containing Polymers at Coarse-Grained Resolutions

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

Subsystem Density Functional Theory Augmented by a Delta Learning Approach to Achieve Kohn–Sham Accuracy

D4FT: A Deep Learning Approach to Kohn-Sham Density Functional Theory

Kernel charge equilibration: efficient and accurate prediction of molecular dipole moments with a machine-learning enhanced electron density model

Machine-learned approximations to Density Functional Theory Hamiltonians

KineticNet: Deep learning a transferable kinetic energy functional for orbital-free density functional theory

Learning local and semi-local density functionals from exact exchange-correlation potentials and energies

NeuralSCF: Neural network self-consistent fields for density functional theory

Deep learning density functionals for gradient descent optimization

Learning Generalized Residual Exchange-Correlation-Uncertain Functional for Density Functional Theory