Kinetic energy densities based on the fourth order gradient expansion: performance in different classes of materials and improvement via machine learning

Abstract

We study the performance of fourth-order gradient expansions of the kinetic energy density (KED) in semi-local kinetic energy functionals depending on the density-dependent variables. The formal fourth-order expansion is convergent for periodic systems and small molecules but does not improve over the second-order expansion (Thomas-Fermi term plus one-ninth of von Weizsäcker term). Linear fitting of the expansion coefficients somewhat improves on the formal expansion. The tuning of the fourth order expansion coefficients allows for better reproducibility of Kohn-Sham kinetic energy density than the tuning of the second-order expansion coefficients alone. The possibility of a much more accurate match with the Kohn-Sham kinetic energy density by using neural networks trained using the terms of the 4th order expansion as density-dependent variables is demonstrated. We obtain ultra-low fitting errors without overfitting. Small single hidden layer neural networks can provide good accuracy in separate KED fits of each compound, while for joint fitting of KEDs of multiple compounds multiple hidden layers were required to achieve good fit quality. The critical issue of data distribution is highlighted. We also show the critical role of pseudopotentials in the performance of the expansion, where in the case of a too rapid decay of the valence density at the nucleus with some pseudopotentials, numeric instabilities arise.

arXiv comment37 pages, 20 figures

References in corpus (17)

• Separable Dual Space Gaussian Pseudo-potentials
• Climbing the Density Functional Ladder: Non-Empirical Meta-Generalized Gradient Approximation Designed for Molecules and Solids
• Ab initio pseudopotentials for electronic structure calculations of poly-atomic systems using density-functional theory
• By-passing the Kohn-Sham equations with machine learning
• Laplacian-level density functionals for the kinetic energy density and exchange-correlation energy
• Orbital-free Bond Breaking via Machine Learning
• Issues and Challenges in Orbital-free Density Functional Calculations
• Non-empirical Generalized Gradient Approximation Free Energy Functional for Orbital-free Simulations
• A Simple Generalized Gradient Approximation for the Non-interacting Kinetic Energy Density Functional
• Pure density functional for strong correlations and the thermodynamic limit from machine learning
• Orbital-Free Density Functional Theory Implementation with the Projector Augmented-Wave Method
• Semilocal Pauli-Gaussian Kinetic Functionals for Orbital-Free Density Functional Theory Calculations of Solids
• Laplacian-level kinetic energy approximations based on the fourth-order gradient expansion: Global assessment and application to the subsystem formulation of density functional theory
• Laplacian-dependent models of the kinetic energy density: Applications in subsystem density functional theory with meta-generalized gradient approximation functionals
• Frank Discussion of the Status of Ground-state Orbital-free DFT
• Visualization and analysis of the Kohn-Sham kinetic energy density and its orbital-free description in molecules
• Orbital-free DFT study of the energetics of vacancy clustering and prismatic dislocation loop nucleation in aluminum

Cited by in corpus (20)

• Perspective on integrating machine learning into computational chemistry and materials science
• Deep learning the Hohenberg-Kohn maps of Density Functional Theory
• Data-driven kinetic energy density fitting for orbital-free DFT: linear vs Gaussian process regression
• Laplacian-Level Quantum Hydrodynamic Theory for Plasmonics
• Deep Neural Network Computes Electron Densities and Energies of a Large Set of Organic Molecules Faster than Density Functional Theory (DFT)
• Random Sampling High Dimensional Model Representation Gaussian Process Regression (RS-HDMR-GPR) for representing multidimensional functions with machine-learned lower-dimensional terms allowing insight with a general method
• Neural network with optimal neuron activation functions based on additive Gaussian process regression
• On the optimization of hyperparameters in Gaussian process regression with the help of low-order high-dimensional model representation
• Non-parametric Local Pseudopotentials with Machine Learning: a Tin Pseudopotential Built Using Gaussian Process Regression
• Easy representation of multivariate functions with low-dimensional terms via Gaussian process regression kernel design: applications to machine learning of potential energy surfaces and kinetic energy densities from sparse data
• Grad DFT: a software library for machine learning enhanced density functional theory
• Automatic Differentiation for Orbital-Free Density Functional Theory
• Efficient Learning of a One-dimensional Density Functional Theory
• KineticNet: Deep learning a transferable kinetic energy functional for orbital-free density functional theory
• Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems
• Machine learning of kinetic energy densities with target and feature averaging: better results with fewer training data
• A machine-learned kinetic energy model for light weight metals and compounds of group III-V elements
• Machine learning-guided construction of an analytic kinetic energy functional for orbital free density functional theory
• Physical machine learning outperforms "human learning" in Quantum Chemistry
• A machine learning approach to predict L-edge x-ray absorption spectra of light transition metal ion compounds