Neural-network Kohn-Sham exchange-correlation potential and its out-of-training transferability

Abstract

We incorporate in the Kohn-Sham self consistent equation a trained neural-network projection from the charge density distribution to the Hartree-exchange-correlation potential $n\to V_{\mathrm{Hxc}}$ for possible numerical approach to the exact Kohn-Sham scheme. The potential trained through a newly developed scheme enables us to evaluate the total energy without explicitly treating the formula of the exchange-correlation energy. With a case study of a simple model we show that the well-trained neural-network $V_{\mathrm{Hxc}}$ achieves accuracy for the charge density and total energy out of the model parameter range used for the training, indicating that the property of the elusive ideal functional form of $V_{\mathrm{Hxc}}$ can approximately be encapsulated by the machine-learning construction. We also exemplify a factor that crucially limits the transferability--the boundary in the model parameter space where the number of the one-particle bound states changes--and see that this is cured by setting the training parameter range across that boundary. The training scheme and insights from the model study apply to more general systems, opening a novel path to numerically efficient Kohn-Sham potential.

arXiv comment7 pages, 6 figures

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Climbing the Density Functional Ladder: Non-Empirical Meta-Generalized Gradient Approximation Designed for Molecules and Solids
• By-passing the Kohn-Sham equations with machine learning
• Finding Density Functionals with Machine Learning
• Understanding and reducing errors in density functional calculations
• Pure density functional for strong correlations and the thermodynamic limit from machine learning
• Density Functional Theory is Not Straying from the Path toward the Exact Functional
• Kohn-Sham calculations with the exact functional
• Guaranteed convergence of the Kohn-Sham equations
• Ground-State Energy as a Simple Sum of Orbital Energies in Kohn-Sham Theory: A Shift in Perspective through a Shift in Potential

Cited by in corpus (26)

• Machine learning and the physical sciences
• Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems
• Opportunities and Challenges for Machine Learning in Materials Science
• Kohn-Sham equations as regularizer: building prior knowledge into machine-learned physics
• Machine Learning the Physical Non-Local Exchange-Correlation Functional of Density-Functional Theory
• Guest Editorial: Special Topic on Data-enabled Theoretical Chemistry
• Inverse Kohn-Sham Density Functional Theory: Progress and Challenges
• Deep learning the Hohenberg-Kohn maps of Density Functional Theory
• Neural functional theory for inhomogeneous fluids: Fundamentals and applications
• Drawing Phase Diagrams of Random Quantum Systems by Deep Learning the Wave Functions
• Learning from the Density to Correct Total Energy and Forces in First Principle Simulations
• Machine Learning Exchange-Correlation Potential in Time Dependent Density Functional Theory
• KS-pies: Kohn-Sham Inversion Toolkit
• Exact constraints and appropriate norms in machine learned exchange-correlation functionals
• Machine learning density functionals from the random-phase approximation
• Why neural functionals suit statistical mechanics
• Hyper-density functional theory of soft matter
• Grad DFT: a software library for machine learning enhanced density functional theory
• Efficient Learning of a One-dimensional Density Functional Theory
• Neural density functionals: Local learning and pair-correlation matching
• Neural density functional theory of liquid-gas phase coexistence
• Why hyperdensity functionals describe any equilibrium observable
• Density functionals and Kohn-Sham potentials with minimal wavefunction preparations on a quantum computer
• Machine-learning Kohn-Sham potential from dynamics in time-dependent Kohn-Sham systems
• Insights into one-body density matrices using deep learning
• Machine learning based prediction of the electronic structure of quasi-one-dimensional materials under strain