Kohn-Sham equations as regularizer: building prior knowledge into machine-learned physics

Abstract

Including prior knowledge is important for effective machine learning models in physics, and is usually achieved by explicitly adding loss terms or constraints on model architectures. Prior knowledge embedded in the physics computation itself rarely draws attention. We show that solving the Kohn-Sham equations when training neural networks for the exchange-correlation functional provides an implicit regularization that greatly improves generalization. Two separations suffice for learning the entire one-dimensional H${}_2$ dissociation curve within chemical accuracy, including the strongly correlated region. Our models also generalize to unseen types of molecules and overcome self-interaction error.

References in corpus (23)

• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Neural Message Passing for Quantum Chemistry
• Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• By-passing the Kohn-Sham equations with machine learning
• Finding Density Functionals with Machine Learning
• Learning data driven discretizations for partial differential equations
• Deep neural network solution of the electronic Schrödinger equation
• Ab-Initio Solution of the Many-Electron Schrödinger Equation with Deep Neural Networks
• Differentiable Programming Tensor Networks
• Pure density functional for strong correlations and the thermodynamic limit from machine learning
• Machine Learning the Physical Non-Local Exchange-Correlation Functional of Density-Functional Theory
• Gedanken Densities and Exact Constraints in Density Functional Theory
• Automatic differentiation in quantum chemistry with an application to fully variational Hartree-Fock
• Quantum deep field: data-driven wave function, electron density generation, and atomization energy prediction and extrapolation with machine learning
• Numerical Methods for the Inverse Problem of Density Functional Theory
• One-dimensional Continuum Electronic Structure with the Density Matrix Renormalization Group and Its Implications For Density Functional Theory
• Neural-network Kohn-Sham exchange-correlation potential and its out-of-training transferability
• Deep learning the Hohenberg-Kohn maps of Density Functional Theory
• Guaranteed convergence of the Kohn-Sham equations
• Deep Learning-Enhanced Variational Monte Carlo Method for Quantum Many-Body Physics
• Automatic differentiation of dominant eigensolver and its applications in quantum physics
• One Dimensional Mimicking of Electronic Structure: The Case for Exponentials

Cited by in corpus (55)

• Machine learning accelerated computational fluid dynamics
• Perspective on integrating machine learning into computational chemistry and materials science
• Machine learning and density functional theory
• Learning the exchange-correlation functional from nature with fully differentiable density functional theory
• Informing Geometric Deep Learning with Electronic Interactions to Accelerate Quantum Chemistry
• SchNetPack 2.0: A neural network toolbox for atomistic machine learning
• Inverse Kohn-Sham Density Functional Theory: Progress and Challenges
• Deep-learning density functional perturbation theory
• Nuclear energy density functionals from machine learning
• Neural functional theory for inhomogeneous fluids: Fundamentals and applications
• DQC: a Python program package for Differentiable Quantum Chemistry
• Evolving symbolic density functionals
• Differentiable quantum chemistry with PySCF for molecules and materials at the mean-field level and beyond
• Overcoming the Barrier of Orbital-Free Density Functional Theory for Molecular Systems Using Deep Learning
• CIDER: An Expressive, Nonlocal Feature Set for Machine Learning Density Functionals with Exact Constraints
• Machine learning the derivative discontinuity of density-functional theory
• Known Operator Learning and Hybrid Machine Learning in Medical Imaging -- A Review of the Past, the Present, and the Future
• Machine learning of a density functional for anisotropic patchy particles
• Linear Jacobi-Legendre expansion of the charge density for machine learning-accelerated electronic structure calculations
• Machine Learning Universal Bosonic Functionals
• Reliable emulation of complex functionals by active learning with error control
• Why neural functionals suit statistical mechanics
• Hyper-density functional theory of soft matter
• Large Scale Distributed Linear Algebra With Tensor Processing Units
• Grad DFT: a software library for machine learning enhanced density functional theory
• Automatic Differentiation for Orbital-Free Density Functional Theory
• Neural density functional theory of liquid-gas phase coexistence
• Neural density functionals: Local learning and pair-correlation matching
• Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems
• Variational principle to regularize machine-learned density functionals: the non-interacting kinetic-energy functional
• Inverse molecular design and parameter optimization with Hückel theory using automatic differentiation
• Mean-Field Density Matrix Decompositions
• Addressing the Band Gap Problem with a Machine-Learned Exchange Functional
• Inverse Hamiltonian design by automatic differentiation
• Why hyperdensity functionals describe any equilibrium observable
• How Well Does Kohn-Sham Regularizer Work for Weakly Correlated Systems?
• Differentiable Physics: A Position Piece
• Flow-Loss: Learning Cardinality Estimates That Matter
• Conditional probability density functional theory
• Physics-Integrated Variational Autoencoders for Robust and Interpretable Generative Modeling
• Machine learning for accuracy in density functional approximations
• AD-NEGF: An End-to-End Differentiable Quantum Transport Simulator for Sensitivity Analysis and Inverse Problems
• Numerical stability and efficiency of response property calculations in density functional theory
• Deep learning density functionals for gradient descent optimization
• Deep learning nonlocal and scalable energy functionals for quantum Ising models
• Using differentiable programming to obtain an energy and density-optimized exchange-correlation functional
• A composable autoencoder-based iterative algorithm for accelerating numerical simulations
• Orbital Mixer: Using Atomic Orbital Features for Basis Dependent Prediction of Molecular Wavefunctions
• INQ, a modern GPU-accelerated computational framework for (time-dependent) density functional theory
• Leveraging Normalizing Flows for Orbital-Free Density Functional Theory
• Calculating potential energy surfaces with quantum computers by measuring only the density along adiabatic transitions
• Solving deep-learning density functional theory via variational autoencoders
• Algorithmic differentiation for plane-wave DFT: materials design, error control and learning model parameters
• Machine learning approach for vibronically renormalized electronic band structures
• Quantum-Enhanced Neural Exchange-Correlation Functionals