Understanding Machine-learned Density Functionals
Chemical Physics
2014-05-28 v2 Machine Learning
Computational Physics
Machine Learning
Abstract
Kernel ridge regression is used to approximate the kinetic energy of non-interacting fermions in a one-dimensional box as a functional of their density. The properties of different kernels and methods of cross-validation are explored, and highly accurate energies are achieved. Accurate {\em constrained optimal densities} are found via a modified Euler-Lagrange constrained minimization of the total energy. A projected gradient descent algorithm is derived using local principal component analysis. Additionally, a sparse grid representation of the density can be used without degrading the performance of the methods. The implications for machine-learned density functional approximations are discussed.
Keywords
Cite
@article{arxiv.1404.1333,
title = {Understanding Machine-learned Density Functionals},
author = {Li Li and John C. Snyder and Isabelle M. Pelaschier and Jessica Huang and Uma-Naresh Niranjan and Paul Duncan and Matthias Rupp and Klaus-Robert Müller and Kieron Burke},
journal= {arXiv preprint arXiv:1404.1333},
year = {2014}
}