Semi-local machine-learned kinetic energy density functional with third-order gradients of electron density

Junji Seino, Ryo Kageyama, Mikito Fujinami, Yasuhiro Ikabata, Hiromi Nakai

Research output: Contribution to journalArticle

17 Citations (Scopus)

Abstract

A semi-local kinetic energy density functional (KEDF) was constructed based on machine learning (ML). The present scheme adopts electron densities and their gradients up to third-order as the explanatory variables for ML and the Kohn-Sham (KS) kinetic energy density as the response variable in atoms and molecules. Numerical assessments of the present scheme were performed in atomic and molecular systems, including first- and second-period elements. The results of 37 conventional KEDFs with explicit formulae were also compared with those of the ML KEDF with an implicit formula. The inclusion of the higher order gradients reduces the deviation of the total kinetic energies from the KS calculations in a stepwise manner. Furthermore, our scheme with the third-order gradient resulted in the closest kinetic energies to the KS calculations out of the presented functionals.

Original languageEnglish
Article number241705
JournalJournal of Chemical Physics
Volume148
Issue number24
DOIs
Publication statusPublished - 2018 Jun 28

Fingerprint

Kinetic energy
Carrier concentration
flux density
kinetic energy
machine learning
gradients
Learning systems
functionals
inclusions
deviation
Atoms
Molecules
atoms
molecules

ASJC Scopus subject areas

  • Physics and Astronomy(all)
  • Physical and Theoretical Chemistry

Cite this

Semi-local machine-learned kinetic energy density functional with third-order gradients of electron density. / Seino, Junji; Kageyama, Ryo; Fujinami, Mikito; Ikabata, Yasuhiro; Nakai, Hiromi.

In: Journal of Chemical Physics, Vol. 148, No. 24, 241705, 28.06.2018.

Research output: Contribution to journalArticle

@article{b8c14ff61d6b499682311e9c27fa4ad5,
title = "Semi-local machine-learned kinetic energy density functional with third-order gradients of electron density",
abstract = "A semi-local kinetic energy density functional (KEDF) was constructed based on machine learning (ML). The present scheme adopts electron densities and their gradients up to third-order as the explanatory variables for ML and the Kohn-Sham (KS) kinetic energy density as the response variable in atoms and molecules. Numerical assessments of the present scheme were performed in atomic and molecular systems, including first- and second-period elements. The results of 37 conventional KEDFs with explicit formulae were also compared with those of the ML KEDF with an implicit formula. The inclusion of the higher order gradients reduces the deviation of the total kinetic energies from the KS calculations in a stepwise manner. Furthermore, our scheme with the third-order gradient resulted in the closest kinetic energies to the KS calculations out of the presented functionals.",
author = "Junji Seino and Ryo Kageyama and Mikito Fujinami and Yasuhiro Ikabata and Hiromi Nakai",
year = "2018",
month = "6",
day = "28",
doi = "10.1063/1.5007230",
language = "English",
volume = "148",
journal = "Journal of Chemical Physics",
issn = "0021-9606",
publisher = "American Institute of Physics Publising LLC",
number = "24",

}

TY - JOUR

T1 - Semi-local machine-learned kinetic energy density functional with third-order gradients of electron density

AU - Seino, Junji

AU - Kageyama, Ryo

AU - Fujinami, Mikito

AU - Ikabata, Yasuhiro

AU - Nakai, Hiromi

PY - 2018/6/28

Y1 - 2018/6/28

N2 - A semi-local kinetic energy density functional (KEDF) was constructed based on machine learning (ML). The present scheme adopts electron densities and their gradients up to third-order as the explanatory variables for ML and the Kohn-Sham (KS) kinetic energy density as the response variable in atoms and molecules. Numerical assessments of the present scheme were performed in atomic and molecular systems, including first- and second-period elements. The results of 37 conventional KEDFs with explicit formulae were also compared with those of the ML KEDF with an implicit formula. The inclusion of the higher order gradients reduces the deviation of the total kinetic energies from the KS calculations in a stepwise manner. Furthermore, our scheme with the third-order gradient resulted in the closest kinetic energies to the KS calculations out of the presented functionals.

AB - A semi-local kinetic energy density functional (KEDF) was constructed based on machine learning (ML). The present scheme adopts electron densities and their gradients up to third-order as the explanatory variables for ML and the Kohn-Sham (KS) kinetic energy density as the response variable in atoms and molecules. Numerical assessments of the present scheme were performed in atomic and molecular systems, including first- and second-period elements. The results of 37 conventional KEDFs with explicit formulae were also compared with those of the ML KEDF with an implicit formula. The inclusion of the higher order gradients reduces the deviation of the total kinetic energies from the KS calculations in a stepwise manner. Furthermore, our scheme with the third-order gradient resulted in the closest kinetic energies to the KS calculations out of the presented functionals.

UR - http://www.scopus.com/inward/record.url?scp=85044043266&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044043266&partnerID=8YFLogxK

U2 - 10.1063/1.5007230

DO - 10.1063/1.5007230

M3 - Article

AN - SCOPUS:85044043266

VL - 148

JO - Journal of Chemical Physics

JF - Journal of Chemical Physics

SN - 0021-9606

IS - 24

M1 - 241705

ER -