Restricted Boltzmann machine learning for solving strongly correlated quantum systems

Yusuke Nomura, Andrew S. Darmawan, Youhei Yamaji, Masatoshi Imada

Research output: Contribution to journalArticle

43 Citations (Scopus)

Abstract

We develop a machine learning method to construct accurate ground-state wave functions of strongly interacting and entangled quantum spin as well as fermionic models on lattices. A restricted Boltzmann machine algorithm in the form of an artificial neural network is combined with a conventional variational Monte Carlo method with pair product (geminal) wave functions and quantum number projections. The combination allows an application of the machine learning scheme to interacting fermionic systems. The combined method substantially improves the accuracy beyond that ever achieved by each method separately, in the Heisenberg as well as Hubbard models on square lattices, thus proving its power as a highly accurate quantum many-body solver.

Original languageEnglish
Article number205152
JournalPhysical Review B
Volume96
Issue number20
DOIs
Publication statusPublished - 2017 Nov 29
Externally publishedYes

Fingerprint

machine learning
Wave functions
Learning systems
wave functions
Hubbard model
Ground state
quantum numbers
Monte Carlo method
Monte Carlo methods
projection
Neural networks
ground state
products

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics

Cite this

Restricted Boltzmann machine learning for solving strongly correlated quantum systems. / Nomura, Yusuke; Darmawan, Andrew S.; Yamaji, Youhei; Imada, Masatoshi.

In: Physical Review B, Vol. 96, No. 20, 205152, 29.11.2017.

Research output: Contribution to journalArticle

Nomura, Yusuke ; Darmawan, Andrew S. ; Yamaji, Youhei ; Imada, Masatoshi. / Restricted Boltzmann machine learning for solving strongly correlated quantum systems. In: Physical Review B. 2017 ; Vol. 96, No. 20.
@article{b47d527ad2dc419d877dea7e51357c6c,
title = "Restricted Boltzmann machine learning for solving strongly correlated quantum systems",
abstract = "We develop a machine learning method to construct accurate ground-state wave functions of strongly interacting and entangled quantum spin as well as fermionic models on lattices. A restricted Boltzmann machine algorithm in the form of an artificial neural network is combined with a conventional variational Monte Carlo method with pair product (geminal) wave functions and quantum number projections. The combination allows an application of the machine learning scheme to interacting fermionic systems. The combined method substantially improves the accuracy beyond that ever achieved by each method separately, in the Heisenberg as well as Hubbard models on square lattices, thus proving its power as a highly accurate quantum many-body solver.",
author = "Yusuke Nomura and Darmawan, {Andrew S.} and Youhei Yamaji and Masatoshi Imada",
year = "2017",
month = "11",
day = "29",
doi = "10.1103/PhysRevB.96.205152",
language = "English",
volume = "96",
journal = "Physical Review B-Condensed Matter",
issn = "2469-9950",
publisher = "American Physical Society",
number = "20",

}

TY - JOUR

T1 - Restricted Boltzmann machine learning for solving strongly correlated quantum systems

AU - Nomura, Yusuke

AU - Darmawan, Andrew S.

AU - Yamaji, Youhei

AU - Imada, Masatoshi

PY - 2017/11/29

Y1 - 2017/11/29

N2 - We develop a machine learning method to construct accurate ground-state wave functions of strongly interacting and entangled quantum spin as well as fermionic models on lattices. A restricted Boltzmann machine algorithm in the form of an artificial neural network is combined with a conventional variational Monte Carlo method with pair product (geminal) wave functions and quantum number projections. The combination allows an application of the machine learning scheme to interacting fermionic systems. The combined method substantially improves the accuracy beyond that ever achieved by each method separately, in the Heisenberg as well as Hubbard models on square lattices, thus proving its power as a highly accurate quantum many-body solver.

AB - We develop a machine learning method to construct accurate ground-state wave functions of strongly interacting and entangled quantum spin as well as fermionic models on lattices. A restricted Boltzmann machine algorithm in the form of an artificial neural network is combined with a conventional variational Monte Carlo method with pair product (geminal) wave functions and quantum number projections. The combination allows an application of the machine learning scheme to interacting fermionic systems. The combined method substantially improves the accuracy beyond that ever achieved by each method separately, in the Heisenberg as well as Hubbard models on square lattices, thus proving its power as a highly accurate quantum many-body solver.

UR - http://www.scopus.com/inward/record.url?scp=85039949714&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85039949714&partnerID=8YFLogxK

U2 - 10.1103/PhysRevB.96.205152

DO - 10.1103/PhysRevB.96.205152

M3 - Article

AN - SCOPUS:85039949714

VL - 96

JO - Physical Review B-Condensed Matter

JF - Physical Review B-Condensed Matter

SN - 2469-9950

IS - 20

M1 - 205152

ER -