A divide-and-conquer method with approximate Fermi levels for parallel computations

Takeshi Yoshikawa; Hiromi Nakai

doi:10.1007/s00214-015-1650-6

A divide-and-conquer method with approximate Fermi levels for parallel computations

Takeshi Yoshikawa, Hiromi Nakai^*

^*Corresponding author for this work

School of Advanced Science and Engineering

Research output: Contribution to journal › Article › peer-review

5 Citations (Scopus)

Abstract

This study proposes two efficient algorithms of the linear-scaling divide-and-conquer self-consistent field (DC-SCF) method for parallel computations. These algorithms minimize the amount of network communication required for determining the common Fermi level by adopting an approximate Fermi level. One algorithm adopts the quasi-converged Fermi level during DC-SCF iterations, while the other uses the quasi-converged electron numbers of individual subsystems. A numerical assessment demonstrates the high parallel efficiency for both methods without loss in accuracy.

Original language	English
Article number	53
Journal	Theoretical Chemistry Accounts
Volume	134
Issue number	5
DOIs	https://doi.org/10.1007/s00214-015-1650-6
Publication status	Published - 2015 May 1

Keywords

Common Fermi level
Divide-and-conquer method
Linear-scaling computation
Parallel computation
Self-consistent field calculation

ASJC Scopus subject areas

Physical and Theoretical Chemistry

Access to Document

10.1007/s00214-015-1650-6

Cite this

@article{239f2e7bd43c4acba41f1adbbdb9f57b,

title = "A divide-and-conquer method with approximate Fermi levels for parallel computations",

abstract = "This study proposes two efficient algorithms of the linear-scaling divide-and-conquer self-consistent field (DC-SCF) method for parallel computations. These algorithms minimize the amount of network communication required for determining the common Fermi level by adopting an approximate Fermi level. One algorithm adopts the quasi-converged Fermi level during DC-SCF iterations, while the other uses the quasi-converged electron numbers of individual subsystems. A numerical assessment demonstrates the high parallel efficiency for both methods without loss in accuracy.",

keywords = "Common Fermi level, Divide-and-conquer method, Linear-scaling computation, Parallel computation, Self-consistent field calculation",

author = "Takeshi Yoshikawa and Hiromi Nakai",

note = "Funding Information: Some of the presented calculations were performed at the Research Center for Computational Science (RCCS), Okazaki Research Facilities, National Institutes of Natural Sciences (NINS). TY is grateful to the Japan Society for the Promotion of Science (JSPS) for a Research Fellowship. Publisher Copyright: {\textcopyright} 2015, Springer-Verlag Berlin Heidelberg.",

year = "2015",

month = may,

day = "1",

doi = "10.1007/s00214-015-1650-6",

language = "English",

volume = "134",

journal = "Theoretical Chemistry Accounts",

issn = "1432-881X",

publisher = "Springer New York",

number = "5",

}

TY - JOUR

T1 - A divide-and-conquer method with approximate Fermi levels for parallel computations

AU - Yoshikawa, Takeshi

AU - Nakai, Hiromi

N1 - Funding Information: Some of the presented calculations were performed at the Research Center for Computational Science (RCCS), Okazaki Research Facilities, National Institutes of Natural Sciences (NINS). TY is grateful to the Japan Society for the Promotion of Science (JSPS) for a Research Fellowship. Publisher Copyright: © 2015, Springer-Verlag Berlin Heidelberg.

PY - 2015/5/1

Y1 - 2015/5/1

N2 - This study proposes two efficient algorithms of the linear-scaling divide-and-conquer self-consistent field (DC-SCF) method for parallel computations. These algorithms minimize the amount of network communication required for determining the common Fermi level by adopting an approximate Fermi level. One algorithm adopts the quasi-converged Fermi level during DC-SCF iterations, while the other uses the quasi-converged electron numbers of individual subsystems. A numerical assessment demonstrates the high parallel efficiency for both methods without loss in accuracy.

AB - This study proposes two efficient algorithms of the linear-scaling divide-and-conquer self-consistent field (DC-SCF) method for parallel computations. These algorithms minimize the amount of network communication required for determining the common Fermi level by adopting an approximate Fermi level. One algorithm adopts the quasi-converged Fermi level during DC-SCF iterations, while the other uses the quasi-converged electron numbers of individual subsystems. A numerical assessment demonstrates the high parallel efficiency for both methods without loss in accuracy.

KW - Common Fermi level

KW - Divide-and-conquer method

KW - Linear-scaling computation

KW - Parallel computation

KW - Self-consistent field calculation

UR - http://www.scopus.com/inward/record.url?scp=84928253310&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84928253310&partnerID=8YFLogxK

U2 - 10.1007/s00214-015-1650-6

DO - 10.1007/s00214-015-1650-6

M3 - Article

AN - SCOPUS:84928253310

SN - 1432-881X

VL - 134

JO - Theoretical Chemistry Accounts

JF - Theoretical Chemistry Accounts

IS - 5

M1 - 53

ER -

A divide-and-conquer method with approximate Fermi levels for parallel computations

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this