Linear-scaling self-consistent field calculations based on divide-and-conquer method using resolution-of-identity approximation on graphical processing units

    Research output: Contribution to journalArticle

    15 Citations (Scopus)

    Abstract

    Graphical processing units (GPUs) are emerging in computational chemistry to include Hartree2Fock (HF) methods and electron-correlation theories. However, ab initio calculations of large molecules face technical difficulties such as slow memory access between central processing unit and GPU and other shortfalls of GPU memory. The divide-and-conquer (DC) method, which is a linear-scaling scheme that divides a total system into several fragments, could avoid these bottlenecks by separately solving local equations in individual fragments. In addition, the resolution-of-the-identity (RI) approximation enables an effective reduction in computational cost with respect to the GPU memory. The present study implemented the DC-RI-HF code on GPUs using math libraries, which guarantee compatibility with future development of the GPU architecture. Numerical applications confirmed that the present code using GPUs significantly accelerated the HF calculations while maintaining accuracy.

    Original languageEnglish
    Pages (from-to)164-170
    Number of pages7
    JournalJournal of Computational Chemistry
    Volume36
    Issue number3
    DOIs
    Publication statusPublished - 2015 Jan 30

    Fingerprint

    Divide and conquer
    Scaling
    Unit
    Approximation
    Processing
    Data storage equipment
    Correlation theory
    Fragment
    Computational chemistry
    Electron correlations
    Computational Chemistry
    Ab Initio Calculations
    Graphics
    Program processors
    Compatibility
    Divides
    Computational Cost
    Molecules
    Electron
    Costs

    ASJC Scopus subject areas

    • Chemistry(all)
    • Computational Mathematics

    Cite this

    @article{3ae9ba7f0dcf4fb6b6618fab0f8b889d,
    title = "Linear-scaling self-consistent field calculations based on divide-and-conquer method using resolution-of-identity approximation on graphical processing units",
    abstract = "Graphical processing units (GPUs) are emerging in computational chemistry to include Hartree2Fock (HF) methods and electron-correlation theories. However, ab initio calculations of large molecules face technical difficulties such as slow memory access between central processing unit and GPU and other shortfalls of GPU memory. The divide-and-conquer (DC) method, which is a linear-scaling scheme that divides a total system into several fragments, could avoid these bottlenecks by separately solving local equations in individual fragments. In addition, the resolution-of-the-identity (RI) approximation enables an effective reduction in computational cost with respect to the GPU memory. The present study implemented the DC-RI-HF code on GPUs using math libraries, which guarantee compatibility with future development of the GPU architecture. Numerical applications confirmed that the present code using GPUs significantly accelerated the HF calculations while maintaining accuracy.",
    author = "Takeshi Yoshikawa and Hiromi Nakai",
    year = "2015",
    month = "1",
    day = "30",
    doi = "10.1002/jcc.23782",
    language = "English",
    volume = "36",
    pages = "164--170",
    journal = "Journal of Computational Chemistry",
    issn = "0192-8651",
    publisher = "John Wiley and Sons Inc.",
    number = "3",

    }

    TY - JOUR

    T1 - Linear-scaling self-consistent field calculations based on divide-and-conquer method using resolution-of-identity approximation on graphical processing units

    AU - Yoshikawa, Takeshi

    AU - Nakai, Hiromi

    PY - 2015/1/30

    Y1 - 2015/1/30

    N2 - Graphical processing units (GPUs) are emerging in computational chemistry to include Hartree2Fock (HF) methods and electron-correlation theories. However, ab initio calculations of large molecules face technical difficulties such as slow memory access between central processing unit and GPU and other shortfalls of GPU memory. The divide-and-conquer (DC) method, which is a linear-scaling scheme that divides a total system into several fragments, could avoid these bottlenecks by separately solving local equations in individual fragments. In addition, the resolution-of-the-identity (RI) approximation enables an effective reduction in computational cost with respect to the GPU memory. The present study implemented the DC-RI-HF code on GPUs using math libraries, which guarantee compatibility with future development of the GPU architecture. Numerical applications confirmed that the present code using GPUs significantly accelerated the HF calculations while maintaining accuracy.

    AB - Graphical processing units (GPUs) are emerging in computational chemistry to include Hartree2Fock (HF) methods and electron-correlation theories. However, ab initio calculations of large molecules face technical difficulties such as slow memory access between central processing unit and GPU and other shortfalls of GPU memory. The divide-and-conquer (DC) method, which is a linear-scaling scheme that divides a total system into several fragments, could avoid these bottlenecks by separately solving local equations in individual fragments. In addition, the resolution-of-the-identity (RI) approximation enables an effective reduction in computational cost with respect to the GPU memory. The present study implemented the DC-RI-HF code on GPUs using math libraries, which guarantee compatibility with future development of the GPU architecture. Numerical applications confirmed that the present code using GPUs significantly accelerated the HF calculations while maintaining accuracy.

    UR - http://www.scopus.com/inward/record.url?scp=84919807931&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84919807931&partnerID=8YFLogxK

    U2 - 10.1002/jcc.23782

    DO - 10.1002/jcc.23782

    M3 - Article

    AN - SCOPUS:84919807931

    VL - 36

    SP - 164

    EP - 170

    JO - Journal of Computational Chemistry

    JF - Journal of Computational Chemistry

    SN - 0192-8651

    IS - 3

    ER -