skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Pushing configuration-interaction to the limit: Towards massively parallel MCSCF calculations

Authors:
ORCiD logo [1];  [1]; ORCiD logo [2]; ORCiD logo [1]; ORCiD logo [3]
  1. Department of Chemistry, Minnesota Supercomputing Institute, and Chemical Theory Center, University of Minnesota, 207 Pleasant Street Southeast, Minneapolis, Minnesota 55455-0431, USA
  2. Department of Chemistry, Aarhus University, Langelandsgade 140, 8000 Aarhus C, Denmark
  3. Computational Research Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA
Publication Date:
Sponsoring Org.:
USDOE
OSTI Identifier:
1420593
Grant/Contract Number:
AC02-05CH11231; SC0008666
Resource Type:
Journal Article: Publisher's Accepted Manuscript
Journal Name:
Journal of Chemical Physics
Additional Journal Information:
Journal Volume: 147; Journal Issue: 18; Related Information: CHORUS Timestamp: 2018-02-14 14:51:44; Journal ID: ISSN 0021-9606
Publisher:
American Institute of Physics
Country of Publication:
United States
Language:
English

Citation Formats

Vogiatzis, Konstantinos D., Ma, Dongxia, Olsen, Jeppe, Gagliardi, Laura, and de Jong, Wibe A. Pushing configuration-interaction to the limit: Towards massively parallel MCSCF calculations. United States: N. p., 2017. Web. doi:10.1063/1.4989858.
Vogiatzis, Konstantinos D., Ma, Dongxia, Olsen, Jeppe, Gagliardi, Laura, & de Jong, Wibe A. Pushing configuration-interaction to the limit: Towards massively parallel MCSCF calculations. United States. doi:10.1063/1.4989858.
Vogiatzis, Konstantinos D., Ma, Dongxia, Olsen, Jeppe, Gagliardi, Laura, and de Jong, Wibe A. 2017. "Pushing configuration-interaction to the limit: Towards massively parallel MCSCF calculations". United States. doi:10.1063/1.4989858.
@article{osti_1420593,
title = {Pushing configuration-interaction to the limit: Towards massively parallel MCSCF calculations},
author = {Vogiatzis, Konstantinos D. and Ma, Dongxia and Olsen, Jeppe and Gagliardi, Laura and de Jong, Wibe A.},
abstractNote = {},
doi = {10.1063/1.4989858},
journal = {Journal of Chemical Physics},
number = 18,
volume = 147,
place = {United States},
year = 2017,
month =
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on November 14, 2018
Publisher's Accepted Manuscript

Save / Share:
  • A massively parallel version of the configuration interaction (CI) section of the COLUMBUS multireference singles and doubles CI (MRCISD) program system is described. In an extension of our previous parallelization work, which was based on message passing, the global array (GA) toolkit has now been used. For each process, these tools permit asynchronous and efficient access to logical blocks of 1- and 2-dimensional (2-D) arrays physically distributed over the memory of all processors. The GAs are available on most of the major parallel computer systems enabling very convenient portability of our parallel program code. To demonstrate the features of themore » parallel COLUMBUS CI code, benchmark calculations on selected MRCI and SRCI test cases are reported for the CRAY T3D, Intel Paragon, and IBM SP2. Excellent scaling with the number of processors up to 256 processors (CRAY T3D) was observed. The CI section of a 19 million configuration MRCISD calculation was carried out within 20 min wall clock time on 256 processors of a CRAY T3D. Computations with 38 million configurations were performed recently; calculations up to about 100 million configurations seem possible within the near future.« less
  • The multireference configuration-interaction (MR-CI) method is used to calculate the binding energy of the He dimer. The convergence of the binding energy to the configuration-set limit (full-CI) is followed by progressively extending the multireference configuration set. Two variants of the Pople size-extensivity correction are applied. The distance dependence of the corrections and hence the effect upon the binding energy turns out to be very small. The effect of orbital optimization is studied and it is shown that it is sufficient to optimize the orbitals used for the multireference space in an atomic multiconfiguration self-consistent field (MCSCF) calculation. In a basismore » of 50 atomic orbitals, the full-CI binding energy of {minus}9.08 K can be reproduced to 0.00 K (0.02 K) in calculations using only 37 (27) reference configurations, built from the atomic 1{ital s}, 2{ital s}, 2{ital p}, and 3{ital s} natural orbitals. Using a very large basis, the 37-reference set gives a best binding energy of {minus}10.87 K, in satisfactory agreement with Aziz's recent semiempirical result of {minus}10.95 K. These findings suggest that the MR-CI method can be developed into an efficient tool for calculating accurate van der Waals interaction energies for larger systems.« less
  • The region around the minimum of the potential-energy curve of Cr{sub 2} has been calculated at the multireference configuration interaction (CI) level including almost 1.3 billion configurations in the CI calculation. The computational techniques as implemented on massively parallel computers which enabled this calculation are described. The calculated results are R{sub e} = 1.72 {angstrom}, D{sub e} = 1.09 eV, and {omega}{sub e} = 338.7 cm{sup {minus}1} as compared to experimental values of R{sub e} = 1.679 {angstrom}, D{sub e} = 1.50 {+-} 0.05 eV, and {omega}{sub e} = 452.34 ({Delta}G{sub 1/2}) cm{sup {minus}1}. The error of 0.4 eV inmore » the dissociation energy can be attributed to relativistic effects following other authors (0.2 eV) and the need for higher angular momentum basis functions in the one-particle set (0.2 eV).« less
  • No abstract prepared.
  • We describe an implementation of the benchmark ab initio electronic structure full configuration interaction model on the Intel Touchstone Delta. Its performance is demonstrated with several calculations, the largest of which (95 million configurations, 418 million determinants) is the largest full-CI calculation yet completed. The feasibility of calculations with over one billion configurations is discussed. A sustained computation rate in excess of 4 GFLOP/s on 512 processors is achieved, with an average aggregate communication rate of 155 Mbytes/s. Data-compression techniques and a modified diagonalization method were required to minimize I/O. The object-oriented design has increased portability and provides the distinctionmore » between local and non-local data essential for use of a distributed-data model.« less