Summary
A matrix-multiplication based implementation of the two-electron integral transformation is compared to the “optimized” algorithm recently presented by Hurley, Huestis and Goddard. In spite of its poorer operation count, the matrix-multiplication based program runs significantly faster on the Alliant FX/8 than the code described by Hurley et al. Our code also uses much less memory, but requires more disk storage. Trade-offs between the requirements for disk storage, central memory, number of computing elements and CPU time are also discussed.
Similar content being viewed by others
References
Hurley JN, Huestis DL, Goddard WA (1988) J Phys Chem 92:4880
Lawson C, Hanson R, Kincaid D, Krough F (1979) ACM Trans Math Software 5:308
Yoshimine M (1971) In: Lester WA (ed) Proceedings of the Conference on Potential Energy Surfaces in Chemistry. Report RA-18, IBM Research Laboratory, San Jose, CA, p 87
Bender CF (1971) J Compt Phys 9:547
Elbert ST (1978) In: Moler C, Shavitt I (eds) Numerical algorithms in chemistry: algebraic methods. LBL 8158, Lawrence Berkeley Laboratory, University of California, Berkeley, p 129
Saunders VR, van Lenthe JH (1983) Mol Phys 48: 923
FX/FORTRAN Language Manual, Alliant Computer Systems Corporation, Littleton, Mass (1987)
Almlöf J (1974) Molecule Program Description. University of Stockholm, Institute of Physics, Report 74-29
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bauschlicher, C.W. An efficient two-electron integral transformation for vector-concurrent computer architectures. Theoret. Chim. Acta 76, 187–193 (1989). https://doi.org/10.1007/BF00527472
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF00527472