Accession Number:

ADA058717

Title:

A Comparison of Three Computer Programs for the Solution of AX =B Where A is Symmetric and Sparse.

Descriptive Note:

Interim rept. Feb-Jun 78,

Corporate Author:

DAVID W TAYLOR NAVAL SHIP RESEARCH AND DEVELOPMENT CENTER BETHESDA MD

Personal Author(s):

Report Date:

1978-08-01

Pagination or Media Count:

17.0

Abstract:

This report documents a continuing investigation of two recent programs for the solution of AXB where A is symmetric and sparse the Yale Sparse Matrix Package and the Munksgaard subroutines. These two programs compute direct solutions in core using triangular decomposition and Gaussian elimination, respectively. Their performance is compared with that of an out-of-core Cholesky algorithm equation solver, CSKYDG2. As would be expected, the two in-core equation solvers are much faster than CSKYDG2. The Yale symmetric subroutines range up to six times faster than the Munksgaard subroutines. All three equation solvers provide the same degree of accuracy. However, the two in-core equation solvers require such enormous amounts of core storage that their use is not recommended on the CDC 6000 series of computers in their present form. While CSKYDG2 requires less core storage, its runs on the CDC 6400 cost more because of the repeated use of the random access storage capability. Author

Subject Categories:

  • Theoretical Mathematics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE