Adapting ITPACK routines for use on a vector computer
The Center for Numerical Analysis of the University of Texas has developed a package of programs, known as ITPACK, for solving large sparse systems of linear algebraic equations using iterative algorithms. The package is primarily designed to handle linear systems where the coefficient matrix is symmetric and positive definite. Included are subroutines based on the Jacobi method, the Symmetric Successive Overrelaxation (SSOR) method, and the RS method for the Reduced System. Each of these methods are accelerated with both Chebyshev and conjugate gradient acceleration. A routine is also included for the Successive Overrelaxationn (SOR) method. All necessary iteration parameters are determined automatically. The current subprograms are designed for use on serial, or scalar, computers.
- Research Organization:
- Texas Univ., Austin (USA). Center for Numerical Analysis
- DOE Contract Number:
- AS05-81ER10954
- OSTI ID:
- 5708792
- Report Number(s):
- CONF-820844-2; ON: DE85012468
- Country of Publication:
- United States
- Language:
- English
Similar Records
The vectorization of ITPACK 2C
The search for high level parallelism for the iterative solution of large sparse linear systems