A Comparative Study of Several Core Storage Schemes for Large Sparse Positive Definite Matrices with Reference to the Cholesky Algorithm

摘要

In the finite element approach to static structural analysis, the solution of the equation KU = P a positive definite system of simultaneous linear equations, is basic. Considerable difficulty may be experienced when K is very large and sparse. The report documents an investigation of several FORTRAN subroutines in order to obtain an efficient Cholesky algorithm subroutine with economical core storage for an in-core solution of KU = P for large sparse K. (Modified author abstract)