|
The main problem for programmers working witn large matrices is
how to keep them in computer's memory. Being written in a
disk-storage, matrix involves a lot of disk exchanges, because
all Matrix-Algorithms use it "row-by-row" and "column-by-column"
at the same time. Here we're proposing an original modification
of known Gauss method for pivot element elimination that uses
internal rows-transposition-vector and works with matrix strictly
"col-by-col".
Our algorithm factorizes the source matrix so that created
factor-matrix may be used several times for quick solution of
Linear System with any Right Parts.
LGSOLV consist of several FORTRAN-written functions, described
in README.DOC file. The file DEMOTEST.EXE is given as example of
using LGSOLV.
This is shareware.
(modified from lgsolv1.inf)
|