ABSTRACT
The matrix solution often requires the most amount of computing in field analysis, and any attempt at making our programs efficient should first be directed to the arena. Two kinds of matrix storage are commonly used in solving the equation, sparse storage where only the nonzero elements of half of the matrix are stored, and profile storage where all elements of a matrix from the first nonzero of each row are stored. The exploitation of the symmetry of a matrix is very straightforward. A sparse matrix is one in which most of the numbers are zero, and it is the exploitation of this that allows to solve large problems by the finite element or finite difference method. Profile storage is used whenever the sparse matrix of coefficients encountered in finite element analysis is decomposed to alter it. Running a large matrix solution program with regular matrix or column-by-column storage will yield a large number of page faults.
