Matrix inversion & solution of simultaneous equations. (single precision)
a.y = x
a is a non-singular n x n matrix y is the vector of n unknowns x is the known vector
the inverse of matrix a the determinant of matrix a the vector of n unknowns
symbol type dimension before after n long no. of unknowns unchanged a float [n][n] matrix inverse y float [n] vector solution *d float - determinant > *jf long - singularity flag iw long [n] - workspace > jf is the singularity flag. If the matrix is non-singular, jf=0 is returned. If the matrix is singular, jf=-1 & d=0.0 are returned. In the latter case, the contents of array a on return are undefined.
Gaussian elimination with partial pivoting.
Very fast.
Fairly accurate - errors 1 to 4 times those of routines optimized for accuracy. P.T.Wallace Starlink 31 October 1993