Computers

The optimqr program will read a description of the sparsity pattern of some system matrix for system of linear equations. It will then apply heuristic branch and bound search to find a near-optimal ordering of the rows and columns of the system matrix. The ordering is written to disk. The codegen.pl program can then read the system ordering, and create a solver written in Fortran 77, that will solve the system using sparse QR factorization (using Givens rotations).

Software by Rasmus Munk Larsen for large and sparse SVD calculations, with versions in Fortran and Matlab.

Fortran 90 modules by Ernst Meese for the compressed sparse row (CSR) and the modified sparse row (MSR) data storage formats, a very general incomplete LU factorisation routine, the Krylov subspace solvers CGS, BiCGSTAB and GMRES.

Fortran 77 code for sparse matrix multiplication, transposition, and format conversion.

Comprised of four numerical (iterative) methods for computing the singular value decomposition (SVD) of large sparse matrices using double precision Fortran 77.

Fortran 77 code that solves sparse systems of linear algebraic equations by Gaussian elimination.