An efficient heuristic ordering algorithm for partial matrix refactorization

An efficient algorithm for ordering a sparse matrix A for partial refactorization is given. The heuristic ordering method is based on simple graphic manipulations. The method locally minimizes the length of the factorization paths while preserving the sparsity of the matrix. The ordering method is also useful in enhancing the sparsity of the inverse of the triangular factor L and A, and in other applications. Claims are substantiated by experiments using test data from several power system networks and comparison with conventional ordering techniques. >