A contribution to the theory and practice of the block Kogbetliantz method for computing the SVD (Q695054)

The Kogbetliantz method is a procedure considered for the singular value decompositon (SVD). It generalizes the Jacobi diagonalization method (initially just for symmetric matrices). Although this method showed a promising potential on parallel computation, a theoretical convergence of the block algorithm was missing up to this paper. The quadratic convergence of the block version of the Kogbetliantz method is proved. Consistent examples and implementation remarks are provided.