Symmetric successive overrelaxation methods for solving the rank deficient linear least squares problem (Q2571978)

Symmetric successive overrelaxation methods (SSOR) are introduced to compute the least squares solution bf minimum norm to linear systems of algebraic equations. The problem is first augmented to a block \(4\times 4\) consistent system, and then subproper SSOR splitting is used. It is shown that the iteration is semiconvergent in an interval of the relaxation parameter. A numerical example is presented to illustrate the procedure.