Numerical aspects of the generalized CG-method applied to least squares problems (Q1114333)

The authors consider the generalized conjugate gradient method of \textit{P. Concus} and \textit{G. M. Golub} [Lect. Notes Econ. Math. Syst. 134, 56-65 (1976; Zbl 0344.65020)] and its adjusted form (AGCG) in comparison to standard conjugate gradient method and the SOR method for linear least squares problems. The numerical experiments performed indicate that AGCG is superior, in terms of CPU time, to generalized and standard gradients and SOR on the problems tried.