Two Storage Schemes for Givens Plane Rotations

Abstract

The main conclusion drawn in the previous chapter was: the Givens transformations (both in the classical form and in the Gentleman version) are best suitable, among the orthogonalization methods, for the case where linear least squares problems with large and sparse matrices are to be handled. Two different implementations of sparse matrix techniques in connection with the Givens plane rotations will be described in this section. Both implementations are very attractive for certain classes of problems and/or for certain computer environments.