Iterative Refinement after the Plane Rotations

Abstract

All numerical results presented in the previous chapter are obtained by the use of iterative refinement (IR) after the calculation of the first solution. However, the classical manner of exploiting sparsity (see Chapter 1) is in fact used in the calculations because the drop-tolerance used is so small (T=10⁻²⁵) that practically no non-zero elements are removed during the decomposition process.