In this section, we discuss the tools to assess the accuracy of computed eigenvalues and corresponding eigenvectors. We only assume the availability of residual vectors for computed eigenvalues and eigenvectors, which are usually available upon exiting a successful computation or are of marginal cost to compute afterwards for the iterative methods discussed in this chapter.
For the treatment of error estimation of the computed eigenvalues, eigenvectors, and invariant subspaces of dense NHEPs, see Chapter 4 of the LAPACK Users' Guide [12].