The role eigenvalues play in forming GMRES residual norms with non-normal matrices (Q2514276)

The authors present explicit expressions for the norms of the residual vectors generated by the generalized minimal residual (GMRES) algorithm applied to a non-normal matrix. Their first goal is to show as precisely as possible how eigenvalues contribute to the computation of residual norms, whereas the second one is to use these expressions in an attempt to enhance insight in when convergence can be suspected to be dominated by the spectrum and when not.