Optimizing the Rayleigh quotient with symmetric constraints and its application to perturbations of structured polynomial eigenvalue problems (Q2132520)

The authors consider a particular constraint Rayleigh quotient maximization problem arising in the analysis of uncertain systems and eigenvalue perturbation theory. A numerically computable upper bound for the problem solution is derived. Formulas for the determination of the structured eigenvalue backward errors of various structured matrix polynomials are obtained. It is proved that the new upper bound is better than the previously known ones. The results are applied to the estimation of structured eigenvalue backward errors of polynomials.