Error bounds for eigenvalue and semidefinite matrix inequality systems (Q2576734)

The authors study error bounds for systems expressed in terms of eigenvalue functions arising frequently in eigenvalue optimization and inequality systems arising from nonlinear semidefinite programming. Basing on some results due to \textit{Z. Wu} and \textit{J. J. Ye} [SIAM J. Optim. 14, No. 3, 621--645 (2003; Zbl 1061.49017)] and by the chain rule due to \textit{B. S. Mordukhovich} and \textit{Y. Shao} [Trans. Am. Math. Soc. 348, No. 4, 1235--1280 (1996; Zbl 0881.49009)], sufficient conditions for the existence of error bounds for the eigenvalue inequality systems in terms of the limiting Fréchet subdifferential are established. From these results, under suitable hypotheses, they show that the eigenvalue inequality systems involving linear combination of eigenvalues have also global error bounds. Sufficient conditions ensuring error bounds for semidefinite matrix inequality systems are given as well.