An implicitly-restarted Krylov subspace method for real symmetric/skew-symmetric eigenproblems (Q414688)

The authors are concerned with the computation of a few eigenvalues and eigenvectors, near a given shift, for large sparse structured generalized eigenvalue problems of the form \(Mx = \lambda N x\), where \(M\) is a symmetric matrix and \(N\) a skew-symmetric one. Their new method improves and generalizes the SHIRA method of \textit{V. Mehrmann} and \textit{D. Watkins} [SIAM J. Sci. Comput. 22, No. 6, 1905--1925 (2001; Zbl 0986.65033)] to the case where the skew-symmetric matrix is singular. Applications and special properties of the new method are illustrated by benchmark problems.