Model-updating for self-adjoint quadratic eigenvalue problems (Q924353)

This paper presents a general method for spectral updating of \[ L(\lambda)=M\lambda^2+D\lambda+K, \] where \(M(>0)\), \(D\), \(K\) are Hermitian, partially based on techniques developed by \textit{M. T. Chu} and \textit{S. Xu} [Spectral decomposition of real symmetric quadratic \(\lambda \)-matrices and its applications, \url{http://www4.ncsu.edu/~mtchu/Research/Papers/spectral_decomp03b.pdf}]. It also includes basic eigenvalue-by-eigenvalue steps in computing the components of a self-adjoint triple. Two basic basic cases are considered: updating a real eigenvalue, and updating a conjugate pair of non-real eigenvalues. Some numerical examples are provided in matrices with low dimensions.