Solution of the linearly structured partial polynomial inverse eigenvalue problem (Q2141570)

This paper deals with the linearly structured partial polynomial inverse eigenvalue problem (LPPIEP) for matrices \(A_i \in \mathbb{R}^{n \times n}\), \(i=0,1,\ldots,k-1\) of such that the matrix polynomial \[ P(\lambda)=\lambda^k I_n + \sum_{i=0}^{k-1} \lambda^{i} A_i \] has \(m\) (\(1 \leq m \leq kn\)) prescribed eigenvalues and eigenvectors. Note that the coefficients of \(P(\lambda)\) may have different linear structures. For instance, some of the matrices can be diagonal matrices, other symmetric tridiagonal, etc. In order to address these situations, the authors study another problem, namely: Problem (LPPIEP-I). Given two positive integers \(k\), \(n\) and a set of partial eigenpairs \((\lambda_j, \phi_j)_{j=1}^m\), \(1 \leq m \leq kn\), construct a monic matrix polynomial \(P(\lambda)\) of degree \(k\) in such a way that matrices \(A_i \in\mathcal L_i\), \(i=0,1,\ldots,k-1\) and \(P(\lambda)\) has the specified set \((\lambda_j, \phi_j)_{j=1}^m\), \(1 \leq m \leq kn\) as its eigenpairs. The authors propose an approach for the original LPPIEP. Their results solve some open problems in the theory of polynomial inverse eigenvalue problems. The proposed solution gives an alternative answer to the inverse eigenvalue problems considered in the literature for a specific structure such as symmetric, skew-symmetric, etc. However, in this paper the framework is generalized to an arbitrary linear structure. In particular, for any given specific linear structure, this paper presents a way that systematically generates a system of linear equations and obtain the solution of the corresponding inverse eigenvalue problem. The contributions made in this paper can be summarized as follows: \begin{itemize} \item[(i)] the proposed methods are capable to solve LPPIEP and LPPIEP-I using a set of \(m\), \(1 \leq m \leq kn\), eigenpairs without imposing any restrictions on it; \item[(ii)] it is possible to study the partial inverse eigenvalue problem for different linear structures at once; \item[(iii)] the authors derive some necessary and sufficient conditions on the eigendata for the existence of solution to these problems; \item[(iv)] explicit analytical expressions for the family of matrix polynomials are provided; \item[(v)] sensitivity of the solutions to the perturbations of the eigendata is also discussed; \item[(vi)] the minimum norm solutions of LPPIEP and LPPIEP-I are obtained. \end{itemize}