Procrustes problem for the inverse eigenvalue problem of normal (skew) <i>J</i> -Hamiltonian matrices and normal <i>J</i> -symplectic matrices (Q6939438)

For a given full-column rank matrix \(X\in\mathbb C^{n\times m}\) and a diagonal \(D\in\mathbb C^{m\times m}\), let \(\mathcal S\) denote the set of all solutions \(A\in\mathbb C^{n\times n}\) to the inverse eigenvalue problem \(AX=XD\).\N\NThe paper is devoted to several versions of the \textit{Procrustes problem}, namely, minimizing the Frobenius norm of \(\widetilde{A}-A\) when \(\widetilde{A}\in\mathbb C^{n\times n}\) is given and \(A\in \mathcal S\cap\mathcal T\). Here \(\mathcal T\) is one of the following three classes: normal \(J\)-Hamiltonian, normal skew \(J\)-Hamiltonian, and normal \(J\)-symplectic matrices. In each case, the uniqueness and existence criteria, the construction algorithm, and numerical examples are provided.