New results on the separation of matrix eigenvalues and the clustering of matrix elements (Q4339732)

The problem of separation of matrix eigenvalues (SMEV) is considered: given \(\gamma \in\mathbb{C} [z_1,z_2]\), establish whether the spectrum of a matrix \(A\in \mathbb{C}^{n \times n}\) is included in the set \(\Gamma= \{z\in\mathbb{C}: \text{Re} \gamma (\overline z,z) >0\}\). Necessary and sufficient conditions are given in the case of separable polynomials (i.e. \(\gamma (z_1,z_2) =\gamma_1(z_1) \gamma_2 (z_2)\), \(\gamma_1, \gamma_2 \in\mathbb{C}[z])\). The problem of clustering of matrix elements is also discussed: an open set in the matrix elements space is determined such that any matrix in this set has its spectrum in a given set \(\Gamma\) corresponding to a separable polynomial \(\gamma\). These results are useful in robust stability analysis as well as in the design of control systems.