Elliptic numerical ranges of \(4\times 4\) matrices (Q2493614)

For an \(n\)-by-\(n\) matrix \(A\), its numerical range \(W(A)\) is, by definition, the subset \(\{\langle Ax, x\rangle: x\in\mathbb{C}^n,\| x\|= 1\}\) of the plane, where \(\langle.,.\rangle\) and \(\|.\|\) denote the standard inner product and norm in \(\mathbb{C}^n\), respectively. It is known that \(W(A)\) is a compact convex set, but its precise shape is in general difficult to determine. For 2-by-2 or 3-by-3 matrices \(A\), when \(W(A)\) is an elliptic disc is completely known. In this paper, the author uses Kippenhahn's theorem on the duality of the boundary of \(W(A)\) and the curve given by the determinantal equation \(\text{det}(x\text{\,Re\,}A+ y\text{\,Im\,}A+ zI_n)= 0\), where \(\text{Re\,}A = (A+ A^*)/2\) and \(\text{Im\,}A= (A- A^*)/(2i)\), to give in Corollaries 7 and 8 an (almost) complete characterization when a 4-by-4 matrix \(A\) has an elliptic disc as its numerical range. The characterization is expressed in terms of the eigenvalues and entries of \(A\), and it is ``complete'' barring some technical condition on when one ellipse is contained in the other.