Elliptical range theorems for generalized numerical ranges of quadratic operators (Q633921): Difference between revisions

The numerical range \(W(A)\) of a bounded linear operator \(A\) on a complex Hilbert space \(H\) with inner product \(\langle\cdot,\cdot\rangle\) is the convex set \(\{\langle Ax,x\rangle: x\in H,\langle x, x\rangle= 1\}\). \(A\) is said to be quadratic if it satisfies \((A- aI)(A- bI)= 0\) for some scalars \(a\) and \(b\). It is known that quadratic operators have a canonical structure under unitary equivalence, and in terms of it their numerical ranges can be described as elliptic discs (or their degenerate forms) in an explicit fashion. In this paper, the authors describe all kinds of generalized numerical ranges of quadratic operators. These include, for a quadratic \(A\), its rank-\(k\) numerical ranges \(\Lambda_k(A)= \{\lambda\in \mathbb{C}: PAP= \lambda P\) for some rank-\(k\) orthogonal projection \(P\}\), the essential numerical range \(W_e(A)= \bigcap\{\overline{A(A+ F)}:\text{rank\,}F< \infty\}\) (for \(A\) on an infinite-dimensional \(H\)), the Davis-Wielandt shell \(DW(A)= \{(\langle Ax,x\rangle,\langle Ax, Ax\rangle): x\in H,\;\langle x,x\rangle= 1\}\), the \(q\)-numerical ranges \(W_q(A)= \{\langle Ax,y\rangle: x,y\in H\), \(\langle x,x\rangle=\langle y,y\rangle= 1\), \(\langle x,y\rangle= q\}\) (for \(0\leq q\leq 1\)), the \(c\)-numerical ranges \[ W_c(A)= \Biggl\{\sum^k_{j=1} c_j\langle Ax_j, x_j\rangle: \{x_1,\dots, x_k\}\text{ orthonormal in }H\Biggr\} \] (for \(c_1\geq\cdots\geq c_k\), \(k\leq\dim H\)), and the \(k\)-numerical ranges \[ W_k(A)= \Biggl\{\sum^k_{j=1}\langle Ax_j, x_j\rangle: \{x_1,\dots, x_k\}\text{ orthogonal in }H\Biggr\} \] (for \(k\leq \dim H\)). In most cases, they are elliptic discs (or their degenerate forms), while \(DW(A)\) is an ellipsoid with or without its interior, and \(W_c(A)\) and \(W_k(A)\) are sums of finitely many elliptic discs.