The spectrum is discontinuous on the manifold of Toeplitz operators (Q1580899): Difference between revisions

For a sequence \(A_n\) of bounded linear operators on a Hilbert space, denote by \(\liminf \sigma(A_n)\) the set of all \(\lambda\in\mathbb{C}\) for which there are \(\lambda_n\in\sigma(A_n)\), the spectrum of~\(A_n\), such that \(\lambda_n\to\lambda\), and by \(\limsup \sigma(A_n)\) the set of all \(\lambda\in\mathbb{C}\) for which there are \(n_1<n_2<\dots\) and \(\lambda_{n_j}\in\sigma(A_{n_j})\) such that \(\lambda_{n_j}\to\lambda\). It~is known that if \(A_n\to A\) in norm, then \(\liminf\sigma(A_n) \subset \limsup\sigma(A_n) \subset\sigma(A)\), where in general both inclusions are strict. The authors show that this is still the case even if \(A_n\) and \(A\) are additionally assumed to be Toeplitz operators: more precisely, they exhibit functions \(c_n\) and \(c\) in \(L^\infty(\mathbb{R})\) such that \(\|c_n-c\|_\infty\to 0\) and \(\liminf\sigma(T_{c_n})\), \(\limsup\sigma(T_{c_n})\) and \(\sigma(T_c)\), where \(T_f\) stands for the Toeplitz operator with symbol~\(f\), are three different sets. This answers in the negative a question raised by \textit{D. R. Farenick} and \textit{W. Y. Lee} [Trans. Am. Math. Soc. 348, No. 10, 4153-4174 (1996; Zbl 0862.47013)].