On the numerical ranges of composition operators induced by mappings with the Denjoy-Wolff point on the boundary (Q2629362)

For an analytic self-map \(\varphi\) of the unit disk \(\mathbb{D}\) into itself, \( a \in \overline{\mathbb{D}}\) is called a Denjoy-Wolf point of \(\varphi\) if the iterates \(\varphi_n=\varphi^n\) of \(\varphi\) converge to \(a\) uniformly on compact subsets of \(\mathbb{D}\). The Denjoy-Wolf theorem [\textit{C. C. Cowen} and \textit{B. D. MacCluer}, Composition operators on spaces of analytic functions. Boca Raton, FL: CRC Press (1995; Zbl 0873.47017)] states that a holomorphic self-map \(\varphi\) of \(\mathbb{D}\) which is not the identity and not an elliptic automorphism of \(\mathbb{D}\), has a Denjoy-Wolf fixed point \(a \in \mathbb{\overline{D}}\). It is unique with \(|\varphi'(a)| \leq 1\) and \(\varphi'(a)\) is positive if \( a \in \partial{D}\). The numerical range of a bounded linear operator \(T\) on a Hilbert space \(H\), denoted by \(W(T)\), is defined as the set \[ W(T)=\{ \langle Tf,f\rangle : \| f \|_{H} = 1 \} \] It is known that \(W(T)\) is a convex subset of \(\mathbb{C}\) and \(\sigma(T) \subseteq \overline{W(T)}\), in particular, \(\sigma(C_{\varphi}) \subseteq \overline{W(C_{\varphi})}\). The paper under review addresses the open question raised by \textit{P. S. Bourdon} and \textit{J. H. Shapiro} [Integral Equations Oper. Theory 44, No. 4, 410--441 (2002; Zbl 1038.47018)] on the zero-inclusion problem: When is zero in the numerical range of a composition operator? In that paper, the authors resolved the zero-inclusion problem except for the class of composition operators \(C_{\varphi}\)induced by injective, non-linear fractional self-mappings \(\varphi\) of \(\mathbb{D}\) of parabolic non-automorphism type. The main result of the present paper states that, for an analytic self-map \(\varphi\) of the unit disk \(\mathbb{D}\) into itself, having the Denjoy-Wolf point \(b \in \partial{\mathbb{D}}\), it follows that \(0\) is an interior point of \(W(C_{\varphi})\), provided that \(J_{\alpha}(z)=e^{\alpha\frac{z+b}{z-b}}\) is not an eigenvector of \(C_{\varphi}\) for some \(\alpha > 0\). As a corollary, the author proves that, if \(\varphi\) is nonlinear-fractional with its Dejoy-Wolff point in \(\partial{\mathbb{D}}\), then \(0\) lies in \(\operatorname{int}(W(C_{\varphi}))\). Furthermore, for the special cases where \(\varphi\) is a self-map of \(\mathbb{D}\) into itself taking \((-1,1)\) into itself and satisfying the hypothesis of the main theorem, the subspace \(A\) of \(H^2(\mathbb{D})\) of functions which are real valued on \((-1,1)\), and the restriction \(\tilde{C}_{\varphi} : A \rightarrow A\) of \(C_{\varphi}\) on \(A\) (which is well-defined as \(A\) is an invariant subspace of \(C_{\varphi}\)), the author proves that \(0\) lies in an open interval of \(W(\tilde{C}_{\varphi})=\{ \langle C_{\varphi}f,f\rangle : \| f\|_{A} = 1 \}\).