The essential norm of a composition operator mapping into the \(Q_{s}\)-space (Q2370735)

Every analytic self-map \(\varphi\) of the open unit disc \(U\) in the complex plane induces the composition operator \(C_\varphi\), defined by \(C_\varphi (f) = f \circ \varphi\), acting on the space of all analytic functions in \(U\). It is a bounded linear operator on the classical Bergman and Hardy spaces. For each \(s > 0\), the \textit{generalized Nevanlinna counting function} is the function \[ N_{\varphi ,s}: U \setminus \{\varphi (0)\} \mapsto \sum_{z \in \varphi^{-1} \{w\}} \left(\log {1 \over | z| }\right)^s \in [0,\infty ), \] where \(z \in \varphi^{-1} \{w\}\) is repeated according to the multiplicity of the zero of \(\varphi - w\) at \(z\) (the classical Nevanlinna counting function \(N_\varphi\) is \(N_{\varphi ,1}\)). For each \(a \in U\), the Möbius self-map of \(U\) interchanging the origin and \(a\) is \(\varphi_a(z) := (a - z)/(1 - \overline{a}z)\). For \(0 < p < \infty\) and \(-1 < \alpha < \infty\), the weighted Bergman space \(A^p_\alpha\) consists of those analytic functions on \(U\) for which \(\| f\| _{A^p_\alpha} := ((\alpha + 1) \int_U | f(z)| ^p (1 - | z| ^2)^\alpha \, dA(z))^{1/p} < \infty\), where \(dA\) is the element of the Lebesgue area normalized such that \(A(U)= 1\). A function \(f\) belongs to the Dirichlet type space \(D^p_\alpha\) if \(f' \in A^p_\alpha\). The classical Dirichlet space \(D\) is \(D_0^2\), the weighted Dirichlet space \(D_\alpha\) is \(D_\alpha^2\), the classical Besov space \(B^p\) is \(D^p_{p-2}\), and \(A^p_\alpha = D^p_{p + \alpha}\). Moreover, for \(0 \leq s < \infty\), the Möbius invariant subspace \(Q_s\) of the weighted Dirichlet space \(D_s\) consists of those \(f\) for which \(\| f\| _{Q_s} := | f(0)| + (\sup_{a \in U} \int_U | f'(z)| ^2 (\log {1 \over | \varphi_a (z)| })^s \, dA(z))^{1/2} < \infty\). The space \(Q_1\) coincides with BMOA, the space of analytic functions of bounded mean oscillation, and if \(s > 1\), then \(Q_s\) reduces to the Bloch space. For a fixed analytic self-map \(\varphi\) of \(U\), \(p > 0\), \(\alpha > -1\), \(s \geq 0\) and \(w \in U\), set \[ m(p,\alpha ,s,w) := {\sup_{b \in U} N_{\varphi \circ \varphi_b ,s} (w) \over (\log {1 \over | w| })^{(2/p)(\alpha + 2)-2}}, \] \[ A(p,\alpha ,s) := \limsup_{| a| \to 1^-} \sup_{b \in U} \int_U | \varphi_a ' (w)| ^{(2/p)(\alpha + 2)} N_{\varphi \circ \varphi_b ,s} (w) \, dA(w) \] and \[ B(p, \alpha ,s) := \limsup_{| w| \to 1^-} m(p,\alpha ,s,w). \] Finally, if \(T:X \to Y\) is a (continuous, linear) operator between two Banach spaces \(X,Y\), the \textit{essential norm} \(\| T\| _e\) of \(T\) is its distance the set \(K(X,Y)\) of compact operators from \(X\) to \(Y\), that is, \(\| T\| _e = \inf \{\| T - L\| : \, L \in K(X,Y)\}\). In the paper under review, the author establishes an asymptotic formula, in terms of a global integral condition, for the essential norm \(\| C_\varphi\| _e\) of the composition operator acting from weighted Bergman spaces or weighted Dirichlet spaces into \(Q_s\), as well as some consequences and related results, so extending or completing some earlier work due to J.\,H.\thinspace Shapiro and P.\,Poggi--Corradini, among others. The proofs use characterizations of bounded and compact composition operators from \(D^p_\alpha\) to \(Q_s\) due to Smith, Zhao, Wirths, Xiao, Lindström, Makhmutov and Taskinen, as well as a generalization of the area formula, the sub-mean value property of the Nevanlinna counting function and Luecking's characterization of Carleson measures in terms of functions in the weighted Bergman spaces. Specifically, the author proves, among other statements, the following: (a) Let \(0 < p \leq 2\), \(-1 < \alpha < \infty\) and \(0 < s < \infty\). Then if \(C_\varphi : D_\alpha^p \to Q_s\) is bounded, then there exists a constant \(c\), depending only on \(p\) and \(\alpha\), such \(A(p,\alpha ,s) \leq \| C_\varphi \| _e^2 \leq c A(p,\alpha ,s)\). (b) Let \(1 < p \leq 2\) and \(-1 < \alpha < \infty\) such that \(p \leq 2 + \alpha\). If \(C_\varphi : D_\alpha^p \to BMOA\) is bounded, then \({B(p,\alpha ,1) \over c} \leq A(p,\alpha ,1) \leq \| C_\varphi\| _e^2 \leq {B(p,\alpha ,1) \over \alpha +1}\), where \(c := c(p,\alpha ) \geq 1\) depends only on \(p\) and \(\alpha\), and satisfies \(c(p,p-2) = 1\). If \(0 \leq s < 1\) and \(C_\varphi : D_\alpha^p \to Q_s\) is bounded, then \(\| C_\varphi\| ^2_e \leq {B(p,\alpha ,s) \over \alpha +1}\). (c) Let \(0 < p \leq 2\) and \(-1 < \alpha < \infty\). Then \(C_\varphi : D_\alpha^p \to BMOA\) is bounded if and only if \(\sup_{w \in U} m(p,\alpha ,1,w) < \infty\); and \(C_\varphi\) is compact if and only if \(\limsup_{| w| \to 1^-} m(p,\alpha ,1,w) = 0\). (d) If \(C_\varphi : D \to BMOA\) is bounded, then \[ \limsup_{| w| \to 1^-} \sup_{b \in U} N_{\varphi \circ \varphi_b} (w) = \| C_\varphi \| _e^2 = \limsup_{| a| \to 1^-} \sup_{b \in U} \int_U | \varphi_a '(w)| ^2 N_{\varphi \circ \varphi_b} (w) \, dA(w) \]