Composition operators induced by smooth self-maps of the real or complex unit balls (Q1019680)

The authors present several results on composition operators \(C_\Phi\) acting on harmonic or holomorphic weighted Bergman spaces defined on the open unit balls of \(\mathbb R^n\) and \(\mathbb C^n\), denoted \(b^p_\alpha(B^n)\) and \(A^p_\alpha(U^n)\), respectively, when the self-map \(\Phi\) is not assumed to be harmonic or holomorphic but only smooth of certain degree. It is shown in the harmonic case that, for \(\alpha>-1\), \(0<p<\infty\) and \(\Psi:B^n\to B^n\) a map in \(C^2(\overline {B}^n)\), there exists a constant \(C>0\) such that \(\|f\circ \Psi\|_{L^p_\alpha}\leq C \|f\|_{b^p_\alpha}\) if and only if the Jacobian \(J_\Psi(\xi)\neq 0\) for all \(\xi\in \Psi^{-1}(\partial B^n)\). In the case of spaces of holomorphic functions, assuming that \(\Phi:U^n\to U^n\) is a map in \(C^4(\overline {U}^n)\), they obtain that \(\|f\circ \Phi\|_{L^p_\alpha}\leq C \|f\|_{A^p_\alpha}\) if and only if \(\operatorname{Rank} M_{\Phi_\eta}(\xi)=2\) for all \(\xi, \eta\in \partial U^n\) satisfying \(\Phi(\xi)=\eta\) and \(\widetilde D_\xi\Re \Phi_{\eta}(\xi)> \widetilde D_{\tau,\tau}\Re \Phi_\eta(\xi)\) for all \(\eta, \xi, \tau\in \partial U^n\) satisfying \(\Phi(\xi)=\eta\) and \(\tau\in \xi^\perp\), where \(M_{\Phi_\eta}(\xi)\) is the real Jacobi matrix of \(\Phi_\eta(\xi)=\langle \Phi(\xi),\eta\rangle\) and \(\widetilde D_\xi\) (and \(\widetilde D_{\tau,\tau}\)) stand for the directional derivatives considered as vectors in \(\partial B^{2n}\), and the orthogonal \(\xi^\perp\) is taken in the real sense. When \(\Phi\) is assumed to be holomorphic, the above mentioned result was shown in [\textit{W.\,R.\thinspace Wogen}, Oper.\ Theory, Adv.\ Appl.\ 35, 249--263 (1988; Zbl 0685.46029)] for Hardy spaces and in [\textit{H.\,Koo} and \textit{W.\,Smith}, J.~Math.\ Anal.\ Appl.\ 329, No.\,1, 617--633 (2007; Zbl 1115.32004)] for weighted Bergman spaces. Also, they appreciate a jump phenomenon in the optimal target spaces, meaning that, if the inducing self-map \(\Phi\) is smooth enough and \(C_\Phi\) does not map \(A^p_\alpha\) or \(b^p_\alpha\) into \(L^p_\alpha\), then actually it does not map the spaces into a larger space \(L^p_\beta\) for \(\alpha<\beta<\alpha+ \varepsilon_0\) for certain \(\varepsilon_0\) depending on \(\alpha\).