Interpolation on non-uniformly separated sequences in a weighted Bergman space (Q392244)

Let \(\varphi\in\mathcal C^2(\mathbb D)\) be a subharmonic function in the unik disk \(\mathbb D\) whose invariant Laplacian \(\tilde \Delta\varphi=(1-|z|^2)^2 \Delta\varphi\) is uniformly bounded above and below (there exist \(C,c>0\) such that \(c\leq \tilde \Delta\varphi\leq C\)). This paper gives a sufficient condition for the interpolation problem by functions in the weighted Bergman space \(A_\varphi^2\), consisting of the holomorphic functions \(f\) in \(\mathbb D\) such that \[ \|f\|=\left(\int_{\mathbb D} |f(z)|^2 \frac{e^{-\varphi(z)}}{1-|z|^2} \frac{d A(z)}{\pi} \right)^{1/2} <\infty . \] Let \(\Gamma\) be a sequence in \(\mathbb D\). For any \(\gamma\in\Gamma\), let \(\rho_\gamma=\frac{1}{2} \inf_{\gamma'\neq\gamma} \rho(\gamma,\gamma')\), where \(\rho(\gamma,\gamma')=|\frac{\gamma-\gamma'}{1-\bar\gamma\gamma'} |\) is the pseudohyperbolic distance in \(\mathbb D\). Let also \(n_\gamma=\#\{\gamma'\in\Gamma : \rho(\gamma,\gamma')<1/2\}\). The sequence \(\Gamma\) is called O-interpolation for \(A_\varphi^2\) if for every sequence of values \(\{a_\gamma\}_{\gamma\in\Gamma}\) such that \[ (1) \qquad\qquad \sum_{\gamma\in\Gamma}|a_\gamma|^2 \frac{e^{-\varphi(\gamma)}}{\rho_\gamma^{2 n_\gamma}} (1-|\gamma|^2) <+\infty, \] there is an \(f\in A_\varphi^2\) such that \(f(\gamma)=a_\gamma\), \(\gamma\in\Gamma\). In the classical definition of interpolation sequences, the values to interpolate satisfy the growth condition \(\sum_{\gamma\in\Gamma}|a_\gamma|^2 e^{-\varphi(\gamma)} (1-|\gamma|^2) <+\infty\), instead of (1). This is motivated by an equivalent formulation in terms of Riesz sequences of normalized reproducing kernels \(\{k_\gamma\}_\gamma\) in the Hilbert space \(A_\varphi^2\); see [\textit{K. Seip}, Interpolation and sampling in spaces of analytic functions. Providence, RI: American Mathematical Society (AMS) (2004; Zbl 1057.30036)]. The definiton forces then the sequence \(\Gamma\) to be uniformly separated, that is, \(\inf_{\gamma\in\Gamma} \rho_\gamma>0\). Here no mention is made of the reasons for choosing the values (1). Interpolation sequences for the standard Bergman spaces, corresponding to weights \(\varphi(z)=\alpha\log(\frac 1{1-|z|^2})\), \(\alpha>0\), were characterized by means of a density condition in [\textit{K. Seip}, Invent. Math. 113, No. 1, 21--39 (1993; Zbl 0789.30025)]. \textit{B. Berndtsson} and \textit{J. Ortega} [J. Reine Angew. Math. 464, 109--128 (1995; Zbl 0823.30023)] used Hörmander's \(L^2\)-estimates for the \(\bar\partial\)-equation to give a different proof of the sufficiency part, which generalizes to weights \(\varphi\) with bounded invariant Laplacian. The main result of this paper is that the density condition of Theorem 4 (a) in Berndtsson and Ortega-Cerdà's paper is also sufficient for O-interpolation. For \(r<1\) consider the function defined by \(\xi_r(\zeta)=c_r\log\frac 1{|\zeta|^2}\) if \(\frac 12<|\zeta|<r\) (where \(c_r\) is chosen so that \(\int_{\mathbb D}\xi_r(\zeta)\frac{dA(\zeta)}{\pi (1-|\zeta|^2)^2}=1\)) and \(\xi_r(\zeta)=0\) otherwise. Consider also the measure \(\nu=\pi\sum_{\gamma\in\Gamma} (1-|\gamma|^2)^2 \delta_\gamma\), where \(\delta_\gamma\) is the Dirac measure at \(\gamma\). The main result is as follows: if there exists \(r<1\) and \(\delta>0\) such that \[ (\nu*\xi_r)(z)< \frac 1{\pi}\tilde\Delta\varphi\; dA * \xi_r(z)-\delta \] for all \(z\in\mathbb D\), then \(\Gamma\) is an O-interpolating sequence for \(A_\varphi^2\). The proof also follows the same scheme of Berndtsson and Ortega-Cerdà.