Strict density inequalities for sampling and interpolation in weighted spaces of holomorphic functions (Q2329293)

Let \(\phi:\mathbb C^n\longrightarrow\mathbb R\) be a plurisubharmonic function satisfying \[ i m \partial\overline\partial|z|^2\leq i\partial\overline\partial\phi\leq iM\partial\overline\partial|z|^2 \] (in the sense of positive currents) for some positive constants \(m,M\). For \(1\leq p\leq\infty\) we denote by \(A^p_{\phi}\) the space of entire functions on \(\mathbb C^n\) equipped with the norm \[ \|f\|^p_{\phi,p}:=\begin{cases}\int_{\mathbb C^n}|f(z)|^pe^{-p\phi(z)}\,dm(z),\quad&p<\infty,\\\sup_{z\in\mathbb C^n}|f(z)|e^{-\phi(z)},\quad&p=\infty,\end{cases} \] where \(m\) denotes the Lebesgue measure. A set \(\Lambda\subset\mathbb C^n\) is called a \textit{sampling} set for \(A^p_{\phi}\) if there are positive constants \(A,B\) such that \[ A\|f\|^p_{\phi,p}\leq\sum_{\lambda\in\Lambda}|f(\lambda)|^pe^{-p\phi(\lambda)}\leq B\|f\|^p_{\phi,p},\quad f\in A^p_{\phi}, \] with the natural modification for \(p=\infty\). A set \(\Lambda\subset\mathbb C^n\) is called a \textit{interpolating} set for \(A^p_{\phi}\) if for every \(a\in\ell^p_{\phi}(\Lambda)\) there is a function \(f\in A^p_{\phi}\) with \(f(\lambda)=a_{\lambda}\), \(\lambda\in\Lambda\). The \textit{weighted Beurling upper density} of \(\Lambda\) is defined as \[ D^+_{\phi}(\Lambda):=\limsup_{r\to\infty}\sup_{z\in\mathbb C^n}\frac{\#(\Lambda\cap B_r(z))}{\int_{B_r(z)}K(w,w)e^{-2\phi(w)}\,dm(w)}, \] where \(K\) is the Bergman kernel of \(A^2_{\phi}\). We define the \textit{weighted Beurling lower density} of \(\Lambda\) by replacing in the equality above \(\limsup\) by \(\liminf\) and \(\sup\) by \(\inf\). The first result of the paper under review is the following. For \(\varepsilon>0\) there exists a set \(\Lambda\subset\mathbb C^n\) that is interpolating for \(A^2_{(1+\varepsilon)\phi}\) and sampling for \(A^2_{(1-\varepsilon)\phi}\). Next, the authors prove that, if \(\Lambda\) is a sampling (resp. an interpolating) set for \(A^p_{\phi}\), then \(D^-_{\phi}(\Lambda)>1\) (resp. \(D^+_{\phi}(\Lambda)<1\)). As a consequence, they deduce that there does not exist a set \(\Lambda\subset\mathbb C^n\) that is simultaneously sampling and interpolating for \(A^2_{\phi}\). Finally, the authors prove the following result. If \((\Lambda_j)\) is a sequence of sets that converges Lipschitz-wise to a sampling (resp. interpolating) set \(\Lambda\) for \(A^p_{\phi}\) (i.e., roughly speaking, there exist maps \(\tau_j:\Lambda\longrightarrow\mathbb C^n\) with \(\Lambda_j=\tau_j(\Lambda)\), \(\tau_j(\lambda)\longrightarrow\lambda\) for \(\lambda\in\Lambda\) and the distances are preserved locally), then almost all \(\Lambda_j\) are sampling (resp. interpolating) sets for \(A^p_{\phi}\).