Product domains, multi-Cauchy transforms, and the \(\bar{\partial}\) equation (Q2281342)

Let \(D\subset \mathbb{C}^1\) be a domain with piecewise smooth boundary \(bD\) and \(f\in C(D)\). A well-known result from one complex variable theory says that the function \(v\) defined as \(v=-\frac{1}{2\pi i}\int_D\frac{f(\zeta)}{\zeta-z}d\overline{\zeta}\wedge d\zeta\) solves the equation \(\overline{\partial}v=fd\overline{z}\). As is well known in the setting of several complex variables there is no analogous universal solution operator for \(\overline{\partial}\). The aim of this paper is the construction of solution operators for the equation \(\overline{\partial}u=f\) on general product domains in \(\mathbb{C}^n\). When the factors are one-dimensional, the operator is a simple integral operator: specific derivatives of \(f\) are integrated against iterated Cauchy kernels. More precisely let \(D = D_1\times \cdot\cdot\cdot \times D_n\) with \(D_j\subset \mathbb{C}^1\). Let \(f=\sum_{j=1}^nf_jd\overline{z_j}\) solves \(\overline{\partial}f=0\) on \(D\). The first main result states the following: Theorem. We have \(\overline{\partial}(T f)=f\) weakly, provided \(f\in \mathcal{B}\). In the above statement \(\mathcal{B}\) is a suitable set of functions and \(T\) is the partial solid Cauchy transform. For the sake of brevity we refer directly to the paper for the corresponding definitions. Concerning \(L^p\) estimates for the solution operator the authors prove the following. Theorem. Let \(p\in [1, \infty]\). Suppose \(f_I\in L^p(D)\) for all \(I\neq \emptyset\) and \(\overline{\partial}f=0\). Then \[\|T(f)\|_{L^p(D)}\leq C \sum_{I\neq \emptyset}\|f_I\|_{L^p(D)}.\] The remaining part of the paper is concerned with the same question but for higher-dimensional factors, that is product domain \(D\) given by \(D=D_1\times \cdot\cdot\cdot \times D_k\), \(D_j\subset \mathbb{C}^{n_j}\). As explained by the authors in this case the strategy is to construct a solution operator \(T\) on \(D\) from given solution operators \(T_j\) on the factors \(D_j\). In particular the following hypothesis, called hypothesis \((C)\), is used. \((C):\) For each \(j = 1,\dots, k\) there is a linear bounded operator \(T_j : L^p_{(0,1)}(D_j)\rightarrow L^p(D_j)\) that solves the \(\overline{\partial}_j\)-equation on \(D_j\) and commutes with all \(\partial/\partial \overline{z}_*\). Above \(\overline{\partial}_j\) denotes the Cauchy-Riemann operator on \(D_j\) while \(\partial/\partial \overline{z}_*\) denotes generically barred derivatives associated to \(D_1,\dots, D_{j-1}, D_{j+1},\dots D_k\). The main result for higher-dimensional factors is the following Theorem. Let \(D=D_1\times \cdot\cdot\cdot \times D_k\), with \(D_j\subset \mathbb{C}^{n_j}\), and \(p\in [1, \infty]\). Assume \((C)\) holds. Then there is a linear operator \(T\) on \[\mathscr{B}^p := \{f\ \overline{\partial}\mathrm{-closed}\ |\ f_I \in L^p(D)\ \mathrm{for\ all}\ I \neq \emptyset \}\] satisfying \(\overline{\partial}T(f) = f\). Furthermore the estimate \[\|T(f)\|_{L^p(D)}\leq C \sum_{|I|\neq \emptyset }\|f_I\|_{L^p(D)}\] holds, for \(C>0\) independent of \(f\).