On the compactness of the \(\overline{\partial}\)-Neumann operator (Q5952887)

The authors study the compactness of the Neumann operator \(N_q\); the interesting results are the cases \(q>1\). More precisely, let \(X\) be a complex manifold of dimension \(n\geq 2\) with a hermitian metric, \(D\) a relatively compact domain in \(\Omega\), \(L^2_{p,q}(\Omega)\) the space of \((p,q)\) forms on \(D\) with square integrable coefficients (with respect to the volume form induced by the given metric) and \(\overline\partial =\overline \partial_q: \text{Dom} \overline \partial_q \subset L_{(p,q)}(D)\to L^2_{p,q+1}(D)\); \(\overline \partial\) is a closed, densely defined operator, and \(\overline \partial =g\) (for \(f\in\text{Dom} \overline \partial)\) if \(\int_D f\wedge \overline\partial \varphi=(-1)^{s+p+1} \int_D\wedge \varphi\) for all \(\varphi\in C^\infty_{n-p,n-q-1} (D)\) with compact support. The complex Laplacian \(\square= \overline\partial \overline\partial^* +\overline\partial^* \overline \partial\) has domain \(\{f\in\text{Dom} \overline \partial\cap \text{Dom} \overline \partial^*: \overline\partial f\in\text{Dom} \overline \partial^* f\in\text{Dom} \overline \partial\}\) and the Neumann operator \(N_{p,q}: L^2_{p,q} (D)\to \text{Dom}\square\) is well defined by \(\square(N_{p,q}f) =f-H_{p,q} (N_{p,q}f) =0\) \(\forall f\in L^2_{p,q} (D)\), where \(H_{p,q}\) is the orthogonal projection onto \(\text{Ker} \square\) in \(L^2_{p,q}(D)\). As the authors deal only with \((0, q)\) forms, we will denote in the sequel \(N_{0q}=N_q\). The main idea is to relate the solvability of the \(\overline\partial\)-equation to the compactness of the \(N_q\). The authors prove an abstract result: Let \(D\Subset X\) as above, \(1 \leq q \leq n\). Suppose that there exist closed vector subspaces \(V_q\subset \text{ker} \overline \partial_q\), \(V_{q+1} \subset\text{ker} \overline \partial_{q+1}\) of finite codimension, and a compact linear operator \(S_k:V_k\to L^2_{0,k=1} (s)\cap \text{Dom} \overline \partial\) such that \(\overline\partial s_kf= f\) for all \(f \in V_k\), for \(k=q\), \(q+1\). Then \(N_q\) is compact. In particular it results that \(N_q\) is compact on any bounded domain in \(\mathbb{C}^n\) for which the \(\overline \partial\)-equation on \((0,q)\) forms and \((0,q+1)\) forms can be solved by integral operators of the form \(s_kf(z)= \int_{\zeta\in D} f(\zeta) \wedge s_K(\zeta,z)\) for \(k=q\) and \(k=q+1\), where the \(s_k\) satisfy some natural conditions. Now, to prove the main applications of this result, namely the case of transversal intersection of strictly \(q_i\)-convex domains \((i= 1,\dots, N \leq n-2)\), the authors give two methods allowing the passage from certain locally defined compact solution operators for \(\overline\partial\) to global compact solution operators. The first is a variation of the bump method inspired from \textit{N. Kerzman} [Commun. Pure Appl. Math. 24, 301-379 (1971; Zbl 0217.13202)], the second is purely local and supposes the existence of some homotype formulas for the local solving compact operators for the \(\overline \partial\)-equation. It happens that this occurs in very general situations; if the domain \(D\) has non-smooth boundary, the authors prove a result which asserts the extension of such homotopy formulas for \(L^2_{(0,q)} (D)\cap \text{Dom} \overline\partial\). Now, the authors give the following interesting result. If \(D\Subset X\) is a \(q\)-convex intersection (in the sense of Ma-Vassiliadou), the Neumann operator \(N_s\) is compact for \(s\geq q\); in particular if \(D\Subset \mathbb{C}^n\) is the transversal intersection of strictly \(q_j\)-convex domains, \(j=1,\dots,N\leq n-2\), then \(N_s\) on \(D\) is compact for \(s\geq q=q_1+\dots q_N-N+1\). If \(D\Subset X\) is strictly \(q\)-convex, then \(N_s\) on \(D\) is compact for \(s\geq q\) (for \(q=1\) this is a result of \textit{G. Henkin} and \textit{A. Jordaon} [Math. Ann. 307, No. 1, 151-168 (1997; Zbl 0869.32009)] who consider also more general boundaries). The authors prove similar results for Neumann operators in Sobolev spaces of forms.