Estimates for the \(\bar\partial\)-Neumann problem and nonexistence of \(C^2\) Levi-flat hypersurfaces in \(\mathbb{C} P^n\) (Q1882596)

The article contains two main results, both of which are significant. Theorem 1: There exists no real Levi-flat hypersurface of class~\(C^2\) in complex projective space \({\mathbb C} P^n\) (\(n\geq2\)). Theorem 2: If \(\Omega\)~is a pseudoconvex domain with class~\(C^2\) boundary in \({\mathbb C}P^n\) (\(n\geq2\)), then the \(\overline{\partial}\)-Neumann operator~\(N\) exists on square-integrable \((p,q)\)-forms; moreover, there exists a positive~\(\epsilon\) (depending on the order of plurisubharmonicity of the distance function of~\(\Omega\)) such that \(N\), the operator~\(\overline{\partial} N\), the operator~\(\overline{\partial}^{*} N\), and the Bergman projection are continuous on the Sobolev space~\(W^s(\Omega)\) when \(0<s<\epsilon\). \textit{Y.-T.~Siu} previously proved theorem~1 under a stronger smoothness hypothesis, first when \(n\geq3\) [Ann. Math. (2) 151, No. 3, 1217--1243 (2000; Zbl 0980.53065)] and then when \(n=2\) [Ann. Math. (2) 156, No. 2, 595--621 (2002; Zbl 1030.53071)]. The real-analytic case was proved when \(n\geq3\) by \textit{A.~Lins Neto} [Ann. Inst. Fourier 49, No. 4, 1369--1385 (1999; Zbl 0963.32022)] and when \(n=2\) by \textit{T.~Ohsawa} [Nagoya Math. J. 158, 95--98 (2000; Zbl 1023.32026)]. The authors offer two proofs of theorem~1, both based on theorem~2, which is itself a fundamental result of independent interest. The first proof, based on solving a \(\overline{\partial}_b\)-problem by extending to the interior, works when \(n\geq3\) and the boundary is smooth of class \(C^{2,\alpha}\) for some positive~\(\alpha\). The second proof is based on a Liouville-type theorem that the authors state as Proposition~4.5: If \(\Omega\)~is a pseudoconcave domain in~\({\mathbb C} P^n\) (\(n\geq2\)) with class~\(C^2\) boundary, then the space of \(\overline{\partial}\)-closed square-integrable \((p,0)\)-forms on~\(\Omega\) is trivial when \(1\leq p\leq n\) and consists of constant functions when \(p=0\). The erratum attempts to fix a gap in the proof of this proposition, but in a subsequent paper [\(\overline{\partial}\)-closed extensions of forms and nonexistence of Levi-flat hypersurfaces in~\({\mathbb C} P^2\), preprint (2004)] the authors reformulate the proposition as a conjecture, stating that it remains unproved. That paper explains how to modify the first proof of theorem~1 to make the proof work for class~\(C^2\) boundary when \(n\geq2\).