The strong Oka's lemma, bounded plurisubharmonic functions and the \(\overline\partial\)-Neumann problem (Q2454926)

The authors, using a strong form of Oka's lemma, obtain, for a complex Hermitian manifold \(M\) and for relatively compact pseudoconvex domains \(\Omega\) with \({\mathcal C}^2\) boundary, existence results of bounded subharmonic functions. More precisely, let \( \delta (x) = d(x, \flat \Omega)\) be a distance function to the boundary \(\flat \Omega\) of \(\Omega\), with respect to the Hermitian metric, such that \(\delta\) satisfies the strong Oka condition, i.e., \(i\partial \overline{\partial} (-\log \delta) \geq c_0 \omega\). Then there exists a \(t_0\), \(0< t_0<1\), such that \[ i\partial \overline{\partial} (-\log \delta) \geq i t_0 \frac{\partial \delta \wedge\overline{\partial}\delta }{\delta^2}. \] From this it easily follows that for \(0<t< t_0\), the function \( \widetilde {\delta} = -\delta^t\) is a strictly plurisubharmonic function on \(\Omega\). This allows simple proofs of the results of \textit{K. Diedrich} and \textit{J. E. Fornaess} [Invent. Math. 39, 129--141 (1977; Zbl 0353.32025)] and \textit{T. Ohsawa} and \textit{N. Sibony} [Nagoya Math. J. 149, 1--8 (1998; Zbl 0911.32027)] which proved the result for the case when \(M\) is \({\mathbb C}^n\) or a Stein manifold, respectively, for the projective space \({\mathbb C} P^n\). The proof relies on the simple observation that if the distance function \(\delta\) satisfies the strong Oka condition, then the condition upon \(\delta\) given by the above results is equivalent to each of the following: (a) \(i \partial \overline{\partial} (-\delta^{t_0}) \geq 0\); (b) for any \(0< t< t_0\) there exists a constant \(C_t>0\) such that \(i \partial \overline{\partial} (-\delta^{t}) \geq C_t \delta^T (\omega + i \frac{\partial \delta \wedge\overline{\partial}\delta }{\delta^2})\), where \( \omega\) is the metric form of \(M\). The auhors study the \(\overline{\partial}\)-Neumann problem on Kähler manifolds with nonnegative curvature. They consider the case when \(M\) is not Stein and there is no strictly plurisubharmonic weight function smooth up to the boundary. For this they establish an \(L^2\) theory without weights. Let \(f\) be a \((p, q)\) form on \(\Omega\), and consider the decomposition of \(f\) into complex normal \(f^\nu\) and tangential part \(f^\tau\). If \[ | f| ^2_A = | f^\tau| ^2 +\frac{| f^\nu| ^2}{| \delta| ^2}, \quad | f| ^2_{A'} = | f^\tau| ^2 +| f^\nu| ^2| \delta| ^2, \] then for any \(t>0\) let \(L^2_{A(p, q)}(\delta^t)\), respectively, \(L^2_{A'(p, q)} (\delta^t)\) be the weighted \(L^2\) spaces on \((p, q)\) forms such that the norms \(\| f\|_{L^2_A(\delta^t)}\), respectively, \(\| f\|_{L^2_{A'}(\delta^t)}\), are finite. Then if \(M\) is a complex Kähler manifold with nonnegative sectional curvature, \(\Omega \Subset M\) is a pseudoconvex domain with \({\mathcal C}^2\) boundary, and \(\delta\) is the distance function which satisfies the strong Oka condition, we have: (1) For any \(f\in L^2_{A'(p, q)}(\Omega)\) (\( 0\leq p\leq n\), \(1\leq q\leq n\)) such that \(\overline\partial f =0 \) in \(\Omega\), there exists \(u\in L^2_{(p, q-1)}\) satisfying \(\overline\partial u=f\) and \( \int_\Omega | u| ^2 \leq C \int_\Omega | f| ^2_{A'}\). (2) The complex Laplacian \(\square_{(p, q)}\) has a closed range and the \(\overline\partial\)-Neumann operator exists for every \(p\), \(q\) with \(0\leq p\leq n\), \(0\leq q \leq n\). (3) For every \(f\in L^2_{(p, q)}(\Omega)\), there are decompositions \[ f = \overline\partial \overline\partial^* N_{(p, q)} f \oplus \overline\partial^* \overline\partial N_{(p, q)} f , \; 1\leq q\leq n, \quad \text{and} \quad f = \overline\partial^* \overline\partial N_{(p, 0)} f \oplus Pf, \] where \(P\) is the orthogonal projection \(\colon L^2_{(p, 0)} \mapsto L^2_{(p, 0)}\cap \text{Ker}(\overline\partial)\). (4) There exists \(t_0\), \(0<t_0\leq 1\) such that \(N\), \(\overline\partial N\), \(\overline\partial^*N\) and the Bergman projection \(P\) are regular on \(W^s_{(p, q)} (\Omega)\) for \(0\leq s< t_0/2\). The positivity of the sectional curvature intervenes by the Bochner-Hörmander-Kohn-Morrey formula. Finally, the authors prove, imposing a stronger assumption for the defining function, namely, if \(\rho\) is a defining function for \(\Omega\) (as above), then \(\rho\) must verify \( i \partial \overline\partial (-\log (-\rho)) \geq \Phi(-\rho) \omega\) in the sense of currents, where \(\Phi\) is a continuous function such that \(\lim_{x\to 0+} \Phi(x) =\infty\), that if \(M\) is as above, \(\Omega\) with \({\mathcal C}^2\) boundary, \(\rho\) is a defining function verifying the above condition for \(\Phi(x) =x^{-2\varepsilon}\), where \(0<\varepsilon<1/2\), the for every \(0\leq s <1/2-\varepsilon\), the \(\overline\partial\) Neumann operator \(N\) is bounded from \(W^{s-\varepsilon}_{(p, q)}(\Omega)\) to \(W^{s+\varepsilon}_{(p, q)}(\Omega)\) for all \(0\leq p\leq n\), \(1\leq q\leq n\). Also, \(N\) is compact in \(W^{s}_{(p, q)}(\Omega)\).