The Bergman projection in \(L^{p}\) for domains with minimal smoothness (Q372160)

The authors study the Bergman projections in strictly pseudoconvex domains with \(\mathcal{C}^2 \) boundary in \(\mathbb C^n\). The Bergman projection and its regularity on pseudoconvex domains in \(\mathbb C^n\) were studied up to now in several settings, as e.g. for strongly pseudoconvex domains in \(\mathbb C^n\), domains in \(\mathbb C^2\) with boundary of finite type, for convex domains in \(\mathbb C^n\) with boundary of finite type; in all those results a sufficient smoothness of the boundary of the domain was assumed, usually higher than only \(\mathcal{C}^2 \)-smoothness. In the present paper it is assumed that the smoothness of the boundaries of the considered domains is minimal in the present context, i.e. it is only \(\mathcal{C}^2\). The main result of the paper is Theorem 6.1, which states the following: Let \(D\) be a bounded strongly pseudoconvex domain in \(\mathbb C^n\) with \(\mathcal{C}^2 \) boundary. Let \(B\) be the orthogonal projection of \(L^2(D)\) to the Bergman space \(\vartheta L^2(D)\), the space of all holomorphic functions in \(L^2(D)\) given by the Bergman kernel \(B(w,z)\), i.e. \[ B(f)(z) = \int _D B(w,z)f(w)dV(w). \] Then the mapping \(f \longrightarrow B(f)\) extends to a bounded mapping from \(L^p(D)\) to \(L^p(D)\), for each \(p\) with \(1<p< +\infty \). A further result concerns the behaviour of the operator \(|B|\), whose kernel is the absolute value \(|B(w,z)|\) of the Bergman kernel \(B(w,z)\). More precisely, let \(|B|\) be the operator \[ |B|(f)(z) = \int _D |B(w,z)|f(w)dV(w). \] This operator is defined on \(L^2(D)\). Then Theorem 6.2 of the paper states that under the assumption on \(D\) as in Theorem 6.1, the operator \(f \rightarrow |B|(f)\) extends to a bounded mapping of \(L^p(D)\) to \(L^p(D)\), for each \(p\) with \(1<p< +\infty \). Another important result is Proposition 7.1; this proposition asserts that if \(D\) is as in Theorem 6.1, i.e. it is a strongly pseudoconvex domain in \(\mathbb C^n\) with \(\mathcal{C}^2 \) boundary, then the set of all functions which are each holomorphic in some neighborhood of \(\overline D\), is dense in \(\vartheta L^p(D)\), the space of all holomorphic functions in \(L^p(D)\), with respect to the \(L^p\)-norm in \(D\), \(1<p<+\infty \). Because of the assumed minimal, i.e. \(\mathcal{C}^2 \), regularity of the boundary of the domain \(D\), which seems to be the minimal lower limit of regularity for strongly pseudoconvex domains, the methods used up to now in former papers for the proofs of similar results for pseudoconvex domains with more regular boundaries are unapplicable directly in the present setting; those proofs use in an essential way the higher order of regularity of the boundary of \(D\). Therefore the proofs of the main results of the present paper are based on a detailed and delicate analysis. In the introduction, the authors describe this as follows: ``One constructs an appropriate family \(\{ B_\varepsilon \} _{\varepsilon >0}\) of non-orthogonal projections. While this family does not approximate \(B\) (in fact the norms of the \(B_\varepsilon \) are in general unbounded as \(\varepsilon \rightarrow 0\)), suitable truncations of the \(B_\varepsilon \) approximate in a specific sense the essential part of \(B\).''