An integral kernel for weakly pseudoconvex domains (Q2377010)

The Levi polynomial (holomorphic quadratic term in the Taylor development of a defining function) is -- for a strictly pseudoconvex domain -- a local holomorphic support function: it yields local supporting holomorphic hypersurfaces at the boundary of the domain. This fact is used to construct the important Ramirez-Henkin integral on strictly pseudoconvex domains, an integral that projects continuous functions on the boundary onto holomorphic functions in the interior, reproduces holomorphic functions, and satisfies estimates exhibiting the complex geometry of the boundary. For general smoothly bounded pseudoconvex domains there are, apart from certain special cases, no local holomorphic support functions. The author therefore modifies the Levi polynomial by a non-holomorphic term in order to obtain a local support function which is in a certain sense nearly holomorphic close to its zeroes, and which satisfies non-isotropic estimates involving -- as in the strictly pseudoconvex case -- the Levi form, and, in addition, a new third order term. This function behaves differently in the different tangent directions and so reflects at least to some extent the complex geometry of the boundary. Combined with the square of the euclidean distance it is then used to build up an integral formula on pseudoconvex domains that reproduces holomorphic functions and satisfies subtle estimates related to the boundary geometry. Along the way the author finds defining functions for a general (smoothly bounded) pseudoconvex domain all of whose level surfaces are again pseudoconvex, and he proves in the strictly pseudoconvex case the fundamental non-isotropic estimate even for arbitrary defining functions. The results of this paper are used in subsequent work of the author [``A pointwise a-priori estimate for the d-bar Neumann problem on weakly pseudoconvex domains'', \url{arXiv:1106.3132}] and could possibly turn out an essential step forward in our understanding of general pseudoconvex domains. The exposition is unusually clear.