Topology of Kähler manifolds with weakly pseudoconvex boundary (Q2286439)

In this paper considered are mainly those Kähler manifolds-with-boundary \((K, J, \omega_0)\) such that \begin{itemize} \item[(a)] Each component \(L\) of \(\partial M\) is compact and admits a defining function \(f_L\) on an open neighborhood \(U_L\) of \(L\) (i.e., \(f_L\) is smooth, \(f_L|_L=0\), \(f_L|_{U_L\setminus L}<0\), and \(\sqrt{-1}\partial\overline\partial f_L\leqslant 0\)), \item[(b)] \(U_L\cap U_{L'}=\empty\) if \(L\neq L'\), and \item[(c)] there is some number \(\varepsilon >0\) such that \(|df_L|>\varepsilon\) for every boundary component \(L\). \end{itemize} In the following we let \(l\) denote the number of the components of \(\partial K\) and suppose that \(K\) has \(n\) distinct nonparabolic ends of \(K\) (\(l\) and \(n\) can be \(0\) or \(\infty\)). (For the notion of distinct nonparabolic ends of a complete Riemannian manifold we refer to p. 279 of \textit{P. Li} and \textit{L.-F. Tam} [J. Differ. Geom. 41, No. 2, 277--318 (1995; Zbl 0827.53033)].) We let \(b^{p,q}(K)=\dim_{\mathbb{C}}H^{p,q}_{\mathrm{Dol}}(K)\) and \(b^k(K)=\dim_{\mathbb{R}}H^k_{\mathrm{DR}}(K)\). The paper proves several results on the Betti numbers and Dolbeault numbers: \begin{itemize} \item[(1)] If \(l\geqslant 1\), then \(b^{1,0}(K) \geqslant l-1\) and \(b^1(K) \geqslant l-1\). If in addition \(n\geqslant 1\), then \(b^{1,0}(K) \geqslant l\) and \(b^1(K) \geqslant l\). \item[(2)] If \(l\geqslant 1\) and \(l + n\geqslant 2\), then each boundary component \(L\) has \(b^1(L) \geqslant 1\) and has a pluriharmonic defining function. In particular, each boundary component is Levi-flat. \item[(3)] If a boundary component \(L\) has finite \(\pi_1(L)\), then \(L\) is the only boundary component of \(K\), all ends of \((K,J,\omega_0)\) are parabolic, and \(b^1(K) = 0\). If \((K,J,\omega_0)\) has no ends, then \(b^{2m-1}(K) = 0\). If \(\dim_{\mathbb{C}}K = 2\) and \((K,J,\omega_0)\) has no parabolic ends, then \(\chi(K)\geqslant 1\). \end{itemize} Now we summarize the main technical results (3.2--3.5), which are related to pluriharmonic functions on \((K, J)\): There exists for every boundary component \(L\) of \(K\) a pluriharmonic function \(h_L: K\rightarrow [0,1]\) satisfying the following conditions: \begin{itemize} \item[(i)] \(h_L|_L =0\) and \(h_L|_{\partial K\setminus L}=0\). \item[(ii)] If \((K, J, \omega_0)\) has no nonparabolic ends and has \(l>0\), then \(\sum_{L} Jdh_L=0\) is the unique linear relation among \(\{Jdh_L\}_L\) (\(L\) running over all boundary components). \item[(iii)] If \(n\geqslant 1\), then there is no linear relation among \(\{Jdh_L\}_L\). In particular, all \(h_L\) are nonconstant. \item[(iv)] We let \(h_c\) denote \(\sum_L c_Lh_L\) for a family of real coefficients \(c=\{c_L\}_L\). If \(l+n\geqslant 2\) and some of the coefficients \(c_L\) are unequal, then \(h_c\) represents a nontrivial class in \(H^1_{DR}(K)\). \item[(v)] If \(h_L\) is nonconstant, then \(b_1(L)\geqslant 1\). \end{itemize} The proofs relies on various results of potential theory as proved or summarized in Section 2 of the paper. Besides, in the proof of (iv), also used is the notion ``distinguishable ends'' introduced by the author.