Boundary manifolds of projective hypersurfaces (Q850016)

Let \(V\) be a hypersurface in the complex projective space \(\mathbb P^l\). The boundary of a closed neighborhood \(N\) of \(V\) is a \((2l-1)\)-dimensional real manifold \(M\), the so-called boundary manifold of \(V\). The topic of the paper under review is the topology of \(M\) in connection with the topology of the complement \(X={\mathbb P}^{l}-V\) of \(V\). There should be a strong connection since \(M\) can be viewed as well as the boundary of \({\mathbb P}^{l}-\text{Int}(N)\) a space homotopy equivalent to \(X\). In fact the inclusion \(M\to X\) is an \((l-1)\)-equivalence, but \(X\) has the homotopy type of a \(CW\)-complex of dimension at most \(l\). The main body of the paper is dedicated to the cohomology ring of \(M\). The authors start by relating the Poincaré polynomials of \(M\) and \(X\): \(P(X,t)=P(X,t)+t^{2l-1} P(X,t^{-1})\). Then given a commutative ring \(R\) the relation between the two cohomology rings \(H^*(M;R)\) and \(H^*(X;R)\) is investigated. The main technical result goes as follows: Assume \(H^*(M;R)\) are free \(R\)-modules and \(H^*(X,M;R)\) is a square-zero subring of \(H^*(X;R)\). Then the graded ring \(H^*(M;R)\) is isomorphic to the double of \(H^*(X;R)\). Using the presence of a mixed Hodge structure on \(H^*(M;{\mathbb C})\) [see \textit{A. Durfee} and \textit{R. Hain}, Math. Ann. 280, No. 1, 69--83 (1988; Zbl 0617.14012)], the authors verify the hypothesis of this theorem for \(R={\mathbb C}\), thus showing that \(H^*(M;{\mathbb C})\) is the double of \(H^*(X;{\mathbb C})\) in the following cases: 1) \(V\) is irreducible; 2) Deligne's mixed Hodge structure on \(H^*(X;{\mathbb C})\) satisfies \(W_{l+1}(H^l(X,{\mathbb C}))=0\). In general, the isomorphism may not hold, as another result of the papers points out: If \(V\) is a plane curve then \(H^*(M;{\mathbb C})\) is the double of \(H^*(X;{\mathbb C})\) if and only if all the irreducible components of \(V\) are rational curves. The case when \(V\) a hyperplane arrangement is considered in detail. Using a result of \textit{B. Z. Shapiro} [Proc. Am. Math. Soc. 117, No. 4, 931--933 (1993; Zbl 0798.32029)] it is shown that \(H^*(M;{\mathbb Z})\) is the double of \(H^*(X;{\mathbb Z})\). Furthermore, an explicit basis for the group \(H^*(M;{\mathbb Z})\) is given and the cup products are computed in this basis. Applications of this computations to the Lusternik-Schnirelmann category \(\text{cat}(M)\) of \(M\) and the topological complexity \(\text{tc}(M)\) of \(M\) are provided: for line arrangements all the possible values of \(\text{cat}(M)\) and \(\text{tc}(M)\) are determined and realized, and it is shown that, the difference \(\text{tc}(M)-\text{cat}(M)\) can be arbitrarly large when \(M\) is the boundary manifold of a hyperplane arrangement. Finally, the resonance varieties of \(H^*(M;{\mathbb C})\) are compared to those of \(H^*(X;{\mathbb C})\), and shown to be more complicated than those. In particular the resonance varieties of \(H^*(M;{\mathbb C})\) may contain singular irreducible components of arbitrarly high degree, for example for a generic hyperplane arrangement. A few results on the asphericity of the boundary manifold are proven in the paper. If \(l\geq 3\) then \(M\) is not aspherical. If \(l=2\) and \(V\) is the union of lines then \(M\) is aspherical if and only if \(V\) is an essential line arrangement.