On contact type hypersurfaces in 4-space (Q2118067)

For a closed, connected, oriented, spin\(^c\), \(3\)-manifold \((Y,{\mathfrak s})\) one has various Heegaard Floer homologies connected by a long exact sequence \(\cdots \to HF^\infty(Y,{\mathfrak s})\to HF^+(Y,{\mathfrak s})\to HF^-(Y,{\mathfrak s})\to HF^\infty(Y,{\mathfrak s})\to\cdots\). These homology groups are infinite-dimensional vector spaces over \({\mathbb Z}/2{\mathbb Z}\), but the reduced groups \(HF^+_{red}=HF^+/im(HF^\infty\to HF^+)\) and \(HF_-=ker(HF^-\to HF^\infty)\) are finite-dimensional. For these reduced groups one obtains an isomorphism \(\delta\colon HF^+_{red}(Y,{\mathfrak s})\to HF^-_{red}(Y,{\mathfrak s})\) from the connecting morphism of the long exact sequence. Moreover there is a morphism \(U\colon HF^+(Y,{\mathfrak s})\to HF^+(Y,{\mathfrak s})\) that fits into another long exact sequence \(\cdots\to \widehat{HF}(Y,{\mathfrak s})\to HF^+(Y,{\mathfrak s})\to HF^+(Y,{\mathfrak s})\to\cdots\). Suppose \(Y\) is equipped with a co-oriented, positive, contact structure \(\xi\). \textit{P. Ozsváth} and \textit{Z. Szabó} [Duke Math. J. 129, No. 1, 39--61 (2005; Zbl 1083.57042)] defined the contact invariant \(\hat{c}(\xi)\in\widehat{HF}(-Y,s_\xi)\), which turned out to be very useful in \(3\)-dimensional contact topology. Its image in \(HF^+(Y,{\mathfrak s})\) is denoted by \(c^+(\xi)\). The paper under review is concerned with the question which smooth, closed, oriented, \(3\)-manifolds can arise as contact type hypersurfaces in \({\mathbb R}^4\) with its standard symplectic structure. This means that there should be a vector field \(v\) defined in a neighborhood of \(Y\) such that \({\mathcal L}_v\omega=\omega\). In this situation \(\alpha=\iota_v\omega\mid_Y\) induces a contact structure \(\xi=ker(\alpha)\). For the standard symplectic \({\mathbb R}^{2n}\), whenever \(W\subset {\mathbb R}^{2n}\) is a smooth, compact, codimension-\(0\) submanifold admitting a Morse function with no critical points of index greater \(n\), then by the work of \textit{K. Cieliebak} and \textit{Y. Eliashberg} [Invent. Math. 199, No. 1, 215--238 (2015; Zbl 1310.32014)] it is a Weinstein domain and its boundary is a contact type hypersurface. The situation is different for \(n=2\), with first examples due to \textit{S. Nemirovski} and \textit{K. Siegel} [ibid. 203, No. 1, 333--358 (2016; Zbl 1335.32008)]. The paper under review introduces an obstruction to contact type embeddings of \(3\)-manifolds using Heegaard Floer homology. If \(\xi\) is overtwisted then \(c^+(\xi)=0\) while if \(\xi\) is strongly fillable then \(c^+(\xi)\not=0\). In the paper under review it is shown that if \((Y,\xi)\) is embedded in \(({\mathbb R}^4,\omega_{std})\) as the boundary of a symplectically convex domain, then \(c^+(\xi)\not=0\) is in the image of \(HF^\infty(-Y,{\mathfrak s}_\xi)\to HF^+(-Y,{\mathfrak s}_\xi)\), equivalently its image in \(HF^+_{red}\) vanishes. This gives an obstruction to being a contact type hypersurface, which the authors use to prove their main result: no \(3\)-dimensional Brieskorn integer homology sphere, oriented as the link of a Brieskorn complete intersection singularity, can arise as a contact type hypersurface in \(({\mathbb R}^4,\omega_{std})\).