An obstruction for smoothing of Gorenstein surface singularities (Q804669)

From MaRDI portal
scientific article
Language Label Description Also known as
English
An obstruction for smoothing of Gorenstein surface singularities
scientific article

    Statements

    An obstruction for smoothing of Gorenstein surface singularities (English)
    0 references
    0 references
    1990
    0 references
    Let \((V,0)\) denote a complex analytic space V with an isolated singularity \(0\in V\). The isolated singularity is called smoothable if there is a flat family \(\phi: {\mathcal V}\to D\) over the unit disk D such that \(\phi^{-1}(0)\) is isomorphic to V and the generic fibre of \(\phi\) is non-singular. Examples of non-smoothable isolated singularities have been detected by complex cobordism (first used by R. Thom for this purpose) by studying the link of the singularity, i.e. the intersection of V and the boundary \(S_{\epsilon}^{2N-1}\) of a small ball \(B_{\epsilon}\) about the singular point of V embedded in \({\mathbb{C}}^ N\). \textit{E. Looijenga} and \textit{J. Wahl} [Topology 25, 261-291 (1986; Zbl 0615.32014)] showed that invariants of the linking pairing on the link of the singularity could also be used to detect non-smoothability. The main result of the present paper is to show that Rokhlin's \(\mu\)- invariant can also be used to detect non-smoothability for surface singularities. The connection is as follows. If \(\dim (V,0)=2\), then the link of \(0\in V\) is a closed 3-manifold which may be regarded as the boundary of a resolution \(\tilde V\) of the singularity. If \((V,0)\) is ``Gorenstein'', meaning that (V,0) is normal and there exists a nonzero holomorphic 2-form \(\omega\) on \(V-\{0\}\), then \((\tilde V,\partial \tilde V)\) inherits a spin structure on the complement \(\tilde V-S\) of a (real) 2-dimensional closed surface \(S\subset \tilde V\) representing the Poincaré dual of \(w_ 2(\tilde V)\). The spin structure allows one to define a quadratic form \(q: H_ 1(S,{\mathbb{Z}}/2)\to {\mathbb{Z}}/2\), whose Arf invariant is denoted Arf(S). Then the \(\mu\)-invariant \(\mu (\partial \tilde V)\equiv \frac{1}{8}(\sigma (\tilde V)-S^ 2)+Arf(S)\) (mod 2), where \(\sigma(\tilde V)\) denotes the signature of \(\tilde V\) and \(S^ 2\) the self-intersection of the homology class represented by S in \(\tilde V\) (note that the formula in the lemma, {\S}1, has a misprint). Theorem 1. Let \((V,0)\) be a 2-dimensional smoothable Gorenstein singularity. Let \(\tilde V\) be a resolution with canonical class K and S a smooth (real) surface in \(\tilde V\) dual to \(K\;(\bmod 2)\). Then \(K^ 2+8p_ g=S^ 2+8Arf(S)\) (mod 16), where \(p_ g\) is the geometric genus of \(\tilde V.\) A corollary (theorem 2) gives a condition for smoothability in terms of the configuration of exceptional curves. Explicit examples of isolated singularities violating this condition are given in {\S}3 and the authors point out that the non-smoothability of these examples is not detected by previously studied invariants.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    non-smoothable isolated singularities
    0 references
    link of the singularity
    0 references
    surface singularities
    0 references
    configuration of exceptional curves
    0 references
    0 references