Non-smoothable curve singularities
Jan Stevens · 2025
Read Non-smoothable curve singularities by Jan Stevens (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-00854.txt.
Introduction
Jan Stevens studies curve singularities that cannot be smoothed in a 2025 math paper. For curve singularities, the size of smoothing parts in the deformation space is a fixed trait, but parts can differ in size.
Two methods can prove a curve is not smoothable. One counts dimensions over a family to show it cannot sit near smooth curves. The other uses an invariant that must vanish for smooth curves. Stevens revisits known cases and computes with algebra software on monomial curves and cones over point sets. New explicit examples appear, even among Gorenstein curves. A cone over a general self-linked point set is shown not smoothable at higher genus.
Catalog record
| CREATOR | Jan Stevens |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-00854
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Non-smoothable curve singularities Jan Stevens Abstract . For curves singularities all smoothing components of the deformation space have the same dimension, but there can be components of different dimensions. We are interested in the question of what the generic singularities are that appear in the fibre over a component. To this end we revisit the known examples of non-smoothable singularities and study their deformations. There are two general methods available to show that a curve is not smoothable. In the first method one exhibits a family of singularities of a certain type and then uses a dimension count to prove that the family cannot lie in the closure of the space of smooth curves. The other method uses the semicontinuity of a certain invariant, related to the Dedekind different. This invariant vanishes for Gorenstein singularities, so in particular for smooth curves. With these methods and computations with computer algebra systems we study monomial curves and cones over point sets in projective space. We also give new explicit examples of non-smoothable singularities. In particular, we find non-smoothable Gorenstein curve singularities. The cone over a general self-associated point set in ℙ g − 2 \mathbb{P}^{g-2} is not smoothable if g g is at least 11, as then the point set can not be a hyperplane section of a canonical curve of genus g g . 2020 Mathematics Subject Classification 14B05 14B07 14D20 14H20 14Q05 32S05 Introduction Deformation spaces of singularities can be very singular [ 43 ] . Reduced curve singularities seem to be better behaved, as the dimension of a smoothing component of the versal deformation, that is a component over which the general fibre is smooth, is an invariant of the singularity [ 9 ] . But if the singularity deforms into a non-smoothable curve singularity, there are in general also components of other dimensions. One of the purposes of this paper is to investigate which singularities can occur as general fibre over a (non smoothing) component. Such singularities may be called generic singularities. To this end we revisit the known examples of non-smoothable singularities and study their deformations. We also give new explicit examples of non-smoothable singularities. In particular, we find non-smoothable Gorenstein curve singularities. The existence of non-smoothable curves was first shown by Mumford [ 26 ] . He constructed families of curves with one irreducible singularity, such that the general element of the family is not smoothable, basically because the family is too large to come from a closure of the moduli space of smooth curves. New examples, of lines through the origin, were given by Pinkham [ 29 ] . These examples were treated and extended by Greuel [ 14 , 15 ] , see also the survey [ 16 ] , using a local argument. The dimension of a smoothing component can be expressed in terms of invariants of the curve singularity C , see Proposition 1.1 ; we call this number the Deligne number e(C) . The idea is now to exhibit a large family of singularities \pi\colon C_{T}\to T with singular section, such that for t\neq 0 the singularity (C_{t},\sigma(t)) is not isomorphic to (C_{0},\sigma(0)) . Here large means that the dimension of T is at least the Deligne number e(C_{0}) . If (T,0) is irreducible, then the general C_{t} is not smoothable, because the image of T in the versal deformation has dimension \dim T\geq e(C_{0}) , but cannot be a smoothing component, as there are no smooth fibres over T . For quasi-homogeneous curves the number e can be expressed in familiar invariants of the curve, making this an effective criterion. This large family argument shows the existence of non-smoothable singularities, but does not give specific examples. The first such examples of monomial curves are due to Buchweitz [ 4 ] . He observed that a necessary condition for smoothability, in fact for deforming into a Gorenstein singularity, is that the length of the Dedekind different is at most 2\delta , something conjectured, or at least seen as a possibility, to hold for all singularities by Herzog [ 17 ] . More generally, Buchweitz defined a k -th normalised Dedekind invariant. Furthermore he showed how to compute this invariant for monomial curves in terms of the semigroup. This has applications to Weierstrass points. A semigroup is the semigroup of a Weierstrass point on a smooth projective curve, if the curve occurs as a deformation of the completion of the corresponding monomial curve in an appropriate weighted projective space, with the Weierstrass point at infinity. The non-smoothability condition is thus also a condition that the semigroup cannot be a Weierstrass semigroup, and this application can be proven directly from Riemann-Roch. In this form Buchweitz’ criterion occurs most often in the literature, but the original criterion applies more generally. These two methods to prove that singularities are not smoothable are the only known general methods. In certain cases one method succeeds, in other cases the other. For not too complicated singularities direct computation of infinitesimal deformations with computer algebra systems, in particular Singular [ 8 ] and Macaulay2 [ 13 ] , can also be used in studying smoothability. Experimentation with such singularities has led to conjectures, which can be found throughout this paper. Not too complicated means in practice quasi-homogeneous. The \mathbb{G}_{m} -actions vastly simplifies computations. Also the occurring invariants of singularities are easier to compute in the quasi-homogeneous case. For monomial curves the generators of the defining ideal are particular simple, as they are binomials. We do give examples of non quasi-homogeneous non-smoothable curve singularities, but they occur as deformations of non-smoothable quasi-homogeneous ones. If a singularity is not smoothable, then by openness of versality every singularity into which it deforms, is also not smoothable. The Buchweitz criterion does not give any non-trivial condition for Gorenstein curves and therefore it cannot be used to show that symmetric semigroups are not Weierstrass semigroups. A double cover construction yields symmetric non-Weierstrass semigroups [ 41 ] . We study deformations of the simplest example and show that it deforms into a Gorenstein curve consisting of lines through the origin, which is not smoothable. This singularity is the cone over a self-associated point set, a concept introduced by Coble [ 7 ] . The condition that the general cone is smoothable is that the general self-associated point set is a hyperplane section of a canonical curve. The large family argument shows that this is in general not the case if the genus of the curves is at least 11 . Besides showing that the general singularity of a certain type is not smoothable, we give explicit examples of non-smoothable curve singularities. We mention in particular a non-smoothable Gorenstein curve singularity (Proposition 4.2 ), an irreducible smoothable but not negatively smoothable quasi-homogeneous curve (Example 2.13 ) and a singularity where the dimension of the base space is less than the Deligne number e (Example 1.9 ). The evidence collected in this paper leads to the conjecture that generic singularities have only smooth branches. The tangents to these branches are not necessarily in general position. 1. Preliminaries 1.1. Let C be a reduced affine curve over an algebraically closed field \mathbf{k} of characteristic 0, lying in \mathbb{A}^{n}