Equivariant geometry of cubic threefolds with non-isolated singularities

Ivan Cheltsov, Lisa Marquand, Yuri Tschinkel, Zhijia Zhang · 2025

Read Equivariant geometry of cubic threefolds with non-isolated singularities by Ivan Cheltsov, Lisa Marquand, Yuri Tschinkel, Zhijia Zhang (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-2505-03986.txt.

Introduction

Mathematicians Ivan Cheltsov, Lisa Marquand, Yuri Tschinkel, and Zhijia Zhang study cubic threefolds with singularities that are not isolated points. Their question is linearizability. When a finite group acts on such a threefold, can the action be rewritten as linear maps through a change of coordinates? The paper takes up this classification problem for singular threefolds, where the standard tools built for smooth varieties do not directly apply.

Catalog record

CREATORIvan Cheltsov, Lisa Marquand, Yuri Tschinkel, Zhijia Zhang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-03986

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Equivariant geometry of cubic threefolds with non-isolated singularities Ivan Cheltsov, Lisa Marquand, Yuri Tschinkel, Zhijia Zhang Abstract . We study linearizability of actions of finite groups on cubic threefolds with non-isolated singularities. 1. Introduction Let k be an algebraically closed field of characteristic zero, X a smooth projective variety over k of dimension n and G\subseteq\mathrm{Aut}(X) a subgroup of automorphisms. The G -action on X is linearizable if it is equivariantly birational to a linear G -action on {\mathbb{P}}^{n} , and unirational if X is G -equivariantly dominated by the projectivization of a linear representation of G . In [ 7 ] , [ 6 ] , [ 5 ] and [ 8 ] , we have addressed the problems of unirationality and linearizability of actions of finite groups on cubic threefolds with isolated singularities. In this paper, we extend the study of birational properties of generically free regular actions of finite groups on singular cubic threefolds to those with non-isolated singularities. Detailed knowledge of degenerations of cubic threefolds, together with their automorphisms, plays an important role in moduli theory, see, e.g., [ 1 ] , [ 4 ] . Indeed, the classification of isolated singularities in [ 18 ] was one of the motivations of our work. Another source of inspiration comes from arithmetic and birational geometry over nonclosed fields, as in [ 10 ] or [ 9 ] . We proceed to describe our strategy: we start by identifying all possibilities for actions and the corresponding normal forms. Roughly, cubic threefolds with non-isolated singularities are of four types, according to the geometry of the singular locus: line, conic, plane, twisted quartic. More precisely, denote by \mathrm{Sing}(X) the singular locus of X . Suppose that \dim(\mathrm{Sing}(X))\geq 1 , and X is not a cone. We observe that the secant variety of \mathrm{Sing}(X) is either X or has dimension \leq 2 ; otherwise, X is reducible. Building on this, the possibilities of \mathrm{Sing}(X) have been classified in [ 19 , Proposition 4.2] : • When \mathrm{dim}(\mathrm{Sing}(X))\geq 2 , then \mathrm{Sing}(X) is a plane. • When \mathrm{dim}(\mathrm{Sing}(X))=1 and \mathrm{Sing}(X) contains a plane curve, then the union of one-dimensional components of \mathrm{Sing}(X) is either a line, a smooth conic, or two lines intersecting at a point. • When \mathrm{dim}(\mathrm{Sing}(X))=1 and \mathrm{Sing}(X) contains a curve not contained in a plane, then \mathrm{Sing}(X) is a smooth rational normal quartic curve and there is a unique such X , known as the chordal cubic , given by x_{1}x_{4}^{2}+x_{2}^{2}x_{5}-x_{1}x_{3}x_{5}-2x_{2}x_{3}x_{4}+x_{3}^{3}=0. We focus on actions not fixing singular points on X , since otherwise, the actions are linearizable. The resulting cases are analyzed using a variety of tools from birational geometry, including the connection between equivariant geometry and geometry over nonclosed fields in [ 12 ] . We also restrict to finite group actions. Our principal results are: • All actions on cubics singular along a line are linearizable, by Theorem 3.1 . • All actions on the chordal cubic are linearizable, by Theorem 5.1 . • All actions on cubics singular along a plane are linearizable, by Theorem 6.2 . • Most actions on cubics singular along a conic are not linearizable, by Theorem 4.3 . The linearizability problem of some actions in this case remains open; see Section 4 for more details. • All actions on cubic threefolds with non-isolated singularities are unirational, by Theorem 4.5 . One of the corollaries of classifications in [ 1 ] , [ 18 ] , [ 19 ] , [ 16 ] is the following: if X\subset{\mathbb{P}}^{4} is a K-unstable rational cubic threefold with isolated singularities and G\subseteq\mathrm{Aut}(X) is a finite subgroup, then the G -action on X is linearizable. This is no longer true for cubic threefolds with non-isolated singularities – there are non-linearizable G -actions on K-unstable cubic 3-folds singular along a conic; this is also the most interesting case in arithmetic considerations in [ 9 ] . Acknowledgments: The first author was supported by Simons Collaboration grant Moduli of varieties . He also thanks CIRM, Luminy, for its hospitality and for providing a perfect work environment. The third author was partially supported by NSF grant 2301983. 2. Tools from birational geometry We recall the basic terminology: a G -action on {\mathbb{P}}^{n} is called linear if it arises from projectivization {\mathbb{P}}(V) of a G -representation V . We do not assume that the action on {\mathbb{P}}(V) is generically free. We will use the following general results. Proposition 2.1 . Let X={\mathbb{P}}_{B}({\mathcal{E}}) be the projectivization of a vector bundle {\mathcal{E}} of rank n+1 over a smooth projective irreducible variety B , and \pi:X\to B the associated {\mathbb{P}}^{n} -bundle. Assume that X carries a regular action of a finite group G such that {\mathcal{E}} is G -linearized, and that the induced action on B is unirational. Then the action on X is G -unirational. Moreover, if the G -action on B is linearizable and generically free then X is linearizable. Proof. By assumption, there exists a G -representation V such that G acts generically freely on {\mathbb{P}}(V) , which in turn admits a dominant G -equivariant rational map to B . Equivariantly resolving indeterminacy of this map \widetilde{{\mathbb{P}}(V)}\to{\mathbb{P}}(V) , we obtain the following G -equivariant diagram: \textstyle{{\mathcal{E}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces\tilde{{\mathcal{E}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \textstyle{B} \textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces\widetilde{{\mathbb{P}}(V)}} where \tilde{{\mathcal{E}}} is the pullback of {\mathcal{E}} to \widetilde{{\mathbb{P}}(V)} . By assumption, the G -action on {\mathbb{P}}_{B}({\mathcal{E}}) lifts to {\mathcal{E}} . It follows that \tilde{{\mathcal{E}}} is a G -linearized vector bundle over \widetilde{{\mathbb{P}}(V)} . By the no-name lemma, see, e.g., [ 13 , Theorem 1] , \tilde{{\mathcal{E}}} is G -equivariantly birational to \widetilde{{\mathbb{P}}(V)}\times{\mathbb{A}}^{n+1} , with trivial action on the second factor, as the G -action on \widetilde{{\mathbb{P}}(V)} is generically free. It follows that X={\mathbb{P}}_{B}({\mathcal{E}}) is dominated by {\mathbb{P}}(W) , for some G -representation W , thus is G -unirational. The second claim is a projective version of the no-name lemma, see, e.g., [ 13 , Theorem 1’] . ∎ Remark 2.2 . This proposition applies in particular to actions on products of projective spaces, e.g., {\mathbb{P}}^{1}\times{\mathbb{P}}^{2} in Section 3 , or to a nontrivial {\mathbb{P}}^{1} -bundle over {\mathbb{P}}^{2} , in Section 5 . In general, e.g, when the G -action on the base B is not generically free, or does not lift to {\mathcal{E}} , the linearization problem remains a challenge. The linearizability of actions on ({\mathbb{P}}^{1})^{2} was recently settled in [ 17 ] ; the case of ({\mathbb{P}}^{1})^{3} is open. The following proposition is a version of [ 8 , Propositions 3.1 and 3.5] . Proposition 2.3 . Let k be an algebraically closed field of characteristic 0. Let X\subset{\mathbb{P}}^{n} , n\geq 3 , be an irreducible cubic hypersurface over k which is not a cone, and G\subseteq\mathrm{Aut}(X) a finite subgroup, acting linearly on {\mathbb{P}}^{n} . Assume that X contains a G -invariant subvariety S , which is G -unirational. Then X is G -unirational. Proof. By [ 12 , Theorem 1.1] , G -unirationality of X is equivalent to the following property: for every field K/k and every G -torsor T over K , the twist {}^{T}X_{K} of X over K via T is K -unirational. Our assumption implies that all such twists are cubic hypersurfaces in {\mathbb{P}}^{n}_{K} , see [ 12 , Lemma 10.1] . Every

Read the complete text