Curves on Hirzebruch surfaces and semistability

Alessio Cela, Carl Lian · 2025

Read Curves on Hirzebruch surfaces and semistability by Alessio Cela, Carl Lian (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-2503-20677.txt.

Introduction

Alessio Cela and Carl Lian study when a marked curve can be made to pass through the count of points one would hope for, on a class of ruled surfaces. Their paper is from two thousand twenty-five, in pure math. For all but the lowest forms of the surface, they show a marked curve never hits that hoped count. One case stands apart. There, the count of maps is set by a known set of degrees tied to the line, which past work had worked out in full.

Catalog record

CREATORAlessio Cela, Carl Lian
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE21 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-20677

Browse the full library · About the Tower of Babel

Full text

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

Curves on Hirzebruch surfaces and semistability Alessio Cela, Carl Lian Abstract . When a ≥ 2 a\geq 2 , we show that a general pointed curve never interpolates through the expected number of points in the Hirzebruch surface ℋ a {\mathcal{H}}_{a} , with one exception. In the exceptional case, the number of such interpolating maps is determined by the geometric Tevelev degrees of ℙ 1 \mathbb{P}^{1} , which have been previously computed. 2020 Mathematics Subject Classification 14H10, 14H60, 14J26, 14N10, 14N35 1. Introduction Let {\mathcal{M}}_{g,n} be the Deligne-Mumford moduli space of genus g , n -pointed curves. Let (C,p_{1}\ldots,p_{n}) be a general, pointed curve of genus g , by which we mean a general point of {\mathcal{M}}_{g,n} . Let X be a smooth, projective variety. Can C be interpolated through n general points of X ? If so, then in how many ways? More precisely, let \beta\in H_{2}(X,\mathbb{Z}) be an effective curve class, and let {\mathcal{M}}_{g,n}(X,\beta) be the moduli space of pointed maps to X in class \beta . Let \tau:{\mathcal{M}}_{g,n}(X,\beta)\to{\mathcal{M}}_{g,n}\times X^{n} be the forgetful morphism. We assume that 2g-2+n>0 . One expects that \tau is generically finite whenever (1) \beta\cdot K_{X}^{\vee}=\dim(X)\cdot(n+g-1). We assume ( 1 ) throughout. Definition 1.1 . Suppose that every irreducible component Z\subset{\mathcal{M}}_{g,n}(X,\beta) dominating {\mathcal{M}}_{g,n}\times X^{n} is generically smooth of the expected dimension. Then, the geometric Tevelev degree {\mathsf{Tev}}^{X}_{g,n,\beta} is by definition the degree of \tau . If no such components Z exist, then {\mathsf{Tev}}^{X}_{g,n,\beta}=0 vacuously. The name “geometric Tevelev degree” was introduced in [ 10 ] after a result of Tevelev [ 28 ] , but in fact, the corresponding virtual invariants go back to work on Vafa-Intriligator formulas, e.g. [ 3 , 27 , 22 ] . See [ 4 , 5 ] for more recent work on the virtual invariants. The geometric invariants, which are by definition enumerative counts as opposed to virtual intersection numbers, tend to be more subtle to compute, see [ 10 , 12 , 8 , 20 , 7 , 19 ] for recent computations and [ 21 , 2 ] for comparisons of virtual and geometric invariants. We also say that {\mathsf{Tev}}^{X}_{g,n,\beta} is well-defined if the hypothesis of Definition 1.1 holds. Given a dominating component Z\subset{\mathcal{M}}_{g,n}(X,\beta) that is generically smooth of the expected dimension, a general curve interpolates through n general points in X via a map corresponding to a point of Z . Furthermore, this map admits no first-order deformations fixing (C,p_{1},\ldots,p_{n}) and the given points on X . If all dominating components Z are generically smooth of the expected dimension, then the total number of interpolating maps is given by {\mathsf{Tev}}^{X}_{g,n,\beta} . A basic obstruction to the existence of dominating components Z\subset{\mathcal{M}}_{g,n}(X,\beta) is that, if f:C\to X is a general point of Z , then f^{*}T_{X} must be semistable, see Lemma 2.2 . Heuristically speaking, if a curve on X is more prone to deform in some tangent directions on X than others, then it has insufficient room to interpolate through n general points when maximally constrained. In particular, if T_{X} is not \beta -semistable (see Definition 2.1 ), then {\mathsf{Tev}}^{X}_{g,n,\beta} , if well-defined, must equal zero. There is a large body of work on (semi)stability of vector bundles and their restrictions to curves, see for instance [ 13 , 26 , 24 , 25 , 15 , 1 , 11 , 16 , 18 , 17 ] . In this note, we determine the geometric Tevelev degrees of Hirzebruch surfaces {\mathcal{H}}_{a}=\mathbb{P}_{\mathbb{P}^{1}}(\mathcal{O}(-a)\oplus\mathcal{O}) with a\geq 2 . We first observe that, with one exception, the semistability obstruction is already enough to rule out the existence of any dominating components Z\subset{\mathcal{M}}_{g,n}({\mathcal{H}}_{a},\beta) . Theorem 1.2 . Let a\geq 2 be an integer, and let {\mathcal{H}}_{a}=\mathbb{P}_{\mathbb{P}^{1}}(\mathcal{O}(-a)\oplus\mathcal{O}) be a Hirzebruch surface. Let C_{0}\subset{\mathcal{H}}_{a} denote the zero-section of {\mathcal{H}}_{a} , corresponding to the map \mathcal{O}(-a)\to\mathcal{O}(-a)\oplus\mathcal{O} . Suppose that n\geq 2 (or that g=0 and n\geq 3 ), and that ( 1 ) holds. Then, no irreducible component Z\subset{\mathcal{M}}_{g,n}({\mathcal{H}}_{a},\beta) dominates {\mathcal{M}}_{g,n}\times X^{n} under the map \tau:{\mathcal{M}}_{g,n}({\mathcal{H}}_{a},\beta)\to{\mathcal{M}}_{g,n}\times{\mathcal{H}}_{a}^{n}, unless a=2 and \beta=m[C_{0}] is a multiple of the zero-section C_{0}\subset{\mathcal{H}}_{a} . In particular, outside of this case, we have {\mathsf{Tev}}^{{\mathcal{H}}_{a}}_{g,n,\beta}=0 . That is, when a\geq 2 , unless a=2 and \beta=m[C_{0}] , a general pointed curve admits no interpolations in class \beta through the maximal expected number of points. This leaves only one interesting case. Theorem 1.3 . We have {\mathsf{Tev}}^{{\mathcal{H}}_{2}}_{g,n,m[C_{0}]}={\mathsf{Tev}}^{\mathbb{P}^{1}}_{g,n,m} for any g,n,m for which n=2m-g+1\geq 3 . The condition n=2m-g+1 is ( 1 ) on both sides. If 2m-g+1<3 and m>0 , then it follows from the Brill-Noether theorem that there are no maps from a general curve to \mathbb{P}^{1} or {\mathcal{H}}_{2} in the required homology classes. The geometric Tevelev degrees of {\mathcal{H}}_{2} are therefore determined by those of \mathbb{P}^{1} , which have been computed in [ 10 , 12 , 8 ] . When a=0 , the geometric Tevelev degrees of \mathbb{P}^{1}\times\mathbb{P}^{1} are trivially determined by those of \mathbb{P}^{1} . The most difficult case is a=1 , in which {\mathcal{H}}_{1}\cong{\mathsf{Bl}}_{q}(\mathbb{P}^{2}) ; some partial results are given in [ 7 ] . Genus 0 invariants of {\mathcal{H}}_{a} in the log setting (with imposed tangencies) are computed in [ 6 ] . 1.1. Conventions • We work over \mathbb{C} for convenience, though Lemma 2.2 holds in arbitrary characteristic. • The projective bundle \mathbb{P}(\mathcal{E}) is the moduli space of lines in the fibers of \mathcal{E} . • Angle brackets \langle-\rangle denote linear span. 1.2. Acknowledgments We thank Roya Beheshti, Gavril Farkas, Eric Larson, Brian Lehmann, Aitor Iribar López, Rahul Pandharipande, Eric Riedl, Jason Starr, Sho Tanimoto, Richard Thomas, and Isabel Vogt for conversations related to this note, and the anonymous referee for their helpful comments. A.C. is supported by SNF grant P500PT-222363. C.L. has been supported by NSF Postdoctoral Fellowship DMS-2001976 and an AMS-Simons travel grant. 2. Semistability and interpolation Definition 2.1 . Let X be a smooth projective variety. Let \mathcal{E} be a vector bundle on X and \beta\in H_{2}(X) be an effective curve class. We say that a non-zero sub-bundle \mathcal{F}\subsetneq\mathcal{E} is \beta -destabilizing if \frac{c_{1}(\mathcal{F})\cdot\beta}{\mathrm{rank}(\mathcal{F})}>\frac{c_{1}(\mathcal{E})\cdot\beta}{\mathrm{rank}(\mathcal{E})}. We say that \mathcal{E} is \beta -semistable if there does not exist a non-zero \beta -destabilizing sub-bundle. If X is a curve, then we take implicitly \beta=[X] , in which case one recovers the standard notion of semistability. The following lemma is well-known. Lemma 2.2 . Let Z\subset{\mathcal{M}}_{g,n}(X,\beta) be a component dominating {\mathcal{M}}_{g.n}\times X^{n} . Assume that Z is generically smooth of the expected dimension, and that ( 1 ) holds. Then, for a general point [f:C\to X] of Z , the restricted tangent bundle f^{*}T_{X} is semistable. Proof. Let (C,p_{1},\ldots,p_{n}) be a general point of {\mathcal{M}}_{g,n} , and let \tau_{C}:{\mathcal{M}}_{\beta}(C,X)\to X^{n} be the pullback of \tau over [(C,p_{1},\ldots,p_{n})]\in{\mathcal{M}}_{g,n} . Let x_{i}=f(p_{i}) for each i=1,\ldots,n . Then, the differential of \tau_{C} at [f] (d\tau_{C})_{f}:H^{0}(C,f^{*}T_{X})\to(T_{x_{i}}X)^{n} is an isomorphism. Suppose that there exists a destabilizing subbundle \mathcal{F}\subset f^{*}T_{X} . The image subspace H^{0}(C,\mathcal{F})\subset H^{0}(C,f^{*}T_{X}) has image of dimension at most n\cdot\text{rk}(\mathcal{F}) , corresponding to the tangent vectors given by the subspaces \mathcal{F}_{p_{i}}\subset T_{x_{i}}X . On the other hand, by Riemann-Roch, \displaystyle h^{0}(C,\mathcal{F}) \displaystyle\geq\int_{C}c_{1}(\mathcal{F})+\text{rk}(\mathcal{F})(1-g) \displaystyle>\frac{\text{rk}(\mathcal{F})}{\dim(X)}\int_{C}c_{1}(f^{*}T_{X})+\text{rk}(\mathcal{F})(1-g) \displaystyle=\frac{\text{rk}(\mathcal{F})}{\dim(X)}\int_{X}c_{1}(T_{X})\cdot\beta+\text{rk}(\mathcal{F})(1-g) \displaystyle=n\cdot\text{rk}(\mathcal{F}), which is a contradiction. \Diamond In particular, if \mathsf{Tev}^{X}_{g,n,\beta} is well-defined and non-zero, then f^{*}T_{X} must be semistable for all [f] enumerated by \mathsf{Tev}^{X}_{g,n,\beta}

Read the complete text