Picard bundles and the degree of irrationality of Jacobians and Pryms

Federico Moretti and Andrés Rojas · 2026

Read Picard bundles and the degree of irrationality of Jacobians and Pryms by Federico Moretti and Andrés Rojas (2026) 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-2605-18469.txt.

Introduction

Federico Moretti and Andrés Rojas study, in a 2026 paper, positivity of twisted rank-g Picard bundles. They work on the g-fold symmetric product of a smooth projective curve. As an application they bound the degree of irrationality of any genus g Jacobian above by 2 to the g. They also bound the degree of irrationality of any (g-1)-dimensional Prym variety above by 2 to the 2g-3.

Catalog record

CREATORFederico Moretti and Andrés Rojas
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49040
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.18469

Browse the full library · About the Tower of Babel

Full text

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

Picard bundles and the degree of irrationality of Jacobians and Pryms Federico Moretti and Andrés Rojas Abstract . For a smooth projective curve of genus g g , we study some positivity properties of (twisted) Picard bundles on the symmetric product. As applications, we prove that the degree of irrationality of any genus g g Jacobian is bounded from above by 2 g 2^{g} , and the degree of irrationality of any ( g − 1 ) (g-1) -dimensional Prym variety is bounded from above by 2 2 ​ g − 3 2^{2g-3} . 1. Introduction Let C be a smooth projective curve of genus g\geq 2 . On its Jacobian JC one can construct several vector bundles, usually referred to as Picard bundles , whose fibers parametrize the cohomology groups of all line bundles of a fixed degree on C . These vector bundles were introduced in classical works of Mattuck [ 17 ] and Schwarzenberger [ 27 ] , and redefined by Mukai [ 19 ] in terms of the Fourier–Mukai transform. Beyond their natural appearance in the determinantal realization of the Brill–Noether loci W^{r}_{d}(C) , they have been explored from several perspectives, including stability [ 14 , 11 ] , cohomological regularity and continuous global generation [ 24 , 25 ] , or potential approaches to the Schottky problem [ 10 ] . In this note we focus on the rank- g Picard bundle F whose fiber over \alpha\in JC admits a canonical identification F|_{\alpha}\cong H^{1}\!\big(C,\mathcal{O}_{C}(-p_{0})\otimes\alpha\big), where p_{0}\in C denotes a fixed base point 1 1 1 For the purposes of this work, the line bundle \mathcal{O}_{C}(-p_{0}) can be replaced by any line bundle of degree -1 . . We investigate some elementary properties of F by passing to the symmetric product C^{(g)} . More precisely, consider the Abel–Jacobi map \pi:C^{(g)}\mathchoice{\longrightarrow}{\rightarrow}{\rightarrow}{\rightarrow}\Pic^{0}(C)=JC,\qquad p_{1}+...+p_{g}\longmapsto\mathcal{O}_{C}(p_{1}+...+p_{g}-gp_{0}) and denote by x:=p_{0}+C^{(g-1)}\subset C^{(g)} the standard divisor defined by p_{0} . We prove that the vector bundle \widetilde{F}:=\pi^{*}F\otimes\mathcal{O}_{C^{(g)}}(x) on C^{(g)} is globally generated and satisfies c_{g}(\widetilde{F})=2^{g} , and we give a precise description of the zero locus of global sections of \widetilde{F} from which we constraint the rank of its determinant map. The main geometric application of our analysis of \widetilde{F} is the following upper bound on the degree of irrationality of Jacobian varieties: Theorem A . For any smooth curve C of genus g\geq 2 , the inequality \irr(JC)\leq 2^{g} holds. The degree of irrationality \irr(X) of an integral projective variety X is the minimal degree of a rational dominant map X\dashrightarrow\mathbb{P}^{\dim(X)} . This birational invariant, generalizing to higher dimension the classical notion of gonality of a curve, has recently been the object of considerable activity (see [ 6 ] for a survey on recent work). However, an essentially complete solution is known in only very few cases, such as hypersurfaces in projective space [ 2 , 7 ] or abelian surfaces [ 8 , 16 ] . For abelian varieties of dimension \geq 3 one has nontrivial lower bounds [ 9 ] , but relevant upper bounds are not known; even for Jacobians, the standard upper bound comes from symmetrizing a map C\mathchoice{\longrightarrow}{\rightarrow}{\rightarrow}{\rightarrow}\mathbb{P}^{1} and gives \irr(JC)\leq\mathrm{gon}(C)^{g}\leq\lfloor\frac{g+3}{2}\rfloor^{g} , see [ 6 , Problem 4.6] . The key point in the proof of Theorem A is the fact that a globally generated vector bundle E of rank \dim(X) with positive top Chern class produces, via a determinantal construction, a rational dominant map to \mathbb{P}^{\dim(X)\\ } whose degree is bounded by the top Chern class of E (see subsection 2.1 , building on results from [ 18 ] ). Applying this general principle to the vector bundle \widetilde{F} on the symmetric product C^{(g)} results in the upper bound \irr(JC)\leq 2^{g} . Notably, the rational maps JC\dasharrow\mathbb{P}^{g} we construct are not rigid. For a fixed divisor \Theta representing the canonical principal polarization on JC , we obtain a large family—parametrized by an open subset of the Grassmannian \Gr(g+1,2g) —of linear projections JC\hookrightarrow|(g+1)\Theta|^{\vee}\dasharrow\mathbb{P}^{g} of degree 2^{g} . It turns out that such projections admit a concrete geometric realization: considering the closed immersion C\hookrightarrow\mathbb{P}^{2g-1}=|\omega_{C}((g+1)p_{0})|^{\vee} and fixing a linear subspace \mathbb{P}^{g}\subset\mathbb{P}^{2g-1} , the corresponding map C^{(g)}\dasharrow\mathbb{P}^{g} sends a general p_{1}+...+p_{g}\in C^{(g)} to the intersection point of \mathbb{P}^{g} with the linear span \langle p_{1},...,p_{g}\rangle\subset\mathbb{P}^{2g-1} . We will address this geometric perspective for arbitrary symmetric products of curves and Brill–Noether loci in future work. In the last part of the paper we exploit a similar viewpoint, namely the use of suitable Picard type vector bundles, to find an upper bound for the degree of irrationality of Prym varieties. Recall that, for a smooth curve C of genus g and an irreducible étale double cover f:\widetilde{C}\mathchoice{\longrightarrow}{\rightarrow}{\rightarrow}{\rightarrow}C , the associated Prym variety \Pr(f) as the connected component of \ker(\Nm_{f}:J\widetilde{C}\mathchoice{\longrightarrow}{\rightarrow}{\rightarrow}{\rightarrow}JC) containing the origin of J\widetilde{C} . It is a principally polarized abelian variety of dimension g-1 . We obtain the following: Theorem B . For any (g-1) -dimensional Prym variety \Pr the inequality \irr(\Pr)\leq 2^{2g-3} holds. In view of our results, and to some extent of our techniques, it would be tempting to speculate upper bounds for the degree of irrationality of a d -dimensional principally polarized abelian variety (A,\Theta) in terms of d and of the minimal integer k such that the class k\frac{\Theta^{d-1}}{(d-1)!} is represented by an effective curve. Of course, a first step in this direction should be an analysis of Picard type bundles on Prym–Tjurin varieties. Acknowledgements. We thank Gabi Farkas and Rob Lazarsfeld for some discussions related to this circle of ideas. We are especially grateful to Martí Lahoz and Joan Carles Naranjo for suggesting the strategy to attack Theorem B . Rojas was supported by the Spanish MICINN project PID2023-147642NB-I00. 2. Preliminaries 2.1. Rational maps via kernel bundles. Our starting point is the observation that, for a given smooth projective variety X of dimension n , one can study rational maps X\dasharrow\mathbb{P}^{n} via the associated kernel sheaf. Namely, given L\in\Pic(X) and a linear system V\in\mathrm{Gr}(n+1,H^{0}(L)) without base divisors, one can consider the reflexive sheaf K sitting in an exact sequence {\lx@inpgf@ignorespaces 0} {\lx@inpgf@ignorespaces K} {\lx@inpgf@ignorespaces V\otimes\mathcal{O}_{X}} {\lx@inpgf@ignorespaces L\otimes\mathcal{I}} {\lx@inpgf@ignorespaces 0,} \scriptstyle{\lx@inpgf@ignorespaces ev} where B:=\underline{\mathop{\mathrm{Spec}}}(\mathcal{O}_{X}/\mathcal{I}) denotes the base locus of the rational map \varphi_{V}:X\dashrightarrow\mathbb{P}^{n}=\mathbb{P}(V^{\vee}) . Dualizing yields an inclusion V^{\vee}\subset H^{0}(K^{\vee}) ; note that V^{\vee} generates K^{\vee} in codimension 1 (away from B ) and also that H^{0}(K)=0 . More generally, following [ 18 ] one can consider good pairs (E,V^{\vee}) where E is a rank n reflexive sheaf with H^{0}(E^{\vee})=0 , and V^{\vee}\in\mathrm{Gr}(n+1,H^{0}(E))/\mathrm{Aut}(E) generates E in codimension 1 . The following holds: Proposition 2.1 ( [ 18 , Proposition 1.5] ) . Let X be a smooth projective variety of dimension n . There is a one-to-one correspondence \big\{\textrm{rational non-degenerate maps }X\dashrightarrow\mathbb{P}^{n}\big\}\longleftrightarrow\big\{\textrm{good pairs }(E,V^{\vee})\big\} \quad\quad\quad\quad\quad\quad\quad\,\,\,\,\,\,(L,V)\longmapsto(K^{\vee},V^{\vee}). Moreover, the fibers of the rational map \varphi_{V}:X\dashrightarrow\mathbb{P}(V^{\vee}) are described by the rule \varphi_{V}^{-1}([s])\cap(X\setminus B)=Z(s)\cap(X\setminus B), where s can be seen both as a point of the target projective space (left-hand side of the equality) and as a global section of the reflexive sheaf E (right-hand side of the equality). For varieties of dimension \geq 3 , computing the degree of a rational map (even understanding if a map is generically finite) using the description above can be difficult. However, the situation gets

Read the complete text