On a conjecture of Kazhdan and Polishchuk
Olivier Debarre · 2024
Read On a conjecture of Kazhdan and Polishchuk by Olivier Debarre (2024) 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-2411-02069.txt.
Description
Abstract . We discuss a conjecture made by Alexander Polishchuk and David Kazhdan at the 2022 ICM about a variety naturally attached to any stable vector bundle of rank 2 and degree 2 g − 1 2g-1 on a smooth projective complex curve of genus g g . 2020 Mathematics Subject Classification 14H60, 14D20, 14H40 1.
Catalog record
| CREATOR | Olivier Debarre |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 18 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2411-02069
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On a conjecture of Kazhdan and Polishchuk Thanks: This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (ERC-2020-SyG-854361-HyperK) Olivier Debarre Abstract . We discuss a conjecture made by Alexander Polishchuk and David Kazhdan at the 2022 ICM about a variety naturally attached to any stable vector bundle of rank 2 and degree 2 g − 1 2g-1 on a smooth projective complex curve of genus g g . 2020 Mathematics Subject Classification 14H60, 14D20, 14H40 1. The conjecture The aim of this note is to discuss the following conjecture made by Alexander Polishchuk and David Kazhdan. It appeared in the survey paper [ BK ] as Conjecture 3.6. Everything is over the field of complex numbers. Conjecture 1 . Let C be a smooth projective curve of genus g\geq 2 and let E be a stable vector bundle on C of rank 2 and degree 2g-1 . Let (1) \mathscr{F}_{E}\coloneqq\{([L],s)\mid[L]\in\Pic^{0}(C),\ s\in{\bf P}(H^{0}(C,E\otimes L))\}. Then, (a) \mathscr{F}_{E} is irreducible, (b) \dim(\mathscr{F}_{E})=g , (c) \mathscr{F}_{E} has rational singularities. When E is a general stable vector bundle on C of rank 2 and degree 2g-1 , the scheme \mathscr{F}_{E} is smooth irreducible of dimension g (Proposition 4 ). The aim of this note is to prove items (a) and (b) above for all stable E and all g\geq 2 . Theorem 2 . Let C be a smooth projective curve of genus g\geq 2 and let E be a stable vector bundle on C of rank 2 and degree 2g-1 . Then \mathscr{F}_{E} is irreducible, normal, and a local complete intersection of dimension g . For any n\geq 3 and g\geq 2 , we construct, on any hyperelliptic curve of genus g , stable vector bundles E of rank n and degree n(g-1)+1 for which the scheme \mathscr{F}_{E} is reducible (see Remark 5 ). So the natural extension of the conjecture to stable vector bundles of higher ranks fails in general (although it holds for E general by Proposition 4 ). After a first version of this note was written, I sent it to Alexander Braverman, who directed me to the eprint [ KP ] , where some progress is made on Conjecture 1 (called Conjecture B in that paper). In particular, the conjecture is proved there for any curve of genus 2 ( [ KP , Theorem 4.20] ) and any nonhyperelliptic curve of genus 3 ( [ KP , Theorem 4.34] ). This first version of the present note also contained a proof of the conjecture for g=2 but, since it was essentially the same as in [ KP , Theorem 4.20] , I decided to include instead in this second version a slightly different proof that makes use of Theorem 2 . Conjecture C(i) in [ KP ] is our Lemma 7 (proved by Brivio and Verra) and Conjecture C(ii) is our Lemma 9 . Conjecture C(ii) implies, by [ KP , Lemma 4.9] , the normality of \mathscr{F}_{E} . 2. Properties of the scheme \mathscr{F}_{E} and the conjecture for general E In this section, C is a smooth projective curve of genus g\geq 2 and E is a vector bundle on C of positive rank n and degree d\coloneqq n(g-1)+1 . The set \mathscr{F}_{E} was defined in ( 1 ). 2.1. General properties of the scheme \mathscr{F}_{E} For all [L]\in\Pic^{0}(C) , one has h^{0}(C,E\otimes L)\geq\chi(C,E\otimes L)=\chi(C,E)=1 , hence the first projection \pr_{1}\colon\mathscr{F}_{E}\longrightarrow\Pic^{0}(C) is dominant. Its fibers are projective spaces, hence \mathscr{F}_{E} is connected. Using a standard argument, we define a scheme structure on \mathscr{F}_{E} and show that it has dimension at least g at every point. Proposition 3 . Let C be a smooth projective curve of genus g\geq 2 and let E be a vector bundle on C of rank n and degree n(g-1)+1 . The scheme \mathscr{F}_{E} has everywhere dimension at least g and it is a local complete intersection at every point where its dimension is exactly g . Proof. Let D be an effective divisor on C of degree g-1 . By Serre duality, we have H^{1}(C,E\otimes L(D))\simeq H^{0}(C,\omega_{C}\otimes E^{\vee}\otimes L^{\vee}(-D))^{\vee}=\Hom(L(D),\omega_{C}\otimes E^{\vee}) and the latter space vanishes for all [L]\in\Pic^{0}(C) by stability of \omega_{C}\otimes E^{\vee} . When [L] varies in \Pic^{0}(C) , the vector spaces H^{0}(C,E\otimes L(D)) are then the fibers of a vector bundle \mathscr{F} on \Pic^{0}(C) of rank \chi(C,E(D))=1+n(g-1) , whereas the vector spaces H^{0}(C,E\otimes L(D)|_{D}) are the fibers of a vector bundle \mathscr{G} on \Pic^{0}(C) of rank m\coloneqq n(g-1) . The kernel at a point [L] of the restriction morphism (2) \alpha\colon\mathscr{F}\longrightarrow\mathscr{G} is H^{0}(C,E\otimes L) . The vector bundles \mathscr{F} and \mathscr{G} have respective ranks m+1 and m . Set \mathscr{X}\coloneqq\{(u,[x])\in\Hom({{\bf C}}^{m+1},{{\bf C}}^{m})\times{\bf P}({{\bf C}}^{m+1})\mid u(x)=0\}. Looking at the projection \mathscr{X}\to{\bf P}({{\bf C}}^{m+1}) , we see that \mathscr{X} is smooth of codimension m in \Hom({{\bf C}}^{m+1},{{\bf C}}^{m})\times{\bf P}({{\bf C}}^{m+1}) . The variety \Pic^{0}(C) is covered by open subsets U with morphisms to \Hom({{\bf C}}^{m+1},{{\bf C}}^{m}) such that \pr_{1}^{-1}(U)=U\times_{\Hom({{\bf C}}^{m+1},{{\bf C}}^{m})}\mathscr{X}. This defines a scheme structure on \mathscr{F}_{E} that makes it into a subscheme of {\bf P}(\mathscr{F}) . Since \mathscr{F}_{E} is not empty, it follows that it has everywhere codimension at most m in {\bf P}(\mathscr{F}) and that it is a local complete intersection at every point where its codimension is exactly m . ∎ 2.2. Brill–Noether loci Let r be a nonnegative integer. The Brill–Noether locus (3) B_{r}(E)\coloneqq\{[L]\in\Pic^{0}(C)\mid h^{0}(C,E\otimes L)\geq r+1\} is the degeneracy subscheme of \Pic^{0}(C) where the morphism \alpha\colon\mathscr{F}\to\mathscr{G} defined in ( 2 ) has corank \geq r ; therefore, it is everywhere of codimension \leq r(r+1) in \Pic^{0}(C) . These loci were studied by Ghione in [ G ] and Lazarsfeld in [ Lz , § 2] : they prove (in our situation) that B_{r}(E) is nonempty when r(r+1)\leq g , connected when r(r+1)<g , and that when B_{r}(E) is everywhere of codimension r(r+1) in \Pic^{0}(C) , it is Cohen–Macaulay and its cohomology class is \Bigl(\prod_{i=0}^{r}\frac{i!}{(r+i)!}\Bigr)\ n^{r(r+1)}\ \theta^{r(r+1)}, where \theta is the canonical principal polarization on \Pic^{0}(C) . 2.3. The conjecture for general vector bundles Stable vector bundles E considered here have a fine moduli space U_{C}(n,d) (since n and d are relatively prime, stability and semistability are equivalent). This is a smooth projective variety of dimension n^{2}(g-1)+1 . When one fixes the determinant, one obtains a smooth projective irreducible fine moduli space SU_{C}(n,d)\subseteq U_{C}(n,d) . For pairs (E,V) , where E is a vector bundle on C and V\subseteq H^{0}(C,E) a one-dimensional space of sections of E , Bradlow, Thaddeus ( [ T ] ), Le Potier ( [ Le ] ), and others introduced a notion of stability depending on a real parameter \sigma ( [ BGMN , Definition 2.1] ). Without going into details, let us just mention that in our situation (where n and d are relatively prime), for \sigma positive and small enough, stability of the pair (E,V) and stability of the bundle E are equivalent, hence are independent of \sigma ( [ BGMN , Proposition 2.5] ). The corresponding moduli space for stable pairs (E,V) , which we denote by G_{C}(n,d) , 1 1 1 It is denoted by G_{0}(n,d,1) , or simply G_{0} , in [ BGMN , Section