The Geometric Syzygy Conjecture in Positive Characteristic

Michael Kemeny and Peter Yi Wei · 2025

Read The Geometric Syzygy Conjecture in Positive Characteristic by Michael Kemeny and Peter Yi Wei (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-2509-00844.txt.

Introduction

Michael Kemeny and Peter Yi Wei prove the geometric syzygy conjecture in positive characteristic. Syzygies are relations among the equations that define a curve. Linear syzygy spaces collect the simplest such relations. The setting is a general smooth curve of a given genus. Genus measures the number of holes in the curve. The curve sits over an algebraically closed field of prime characteristic.

They show all linear syzygy spaces are spanned by minimal-rank syzygies. That holds provided the prime is large enough. The bound is at least twice the genus minus four. The result settles the geometric form of the conjecture here. It extends a circle of questions about equations of curves.

Catalog record

CREATORMichael Kemeny and Peter Yi Wei
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-00844

Browse the full library · About the Tower of Babel

Full text

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

The Geometric Syzygy Conjecture in Positive Characteristic Michael Kemeny and Peter Yi Wei Abstract . We show the geometric syzygy conjecture in positive characteristic. Specifically, if C C is a general smooth curve of genus g g defined over an algebraically closed field 𝔽 \mathbb{F} of characteristic p p , then all linear syzygy spaces K i , 1 ​ ( C , ω C ) \mathrm{K}_{i,1}(C,\omega_{C}) , for any integer i i are spanned by syzygies of minimal rank i + 1 i+1 provided p ≥ 2 ​ g − 4 p\geq 2g-4 . 1. Introduction Our aim in this paper is to prove the Geometric Syzygy Conjecture for canonical curves of genus g\geq 4 over algebraically closed fields \mathbb{F} of characteristic p\geq 2g-4 . Suppose that \phi\,:\,X\subseteq\mathbb{P}^{n}_{\mathbb{F}} , where \mathbb{F} is an algebraically closed field of arbitrary characteristic. A huge amount of work has gone into studying the structure of the homogeneous ideal I_{X/\mathbb{P}^{n}} considered as a graded S:=\mathrm{Sym}\,H^{0}(\mathcal{O}_{\mathbb{P}^{n}}(1)) module. In order to do this, one commonly assumes that \phi is the embedding defined by a very ample line bundle L on X and furthermore, that \phi is projectively normal , i.e. that the homogeneous coordinate ring of X can be identified with the global section ring \Gamma_{L}(X):=\bigoplus_{q\in\mathbb{Z}}H^{0}(X,L^{q}), considered as a graded S module, and we have a surjection S\twoheadrightarrow\Gamma_{L}(X) . Assuming this, I_{X/\mathbb{P}^{n}} is the kernel of this surjection, and by Hilbert’s syzygy theorem, we have a finite, minimal graded free resolution 0\to F_{k}\to\cdots\to F_{2}\to F_{1}\to I_{X/\mathbb{P}^{n}}\to 0,\textrm{ of length }k\leq n. The F_{i} are all free, graded S modules and so we can write them in terms of \mathbb{F} vector spaces \mathrm{K}_{i,j}(X,L) , called the Koszul spaces as F_{i}=\bigoplus_{j\in\mathbb{Z}}\mathrm{K}_{i,j}(X,L)\otimes_{\mathbb{F}}S(-i-j). A huge amount of work has gone into understanding this free resolution. In the case where X is a smooth, canonical curve C of genus g , we have in many cases a rather good understanding of the graded modules F_{i} . A conjecture regarding these graded modules was first made by Mark Green in 1984, known as Green’s Conjecture [ G1 ] . Whilst this is still in general open, in landmark papers by Voisin [ V1 ] , [ V2 ] it has been proven under the assumption that the canonical curve C is a general smooth curve and \mathbb{F}=\mathbb{C} . In characteristic p (for a very near optimal bound), this result was proven in [ AFPRW ] , and the bound subsequently improved [ RS ] . The upshot is that, at in characteristic zero or when \mathrm{char}(\mathbb{F})\geq\frac{g-1}{2} then, assuming C is general, we have a complete understanding of the graded modules F_{i} , [ RS ] . However, we do not have a complete understanding of the maps in the minimal free resolution of I_{C/\mathbb{P}^{g-1}} , as opposed to whether or not they vanish. In this direction, Schreyer formulated the minimal syzygy conjecture , which tells us that the modules F_{i} are spanned by elements which arise as restrictions of scrolls in \mathbb{P}^{g-1} containing the curve (see [ AN , chapter 3] ). We can describe the minimal free resolutions of these scrolls giving rise to these elements explicitly using the Eagon–Northcott complex [ S ] . To make this precise, we first need the notion of rank of a syzygy, which can be given in terms of the maps in the minimal free resolution, [ vB , §2] . We give the simple definition of rank from [ AN , chapter 3] . Assume a polarized variety (X,L) is normally generated. Then we define the rank of the syzygy \alpha\neq 0\in\mathrm{K}_{i,1}(X,L) to be the dimension of the smallest vector space V\subseteq H^{0}(X,L) such that \alpha\neq 0\in\mathrm{K}_{i,1}(X,L,V)\subseteq\mathrm{K}_{i,1}(X,L) . See [ Ke2 , §1] for a discussion on these definitions. The rank of \alpha\in\mathrm{K}_{i,1}(X,L) is always at least i+1 . Those syzygies of the minimal rank are particularly simple. They all arise from a rational normal scroll, [ AN , §3.2] . Moreover, syzygies of minimal rank can be explicitly constructed [ Ke2 , Prop. 1.9] . To be explicit, if \alpha\neq 0 is a syzygy of minimal rank, there exists a line bundle M on X with h^{0}(M)=2 and h^{0}(L\otimes M^{\vee})=i+1 such that \alpha is in the 1 -dimensional space \bigwedge^{i+1}H^{0}(L\otimes M^{\vee})\otimes\frac{H^{0}(M)}{\mathbb{F}\langle s\rangle}\subseteq\mathrm{K}_{i,1}(X,L) where s\neq 0\in H^{0}(M) , and the inclusion to \mathrm{K}_{i,1}(X,L) is induced by multiplication by s . Related to this, Schreyer has conjectured Conjecture 1.1 (Geometric Syzygy Conjecture) . Let C be a general curve of genus g . Then all linear syzygy spaces \mathrm{K}_{i,1}(C,\omega_{C}) are spanned by syzygies of minimal rank i+1 . Over \mathbb{C} , this conjecture was proven for g\leq 8 [ vB2 ] and for general genus [ Ke2 ] . In positive characteristic p>g , the conjecture was proven in the important special case g=2k and the last syzygy space \mathrm{K}_{k-1,1}(C,\omega_{C}) ( [ W ] ). The aim of this paper is to prove it in full over arbitrary fields \mathbb{F} of characteristic at least 2g-4 . Theorem 1.1 . The Geometric Syzygy Conjecture holds for general curves of genus g\geq 4 defined over an algebraically closed field \mathbb{F} of characteristic p\geq 2g-4 . As in [ Ke2 ] , the technique we use is to reduce to the case g=2k and the last linear syzygy space. However, the major difficulty in imitating the method of [ Ke2 ] is that we no longer can ensure that a general polarized K3 surface contains nodal , rational curves in the linear system of a generator of \mathrm{Pic}(X) . We overcome this obstacle by observing that the space of integral rational curves of a fixed degree is irreducible, which ensures the existence of suitable nodal rational curves that need not lie on a K3 surface. In this paper, \mathbb{F} denotes an algebraically closed field of characteristic p . Acknowledgements MK was supported by NSF grant DMS-2100782. 2. Preliminaries 2.1. Koszul cohomology Let V be a finite-dimensional vector space over a field \mathbb{F} , and let S_{V}:=\bigoplus_{d\geq 0}\mathrm{Sym}^{d}(V) denote the symmetric algebra of V . Let M=\oplus_{j\geq 0}M_{j} be a graded S_{V} -module. Definition 1 . The Koszul cohomology \mathrm{K}_{i,j}(M,V) is defined as the middle cohomology of the complex: \bigwedge^{i+1}V\otimes M_{j-1}\to\bigwedge^{i}V\otimes M_{j}\to\bigwedge^{i-1}V\otimes M_{j+1}. Koszul cohomology, introduced by Green, is a fundamental tool in the study of projective geometry. Let X be a projective variety, L a line bundle and B a coherent sheaf on X . We define the graded S_{L}:=S_{H^{0}(X,L)} -module \Gamma_{X}(B,L):=\bigoplus_{j\in\mathbb{Z}}H^{0}(X,L^{\otimes j}\otimes B). We then define \mathrm{K}_{i,j}(X,B,L):=\mathrm{K}_{i,j}(\Gamma_{X}(B,L),H^{0}(X,L)) . Given a subspace V\subseteq H^{0}(X,L) , we may regard \Gamma_{X}(\mathcal{O}_{X},L) as a S_{W} -module and define \mathrm{K}_{i,j}(X,B,L;V):=\mathrm{K}_{i,j}(\Gamma_{X}(B,L),V). When B=\mathcal{O}_{X} , we write \mathrm{K}_{i,j}(X,L):=\mathrm{K}_{i,j}(X,B,L) , and similarly \mathrm{K}_{i,j}(X,L;V) . 2.2. Kernel bundle description Let L be a very ample line bundle on a variety X . Let V\subseteq H^{0}(X,L) be a base-point free subspace of dimension r that embeds X\hookrightarrow\mathbb{P}^{r-1} . The case when V=H^{0}(X,L) corresponds to the complete linear series. The associated kernel bundle M_{V} is a vector bundle of rank r-1 defined by the following exact sequence 0\to M_{V}\to V\otimes\mathcal{O}_{X}\to L\to 0. The we have the following three lemmas. Lemma 2.1 . Suppose L is a globally generated line bundle on X . B is a coherent

Read the complete text