Some evidence for the existence of Ulrich bundles
Ştefan Deaconu · 2025
Read Some evidence for the existence of Ulrich bundles by Ştefan Deaconu (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-2502-18250.txt.
Introduction
Stefan Deaconu gathers evidence for a hard existence question in a 2025 mathematics paper. The question asks whether every smooth projective variety carries a special kind of vector bundle called an Ulrich bundle. Deaconu poses the problem in weaker forms, asking for a solution counted in an algebraic ledger called the Grothendieck group, or built inside a derived category. He shows a ledger-level answer always exists if a certain bundle is algebraic, and sketches a formal route to such sheaves on surfaces.
Catalog record
| CREATOR | Ştefan Deaconu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18250
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Some evidence for the existence of Ulrich bundles Ştefan Deaconu Abstract . The question of existence of Ulrich bundles on nonsingular projective varieties is posed here in weaker terms: either to find a K–theoretic solution, or to find one in the derived category of the variety. We observe that if any motivic vector bundle is algebraic, there is always a solution in the Grothendieck group. Also, by considering the derived problem, it is noted a formal way of producing Ulrich sheaves on a surface. 1. Introduction In general, by k it is meant an algebraically closed field. Definition 1.1 . An Ulrich bundle on a projective variety (X,\mathcal{O}_{X}(1)) of dimension d over k is a locally free sheaf \mathcal{E} such that \displaystyle H^{i}(X,\mathcal{E}\otimes\mathcal{O}_{X}(j))=0\text{ for all }-d\leq j\leq-1\text{ and all }i\in\mathbb{N}. There are two others equivalent definitions for an Ulrich bundle [ Be , Theorem 1.1] . The guiding problem is to find such bundles on any nonsingular projective variety over k (this geometric form of the problem was first raised in [ EiSc ] ). In fact, the guiding problem is the fundamental problem of the extrinsic point of view: to pin down equations of varieties by resolving sheaves. For curves, this was accomplished by Petri [ Mu2 , A.I] , [ ArCoGrHa , III.3] , but in general less is known. Now, the idea is that in the presence of an Ulrich bundle, the Chow form can (almost) be determined linearly. For instance: Example 1.1 . If X is a hypersurface of degree d in some projective space \mathbb{P}_{k}^{N} , and f=0 is its equation, then the existence of an Ulrich bundle of rank r on X allow us to write f^{r}=det(l_{ij}) , where (l_{ij}) is a d\times d matrix of linear forms on \mathbb{P}_{k}^{N} ( [ Be , Proposition 1] , or [ EiSc , Section 3] for more). We note that in the case of curves, one can use invariant polynomials in the Chow form of the curve in order to embed the moduli space of curves in a projective space [ Mu2 , A.II] . Thus, if any nonsingular projective variety is to carry an Ulrich bundle, one might hope that this will improve our understanding of moduli in larger dimensions. The purpose of this note is to give an evidence to this question in the form of a weak (K–theoretic) solution (Corollary 3.2 ). More precisely, the idea is this: one first notice that, over k=\mathbb{C} , a topological solution to the problem can be produced easy enough, so then the question is to find a more algebraic language allowing to use homotopy reasons; such a language was introduced by Morel and Voevodsky, which was used to establish a notion of motivic vector bundle as presented in [ AsFa ] . One can always pass from a vector bundle to a motivic one, and under the assumption that this process produce all motivic bundles (only in the smooth context), a K–theoretic Ulrich bundle can be found. In the final part, the notion of an Ulrich sheaf is introduced in derived version, and it is noticed that a derived Ulrich sheaf on a surface always gives an Ulrich bundle (Corollary 3.4 ). 2. Preliminaries By K_{0,\mathbb{Q}} we shall denote the functor X\mapsto K_{0}(X)\otimes_{\mathbb{Z}}\mathbb{Q} , associating to a noetherian scheme X the Grothendieck group of the category of coherent sheaves on X , extended over the rationals. Also, K^{0}_{\mathbb{Q}} will denote the functor given by X\mapsto K^{0}(X)\otimes\mathbb{Q} , where K^{0}(X) is the Grothendieck group of the category of locally free sheaves on X (see [ Ma ] ). Recall the following fundamental fact: Theorem 2.1 . (Kleiman, [ Ma , 1.9] ) If X is a regular noetherian scheme having an ample invertible sheaf on it, the tautological map \displaystyle K^{0}(X)\rightarrow K_{0}(X) is an isomorphism of K^{0}(X) –modules. Definition 2.1 . For a scheme X , by a Jouanolou device it is meant an affine vector bundle torsor p:\tilde{X}\rightarrow X . In order to present the existence result concerning the notion of Jouanolou device in full generality, recall that a quasi-compact and quasi-separated scheme X is called divisorial if it has an ample family of invertible sheaves \{\mathcal{L}_{i}\}_{i} , i.e. the distinguished open subsets D(f)=\{f\neq 0\},f\in\text{H}^{0}(X,\mathcal{L}_{i}^{\otimes n}) , form a base for the Zariski topology (this class of varieties were introduced first in S.G.A 6). This said, we have: Lemma 2.1 . (Jouanolou-Thomason, [ We , 4.4] ) Any divisorial quasiseparated and quasicompact scheme admits a Jouanolou device. This trick was brought in Algebraic Geometry by Jouanolou from a construction in Algebraic Topology (as with many others ideas) in order to extend the K –spectrum from commutative rings to schemes (according to [ We ] ). Example 2.1 . For instance, the standard Jouanolou device for a projective variety over k is constructed as follows. First, one observe that it suffices to construct it on a projective space (by pulling back through a finite morphism, according to Lemma 3.2 below). For X=\mathbb{P}_{k}^{n} , take \widetilde{X} to be the affine variety given by the complement of the incidence divisor in \mathbb{P}^{n}_{k}\times(\mathbb{P}^{n}_{k})^{\vee} , together with the projection onto the first factor. This is affine indeed: it is the complement of the hypersurface given by the equation x_{0}y_{0}+\ldots+x_{n}y_{n}=0 , which we see via the Segre embedding \mathbb{P}^{n}_{k}\times\mathbb{P}^{n}_{k}\rightarrow\mathbb{P}^{n^{2}+2n}_{k} that this hypersurface is the preimage of a hyperplane [ Ha , I.Ex.2.14] ; moreover, each fiber is given by the hyperplanes not containing a given point, so it is isomorphic to \mathbb{A}_{k}^{n} . For another variant: take \widetilde{\mathbb{P}_{k}^{n}} to be the variety given by all (n+1)\times(n+1) matrices which are idempotent and have rank one. Using the characteristic polynomial, we see that this is affine, and seeing each matrix as a linear transformation, we fiber it over \mathbb{P}_{k}^{n} by sending a matrix to its image (makes sense, since these matrices have rank 1 ). Theorem 2.2 . [ AsFa , 5.1.2] Assume \text{dim}(X)\leq 3 . If p:\tilde{X}\rightarrow X is a Jouanolou device, then any locally free sheaf on \tilde{X} admits a descent datum relative to p . Remark 2.1 . It is hoped that Theorem 2.2 can be extended to any dimension [ AsFa , 5.1.1] . Just to briefly mention why is this important, given a nonsingular projective variety X over k , set \mathcal{V}^{r}_{\text{alg}}(X) to be the collection of locally free sheaves on X_{\text{Zar}} , of rank r , and \mathcal{V}^{r}_{\text{mot}}(X) to be that of motivic locally free sheaves of rank r on X , which is the collection of \mathbb{A}^{1} –homotopy classes of maps from X_{\text{an}} to the corresponding infinite grassmanian [ AsFa , 3.6.1] . Also, let \mathcal{V}^{r}_{\text{alg}}(\tilde{X}) and \mathcal{V}^{r}_{\text{mot}}(\tilde{X}) be defined similarly for \tilde{X} , where p:\tilde{X}\rightarrow X is a Jouanolou device for X (see Lemma 2.1 ). Consider the following diagram: {\lx@inpgf@ignorespaces\mathcal{V}^{r}_{\text{alg}}(X)} {\lx@inpgf@ignorespaces\mathcal{V}^{r}_{\text{mot}}(X)} {\lx@inpgf@ignorespaces\mathcal{V}^{r}_{\text{alg}}(\tilde{X})} {\lx@inpgf@ignorespaces\mathcal{V}^{r}_{\text{mot}}(\tilde{X})} \scriptstyle{\lx@inpgf@ignorespaces p_{X}} \scriptstyle{\lx@inpgf@ignorespaces p^{*}} \scriptstyle{\lx@inpgf@ignorespaces p^{*}_{\text{mot}}} \scriptstyle{\lx@inpgf@ignorespaces p_{\tilde{X}}} Since \tilde{X} is affine, the Morel–Schlichting theorem asserts that \mathcal{V}^{r}_{\text{alg}}(\tilde{X})\simeq\mathcal{V}^{r}_{\text{mot}}(\tilde{X}) functorially [ AsFa , 3.4.3] . Also, \mathcal{V}^{r}_{\text{mot}}(X)=\mathcal{V}^{r}_{\text{mot}}(\tilde{X}) (by definition of the motivic vector bundles, since \tilde{X} is an affine torsor over X ). Therefore, the meaning of the positive answer to [ AsFa , 5.1.1] would be that any motivic vector bundle on