The period–index problem for hyper-Kähler varieties via hyperholomorphic bundles

James Hotchkiss, Davesh Maulik, Junliang Shen, Qizheng Yin, Ruxuan Zhang · 2025

Read The period–index problem for hyper-Kähler varieties via hyperholomorphic bundles by James Hotchkiss, Davesh Maulik, Junliang Shen, Qizheng Yin, Ruxuan Zhang (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-2502-09774.txt.

Introduction

James Hotchkiss, Davesh Maulik, Junliang Shen, Qizheng Yin, and Ruxuan Zhang tighten a bound in 2025 on a puzzle about twisted bundles over a rich class of shapes. The period-index puzzle asks how far apart two measures of a twist can drift. Using special bundles built by Markman, the team show the shape's size caps the drift for shapes of the K3-family kind. When the shape's curve rank runs at least two, most twists obey a cap of half that size, backing a guess in the field.

Catalog record

CREATORJames Hotchkiss, Davesh Maulik, Junliang Shen, Qizheng Yin, Ruxuan Zhang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-09774

Browse the full library · About the Tower of Babel

Full text

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

The period–index problem for hyper-Kähler varieties via hyperholomorphic bundles James Hotchkiss, Davesh Maulik, Junliang Shen, Qizheng Yin, Ruxuan Zhang Abstract . We prove new bounds for the period–index problem for hyper-Kähler varieties of K ​ 3 [ n ] K3^{[n]} -type using projectively hyperholomorphic bundles constructed by Markman. We show that dim ⁡ ( X ) \mathrm{dim}(X) is a bound for any X X of K ​ 3 [ n ] K3^{[n]} -type. We also show that 1 2 ​ dim ​ ( X ) \frac{1}{2}\mathrm{dim}(X) is a bound for most Brauer classes when the Picard rank of X X is at least two, providing evidence for a conjecture of Huybrechts. 0. Introduction Throughout, we work over the complex numbers {\mathbb{C}} . 0.1. The period–index problem Let X be a nonsingular projective variety, and let \alpha\in\mathrm{Br}(X) be a class in the Brauer group of X . Two basic invariants of \alpha are its period and its index . The period \mathrm{per}(\alpha) is nothing more than the order of \alpha in \mathrm{Br}(X) , which is a torsion group, while the index is given by \mathrm{ind}(\alpha)=\gcd\{\sqrt{\mathrm{rk}(\mathcal{A})}\} , as \mathcal{A} runs over the Azumaya algebras of class \alpha . It is an elementary fact that \mathrm{per}(\alpha) and \mathrm{ind}(\alpha) share the same prime factors, and that \mathrm{per}(\alpha) divides \mathrm{ind}(\alpha) . In this paper, we are concerned with the period–index problem for the variety X , which is the problem of determining the least positive integer e(X) such that the inequality (1) \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{e(X)} holds for all \alpha\in\mathrm{Br}(X) . The motivation for the problem comes from the theory of central simple algebras over the function field \mathbb{C}(X) : There is a natural inclusion \mathrm{Br}(X)\subset\mathrm{Br}(\mathbb{C}(X)) , and the index of a class \alpha\in\mathrm{Br}(X) coincides with \deg(D):=\sqrt{\mathrm{dim}_{\mathbb{C}(X)}(D)} , where D is the unique division algebra over \mathbb{C}(X) of class \alpha , viewed as an element of \mathrm{Br}(\mathbb{C}(X)) . Determining the degree of a central division algebra over a field in terms of its period is a fundamental classical problem, originating in the study of Brauer groups of local and global fields from the early part of the twentieth century; for an introduction, see [ 1 , Section 4] or [ 3 ] . If X is a surface, then \mathrm{per}(\alpha)=\mathrm{ind}(\alpha) for all \alpha\in\mathrm{Br}(X) by a theorem of de Jong [ 5 ] . In higher dimensions, however, little is known, beyond the fact (proved in [ 11 ] ) that a bound as in ( 1 ) exists. From the longstanding period–index conjecture for the function field \mathbb{C}(X) from [ 4 ] , one expects (2) \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{\mathrm{dim}(X)-1} for Brauer classes on all nonsingular projective varieties X ; moreover, there are varieties of each dimension for which the bound ( 2 ) is sharp [ 2 ] . In [ 8 ] , the period–index conjecture ( i.e. , \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{2} ) was proven for Brauer classes on abelian threefolds. In [ 10 ] , Huybrechts studied the period–index problem for hyper-Kähler varieties, and proved the following result concerning Lagrangian fibrations. It shows that if X admits a Lagrangian fibration, then e(X) can be far smaller than the exponent \mathrm{dim}(X)-1 occuring in the period–index conjecture, at least if \mathrm{per}(\alpha) avoids finitely many primes. Theorem 0.1 (Huybrechts [ 10 ] ) . Let X be a hyper-Kähler variety which admits a Lagrangian fibration. Then there exists an integer N_{X} such that (3) \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{\frac{\mathrm{dim}(X)}{2}} for all \alpha\in\mathrm{Br}(X) with \mathrm{per}(\alpha) coprime to N_{X} . An explicit description of N_{X} was given in [ 10 , (0.1)] . Huybrechts further conjectured that the equation ( 3 ) holds with N_{X}=1 for any hyper-Kähler variety X [ 10 , Conjecture 0.2] , and verified it for the Hilbert scheme of points on a K3 surface [ 10 , Theorem 0.4] . The purpose of this note is to study the period–index problem for hyper-Kähler varieties of K3^{[n]} -type. Using Markman’s projectively hyperholomorphic bundles from [ 14 ] , we are able to obtain new bounds for the period–index problem. For varieties of K3^{[n]} -type, our results also recover and provide a new proof of Theorem 0.1 . The idea of applying projectively hyperholomorphic bundles to classical problems about Brauer groups goes back to the work of Huybrechts and Schröer [ 12 ] on Grothendieck’s question for Brauer groups of analytic K3 surfaces. 0.2. Hyper-Kähler varieties of K3^{[n]} -type Throughout Section 0.2 , we consider X a hyper-Kähler variety of K3^{[n]} -type, i.e. , it is deformation equivalent to the Hilbert scheme of n points on a K3 surface. As the K3 surface case has been dealt with, we assume n\geq 2 . Our first result provides a uniform bound for any X of K3^{[n]} -type. For notational convenience, we set I_{X}:=\mathrm{Index}\left(T(X)\hookrightarrow H^{2}(X,Z)/\mathrm{Pic}(X)\right) where T(X)\subset H^{2}(X,{\mathbb{Z}}) is the transcendental lattice. Note that the embedding of the lattices on the right-hand side is always of finite index. Theorem 0.2 . Assume that X admits a primitive polarization of degree 2h . We have (4) \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{\mathrm{dim}(X)} for all \alpha\in\mathrm{Br}(X) with \mathrm{per}(\alpha) coprime to n!hI_{X} . In Section 2.3 , we discuss numerical obstructions to proving Huybrechts’ conjecture ( 3 ) using our method. In particular, our approach is not expected to be sufficient to prove Huybrechts’ conjecture for the Picard rank 1 case. Remark 0.3 . To the best of our knowledge, ( 4 ) gives the best bound currently known for K3^{[n]} -type varieties of Picard rank 1 . In [ 10 , Section 2.6] , Huybrechts proved that the Fano variety of lines F(Y) of a nonsingular cubic 4 -fold Y , which is a variety of K3^{[2]} -type, satisfies \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{5},\quad\alpha\in\mathrm{Br}(F(Y)) if \mathrm{per}(\alpha) is coprime to some uniform integer depending on Y . In this case, Theorem 0.2 reads \mathrm{ind}(\alpha)\mid\mathrm{per}(\alpha)^{4} when \mathrm{per}(\alpha) avoids finitely many prime factors. When X has higher Picard rank, much better bounds can be obtained. For further discussions, we need to introduce some notation. Recall the explicit description of the Brauer group \mathrm{Br}(X)=\left(\frac{H^{2}(X,{\mathbb{Z}})}{\mathrm{Pic}(X)}\right)\otimes{\mathbb{Q}}/{\mathbb{Z}}, which allows us to present a Brauer class in the form of a “ B -field”: (5) \alpha=\left[\frac{{\mathcal{B}}}{\ell}\right]. Here \ell is a positive integer, and {\mathcal{B}} is an element of the transcendental lattice T(X)\subset H^{2}(X,{\mathbb{Z}}) which is not divisible by any prime factor of \ell . We say that the Brauer class ( 5 ) is non-special if \mathrm{gcd}(q({\mathcal{B}}),\ell)=1, where q({\mathcal{B}})\in{\mathbb{Z}} is the norm with respect to the Beauville–Bogomolov–Fujiki (BBF) form q(-,-) on H^{2}(X,{\mathbb{Z}}) . If \mathrm{per}(\alpha) is coprime to I_{X} , we further have \mathrm{per}(\alpha)=\ell. The following result is our main tool of constructing Brauer classes satisfying ( 3 ). Theorem 0.4 . For a Brauer class \alpha=\left[\frac{{\mathcal{B}}}{\ell}\right] with \mathrm{per(\alpha)} coprime to n!I_{X} , if there exist classes {\mathcal{D}},{\mathcal{D}}^{\prime}\in\mathrm{Pic}(X) and a positive integer d with \mathrm{gcd}(d,\ell)=1 satisfying that (i) q({\mathcal{D}})\equiv-q({\mathcal{B}})d^{2}\mod\ell , (ii) \mathrm{gcd}\left(q({\mathcal{D}},{\mathcal{D}}^{\prime}),q({\mathcal{B}}),\ell\right)=1 , then ( 3 ) holds for \alpha . As an application, Theorem 0.4 implies immediately Theorem 0.1 for hyper-Kähler varieties of K3^{[n]} -type; see Section 2.5 . The next theorem is also a consequence of Theorem 0.4 , which verifies ( 3 ) for most Brauer classes when the Picard rank is at least 2 . Theorem 0.5 . If X has Picard rank at least 2 , then there exists an integer N_{X}

Read the complete text