Universal Brauer–Severi varieties
Frank Gounelas and Daniel Huybrechts · 2025
Read Universal Brauer–Severi varieties by Frank Gounelas and Daniel Huybrechts (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-2510-20474.txt.
Introduction
Frank Gounelas and Daniel Huybrechts build universal Brauer-Severi forms of fixed span and rank and study their shape. They find the cohomology of these forms and their Brauer and Picard groups. They show the forms are, in most cases, simply linked. As a use, they reframe a known result of de Jong and Starr on discriminant avoidance in this new tongue.
Catalog record
| CREATOR | Frank Gounelas and Daniel Huybrechts |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 277 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-20474
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Universal Brauer–Severi varieties Frank Gounelas and Daniel Huybrechts Abstract . We construct universal Brauer–Severi varieties of fixed period and index and study their geometry. We determine their cohomology and their Brauer and Picard groups and show that they are almost always simply connected. As an application, we reinterpret the discriminant avoidance result of de Jong and Starr in terms of universal Brauer–Severi varieties. 0 0 footnotetext: Both authors are supported by the ERC Synergy Grant HyperK (ID 854361). The purpose of this article is to give a concrete geometric meaning to certain constructions in the literature. The main motivation is to eventually be able to use global methods in algebraic geometry to approach algebraic problems concerning Brauer groups of function fields. More concretely, we construct universal Brauer–Severi varieties {Q}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B} depending on a fixed degree n and period d such that any other Brauer–Severi variety with the same invariants is obtained by pull-back. We phrase this here as follows, see later for more precise versions. Theorem A . Fix three integers m , n , and d with d\mid n\mid d^{e} for some e . Then there exists a Brauer–Severi variety {Q}\lx@xy@svg{\hbox{\raise 2.55554pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B} of relative dimension n-1 , index n , and period d over a simply connected, quasi-projective, smooth variety {B} such that {\rm Pic}({B})\simeq\mathbb{Z}~\text{ and }~{\rm Br}({B})\simeq\lx@nobreakspace\mathbb{Z}/d\mathbb{Z}. and with the following universality property: If P\lx@xy@svg{\hbox{\raise 2.55554pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces X is any Brauer–Severi variety of relative dimension n-1 and period dividing d over a quasi-projective variety X of dimension \leq m , then P is isomorphic to the pull-back of {Q} under a classifying morphism X\lx@xy@svg{\hbox{\raise 2.55554pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.55554pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B} . All varieties will be considered over a fixed field k . The universality of our construction holds without any further assumption, but the description of the fundamental group and of the Brauer group assumes k to be algebraically closed. Note that in the above statement, the classifying map X\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B} is not necessarily unique. Also, there is in fact an ascending sequence of universal Brauer–Severi varieties {Q}_{0}\subset{Q}_{1}\subset\cdots over a sequence of closed immersions {B}_{0}\subset{B}_{1}\subset\cdots . The above result is valid for any {Q}_{i}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B}_{i} with i\gg 0 . For the minimal universal Brauer–Severi variety {Q}_{0}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B}_{0} the fundamental group is non-trivial and the Picard group is cyclic of finite order. Theorem B . For k=\mathbb{C} the singular cohomology of {B}_{i} is given as \textstyle{H^{\ast}({B}_{i},\mathbb{Q})\simeq\bigwedge^{\ast}\langle\xi_{2n+1},\ldots,\xi_{2N-1}\rangle\otimes H^{*}({\rm Gr}(N,N+i),\mathbb{Q}).} Here, N={n+d-1\choose n-1} and \xi_{2j-1}\in H^{2j-1}({B}_{i},\mathbb{Q}) are classes of weight 2j and type (j,j) in the mixed Hodge structure of B_{i} . Pulling back the classes \xi_{2j-1} under the classifying morphism \psi\colon X\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B}_{i} in Theorem A leads to classes in \psi^{\ast}\xi_{2j-1}\in H^{2j-1}(X,\mathbb{Q}) which we view as obstructions to extend the Brauer–Severi variety P\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces X . More precisely, if P\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces X extends to a Brauer–Severi variety \bar{P}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces\bar{X} over a smooth complex projective variety \bar{X} , then \psi^{\ast}(\xi_{2j-1})=0 , cf. Remark 3.7 . Ultimately, the computation describes the cohomology of the classifying space of the group that governs Brauer–Severi varieties of relative dimension n-1 with a fixed relative {\cal O}(d) , cf. Remarks 1.2 & 3.9 . However, for geometric considerations the finite approximations {B}_{i} are crucial. The idea to use universal structures in the algebraic theory of central division algebras is not new, but the geometric emphasis on Brauer–Severi varieties is. For example, Amitsur [ 1 ] constructed the universal division algebra {\rm UD}(n,k) of index n over a certain field extension of k . The notion has been studied intensely in the algebra literature, see [ 6 ] for a recent survey. At least morally, there is a link between the universal Brauer–Severi variety constructed here and {\rm UD}(n,k) . More precisely, the division algebra associated to the generic fibre {Q}_{\eta} over the function field K({B}) is obtained by specialisation of {\rm UD}(n,k) . However, we are not aware of a geometric meaningful interpretation of this link. Variants of the universal division algebra taking into account also the period have been introduced and studied by Saltman [ 43 , 44 ] , see also [ 47 ] . The construction of {Q}\lx@xy@svg{\hbox{\raise 2.5pt\hbox{\kern 1.0pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&\crcr}}}\ignorespaces{\hbox{\kern-1.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 14.0pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 1.0pt\raise-2.5pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces{B} lends itself naturally to global considerations. For example, the B_{i} , i\gg 0 , admit compactifications by projective varieties with a boundary of arbitrarily high codimension which suggests the use of cohomological methods. Theorem B is a first step in this direction and we plan to come back to this in a future paper. To demonstrate the usefulness of the notion of universal Brauer–Severi varieties for geometric problems, we revisit in § 4 work of de Jong and Starr. Their article [ 17 ] was in fact the starting point for this paper. To provide some background, recall that the period {\rm per}(\alpha) of a Brauer class \alpha is simply its order and the index {\rm ind}(\alpha) is the degree of the central division algebra representing \alpha (at the generic point). It is known classically that {\rm per}(\alpha)\mid{\rm ind}(\alpha)\mid{\rm per}(\alpha)^{e_{\alpha}} for some e_{\alpha} and Colliot-Thélène [ 11 ] conjectured that one can choose e_{\alpha}=\dim(X)-1 for all \alpha\in{\rm Br}(K(X)) with X a variety over an algebraically closed field. The conjecture has been proved for unramified classes on projective surfaces by de Jong [ 15 ] with further contributions by de Jong and Starr [ 17 ] and Lieblich [ 33 , 34 ] . Despite some recent results for unramified classes [ 16 , 28 ] , the conjecture is essentially completely open in higher dimensions. However, de Jong and Starr managed to prove that once the conjecture is settled for unramified classes, so for \alpha\in{\rm Br}(X) with X projective, it also holds for all ramified classes \alpha\in{\rm Br}(K(X)) . Their result, not explicitly stated in [ 17 ] , runs under the name of ‘discriminant avoidance’ and will be recalled as Theorem 4.1 . We provide a proof that roughly follows the original arguments in [ 17 ] but in the language of universal Brauer–Severi varieties. Acknowledgement: The text grew out of a talk in a seminar on Brauer groups in 2025. The original goal was to present the discriminant avoidance result of de Jong and Starr [ 17 ] . Looking for alternative ways to present the material, we came up