Perverse Sheaves on the Stack of G G -Zips

Christopher Lang · 2025

Read Perverse Sheaves on the Stack of G G -Zips by Christopher Lang (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-2505-10362.txt.

Introduction

Christopher Lang explains, in solo 2025 work, how to compute the simple twisted sheaves on a stack built from a group's zip forms. Such stacks sort paired flags in a group's world, a shape that codes deep traits of curved forms. Twisted sheaves are the stack's core build bits, sorted by how they twist at its seams. Lang lays out the full count path step by step. Worked cases then show the path in play across a spread of groups.

Catalog record

CREATORChristopher Lang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-10362

Browse the full library · About the Tower of Babel

Full text

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

Perverse Sheaves on the Stack of G G -Zips Christopher Lang Abstract 1 Introduction If we have a Shimura variety {\mathcal{S}} of abelian type associated to a Shimura datum (\mathbb{G},X) , one can look at its reduction {\mathcal{S}}_{0} at a prime p of good reduction and get a smooth map \zeta:{\mathcal{S}}_{0}\to G\text{-}\mathrm{Zip}^{\mu}, see [ SZ22 ] . Here, G is the base change to \mathbb{F}_{q} of an integral model of the adjoint of \mathbb{G} , and \mu is the reduction of a cocharacter from the Shimura datum (\mathbb{G},X) . The idea is that the stack of G -zips is easier to study than Shimura varieties. Let’s explain why that is and let’s give a rough definition of G -zips in the process. If the Shimura variety is of PEL type, so that it is the moduli space of abelian varieties with additional structure, the map \zeta associates to an abelian variety its first de Rham cohomology, which naturally comes equipped with the Hodge filtration and the conjugate filtration, and we have Frobenius-linear isomorphisms between the graded pieces of the two filtrations. Such an object, namely a vector space with two filtrations and Frobenius-linear isomorphisms of the graded pieces, is called an F-zip. These are objects of linear algebra, and hence easier to understand than Shimura varieties. Now, a G -zip is just an F-zip that is equipped with a structure of the group G . By that we mean an exact tensor functor from the category of representations of G to the category of F-zips. The connected components of the stack of G -zips are classified by conjugacy classes of cocharacters, and one such connected component is called G\text{-}\mathrm{Zip}^{\mu} , the G -zips of type \mu . The fibres of the map \zeta are the Ekedahl-Oort strata of the Shimura variety. Furthermore, the stack of G -zips of type \mu can be identified as a quotient stack [E_{\mu}\backslash G] , where E_{\mu} is a subgroup of G\times G acting on G by (g_{1},g_{2}).g=g_{1}gg_{2}^{-1} . The underlying topological space of G\text{-}\mathrm{Zip}^{\mu} can be explicitly computed using the Weyl group of G , and the stabilizers of the quotient stack are semi-direct products of a finite group of Lie type with a connected unipotent group. The idea is now that if you are interested in some type of object over the Shimura variety, a good starting point can be to look at this type of object over the stack of G -zips and then pull it back to the Shimura variety. For example you can do this for the category of perverse sheaves: On the stack of G -zips of type \mu , these are E_{\mu} -equivariant perverse sheaves on G , and since the map \zeta is smooth, we obtain a shifted pullback functor to the category of perverse sheaves on the Shimura variety. Perverse Sheaves on G -Zips We compute the simple perverse sheaves on the stack of G -zips of type \mu:\mathbb{G}_{m}\to G in several examples. It follows from the general theory that these can be computed using the following principle: Theorem 1 (Theorem 2.3 ) . Given a connected reductive group G over \mathbb{F}_{q} and a cocharacter \mu:\mathbb{G}_{m,\overline{\mathbb{F}}_{q}}\to G_{\overline{\mathbb{F}}_{q}} , the simple perverse sheaves on the stack of G -zips of type \mu are given by pairs (w,\theta) , where w is an orbit of the E_{\mu} action on G and \theta is an irreducible \overline{\mathbb{Q}}_{l} -representation of the group of connected components of the stabilizer of this group action at w . The set of orbits is {}^{I}W , where W is the Weyl group of G and I is the type of \mu , and the group of connected components of the stabilizer of this group action at w is known to be \Pi_{w}\coloneqq\left\{\,h\in H_{w}(\overline{\mathbb{F}}_{p})\ ;\ h=\varphi\big(\dot{y}\dot{w}h(\dot{y}\dot{w})^{-1}\big)\,\right\}, where H_{w} is some Levi-subgroup of G_{\overline{\mathbb{F}}_{q}} and y\in G depends only on I . This can be understood as a group over \overline{\mathbb{F}}_{p} together with a twisted Frobenius action, which yields a descent datum to \mathbb{F}_{q} , and the group above are just the \mathbb{F}_{q} -valued points thereof. Hence one needs to understand the irreducible \overline{\mathbb{Q}}_{l} -representations of certain finite groups of Lie type, and these can in principle be understood with Deligne-Lusztig theory. We list some examples: • If one takes the constant cocharacter, there is only one orbit and the corresponding group is \Pi_{w}=G(\mathbb{F}_{q}) . • For G=\GL_{n} and the cocharacter that corresponds to a Borel subgroup, the set of orbits is W=\mathrm{S}_{n} , the group H_{w} is a torus and \Pi_{w} is a product of some unit groups of finite fields. In particular, these are abelian groups, so their irreducible representations can be written down explicitly. • For the symplectic group \Sp_{2n} and the cocharacter that corresponds to the parabolic with two n\times n -blocks on the diagonal, the set of orbits is \{0,1\}^{n} , and for the closed orbit w=\id we have \Pi_{\id}=\U_{n}(\mathbb{F}_{q}) , while for the open orbit we have \Pi_{w}=\GL_{n}(\mathbb{F}_{q}) . The other \Pi_{w} we computed for n=2 , where they are again abelian groups. • If we take the general unitary group G=\GU_{n} with a cocharacter \mu of type (1,n-1) , then the set of orbits is totally ordered of size n . For each orbit w , the group \Pi_{w} is an extension of \GU_{n_{w}}(\mathbb{F}_{q}) by some abelian group, where 0\leq n_{w}<n is an integer depending on w . Notations • We once and for all fix a prime p and an algebraic closure \overline{\mathbb{F}}_{p} of \mathbb{F}_{p} , and if q is a power of p , we view \mathbb{F}_{q} as a subfield of \overline{\mathbb{F}}_{q}\coloneqq\overline{\mathbb{F}}_{p} . • If we have a scheme X over S and a morphism T\to S , we write X_{T}\coloneqq X\underset{S}{\otimes}T . If T=\Spec A , we also write X_{A}\coloneqq X_{T} . • The natural numbers \mathbb{N} contain 0 . • The symbol \varphi will always denote the q -Frobenius morphism whenever we live over \mathbb{F}_{q} . For matrix groups this means raising the matrix entries to the q -th power and is denoted by M^{(q)} , and we also get an induced Frobenius on Weyl groups, compare [ PWZ11 , (3.11)] , which we will again denote by \varphi . • If we have a group G and x\in G , we write \mathrm{int}(x):G\to G,\quad g\mapsto{}^{x}g\coloneqq xgx^{-1}. We also write {}^{x}A\coloneqq xAx^{-1} for some subset A\subseteq G . • \mathbbl{1}_{n} denotes the identity n\times n -matrix, \diag(a_{1},\dots,a_{n}) the diagonal n\times n -matrix with the a_{i} on the diagonal and \mathrm{antidiag}_{n}(c) the n\times n -matrix with a_{ij}=c if i+j=n+1 and 0 otherwise. • All reductive groups G are assumed to be connected. As we basically always consider reductive groups over finite fields, we can choose a maximal torus and a Borel subgroup T\subseteq B\subseteq G , so that we get the Weyl group with its simple reflections. We write \mathcal{R}_{u}P for the unipotent radical of a parabolic subgroup of G , and \Par_{I} for the scheme of parabolic subgroups of G of type I , where I is a subset of the simple reflections. • Given a group scheme G over L and a field extension L/K , we write \Res_{L/K}G:\mathcal{S}\textup{ch}/K\to\mathcal{S}\textup{ets},\quad S\mapsto G(S_{L}). • If we

Read the complete text