Uniform stratified vanishing and equidistribution on 𝔾 m d \mathbb{G}_{m}^{d}

Amadou Bah, K. V. Shuddhodan · 2026

Read Uniform stratified vanishing and equidistribution on 𝔾 m d \mathbb{G}_{m}^{d} by Amadou Bah, K. V. Shuddhodan (2026) 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-2608-00304.txt.

Introduction

Bah and Shuddhodan prove a uniform generic vanishing theorem for perverse sheaves on G_m^d over finite fields. The setting is a d-fold product of the multiplicative group. Constants in the theorem depend only on dimension and complexity in Sawin's sense. Stratified vanishing is the core claim. Uniformity across sheaves is its strength.

An equidistribution theorem follows as a corollary. It extends Katz's theorem for G_m to the higher product. The corollary covers sequences of perverse sheaves of bounded complexity. Those sheaves share a common tannakian monodromy group. Fields may vary in characteristic across the sequence. As field sizes grow, Frobenius classes equidistribute in a maximal compact subgroup's class space.

Catalog record

CREATORAmadou Bah, K. V. Shuddhodan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE130 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2608.00304

Browse the full library · About the Tower of Babel

Full text

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

Uniform stratified vanishing and equidistribution on 𝔾 m d \mathbb{G}_{m}^{d} Amadou Bah, K. V. Shuddhodan Abstract . We prove a uniform stratified generic vanishing theorem for perverse sheaves on 𝔾 m d \mathbb{G}_{m}^{d} over finite fields, with constants depending only on the dimension and on the complexity à la Sawin. As a corollary, we prove an equidistribution theorem for 𝔾 m d \mathbb{G}_{m}^{d} extending Katz’s theorem for 𝔾 m \mathbb{G}_{m} . The corollary applies to sequences of perverse sheaves of bounded complexity with common tannakian monodromy group, over finite fields of varying characteristic. As the cardinalities of the fields tend to infinity, the Frobenius conjugacy classes attached to the multiplicative characters equidistribute in the space of conjugacy classes of a maximal compact subgroup. 2020 Mathematics Subject Classification Primary 14F20; Secondary 11T23, 14G15, 11L05 1. Introduction Let k=\mathbb{F}_{q} be a finite field of characteristic p , let \bar{k} be a fixed algebraic closure of k , let \ell\neq p be a prime, and let \mathbb{G}_{m}^{d}/k be the split d -dimensional torus. We write \mathbb{G}_{m}^{d}/\bar{k} for its geometric base change and use unadorned \mathbb{G}_{m}^{d} for this geometric torus. For n\geq 1 , k_{n}=\mathbb{F}_{q^{n}} denotes the extension of k of degree n inside \bar{k} . Through the Lang isogeny, a character \chi:\mathbb{G}_{m}^{d}(k_{n})\to\overline{\mathbb{Q}_{\ell}}^{\times} determines a rank-one Kummer local system \mathcal{L}_{\chi} on \mathbb{G}_{m}^{d} . If M_{0} is perverse on \mathbb{G}_{m}^{d}/k and M is its geometric base change, we set M_{\chi}=M\otimes\mathcal{L}_{\chi} . To this we can associate a character sum S(M_{0};k_{n},\chi)=\sum_{x\in\mathbb{G}_{m}^{d}(k_{n})}t_{M_{0}}(x;k_{n})\,\chi(x), where t_{M_{0}} is the trace function of M_{0} . This class includes Gauss sums, Kloosterman sums, and their multivariable analogues. By the Grothendieck–Lefschetz trace formula, the sum is the alternating Frobenius trace on R\Gamma_{c}(\mathbb{G}_{m}^{d},M_{\chi}) . Generic vanishing theorems describe how this cohomology varies with \chi . Tannakian methods turn this description into equidistribution theorems for the sums. (1) Gabber–Loeser proved generic vanishing on tori [ GL96 , Cor. 2.3.2] . Assuming resolution of singularities and simplification of ideals for varieties of dimension <d , they proved that the locus where the forget-supports morphism R\Gamma_{c}(\mathbb{G}_{m}^{d},M_{\chi})\to R\Gamma(\mathbb{G}_{m}^{d},M_{\chi}) is not an isomorphism is contained in a finite union of translated algebraic cotori of the character scheme \widehat{\mathbb{G}_{m}^{d}} [ GL96 , Th. 4.1.1’] . 1 1 1 The \overline{\mathbb{Q}_{\ell}} -scheme of \overline{\mathbb{Q}_{\ell}}^{\times} -characters of \mathbb{G}_{m}^{d} , whose \overline{\mathbb{Q}_{\ell}} -points parametrize rank-one Kummer local systems. The construction is recalled in § 2.5 . Forey–Fresán–Kowalski removed these hypotheses [ FFK21 , Th. 2.12] . (2) For d=1 , Katz developed a tannakian theory on \mathbb{G}_{m} and proved equidistribution theorems for these character sums [ Kat12 ] . (3) Forey, Fresán, and Kowalski [ FFK21 ] extended the tannakian framework to every connected commutative algebraic group over a finite field. For a fixed generically unramified, arithmetically semisimple 2 2 2 A perverse sheaf on \mathbb{G}_{m}^{d}/k is arithmetically semisimple if it is semisimple as a perverse sheaf over k [ FFK21 , §1.12] . On a torus every arithmetically semisimple object is generically unramified [ FFK21 , Th. 3.27] , and Zurbuchen [ Zur26 ] has recently proved generic unramifiedness for every perverse sheaf on a connected commutative algebraic group. object M_{0} pure of weight zero, they proved that the classes \Theta_{M_{0},k_{n}}(\chi) attached to the unramified characters equidistribute on average as n\to\infty [ FFK21 , Th. 4.11] . In the equidistribution results above, the object is fixed and the field varies through its finite extensions. This article concerns equidistribution along arbitrary sequences of finite fields of growing cardinality, allowing the characteristic to vary. Katz proved such a theorem in dimension one, for forms of \mathbb{G}_{m} [ Kat12 , Th. 28.1] (see also the Bourbaki seminar of Fresán [ Fre19 ] ). For arbitrary powers of the additive group, an equidistribution theorem over prime fields follows from the result for lisse sheaves of [ SFFK23 , Th. 7.22] , see [ FFK21 , Rem. 4.20(1)] . Forey–Fresán–Kowalski developed the corresponding theory for multiplicative characters of connected commutative algebraic groups, including their horizontal equidistribution theorem [ FFK21 , Th. 4.19] . For tori of dimension at least two, this theorem was left open [ FFK21 , Rem. 4.20(2)] . In the present article we prove it for \mathbb{G}_{m}^{d} in every dimension. 1.1. Statements 1.1.1. Uniform stratified generic vanishing We fix notation. For K\in D^{b}_{c}(\mathbb{G}_{m}^{d},\overline{\mathbb{Q}_{\ell}}) , c_{u}(K) denotes Sawin’s complexity relative to the embedding u_{d} fixed in § 2.1 and recalled in § 2.3 . We denote by \widehat{\mathbb{G}_{m}^{d}} the Gabber–Loeser scheme of \overline{\mathbb{Q}_{\ell}}^{\times} -characters of \mathbb{G}_{m}^{d} (§ 2.5 ). Its \overline{\mathbb{Q}_{\ell}} -points parametrize the Kummer sheaves \mathcal{L}_{\chi} , and we put K_{\chi}=K\otimes\mathcal{L}_{\chi} . Applied to the structure morphism of \mathbb{G}_{m}^{d} and the complex K_{\chi} , the transformation ( 3 ) gives the morphism R\Gamma_{c}(\mathbb{G}_{m}^{d},K_{\chi})\xrightarrow{\varphi_{K,\chi}}R\Gamma(\mathbb{G}_{m}^{d},K_{\chi}). Following Gabber–Loeser, let \Delta(K)\subseteq\widehat{\mathbb{G}_{m}^{d}} be the support of the cone of the comparison morphism between the two Mellin transforms of K , as recalled in (GL2) of § 2.8 . By [ GL96 , Prop. 3.3.4(i)] , \Delta(K)(\overline{\mathbb{Q}_{\ell}})=\{\chi\in\widehat{\mathbb{G}_{m}^{d}}(\overline{\mathbb{Q}_{\ell}}):\varphi_{K,\chi}\text{ is \emph{not} an isomorphism}\}. For n\geq 1 we write \widehat{\mathbb{G}_{m}^{d}(k_{n})}=\operatorname{Hom}\bigl(\mathbb{G}_{m}^{d}(k_{n}),\overline{\mathbb{Q}_{\ell}}^{\times}\bigr) , and for a subset Z\subseteq\widehat{\mathbb{G}_{m}^{d}}(\overline{\mathbb{Q}_{\ell}}) we put Z(k_{n})=Z\cap\widehat{\mathbb{G}_{m}^{d}(k_{n})} . If A\in\operatorname{Perv}(\mathbb{G}_{m}^{d},\overline{\mathbb{Q}_{\ell}}) , let X_{w}(A)\subseteq\widehat{\mathbb{G}_{m}^{d}}(\overline{\mathbb{Q}_{\ell}}) be the weakly unramified locus, defined by the conditions of [ FFK21 , Def. 3.1] . 3 3 3 For the perverse sheaf A_{\chi} on the affine torus, Artin vanishing and its Verdier-dual form [ BBDG18 , Th. 4.1.1, Cor. 4.1.2] show that \varphi_{A,\chi} is an isomorphism exactly when the conditions of loc. cit. hold. Thus \Delta(A)(\overline{\mathbb{Q}_{\ell}}) is the complement of the weakly unramified locus. (1) X_{w}(A)=\widehat{\mathbb{G}_{m}^{d}}(\overline{\mathbb{Q}_{\ell}})\setminus\Delta(A)(\overline{\mathbb{Q}_{\ell}}). Theorem 1.1 (uniform point count for \Delta(K) ) . For every d\geq 1 there is a nondecreasing function \Phi^{\Delta}_{d}:\mathbb{R}_{\geq 0}\to\mathbb{R}_{\geq 0} with the following property. For every C\geq 0 , every finite field k=\mathbb{F}_{q} , every prime \ell\neq\operatorname{char}k , every K\in D^{b}_{c}(\mathbb{G}_{m}^{d},\overline{\mathbb{Q}_{\ell}}) with c_{u}(K)\leq C , and every n\geq 1 , |\Delta(K)(k_{n})|\leq\Phi^{\Delta}_{d}(C)\,q^{n(d-1)}. For A\in\operatorname{Perv}(\mathbb{G}_{m}^{d},\overline{\mathbb{Q}_{\ell}}) and 0\leq i\leq d , put (2) \displaystyle\mathcal{V}_{d,i}(A)=\bigl\{\chi\in\widehat{\mathbb{G}_{m}^{d}}(\overline{\mathbb{Q}_{\ell}}):{} \displaystyle\mathrm{H}^{i}_{c}(\mathbb{G}_{m}^{d},A_{\chi})\neq 0\ \text{or}\ \mathrm{H}^{-i}_{c}(\mathbb{G}_{m}^{d},A_{\chi})\neq 0\ \text{or} \displaystyle\mathrm{H}^{i}(\mathbb{G}_{m}^{d},A_{\chi})\neq 0\ \text{or}\ \mathrm{H}^{-i}(\mathbb{G}_{m}^{d},A_{\chi})\neq 0\bigr\}. For i>d , we use the convention \mathcal{V}_{d,i}(A)=\varnothing . Theorem 1.2 (bounded-complexity stratified point count) . For every d\geq 1 there is a nondecreasing function \Psi_{d}:\mathbb{R}_{\geq 0}\to\mathbb{R}_{\geq 0} such that for every C\geq 0 , every finite field k=\mathbb{F}_{q} , every prime \ell\neq\operatorname{char}k , every A\in\operatorname{Perv}(\mathbb{G}_{m}^{d},\overline{\mathbb{Q}_{\ell}}) with c_{u}(A)\leq C , every 0\leq i\leq d , and every n\geq 1 , |\mathcal{V}_{d,i}(A)(k_{n})|\leq\Psi_{d}(C)\,q^{n(d-i)}. Forey–Fresán–Kowalski propose this bounded-complexity refinement of their stratified vanishing theorem [ FFK21 , Th. 2.3, Rem. 2.4] . 4 4 4 For tori, the non-uniform stratified vanishing theorem is [ GL96 , Th. 3.4.1, Prop. 6.3.2, Th. 6.3.3] , again conditional on the hypotheses recalled above, removed in [ FFK21 , Cor. 2.16] . Its case i=1 bounds \mathcal{V}_{d,1}(A) , a subset of \Delta(A)(\overline{\mathbb{Q}_{\ell}}) , and is subsumed by Theorem 1.1 , which bounds the larger set \Delta(A)(\overline{\mathbb{Q}_{\ell}}) . 1.1.2. Equidistribution The remaining statements use the tannakian framework of [ FFK21 ] . The categories \operatorname{Perv}_{\mathrm{int}}(\mathbb{G}_{m}^{d}) , \operatorname{Perv}_{\mathrm{int}}^{\mathrm{ari}}(\mathbb{G}_{m}^{d}/k) , and the tannakian subcategory \langle M\rangle of (\operatorname{Perv}_{\mathrm{int}}(\mathbb{G}_{m}^{d}),\ast_{\mathrm{int}}) generated by an object M are recalled in §§ 2.6 – 2.9 . For M_{0}\in\operatorname{Perv}_{\mathrm{int}}^{\mathrm{ari}}(\mathbb{G}_{m}^{d}/k) , with geometric base change M=M_{0,\bar{k}} , let X(M_{0})=X(M)\subseteq X_{w}(M) denote the unramified locus of [ FFK21 , Def. 3.25, §3.8] . Its definition in terms of fibre functors on \langle M\rangle is recalled in

Read the complete text