Non-abelian p p -curvature and a non-abelian Katz’s formula

Michael Barz · 2026

Read Non-abelian p p -curvature and a non-abelian Katz’s formula by Michael Barz (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-2604-20054.txt.

Introduction

Barz extends Katz's formula to non-abelian p-curvature, relating the Kodaira-Spencer map to the p-curvature of the Gauss-Manin connection. Following Lam and Litt's non-abelian Hodge variant, he gives a conceptual proof via sheared de Rham stacks, building on work by Bhatt, Kanaev, Vologodsky, Zhang, and Drinfeld. The account explains concrete phenomena of Katz and Lam-Litt in unified stacky terms, proving a slightly different statement than Lam-Litt. No background in de Rham stack theory is assumed.

Catalog record

CREATORMichael Barz
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75501
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.20054

Browse the full library · About the Tower of Babel

Full text

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

Non-abelian p p -curvature and a non-abelian Katz’s formula Michael Barz Abstract . Let k k be a field of characteristic p , p, and f : X → S f:X\to S a smooth proper morphism of smooth k k -schemes. Katz’s formula [ 11 ] gives a relationship between the Kodaira–Spencer map of f f , and an invariant called the p p -curvature of the Gauss–Manin connection associated to f . f. Recently, Lam–Litt [ 12 ] proved a variant of Katz’s formula in non-abelian Hodge theory, and suggested that it should be possible to give a more conceptual proof of their formula using the stacky approach to p p -adic Hodge theory. In this article, we realize their suggestion, explaining how this rather concrete phenomena observed by Katz [ 11 ] and Lam–Litt [ 12 ] can be explained in a conceptual way using sheared de Rham stacks , as developed by Bhatt–Kanaev–Vologodsky–Zhang [ 3 ] and Drinfeld [ 10 ] (though we prove a slightly different statement than Lam–Litt [ 12 ] ). We do not assume the reader has any background in the theory of de Rham stacks, and along the way we give some general constructions related to non-abelian connections which may be of use in other contexts. But no matter how advanced the system, no matter how precise, unless we have the will to communicate, there’s no connection. Dance Dance Dance , Murakami 1. Introduction Let k be a field of characteristic p, and f:X\to S a smooth proper morphism of smooth k -schemes. In this situation, one can consider the relative de Rham cohomology \mathcal{H}^{n}_{\dR}(X/S) (viewed as a vector bundle on S ). This vector bundle \mathcal{H}^{n}_{\dR}(X/S) admits much additional structure. For example, it famously admits the Gauss–Manin connection \nabla_{\GM}:\mathcal{H}^{n}_{\dR}(X/S)\to\Omega^{1}_{S}\otimes\mathcal{H}^{n}_{\dR}(X/S), a certain flat connection playing an important role in the study of periods. The relative de Rham cohomology \mathcal{H}^{n}_{\dR}(X/S) also carries two natural filtrations, the Hodge filtration and the conjugate filtration. This conjugate filtration is specific to characteristic p, and we will discuss it in Section 4 . Katz, in his work [ 11 ] on the Grothendieck–Katz p -curvature conjecture, discovered a remarkable relationship between how \nabla_{\GM} acts on the associated gradeds of these two filtrations; more precisely, Katz’s formula [ 11 ] relates the Kodaira–Spencer class of f with the p -curvature of \nabla_{\GM}. Katz’s formula plays a crucial role in his proof [ 11 ] of the Grothendieck–Katz p -curvature conjecture for Gauss–Manin connections. In this note, we will explain a simple proof of a non-abelian variant of Katz’s formula. This was inspired by recent work of Lam–Litt [ 12 ] , who proved a non-abelian variant of the Grothendieck–Katz p -curvature conjecture using (amongst other tools) a non-abelian Katz’s formula. Lam–Litt [ 12 ] , in their Remark 3.8.2, suggested that their version of Katz’s formula should follow from the theory of syntomification ; this suggestion inspired our article. Remark 1.1 . The non-abelian variant of the Grothendieck–Katz p -curvature conjecture is generally called the Ekedahl–Shepherd-Barron–Taylor conjecture; this conjecture first appeared in print in Bost’s article [ 6 ] . The main novelty of this paper is that, by systematically using the sheared de Rham stack of Bhatt–Kanaev–Mathew–Vologodsky–Zhang [ 3 ] and Drinfeld [ 10 ] , we can give an especially simple proof of the non-abelian Katz’s formula, and simplify the construction of the objects appearing in the formula. Remark 1.2 . In this note, we work entirely over k. So, to us a prestack will mean a functor \Alg_{k}\to\Gpd, and a stack will be a prestack obeying fpqc descent. 1.1. Non-abelian connections As before, let S be a smooth k -scheme. In this situation, one can define the sheared de Rham stack (S/k)^{\widehat{\dR}}. This is a stack (by which we mean fpqc sheaf \Alg_{k}\to\Gpd ) defined by Bhatt–Mathew–Kanaev–Vologodsky–Zhang [ 3 ] and Drinfeld [ 10 ] (based on earlier work of Bhatt–Lurie [ 4 ] , Drinfeld [ 7 ] , and Simpson [ 16 ] ) having the remarkable property that vector bundles on (S/k)^{\widehat{\dR}} are the same as vector bundles with flat connection on S. Remark 1.3 . The notation (S/k)^{\widehat{\dR}} indicates that we are taking the de Rham stack relative to k ; recall that a flat connection on a vector bundle \mathcal{E} on S is a map \nabla:\mathcal{E}\to\Omega^{1}_{S/k}\otimes\mathcal{E} obeying certain axioms. The k in (S/k)^{\widehat{\dR}} is the same as the k in \Omega^{1}_{S/k}. Remark 1.4 . There is a ‘non-sheared’ variant of the sheared de Rham stack, discovered first, which only connections with nilpotent p -curvature. We will not be using this non-sheared variant anywhere in this note. One of the basic insights of Simpson is to try and study non-linear differential equations using stacks over (S/k)^{\widehat{\dR}} instead of studying quasicoherent sheaves on (S/k)^{\widehat{\dR}}. Definition 1.5 . Let E be a prestack with a morphism E\to S. A non-abelian connection on E/S is a prestack E^{\prime} together with a Cartesian diagram {\lx@inpgf@ignorespaces E} {\lx@inpgf@ignorespaces E^{\prime}} {\lx@inpgf@ignorespaces S} {\lx@inpgf@ignorespaces(S/k)^{\widehat{\dR}}.} 1.2. The non-abelian Gauss–Manin connection Let f:X\to S be a smooth proper morphism of smooth k -schemes, as before. We now introduce the non-abelian analogue of the Gauss–Manin connection. To start, we replace the relative de Rham cohomology \mathcal{H}^{q}_{\dR}(X/S) (a vector bundle on S ) with the de Rham moduli stack of X over S. Definition 1.6 . Fix n\geq 1. We define \mathcal{M}_{\dR}(X/S,n) to be the mapping stack \Map_{S}((X/S)^{\widehat{\dR}},B\SL_{n}). In other words, \mathcal{M}_{\dR}(X/S,n) is the S -stack such that, for T\to S an S -scheme, a T -valued point of \mathcal{M}_{\dR}(X/S,n) is just a rank n vector bundle \mathcal{E} on X\times_{S}T, together with a flat connection \nabla:\mathcal{E}\to\Omega^{1}_{(X\times_{S}T)/T}\otimes\mathcal{E} and an isomorphism \xi:\det(\mathcal{E})\simeq\mathcal{O}_{X\times_{S}T} which is horizontal (that is, it intertwines the connection \det(\nabla) on \det\mathcal{E} with the trivial connetion d on \mathcal{O} ). Remark 1.7 . We study \SL_{n} -bundles rather than \GL_{n} -bundles to better match with Lam–Litt [ 12 ] , though for Katz’s formula the difference between \SL_{n} and \GL_{n} does not seem to matter. In the body of the paper, we will show how to use de Rham stacks to construct a non-abelian connection on \mathcal{M}_{\dR}(X/S,n). This connection is furnished by the Cartesian diagram {\lx@inpgf@ignorespaces\mathcal{M}_{\dR}(X/S,n)} {\lx@inpgf@ignorespaces\Map_{(S/k)^{\widehat{\dR}}}((X/k)^{\widehat{\dR}},B\SL_{n})} {\lx@inpgf@ignorespaces S} {\lx@inpgf@ignorespaces(S/k)^{\widehat{\dR}}.} as we discuss in Section 3 . In the body of this paper, we will explain how to construct non-abelian variants of the Hodge and conjugate filtrations on the non-abelian Gauss–Manin connection. We can now, somewhat vaguely, state our main theorem; it will be stated more precisely in Section 5 . Theorem 1.8 (Katz’s formula, imprecise statement) . The non-abelian Hodge filtration on \mathcal{M}_{\dR}(X/S,n) gives rise to a non-abelian Higgs field \Theta_{X/S}:\pi^{*}_{\operatorname{Dol}}T_{S/k}(-1)\to\mathcal{T}_{\mathcal{M}_{\operatorname{Dol}}/(S\times B\mathbb{G}_{m})}, where \mathcal{M}_{\operatorname{Dol}}:=\Map_{S\times B\mathbb{G}_{m}}((X/S)^{\operatorname{Hodge}},B\SL_{n}), with \pi_{\operatorname{Dol}}:\mathcal{M}_{\operatorname{Dol}}\to S\times B\mathbb{G}_{m} the structure map Roughly, this \Theta_{X/S} is obtained by taking the associated graded of the non-abelian Hodge filtration. Let S^{\prime} denote the Frobenius twist of S relative to k. Then the non-abelian conjugate filtration on \mathcal{M}_{\dR}(X/S,n) similarly gives rise to a certain stack \pi_{\operatorname{Dol,c}}:\mathcal{M}_{\operatorname{Dol,c}}\to S\times B\mathbb{G}_{m}, analogous to the Dolbeault moduli stack, and a morphism \psi_{X/S}|_{\lambda=0}:\pi_{\operatorname{Dol,c}}^{*}F_{S/k}^{*}T_{S^{\prime}/k}(+1)\to\mathcal{T}_{\mathcal{M}_{\operatorname{Dol,c}}/(S\times B\mathbb{G}_{m})}. There is a Cartesian diagram {\lx@inpgf@ignorespaces\mathcal{M}_{\operatorname{Dol,c}}} {\lx@inpgf@ignorespaces\mathcal{M}_{\operatorname{Dol}}} {\lx@inpgf@ignorespaces S\times B\mathbb{G}_{m}} {\lx@inpgf@ignorespaces S\times B\mathbb{G}_{m},} \scriptstyle{\lx@inpgf@ignorespaces\varphi} \scriptstyle{\lx@inpgf@ignorespaces(F_{\abs},[-1])} and for the map \varphi:\mathcal{M}_{\Dol,c}\to\mathcal{M}_{\Dol} in this diagram, we have

Read the complete text