Note on a certain category of mod p p representations

Reinier Sorgdrager · 2025

Read Note on a certain category of mod p p representations by Reinier Sorgdrager (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-2503-17773.txt.

Introduction

Reinier Sorgdrager, a solo author, studies a special class of representations in a 2025 paper. The setting is the mod p Langlands program, which links number theory to symmetry groups over p-adic fields. Two groups are treated: a matrix group and a quaternion unit group. Other researchers had carved out a full subcategory of smooth representations for special study.

His result says membership in that subcategory can be read off locally. Whether a representation belongs is fully decided by how it behaves when restricted to any small open subgroup, however tiny. The paper is a short, technical note for specialists in the Langlands program.

Catalog record

CREATORReinier Sorgdrager
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-17773

Browse the full library · About the Tower of Babel

Full text

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

Note on a certain category of mod p p representations Reinier Sorgdrager Abstract Let p > 3 p>3 be a prime number, f ≥ 1 f\geq 1 an integer. We consider a certain full subcategory 𝒞 \mathcal{C} of the category of smooth admissible mod p p representations of either GL 2 ⁡ 𝐐 p f \GL_{2}\mathbf{Q}_{p^{f}} or of the group of units of the quaternion algebra over 𝐐 p f \mathbf{Q}_{p^{f}} . This category was introduced in the context of the mod p p Langlands program by [ BHH+23a ] in the GL 2 \GL_{2} -case and by [ HW24 ] in the quaternion case. We prove that whether a smooth admissible mod p p representation π \pi (with central character) belongs to 𝒞 \mathcal{C} is completely determined by the restriction of π \pi to an arbitrarily small open subgroup. Introduction Let f\geq 1 and K=\mathbf{Q}_{p^{f}} , for p a prime bigger than 3 . Let \mathbf{F} be a finite field containing \mathbf{F}_{p^{f}} . We consider a p -adic Lie group G which is either • \GL_{2} -case : the group I_{1}/Z_{1} , where I_{1}=\begin{pmatrix}1+p\mathcal{O}_{K}&\mathcal{O}_{K}\\ p\mathcal{O}_{K}&1+p\mathcal{O}_{K}\end{pmatrix}\subset\GL_{2}K is the upper-triangular pro- p Iwahori subgroup and Z_{1} , the center, its subgroup of diagonal matrices, or • quaternion case : the group U_{1}/Z_{1} , where U_{1}=1+\Pi\mathcal{O}_{D}\subset D^{\times} is the subgroup of the units of the quaternion algebra D over K of invariant 1/2 of elements that are 1 modulo a uniformizer \Pi (that we normalize to square to p ); the group Z_{1}=1+p\mathcal{O}_{K} then denotes the center. In both cases the p -adic Lie group G is pro- p of dimension 3f . In [ BHH+23a ] and [ HW24 ] a certain category \mathcal{C} of mod p representations \pi of the bigger group G_{\text{big}} ( \GL_{2}K or D^{\times} depending on which case we are in) is introduced whose definition is given in terms of certain invariants that can be attached to \pi|_{G} 1 1 1 The restriction to G will make sense because the center by which one mods out to obtain G will always act trivially on these \pi (as they will be required to have a central character). . The main purpose of this note is to show that in fact being in \mathcal{C} can be detected in terms of invariants attached to the restriction to an arbitrarily small open subgroup. It is the statement of the following theorem: Theorem 1 . Let \pi and \pi^{\prime} be smooth representations of G_{\text{big}} ( \GL_{2}K or D^{\times} ) with same fixed central character. Suppose \pi^{\prime}\in\mathcal{C} and \pi|_{H}\cong\pi^{\prime}|_{H} for some open subgroup H of G_{\text{big}} . Then \pi\in\mathcal{C} . Warning 2 . We immediately warn the reader that what we call the category \mathcal{C} in the \GL_{2} -case is actually not the same as the category denoted by “ \mathcal{C} ” in [ BHH+23a ] , but a full subcategory thereof, which is explained by the paragraph at the top of page 115 of loc. cit. . The category \mathcal{C} is of interest in the mod p Langlands program in the \GL_{2} -case because it turns out that candidates \pi_{\overline{\rho}} for a conjectured mod p Langlands correspondence from mod p \Gal(\overline{\mathbf{Q}}_{p}/K) -representations to mod p \GL_{2}K -representations \overline{\rho}\mapsto\pi_{\overline{\rho}} (restricted to sufficiently generic \overline{\rho} ) are all in \mathcal{C} : inspired by Emerton’s local-global compatibility such candidates are constructed via global methods, more precisely via the mod p cohomology of Shimura varieties, and in [ BHH+23b ] it is proven that the resulting mod p representation of \GL_{2}K is in \mathcal{C} (see the proof of Thm. 5.3.5 in loc. cit. and their Thm. 6.4.7, see also [ Wan23 , Thm. 5.1] ). The paper [ HW24 ] takes an interest in D^{\times} motivated by the search for a mod p Jacquet-Langlands correspondence. In the case f=1 they similarly prove that mod p representations of D^{\times} coming from a sufficiently generic mod p representation of \Gal(\overline{\mathbf{Q}}_{p}/\mathbf{Q}_{p}) , via a construction using the cohomology of a suitable Shimura variety, is in \mathcal{C} (the quaternion version this time), see Thm. 6.14 of loc. cit. . We remark that (for both groups) \pi being in \mathcal{C} implies that its Gelfand-Kirillov dimension is \leq f , see the proof of [ BHH+23b , Thm. 5.3.5] . Let us now give some details on the definition of the category \mathcal{C} . In both cases the category \mathcal{C} is a full subcategory of the category of smooth admissible representations of G_{\text{big}} over \mathbf{F} that have a fixed central character (which we abusively omit from the notation). An object \pi of this ambient category has Z_{1} acting trivially on it, because Z_{1} is pro- p and central and acting via a character to \mathbf{F}^{\times} , which is therefore necessarily trivial. This means \pi is also a representation of G . Its dual \pi^{\vee} is then a pseudocompact \mathbf{F}\llbracket G\rrbracket -module, where \mathbf{F}\llbracket G\rrbracket is the completed group ring, i.e. the limit of all group rings over \mathbf{F} of quotients of G by normal open subgroups. Since G is pro- p , the completed group ring is local, having a unique (two-sided) maximal ideal \mathfrak{m}_{G} . One can form the associated graded ring \gr\mathbf{F}\llbracket G\rrbracket\coloneqq\bigoplus_{i\geq 0}\mathfrak{m}_{G}^{i}/\mathfrak{m}_{G}^{i+1} which acts on \gr\pi^{\vee}\coloneqq\bigoplus_{i\geq 0}\mathfrak{m}_{G}^{i}\pi^{\vee}/\mathfrak{m}_{G}^{i+1}\pi^{\vee} , making the latter into a graded module. Next, a certain two-sided ideal J_{\mathcal{C}}\subset\gr\mathbf{F}\llbracket G\rrbracket is identified: in the \GL_{2} -case it is the ideal I_{G} defined just before [ BHH+23b , Thm. 5.3.4] and in the quaternion case it is the ideal I_{D} defined just before [ HW24 , Cor. 2.10] . The definition of \mathcal{C} is now straightforward: its objects are all \pi for which \gr\pi^{\vee} is killed by a power of J . It turns out to be an Abelian subcategory of the smooth admissible representations with the fixed central character. Let us briefly say a few words about our strategy. It is roughly to associate to any \pi|_{H} some graded module ( \text{gr}_{N,\text{res}}\pi|_{H}^{\vee} in notation that will make sense later) over some graded subalgebra of \gr\mathbf{F}\llbracket G\rrbracket and to show that \pi\in\mathcal{C} is equivalent to this graded module being killed by some power of a certain ideal. We will see that we can define the \mathfrak{m}_{G} -adic filtration on \pi^{\vee} giving rise to \gr\pi^{\vee} in a language that is more amenable to restricting to subgroups. Namely, we can define the \mathfrak{m}_{G} -adic filtration in terms of p -valuations and ordered bases. The relevant concepts for this will be introduced in the next two sections. Afterwards, we define the new association of graded modules \text{gr}_{N,\text{res}}(-) . In order to prove Theorem 1 we then establish a series of lemmas that phrase the condition of being in \mathcal{C} in terms of the new graded modules and conclude. We note that we do not need to know the precise description of the ideal J_{\mathcal{C}} . All that matters is that \gr\mathbf{F}\llbracket G\rrbracket/J_{\mathcal{C}} is commutative. Therefore our Theorem 1 generalizes to any category \mathcal{C}^{\prime} defined in the same way but with respect to a different such ideal J_{\mathcal{C}^{\prime}} . p -valuations We recall the definition of a p -valuation. Definition 3 . Let H be a group. A p -valuation on H is a map \omega\colon H\to(\tfrac{1}{p-1},\infty] such that for all

Read the complete text