Depth Preservation and Close-Field Transfer in the Local Langlands Correspondence

Manish Mishra · 2025

Read Depth Preservation and Close-Field Transfer in the Local Langlands Correspondence by Manish Mishra (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-2509-04997.txt.

Introduction

Depth, a measure of ramification, should be preserved when the local Langlands correspondence matches representations with parameters. In short, the match should not change the depth. Manish Mishra revises that notion for tori in a solo 2025 mathematics paper. The new definition restores that preservation under the match over a non-archimedean local field.

Mishra uses it to transfer broad harmonic analysis results from characteristic zero to characteristic p. For fields of large enough characteristic, he proves block-by-block equivalences with a nearby field of characteristic zero. A match in characteristic zero then yields a canonical one in characteristic p. Results on close fields, Hecke algebras, and regular supercuspidal cases support the build.

Catalog record

CREATORManish Mishra
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE137 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-04997

Browse the full library · About the Tower of Babel

Full text

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

Depth Preservation and Close-Field Transfer in the Local Langlands Correspondence Manish Mishra Abstract We introduce a revised notion of depth for Langlands parameters for a connected reductive G G defined over a nonarchimedean local field F F that restores depth preservation under the local Langlands correspondence (LLC)—in particular for all tori. We leverage that preservation to derive structural results that, taken together, yield a canonical transfer of broad harmonic-analytic results from characteristic 0 0 to characteristic p p . When F F has suitably large positive characteristic, we prove a block-by-block equivalence: each Bernstein block of G ⁡ ( F ) G(F) is equivalent to a corresponding block for some G ′ ​ ( F ′ ) G^{\prime}(F^{\prime}) with F ′ F^{\prime} of characteristic 0 0 ℓ \ell -close to F F ; using this, we show that a LLC in characteristic 0 0 corresponds canonically to a LLC in characteristic p p . For regular supercuspidals we give a direct, more structured construction via Kaletha. Along the way we recover and extend results on ℓ \ell -close fields—introducing a depth-transfer function generalizing the normalized Hasse–Herbrand function, proving truncated isomorphisms for arbitrary tori and parahorics, establishing a depth and supercuspidality preserving Kazhdan-type Hecke-algebra isomorphism for arbitrary maximal parahorics of arbitrary connected reductive groups; and a generalized Cartan decomposition for arbitrary maximal parahorics—thereby subsuming several earlier results in the literature. Collectively, the results let one work in characteristic 0 0 without loss of generality for a wide swath of harmonic analysis on p p -adic groups. 1 Introduction The depth of representations and parameters serves as a fundamental invariant in the harmonic analysis of p-adic groups and the local Langlands correspondence (LLC). The central contribution of this work is the introduction of a new notion of depth of a Langlands parameter which rectifies a key deficiency in the standard notion - the preservation of depth. Leveraging this foundation, we establish a canonical transfer of harmonic analysis results from non-archimedean local fields of characteristic zero to those over local function fields, including the local Langlands correspondence. For a non-archimedean local field F with maximal ideal \mathfrak{p}_{F}\subset\mathcal{O}_{F} , local class-field theory interprets the filtration 1+\mathfrak{p}_{F}^{\,r}\;\subset\;F^{\times},\qquad r\geq 0, as the upper–numbering filtration \{W_{F}^{\,s}\}_{s\geq 0} of the Weil group. The Hasse–Herbrand function \phi_{E/F} transfers this numbering across an extension E/F , matching the depth‐ r elements of E^{\times} with the s:=\phi_{E/F}(er) -th filtration subgroup W_{F}^{s} , where e denotes the ramification index of the extension. The correspondence is rigorously captured in the local Langlands correspondence for the induced torus T=\Res_{E/F}\mathbb{G}_{m} : a depth- r character of T(F) corresponds to its standard depth \Phi_{T}(r):=\phi_{E/F}(er) Langlands parameter. Consequently, when E/F is wildly ramified, this standard depth fails to respect depth preservation limiting its compatibility with harmonic analysis. To resolve this, we work with Yu’s minimal congruence filtration subgroups (see § 2 ) and we introduce the following notion of the depth of a Langlands parameter \varphi of an arbitrary F -torus T , \depth_{T}(\varphi)\;:=\;\sup_{f:R\to T}\,\depth_{R}\!\bigl(\widehat{f}\circ\varphi\bigr). Here, the supremum is taken over all induced F -tori R and all F -morphisms f:R\to T and where depth in the case of an induced torus is defined in such a way as to make its LLC depth preserving. When T is tamely ramified, this notion of depth reduces to the standard notion. For arbitrary G (Definition 6.8 ), from \varphi we extract a torus-valued toral parameter and a finite residual Weyl parameter , and set \depth_{G}(\varphi):=\max\{\text{depth of the toral parameter},\ \text{depth of the Weyl parameter}\}, In the case of a tame Langlands parameter (see Definition 6.8 ) this notion coincides with the standard notion. We note that Langlands parameters which are not tame can exist plentifully even when G is tamely ramified. To control depth bounds for connected reductive group G/F , we attach to each vertex x of the (reduced) Bruhat–Tits building a depth-transfer function (§ 6.4 ), \Phi_{G,x}\colon\mathbb{R}_{\geq 0}\longrightarrow\mathbb{R}_{\geq 0}, which simultaneously accounts for the ramification of all maximally unramified elliptic maximal tori associated to x and of all root subgroups (§ 6.4 ). For a torus, one has \Phi_{T}=\phi^{\mathrm{norm}}_{E/F} , so \Phi_{G,x} can be considered a generalization of the normalized Hasse–Herbrand function. The new depth and depth-transfer function enable a uniform treatment of depth‐ r phenomena across the representation theory of G and its behavior under Deligne–Kazhdan close‐field transfer. Throughout this article, we work with Yu’s minimal congruence filtration in place of Moy-Prasad filtrations. Our first theorem (Theorem 4.9 ) establishes depth preservation for tori without any tameness hypothesis. Depth- r local Langlands for tori Theorem A (Depth- r LLC for tori) . Let T/F be an arbitrary torus and let r\geq 0 . The local Langlands correspondence for tori \mathcal{L}_{T}\colon\Hom\!\bigl(T(F),\mathbb{C}^{\times}\bigr)\xrightarrow{\sim}H^{1}(W_{F},\widehat{T}) preserves depth, i.e., for every character \chi of T(F) , we have, \operatorname{dep}\bigl(\mathcal{L}_{T}(\chi)\bigr)\;=\;\!\operatorname{dep}(\chi), where the depth of a Langlands parameter is as defined in Definition 4.2 . Theorem A extends Yu’s depth-preservation theorem ( T tame, r>0 ) and the induced-torus result of [ MP15 ] ( T=\Res_{E/F}\mathbb{G}_{m} ). Using Deligne’s isomorphism on the Galois side of this correspondence, we obtain (Theorem 5.3 ), Truncated isomorphisms over \ell -close fields Theorem B (Truncated-torus close field isomorphisms) . Let F and F^{\prime} be \ell -close and let T/F be a torus with transfer T^{\prime}/F^{\prime} . If 0\leq r\leq\Phi_{T}(\ell) , there is a canonical isomorphism, T(F)\big/\!T(F)_{r}\;\xrightarrow{\;\sim\;}T^{\prime}(F^{\prime})\big/\!T^{\prime}(F^{\prime})_{\,r}. When r\in\mathbb{Z}_{\geq 0} , this recovers the main result of Aubert–Varma [ AV25 ] . Before stating our next result, we first recall a result of Chai–Yu [ CY01 ] . Let F and F^{\prime} be \ell -close local fields, let T/F be a torus, and let T^{\prime}/F^{\prime} be its Deligne transfer. Write \mathscr{T} and \mathscr{T}^{\prime} for the respective Néron models. Then for any given integer m , there is a suitably large \ell such that there is a canonical isomorphism of smooth group schemes \mathscr{T}\,\times_{\mathcal{O}_{F}}\mathcal{O}_{F}/\mathfrak{p}_{F}^{\,m}\;\xrightarrow{\;\sim\;}\mathscr{T}^{\prime}\times_{\mathcal{O}_{F^{\prime}}}\mathcal{O}_{F^{\prime}}/\mathfrak{p}_{F^{\prime}}^{\,m}. Consequently, one has a canonical isomorphism, \mathscr{T}(\mathcal{O}_{F}/\mathfrak{p}_{F}^{\,m})\cong\mathscr{T}^{\prime}(\mathcal{O}_{F^{\prime}}/\mathfrak{p}_{F^{\prime}}^{\,m}). Building on this, by showing that Bruhat–Tits root-group filtration schemes behave just as well, Ganapathy [ Gan19 , Cor. 6.3] extends this result to parahoric group schemes. Our next Theorem dispenses with the Chai–Yu input which is the main ingredient in [ AV25 ] and [ Gan19 ] . Working with Yu’s minimal-congruence filtration, we prove (Theorem 6.7 ), Theorem C (Truncated-parahoric close field isomorphisms) . Let G/F be connected reductive, x\in\mathcal{B}^{\mathrm{red}}(G,F) , and let (G^{\prime},x^{\prime}) be the Deligne transfer over an \ell -close field F^{\prime} . For every 0\leq r\leq\Phi_{G,x}(\ell) , there is a canonical isomorphism, G(F)_{x,0}\big/\!G(F)_{x,r}\;\xrightarrow{\;\sim\;}G^{\prime}(F^{\prime})_{x^{\prime},0}\big/\!G^{\prime}(F^{\prime})_{x^{\prime},r}. In particular, by allowing real r instead of integers, our result extends a key aspect of the main results of Chai-Yu [ CY01 ] and Ganapathy [ Gan19 ] and which is also sufficient for most harmonic analysis applications. After extending the Cartan decomposition to arbitrary parahorics (see Theorem H below), by following the method in [ Gan19 ] , we obtain (Theorem 8.1 ), Close field Hecke algebra isomorphisms Theorem D (Kazhdan Hecke-algebra isomorphism) . Fix m\in\mathbb{Q}_{\geq 0} . There is an integer \ell(m)\!\gg\!0 depending only on (G,x,m) such that, whenever F and F^{\prime} are \ell -close with \ell\geq\ell(m) , the assignment coming from Theorem C induces an algebra isomorphism \kappa_{\ell,x}\;:\;\mathcal{H}\!\bigl(G,G(F)_{x,m}\bigr)\;\xrightarrow{\;\sim\;}\mathcal{H}\!\bigl(G^{\prime},G^{\prime}(F^{\prime})_{x^{\prime},m}\bigr). Moreover, under \kappa_{\ell} every supercuspidal representation of G(F) maps

Read the complete text