The ρ \rho -Schwartz space

Jayce R. Getz, Armando Gutiérrez Terradillos, Farid Hosseinijafari, Aaron Slipper, Guodong Xi, HaoYun Yao, Alan Zhao · 2025

Read The ρ \rho -Schwartz space by Jayce R. Getz, Armando Gutiérrez Terradillos, Farid Hosseinijafari, Aaron Slipper, Guodong Xi, HaoYun Yao, Alan Zhao (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-2512-00182.txt.

Introduction

Jayce Getz and co-authors develop a new function space for harmonic analysis on reductive groups in 2025. The space is named in the title after a representation of the group's Langlands dual. Suitable assumptions on that representation shape the whole construction.

Spaces of test functions control which integrals and transforms are well behaved in the Langlands program. That program is a vast web of conjectures linking number theory and representation theory. A well-chosen Schwartz space can make trace formulas and period integrals work more smoothly. The work is deep pure mathematics for specialists in automorphic forms. The assumptions on the representation decide which groups and fields the theory covers.

Catalog record

CREATORJayce R. Getz, Armando Gutiérrez Terradillos, Farid Hosseinijafari, Aaron Slipper, Guodong Xi, HaoYun Yao, Alan Zhao
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE167 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-00182

Browse the full library · About the Tower of Babel

Full text

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

The ρ \rho -Schwartz space Thanks: Getz is thankful for partial support provided by NSF grant DMS-2400550, and Slipper is supported by NSF grant DMS-2231514. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. Gutiérez Terradillos is supported by a research grant (VIL54509) from VILLUM FONDEN Jayce R. Getz, Armando Gutiérrez Terradillos, Farid Hosseinijafari, Aaron Slipper, Guodong Xi, HaoYun Yao, Alan Zhao Abstract . Let G G be a reductive group over a local field F F and let ρ : G L → GL V ρ ​ ( ℂ ) \rho:{}^{L}G\rightarrow\mathrm{GL}_{V_{\rho}}(\mathbb{C}) be a representation of its L L -group satisfying suitable assumptions. Braverman, Kazhdan and Ngô conjectured that one has a ρ \rho -Schwartz space 𝒮 ρ ​ ( G ​ ( F ) ) \mathcal{S}_{\rho}(G(F)) of functions on G ⁡ ( F ) G(F) that admits a Fourier transform and satisfies certain desiderata. We construct the Fourier transform for arbitrary fields. Over non-Archimedean fields we construct the Schwartz space, and in the Archimedean case we construct an approximation to it. This proves a large portion of their conjectures. Our methods are spectral in nature. 2020 Mathematics Subject Classification Primary 11F70; Secondary 22E30, 22E35 1. Introduction Let G be a (connected) reductive group over a local field F and let \rho:{}^{L}G\rightarrow\mathrm{GL}_{V_{\rho}}(\mathbb{C}) be a representation of its L -group. We assume for the introduction that we are given an isomorphism d:G/G^{\mathrm{der}}\tilde{\,\rightarrow\,}\mathbb{G}_{m}, and that the dual map \rho\circ d^{\vee}:\mathbb{C}^{\times}\rightarrow\mathrm{GL}_{V_{\rho}}(\mathbb{C}) is the inclusion of \mathbb{C}^{\times} into the center. It was conjectured by Braverman and Kazhdan [ BK00 , Ngô14 ] that there exists a Schwartz space attached to \rho. To state the conjecture precisely, let \mathcal{R}:G(F)\times G(F)\rightarrow\mathrm{Aut}(L^{2}(G(F))) denote the regular representation and let \psi:F\rightarrow\mathbb{C}^{\times} be a nontrivial character. For any function h:G(F)\rightarrow\mathbb{C} write h^{\vee}(g):=h(g^{-1}). Conjecture 1.1 (Braverman-Kazhdan, Ngô) . There is a G(F)\times G(F) -invariant subspace \displaystyle\mathcal{S}_{\rho}<C^{\infty}(G(F))\cap L^{2}(G(F)) that is dense in L^{2}(G(F)) and equipped with a unitary operator \mathcal{F}_{\rho,\psi}:\mathcal{S}_{\rho}\rightarrow\mathcal{S}_{\rho} that satisfies \displaystyle\mathcal{F}_{\rho,\psi}\circ\mathcal{R}(g_{1},g_{2}) \displaystyle=\mathcal{R}(g_{2},g_{1})\circ\mathcal{F}_{\rho,\psi}, \displaystyle\mathcal{F}_{\rho,\psi}\circ\mathcal{F}_{\rho,\overline{\psi}} \displaystyle=\mathrm{id}_{\mathcal{S}_{\rho}}. Moreover, for all f\in\mathcal{S}_{\rho} and tempered representations \pi , the operator \pi\left(f\right) is well-defined and (1.0.1) \displaystyle\pi(\mathcal{F}_{\rho,\psi}(f)^{\vee})=\gamma(\tfrac{1}{2},\pi,\rho,\psi)\pi(f). Here the \gamma -factor is defined using the local Langlands correspondence as in ( 3.0.4 ). Since \psi is usually fixed, we often drop it from notation, e.g. \mathcal{F}_{\rho}:=\mathcal{F}_{\rho,\psi}. We refer to \mathcal{F}_{\rho} as the \rho -Fourier transform , or simply the Fourier transform. Remark 1.2 . In the non-Archimedean case L. Lafforgue has constructed a space of functions that satisfies a weak analogue of Conjecture 1.1 [ Laf14 ] . Roughly his space of functions is designed to satisfy Conjecture 1.1 at the level of traces (not operators). Our first main theorem is that the conjecture above is true: Theorem 1.3 . Assume the desiderata on the local Langlands correspondence explained in § 3 below. Then Conjecture 1.1 is true; we may take \mathcal{S}_{\rho}=\mathcal{C}(G(F)), the Harish-Chandra space of G(F). The space \mathcal{C}(G(F)) is usually referred to as the Harish-Chandra Schwartz space, but we omit “Schwartz” for brevity and to distinguish it from other spaces we consider. Remark 1.4 . The desiderata on the local Langlands correspondence we require are rather weak, and are known in many cases as we explain in § 3 . For example, if one restricts attention to G=\mathrm{GL}_{n} then everything we require about the local Langlands correspondence is known. The space \mathcal{C}(G(F)) is too large for many purposes. Thus one is led to introduce additional constraints on \mathcal{S}_{\rho}. Let \mathcal{C}(\pi) denote the space of smooth matrix coefficients of \pi. For f\in C^{\infty}(G(F)) and c\in\mathcal{C}(\pi) , let \displaystyle Z(s,f,c)=\int_{G(F)}f(g)|d(g)|^{s}c(g)dg, whenever this function converges absolutely. We refer to Z(s,f,c) as a local zeta integral . For a<b , let V_{[a,b]}:=\{s\in\mathbb{C}:a\leq\mathrm{Re}(s)\leq b\}. Conjecture 1.5 (Braverman-Kazhdan, Ngô) . One can choose \mathcal{S}_{\rho} as in Conjecture 1.1 that additionally satisfies the following for all f\in\mathcal{S}_{\rho} and tempered representations \pi of G(F): ( \mathcal{S} 1 ) For all c\in\mathcal{C}(\pi) the integral Z(s,f,c) converges for \mathrm{Re}(s) sufficiently large and \displaystyle\frac{Z(s,f,c)}{L(\tfrac{1}{2}+s,\pi,\rho)} is holomorphic as a function of s. Moreover Z(-s,\mathcal{F}_{\rho}(f),c^{\vee})=\gamma(\tfrac{1}{2}+s,\pi,\rho,\psi)Z(s,f,c) as meromorphic functions in s. ( \mathcal{S} 2 ) For Archimedean F, if p\in\mathbb{C}[s] is chosen so that s\mapstochar\rightarrow p(s)L(s,\pi,\rho) has no pole on V_{[a,b]} , then s\mapstochar\rightarrow p(s)Z(s,f,c) is bounded on V_{[a,b]} . ( \mathcal{S} 3 ) For non-Archimedean F and unramified \psi there is a function b_{\rho}\in\mathcal{S}_{\rho} such that \mathcal{F}_{\rho}(b_{\rho})=b_{\rho}, Z(s,b_{\rho},c) is nonzero only if \pi is unramified, and Z(s,b_{\rho},c^{\circ})=L(s,\pi,\rho) when c^{\circ} is the zonal spherical function. ( \mathcal{S} 4 ) If F is non-Archimedean then the space \mathcal{S}_{\rho} consists of functions that are almost compactly supported. Conjectures 1.1 and 1.5 are a large part of the local conjectures of Braverman-Kazhdan and Ngô on Schwartz spaces [ BK00 , Ngô14 , Ngô20 ] . A nice summary is given in [ LN24 , Conjecture 2.1.1] . To compare the conjectures, we point out that for technical reasons we only consider tempered representations in this paper, whereas in loc. cit. the authors consider all irreducible admissible representations. Moreover in loc. cit. the authors ask for an interpretation of \mathcal{F}_{\rho} as a convolution satisfying certain properties. We omit a discussion of this in the introduction (but see § 4.1 below). We also prove Conjecture 1.5 : Theorem 1.6 . If the desiderata on the local Langlands correspondence in § 3 are valid then Conjecture 1.5 holds. Let us simultaneously outline the argument and the sections of this paper. We point out at the outset that we work with more general G and \rho than those considered above in the body of the paper. In § 2 we collect various preliminaries, culminating with a description of the matrical Harish-Chandra Plancherel formula in § 2.8 . We then turn to a statement of a coarse version of the local Langlands correspondence in § 3 . We also prove uniform bounds on \gamma -factors in the Archimedean case in § 3.3 . These are important because we treat functions that are not necessarily finite under a maximal compact subgroup when F is Archimedean. We then construct the Fourier transform in § 4 and prove Conjecture 1.1 in Theorem 4.7 . Essentially we construct the Fourier transform using the Harish-Chandra Plancherel theorem by declaring that ( 1.0.1 ) is valid. We construct the kernel function for the Fourier transform spectrally in § 4.1 , define the basic function in § 4.2 , and prove several estimates for later use in § 4.3 . In § 5 we construct the asymptotic \rho -Schwartz space \mathcal{S}_{\rho}^{\mathrm{as}}(G(F))<\mathcal{C}(G(F)). Briefly, it is the inverse image under the Harish-Chandra Plancherel isomorphism of meromorphic sections with poles no worse than the Langlands L -functions attached to \rho. We refer to \mathcal{S}_{\rho}^{\mathrm{as}}(G(F)) as the asymptotic \rho -Schwartz space due to the fact that it is formally similar to the asymptotic Hecke algebra [ BK18 ] , as pointed out to us by Beuzart-Plessis. The asymptotic Hecke algebra satisfies all of the desiderata of Conjecture 1.5 except ( S 4 ). At this point we narrow our focus to the non-Archimedean case and cut out a subspace of \mathcal{S}_{\rho}^{\mathrm{as}}(G(F)) to construct the Schwartz space \mathcal{S}_{\rho}(G(F)). A clue to the correct definition is given by Example 6.2

Read the complete text