betaFeedback

WorksMathematics

Kloosterman Sums and Plücker Height Zeta Functions

Abstract

We establish nontrivial bounds for Kloosterman sums over pp-adic fields associated with the longest Weyl element on arbitrary split reductive groups, at arbitrary level. As an application, we obtain the regularity of Bessel distributions (of Whittaker models) for all such groups.

Introduction

Bounding Kloosterman sums is a fundamental problem in number theory, with many important applications.

In the classical GL2\mathrm{GL}_2 case, Salié [24] obtained early estimates. Weil [Wei48] established the celebrated square-root bound. Such estimates played an important role in Selberg’s work on the spectral theory of automorphic forms.

The study of Kloosterman sums beyond GL2\mathrm{GL}_2 was pioneered by Bump, Friedberg and Goldfeld [2]. Subsequently, Friedberg [11] and Stevens [25] simultaneously and independently developed the theory for general GLn\mathrm{GL}_n. Related developments for orthogonal groups were carried out by Cogdell, Li, Piatetski-Shapiro and Sarnak [6]. More recently, significant progress has been made for particular families of groups by Assing and Blomer [1], Blomer and Man [3], Man [19, 20], Linn [18], Mujdei [23], Huang (in the appendix of [12]) and Miao [22].

A uniform theory of Kloosterman sums for general split groups was initiated by Dąbrowski [8] and further developed by Dąbrowski and Reeder [9]. In particular, the latter obtained an explicit formula for the generating function of the cardinalities of Kloosterman sets, yielding general trivial bounds. Our recent works [4 CLYY26] further investigate Kloosterman sums on split groups and their connections with Bessel distributions.

In this paper, we establish nontrivial bounds for Kloosterman sums associated with the longest Weyl element on arbitrary split reductive groups at any level.

Our approach to obtain the power-saving counting estimates is through a representation-theoretic analysis of Plücker height zeta functions, using the Casselman-Shalika formula for spherical Whittaker functions and the Weyl dimension formula. This avoids explicit calculations of Plücker relations for individual groups.

Let GG be a split reductive group over a pp-adic field FF with derived subgroup GderG^{\mathrm{der}}. Let B=ANB = AN be a Borel subgroup of GG with AA a maximal torus of GG and NN the unipotent radical of BB. Denote by δ\delta the modulus character of BB. We choose a representative w0∈NG(A)w_0 \in N_G(A) of the longest Weyl element. Let ψ0\psi_0 be the generic character on NN given as in § 2. For an open compact subgroup KK of GG such that ψ0∣N∩K=1\psi_0|_{N\cap K}=1, we consider the following Kloosterman sum

Kl⁡(t)=Kl⁡(t;ψ0,w˙0,K)=∑x∈XK(w˙0t)ψ0(u(x))ψ0(u′(x)),t∈A.\operatorname{Kl}(t)=\operatorname{Kl}(t;\psi_0,\dot{w}_0,K)=\sum_{x\in X_K(\dot{w}_0t)}\psi_0(u(x))\psi_0(u'(x)),\qquad t\in A.

Here, the Kloosterman set is

XK(w˙0t)=(N∩K)\(Nw˙0tN∩K)/(N∩K).X_K(\dot{w}_0t)=(N\cap K)\backslash(N\dot{w}_0tN\cap K)/(N\cap K).

together with two maps

u:XK(w˙0t)→(N∩K)\N,u′:XK(w˙0t)→N/(N∩K).u:X_K(\dot{w}_0t)\to(N\cap K)\backslash N,\qquad u':X_K(\dot{w}_0t)\to N/(N\cap K).

In the above, if [n1w˙0tn2]∈XK(w˙0t)[n_1\dot{w}_0tn_2]\in X_K(\dot{w}_0t) with n1,n2∈Nn_1,n_2\in N and t∈At\in A, then

u([n1w˙0tn2])=[n1],u′([n1w˙0tn2])=[n2].u([n_1\dot{w}_0tn_2])=[n_1],\qquad u'([n_1\dot{w}_0tn_2])=[n_2].

We say that the Kloosterman sum Kl⁡\operatorname{Kl} has a nontrivial bound ([4 Definition 3.17]) if there exist constants ε>0\varepsilon>0 and C>0C>0 (possibly depending on pp) such that

∣Kl⁡(t)∣δ1/2−ε(t)≤C,t∈A∩Gder.|\operatorname{Kl}(t)|\delta^{1/2-\varepsilon}(t)\leq C,\qquad t\in A\cap G^{\mathrm{der}}.

The following is the main result of this paper.

Theorem 1.1. Let GG be a split reductive group over FF. Then for every compact open subgroup K⊂GK\subset G such that ψ0∣N∩K=1\psi_0|_{N\cap K}=1, the Kloosterman sum Kl⁡(⋅,ψ0−1,w˙0,K)\operatorname{Kl}(\cdot,\psi_0^{-1},\dot{w}_0,K) has a nontrivial bound.

As an application of Theorem 1.1, we obtain the regularity of Bessel distributions BπB_\pi on GG as follows (See [4] for the survey of this problem).

Denote by S(G)S(G) the space of Schwartz functions on GG. Let π\pi be an irreducible smooth admissible representation on GG with π∗\pi^* denoting its linear dual. Let π∨\pi^\vee be the contragredient representation of π\pi, that is, the smooth part of π∗\pi^*. The group GG then acts on all these spaces and these actions induce actions of S(G)S(G). Note that for any ℓ∈(π∨)∗\ell\in(\pi^\vee)^* and f∈S(G)f\in S(G), π∨(f)ℓ\pi^\vee(f)\ell is smooth so that π∨(f)ℓ∈(π∨)∨=π\pi^\vee(f)\ell\in(\pi^\vee)^\vee=\pi. For any ℓ1∈π∗\ell_1\in\pi^* and ℓ2∈(π∨)∗\ell_2\in(\pi^\vee)^*, the relative character associated to ℓ1\ell_1 and ℓ2\ell_2 is defined as

Bℓ1,ℓ2(f)=ℓ1(π∨(f)ℓ2),f∈S(G).B_{\ell_1,\ell_2}(f)=\ell_1(\pi^\vee(f)\ell_2),\qquad f\in S(G).

When π\pi is generic with respect to ψ0\psi_0, and both ℓ1,ℓ2\ell_1,\ell_2 are Whittaker functionals, the associated relative character Bπ=Bℓ1,ℓ2B_\pi=B_{\ell_1,\ell_2} is called Bessel distribution attached to π\pi.

Theorem 1.2. Suppose GG is a split reductive group over FF. For any irreducible admissible generic representation π\pi of GG, the Bessel distribution BπB_\pi with respect to the generic character ψ0\psi_0 is regular.

Proof. By Theorem [4 Theorem 3.24], it is enough to prove that for any subset II of the set of simple roots Δ\Delta for GG, the Kloosterman sum Kl⁡MI\operatorname{Kl}_{M_I} on the standard Levi MIM_I associated to GG with respect to the triple (ψ0∣NMI,w˙I,KI)(\psi_0|_{N_{M_I}},\dot{w}_I,K_I) has a nontrivial bound. Here, KIK_I is some admissible group of MIM_I ([4 Definition 2.25]) and {w˙I}I⊂Δ\{\dot{w}_I\}_{I\subset\Delta} is a system of representatives of {wI}I⊂Δ\{w_I\}_{I\subset\Delta}. Applying Theorem 1.1 to each MIM_I, we obtain the regularity. □

We sketch the proof Theorem 1.1. For this, we assume GG is semisimple and simply connected. Stevens’ approach [25] is a basic method for bounding Kloosterman sums. By considering the action of AA on the Kloosterman set, the contribution of each orbit in the Kloosterman sum on GG can be decomposed into a product of Kloosterman sums on GL2G_{L_2}. Applying the Weil bound for these Kloosterman sums on GL2G_{L_2}, the problem is reduced to a counting problem for the Kloosterman set.

Plücker coordinates are generators of F[N\G]F[N\backslash G]. Based on the homogeneity of Plücker coordinates, the counting problem is reduced to that of the Plücker-constrained sets SrS_r, r∈Z≥0mr\in\mathbb{Z}_{\geq0}^m with mm the rank of GG, which are subsets of affine spaces defined by conditions on Plücker coordinates (Definition 3.6). This reduction is the main result of [CLYY26] (Theorem 5.1).

For each r∈Z≥0mr\in\mathbb{Z}_{\geq0}^m with R=∑iriR=\sum_i r_i, we have (Proposition 3.8)

∣Sr∣≤∑0≤ℓ≤Rbℓ,bℓ=#{n∈D/N(O)∣h(n)=ℓ}.|S_r|\leq\sum_{0\leq\ell\leq R}b_\ell,\qquad b_\ell=\#\{n\in\mathcal{D}/N(\mathcal{O})\mid h(n)=\ell\}.

Here, D\mathcal{D} consists of elements in NN with integral entries for simple roots and hh is the Plücker height (Definition 3.3).

We define the Plücker height zeta function (Definition 4.1)

ZPl(s)=∑ℓ≥0bℓq−ℓs.(1)Z_{\mathrm{Pl}}(s)=\sum_{\ell\geq0}b_\ell q^{-\ell s}. \tag*{(1)}

The series is absolutely convergent for Re⁡(s)>1\operatorname{Re}(s)>1.

A simple Fourier analysis gives (Lemma 4.5)

ZPl(s)=∫OmJs(c) dc,Js(c)=∫Nfs(w˙0n)ψc(n)‾ dn.Z_{\mathrm{Pl}}(s)=\int_{\mathcal{O}_m}J_s(c)\,dc,\qquad J_s(c)=\int_N f_s(\dot{w}_0n)\overline{\psi_c(n)}\,dn.

Here, fsf_s is the normalized spherical section of certain unramified principal series on GG.

The Casselman–Shalika formula expresses the spherical Whittaker function JsJ_s on GG in terms of Weyl characters of finite-dimensional representations on the Langlands dual group Gad∨G_{\mathrm{ad}}^\vee of the adjoint quotient GadG_{\mathrm{ad}} of GG (Theorem 4.6). We obtain that ZPl(s)Z_{\mathrm{Pl}}(s) extends holomorphically to the half-plane (Theorem 4.4(1))

Re⁡(s)>1−1MG.\operatorname{Re}(s)>1-\frac{1}{M_G}.

Here, MG=max⁡iaiM_G=\max_i a_i with 2ρ=∑i=1maiαi2\rho=\sum_{i=1}^{m}a_i\alpha_i where 2ρ2\rho is the sum of positive roots and αi\alpha_i, 1≤i≤m1\leq i\leq m are the simple roots.

Moreover, the Weyl dimension formula gives the bounding of ZPl(s)Z_{\mathrm{Pl}}(s) for ss near its convergence boundary (Theorem 4.4(2)). This estimate for ZPl(s)Z_{\mathrm{Pl}}(s) implies the following bound for Plücker constrained sets SrS_r (Theorem 4.8)

∣Sr∣≪q(1−12MG)R,R=∑iri.|S_r|\ll q^{\left(1-\frac{1}{2M_G}\right)R},\qquad R=\sum_i r_i.

By Theorem 5.1, this gives a nontrivial bound for Kloosterman sums on GG with level KdK_d, d≫0d\gg0 (Corollary 5.2). Here, {Kd}d\{K_d\}_d is the family of principal congruence subgroups of GG.

By the germ expansion of Kloosterman integrals (a generalization of Kloosterman sums) in [4], together with the corresponding bounds for standard Levi subgroups, we obtain non-trivial bounds for Kloosterman sums with arbitrary level.

It is instructive to compare our approach with that of Dąbrowski and Reeder [9 Section 2], whose method yields the trivial bound for Kloosterman sums.

Acknowledgements. L. Cai is supported by the National Key R&D Program of China No. 2023YFA1009702 and National Natural Science Foundation of China, No. 12371012. J. Chai is supported by National Natural Science Foundation of China, No. 12571010 and No. 12671039.

Use of artificial intelligence. ChatGPT was used for exploratory computations, editing, searching for references and for verification of results.

Notations and conventions

Let FF be a pp-adic field. Denote by O\mathcal{O} the ring of integers, p\mathfrak{p} the maximal ideal of O\mathcal{O} and qq the cardinality of its residue field. Denote by ∣⋅∣|\cdot| the pp-adic absolute value on FF with vv the additive valuation. We use ϖ\varpi to denote a fixed uniformizer of FF.

Let GG be a split (connected) reductive group over FF. We shall make no distinction between an algebraic group and its FF-points.

We fix a maximal torus AA of GG and a Borel subgroup B=ANB=AN of GG. Denote by ZZ the center of GG. Let Φ+\Phi^+ be its positive roots and Δ\Delta its simple roots. Put L=∣Φ+∣L=|\Phi^+| and m=∣Δ∣m=|\Delta|. Let α1,…,αm∈Δ\alpha_1,\ldots,\alpha_m\in\Delta be the simple roots and ω1,…,ωm\omega_1,\ldots,\omega_m be the fundamental weights.

Denote by

2ρ=∑γ∈Φ+γ.2\rho=\sum_{\gamma\in\Phi^+}\gamma.

Then

ρ=∑iωi.\rho=\sum_i\omega_i.

Denote by X∗(A)X^*(A) and X∗(A)X_*(A) the groups of characters and cocharacters on AA. For any a∈Aa\in A, there is a unique cocharacter ν(a)∈X∗(A)\nu(a)\in X_*(A) such that

⟨χ,ν(a)⟩=v(χ(a)),χ∈X∗(A).\langle\chi,\nu(a)\rangle=v(\chi(a)),\qquad\chi\in X^*(A).

Denote by δ\delta the modulus character of BB. Then

δ(a)=q−⟨2ρ,ν(a)⟩,a∈A.\delta(a)=q^{-\langle2\rho,\nu(a)\rangle},\qquad a\in A.

By the existence theorem for split reductive group schemes (see [7 Exposé XXV, Theorem 1.1 and Corollary 1.2]), there exists a split reductive group scheme GZG_{\mathbb{Z}} over Z\mathbb{Z}, a split maximal torus AZA_{\mathbb{Z}} and a Borel subgroup BZ=AZNZB_{\mathbb{Z}}=A_{\mathbb{Z}}N_{\mathbb{Z}}, together with an isomorphism

GZ⊗ZF→∼GG_{\mathbb{Z}}\otimes_{\mathbb{Z}}F \xrightarrow{\sim} G

such that under this isomorphism, AZ⊗ZFA_{\mathbb{Z}}\otimes_{\mathbb{Z}}F and BZ⊗ZFB_{\mathbb{Z}}\otimes_{\mathbb{Z}}F go to AA and BB, respectively.

We fix a Chevalley system for GZG_{\mathbb{Z}} ([7 Exposé XXIII, Proposition 6.2])

Xγ:Ga,Z→∼Nγ,Z,γ∈Φ.X_{\gamma}:\mathbb{G}_{\mathrm{a},\mathbb{Z}}\xrightarrow{\sim}N_{\gamma,\mathbb{Z}},\qquad\gamma\in\Phi.

We shall use the same notation XγX_{\gamma} for their base changes to FF.

Denote by G=GZ⊗ZO\mathcal{G}=G_{\mathbb{Z}}\otimes_{\mathbb{Z}}\mathcal{O}. Then G\mathcal{G} is a split reductive O\mathcal{O}-model of GG. Denote by A\mathcal{A}, B\mathcal{B}, and N\mathcal{N} the corresponding subgroup schemes obtained by base change from AZA_{\mathbb{Z}}, BZB_{\mathbb{Z}} and NZN_{\mathbb{Z}} respectively. In the following, all notations involving O\mathcal{O}-point refer to this model.

Let W=NG(A)/AW=N_G(A)/A be the Weyl group. For each simple root αi\alpha_i, let wi∈Ww_i\in W be the corresponding simple reflection. Consider

w˙i=Xαi(1)X−αi(−1)Xαi(1).\dot{w}_i=X_{\alpha_i}(1)X_{-\alpha_i}(-1)X_{\alpha_i}(1).

Then w˙i∈NGZ(AZ)(Z)\dot{w}_i\in N_{G_{\mathbb{Z}}}(A_{\mathbb{Z}})(\mathbb{Z}) represents wiw_i. For each w∈Ww\in W, choose a reduced expression w=wk1⋯wkrw=w_{k_1}\cdots w_{k_r} and set

w˙=w˙k1⋯w˙kr.\dot{w}=\dot{w}_{k_1}\cdots\dot{w}_{k_r}.

The definition of w˙\dot{w} is independent on the choice of the above reduced expression of ww. In particular, we fix a representative w˙0\dot{w}_0 of the longest Weyl element w0w_0 which satisfies

w˙0∈NGZ(AZ)(Z)⊂NG(A)(F)∩G(O).\dot{w}_0\in N_{G_{\mathbb{Z}}}(A_{\mathbb{Z}})(\mathbb{Z})\subset N_G(A)(F)\cap G(\mathcal{O}).

We use this representative throughout the paper.

We fix an ordering Φ+={γ1,…,γL}\Phi^{+}=\{\gamma_1,\ldots,\gamma_L\} by increasing height. With respect to this ordering, every element in NN can be written uniquely as

n(x)=∏γ∈Φ+Xγ(xγ),x=(xγ)γ∈Φ+∈FΦ+.n(\mathbf{x})=\prod_{\gamma\in\Phi^{+}}X_{\gamma}(x_{\gamma}),\qquad\mathbf{x}=(x_{\gamma})_{\gamma\in\Phi^{+}}\in F^{\Phi^{+}}.

If n=n(x)n=n(\mathbf{x}) for some x∈FΦ+\mathbf{x}\in F^{\Phi^{+}}, then for any γ∈Φ+\gamma\in\Phi^{+}, we write xγ(n)x_{\gamma}(n) for xγx_{\gamma}.

We fix a nontrivial character ψF\psi_F on FF. We assume that ψF∣O=1\psi_F|_{\mathcal{O}}=1, and that ψF∣ϖ−1O≠1\psi_F|_{\varpi^{-1}\mathcal{O}}\ne1. For each c=(ci)i∈Fm\mathbf{c}=(c_i)_i\in F^m, denote by

ψc(n)=ψF(∑icixαi(n)).\psi_{\mathbf{c}}(n)=\psi_F\left(\sum_i c_i x_{\alpha_i}(n)\right).

We shall write ψ0=ψc\psi_0=\psi_{\mathbf{c}} if ci=1c_i=1 for all ii.

The Plücker height

In this section, we assume GG is semisimple and simply connected. For each i=1,…,mi=1,\ldots,m, let (πi,Vi)(\pi_i,V_i) be an irreducible (algebraic) representation of GG with highest weight ωi∈X∗(A)\omega_i\in X^*(A).

Lemma 3.1. There exists a G(O)G(\mathcal{O})-stable lattice Λi\Lambda_i in ViV_i with a O\mathcal{O}-basis {eij}j\{e_{ij}\}_j of Λi\Lambda_i such that eije_{ij} are weight vectors and the basis containing a generator ei+e_i^+ of the highest-weight space.

Proof. By [15 Lemma in I.10.4], there is a GG-stable lattice Λi\Lambda_i in ViV_i and hence is a G(O)G(\mathcal{O})-stable lattice of ViV_i. Consider the restriction of the action GG on Λi\Lambda_i to the split torus AA. As the group AA is diagonalizable, by [15 (3) in I.2.11], Λi\Lambda_i decomposes into A(O)A(\mathcal{O})-eigen subspace:

Λi=⨁μ∈X∗(A)Λi,μ.\Lambda_i=\bigoplus_{\mu\in X^*(A)}\Lambda_{i,\mu}.

Here, X∗(A)=Hom⁡(A,Gm,O)X^*(A)=\operatorname{Hom}(A,\mathbb{G}_{\mathrm{m},\mathcal{O}}) and Λi,μ\Lambda_{i,\mu} consists of v∈Λiv\in\Lambda_i such that πi(a)v=μ(a)v\pi_i(a)v=\mu(a)v for all a∈A(O)a\in A(\mathcal{O}).

After extending scalars to FF, this is the weight-space decomposition of $V_i

Λi,μ⊗OF=Vi,μF,Λi,μ=Λi∩Vi,μF.\Lambda_{i,\mu}\otimes_{\mathcal{O}}F=V_{i,\mu_F},\qquad\Lambda_{i,\mu}=\Lambda_i\cap V_{i,\mu_F}.

Here we use the following canonical isomorphism ([17 Fact B.1.2])

X∗(A)→∼X∗(A),μ↦μF:=μ⊗F.X^*(A)\xrightarrow{\sim}X^*(A),\qquad\mu\mapsto\mu_F:=\mu\otimes F.

The highest-weight space Vi,ωiV_{i,\omega_i} is one-dimensional, so its lattice Λi,ωi\Lambda_{i,\omega_i} has rank one. Choose an O\mathcal{O}-basis in each Λi,μ\Lambda_{i,\mu}, choosing a generator ei+e_i^+ for the highest-weight summand. The union of these bases is a weight basis of ViV_i and an O\mathcal{O}-basis of Λi\Lambda_i. □

For each ii, denote by Λi∗=Hom⁡O(Λi,O)\Lambda_i^* = \operatorname{Hom}_{\mathcal{O}}(\Lambda_i,\mathcal{O}) the dual lattice of Λi\Lambda_i. A linear functional ℓi∈Λi∗\ell_i \in\Lambda_i^* is called primitive if there is no ℓ′∈Λi∗\ell' \in\Lambda_i^* such that ℓ=ϖℓ′\ell= \varpi\ell'.

Lemma 3.2. For each ii, there exists a unique primitive linear functional ℓi∈Λi∗\ell_i \in\Lambda_i^* such that

  • ℓi∘πi(n)=ℓi\ell_i \circ\pi_i(n) = \ell_i for any n∈Nn \in N.

  • ℓi∘πi(a)=w˙0ωi(a)ℓi\ell_i \circ\pi_i(a) = \dot{w}_0\omega_i(a)\ell_i for any a∈Aa \in A.

  • ℓi(πi(w˙0)ei−)=1\ell_i(\pi_i(\dot{w}_0)e_i^-) = 1.

Proof. We give a construction of the required ℓi\ell_i. Consider the weight decomposition in Lemma 3.1

Λi=⨁μΛi,μ.\Lambda_i = \bigoplus_{\mu} \Lambda_{i,\mu}.

Denote by λi=w˙0ωi\lambda_i = \dot{w}_0\omega_i. Then λi\lambda_i is the lowest weight of ViV_i and Λi,λi=Oei−\Lambda_{i,\lambda_i} = \mathcal{O}e_i^- is of rank one. Consider ℓi∈Λi∗\ell_i \in\Lambda_i^* such that

ℓi(ei−)=1,ℓi∣Λi,μ=0,μ≠λi.\ell_i(e_i^-) = 1,\qquad\ell_i|_{\Lambda_{i,\mu}} = 0,\quad\mu\ne\lambda_i.

For any weight vector vμ∈Vμv_\mu\in V_\mu and n∈Nn \in N, we have

π(n)vμ=vμ+∑ν>μvν.\pi(n)v_\mu= v_\mu+ \sum_{\nu>\mu} v_\nu.

In particular, for any μ>λi\mu> \lambda_i, ℓi(π(n)vμ)=0=ℓi(vμ)\ell_i(\pi(n)v_\mu) = 0 = \ell_i(v_\mu) and ℓi(πi(n)vλi)=ℓi(vλi)\ell_i(\pi_i(n)v_{\lambda_i}) = \ell_i(v_{\lambda_i}). Therefore, ℓi\ell_i is NN-invariant.

As the torus AA acts on VλiV_{\lambda_i} via λi=w˙0ωi\lambda_i = \dot{w}_0\omega_i, ℓiπi(a)=λi(a)ℓi\ell_i\pi_i(a) = \lambda_i(a)\ell_i for any a∈Aa \in A.

We have πi(w˙0)ei+∈Vλi\pi_i(\dot{w}_0)e_i^+ \in V_{\lambda_i}. As we require that w˙0∈G(O)\dot{w}_0 \in G(\mathcal{O}),

πi(w˙0)ei+=uei−,u∈O×\pi_i(\dot{w}_0)e_i^+ = u e_i^-,\qquad u \in\mathcal{O}^{\times}

so that ℓi(πi(w˙0)ei+)=u\ell_i(\pi_i(\dot{w}_0)e_i^+) = u. Modifying ℓi\ell_i to u−1ℓiu^{-1}\ell_i, we obtain the required ℓi∈Λi∗\ell_i \in\Lambda_i^*. The uniqueness of such linear functional is clear. □\square

Definition 3.3 (The Plücker height). For each ii, we choose a G(O)G(\mathcal{O})-stable lattice Λi\Lambda_i of ViV_i and an O\mathcal{O}-basis {eij}j\{e_{ij}\}_j of Λi\Lambda_i as in Lemma 3.1. We also choose a primitive linear functional ℓi∈Λi∗\ell_i \in\Lambda_i^* as in Lemma 3.2. Denote by

pij(n)=ℓi(πi(w˙0n)eij),Hi(n)=max⁡j∣pij(n)∣,H(n)=∏iHi(n).p_{ij}(n) = \ell_i(\pi_i(\dot{w}_0n)e_{ij}),\qquad H_i(n) = \max_j |p_{ij}(n)|,\qquad H(n) = \prod_i H_i(n).

Proposition 3.4. We have the following.

  1. For each ii, Hi(n)≥1H_i(n) \ge1 and Hi(nn0)=Hi(n)H_i(nn_0) = H_i(n) for any n∈Nn \in N and n0∈N(O)n_0 \in N(\mathcal{O}). In particular, for each n∈Nn \in N, there exists a unique positive integer hi(n)≥0h_i(n) \ge0 with Hi(n)=qhi(n)H_i(n) = q^{h_i(n)} and we denote by h(n)=∑ihi(n)h(n) = \sum_i h_i(n).

  1. Let Gad=G/ZG_{\mathrm{ad}} = G/Z be the adjoint quotient associated to GG. Denote by Aad=A/ZA_{\mathrm{ad}} = A/Z and Bad=B/ZB_{\mathrm{ad}} = B/Z. For each s∈Cs \in\mathbb{C}, consider the induced representation on GadG_{\mathrm{ad}}

I(s−1)=IBadGad(s−1)/2={f:Gad→C|f(bg)=δs/2(b)f(g), ∀b∈Bad, g∈Gad}.I(s-1) = I_{B_{\mathrm{ad}}}^{G_{\mathrm{ad}}}(s-1)/2 = \left\{f:G_{\mathrm{ad}}\to\mathbb{C}\middle|f(bg)=\delta^{s/2}(b)f(g),\ \forall b\in B_{\mathrm{ad}},\ g\in G_{\mathrm{ad}}\right\}.

Let fs∈I(s−1)f_s \in I(s-1) be the unique spherical vector with fs(1)=1f_s(1) = 1. Then

fs(w˙0n)=H(n)−s,n∈N.f_s(\dot{w}_0n) = H(n)^{-s},\qquad n\in N.

Remark 3.5. Here, the reason we prefer GadG_{\mathrm{ad}} to GG is for the Fourier analysis of the zeta function. See Theorem 4.4 (1).

The proof of Proposition 3.4 (1) The highest-weight vector ei+e_i^+ is contained in the integral weight basis of Λi\Lambda_i chosen in Lemma 3.1, say eij0=ei+e_{ij_0} = e_i^+. The unipotent subgroup NN of GG fixes this vector. Hence for every n∈Nn \in N,

pij0(n)=ℓi(πi(w˙0n)ei+)=ℓi(πi(w˙0)ei+)=1.p_{ij_0}(n) = \ell_i(\pi_i(\dot{w}_0n)e_i^+) = \ell_i(\pi_i(\dot{w}_0)e_i^+) = 1.

Since Hi(n)=max⁡j∣pij(n)∣H_i(n) = \max_j |p_{ij}(n)|, this gives Hi(n)≥∣pij0(n)∣=1H_i(n) \ge|p_{ij_0}(n)| = 1. In particular, we can write Hi(n)=qhi(n)H_i(n) = q^{h_i(n)} for some hi(n)≥0h_i(n) \ge0.

Let n0∈N(O)n_0 \in N(\mathcal{O}). Because Λi\Lambda_i is G(O)G(\mathcal{O})-stable, πi(n0)\pi_i(n_0) preserves Λi\Lambda_i. Let MM be the matrix of πi(n0)\pi_i(n_0) with respect to the O\mathcal{O}-basis {eij}j\{e_{ij}\}_j. Then the entries of MM are contained in O\mathcal{O}. For any n∈Nn \in N, we have

(pi1(nn0),…,pidi(nn0))=(pi1(n),…,pidi(n))M(p_{i1}(nn_0),\ldots,p_{id_i}(nn_0)) = (p_{i1}(n),\ldots,p_{id_i}(n))M

with di=dim⁡Vid_i=\dim V_i. This implies that for each jj

∣pij(nn0)∣≤Hi(n)|p_{ij}(nn_0)|\le H_i(n)

and

Hi(nn0)≤Hi(n).H_i(nn_0)\le H_i(n).

Similarly, as πi(n0−1)\pi_i(n_0^{-1}) preserves Λi\Lambda_i, for any n∈Nn\in N, we have

Hi(nn0−1)≤Hi(n).H_i(nn_0^{-1})\le H_i(n).

In particular, HiH_i is right N(O)N(\mathcal{O})-invariant.

(2) For any n∈Nn\in N, consider the Iwasawa decomposition

w˙0n=uak,u∈N,a∈A,k∈G(O).\dot{w}_0n=uak,\qquad u\in N,a\in A,k\in G(\mathcal{O}).

Then for each i,ji,j,

pij(n)=(w0ωi)(a)ℓi(πi(k)eij).p_{ij}(n)=(w_0\omega_i)(a)\ell_i(\pi_i(k)e_{ij}).

Similar argument as in the proof of (1),

max⁡j∣ℓi(πi(k)eij)∣=max⁡j∣ℓi(eij)∣.\max_j|\ell_i(\pi_i(k)e_{ij})|=\max_j|\ell_i(e_{ij})|.

As ℓi\ell_i is primitive,

max⁡j∣ℓi(eij)∣=1.\max_j|\ell_i(e_{ij})|=1.

In particular,

Hi(n)=max⁡j∣pij(n)∣=∣w0ωi(a)∣,H(n)=∏iHi(n)=∏i∣w0ωi(a)∣.H_i(n)=\max_j|p_{ij}(n)|=|w_0\omega_i(a)|,\qquad H(n)=\prod_iH_i(n)=\prod_i|w_0\omega_i(a)|.

Therefore,

h(n)=−∑i⟨w0ωi,ν(a)⟩=⟨ρ,ν(a)⟩.h(n)=-\sum_i\langle w_0\omega_i,\nu(a)\rangle=\langle\rho,\nu(a)\rangle.

We obtain

fs(w˙0n)=δ(a)s/2=q−s⟨ρ,ν(a)⟩=H(n)−s.f_s(\dot{w}_0n)=\delta(a)^{s/2}=q^{-s\langle\rho,\nu(a)\rangle}=H(n)^{-s}.

Consider the Plücker coordinate polynomials

pij(x)=pij(n(x)),x∈FΦ+.p_{ij}(x)=p_{ij}(n(x)),\qquad x\in F^{\Phi^+}.

Definition 3.6 (Plücker-constrained set). We fix a section R⊂F\mathcal{R}\subset F of F/OF/\mathcal{O} containing 00. We also choose polynomials qγ∈F[x]q_\gamma\in F[x] for each γ∈Φ+\gamma\in\Phi^+ satisfying

qαi=xαi,qγ=xγ+fγ,fγ∈F[xβ:β<γ].q_{\alpha_i}=x_{\alpha_i},\qquad q_\gamma=x_\gamma+f_\gamma,\qquad f_\gamma\in F[x_\beta:\beta<\gamma].

For r=(r1,…,rm)∈Z≥0mr=(r_1,\ldots,r_m)\in\mathbb{Z}_{\geq0}^m, define the Plücker-constrained set

Sr={x=(xγ)γ∈Φ+∣xαi=0, v(pij(x))≥−ri for all i,j, qγ(x)∈R for all γ∈Φ+∖Δ}.S_r=\left\{x=(x_\gamma)_{\gamma\in\Phi^+}\mid x_{\alpha_i}=0,\ v(p_{ij}(x))\ge-r_i\text{ for all }i,j,\ q_\gamma(x)\in\mathcal{R}\text{ for all }\gamma\in\Phi^+\setminus\Delta\right\}.

The main result of [CLYY26] is that the bounding of Plücker-constrained sets implies the bounding of Kloosterman sums. See Theorem 5.1.

Consider the right N(O)N(\mathcal{O})-invariant space

D={n∈N∣xαi(n)∈O,1≤i≤m}.\mathcal{D}=\left\{n\in N\mid x_{\alpha_i}(n)\in\mathcal{O},\quad1\le i\le m\right\}.

Lemma 3.7. For each positive integer ℓ≥0\ell\ge0, there are only finitely many n∈D/N(O)n\in\mathcal{D}/N(\mathcal{O}) with the Plücker height h(n)=ℓh(n)=\ell.

Proof. We shall give a proof in Lemma 4.2.

:::

For each positive integer ℓ≥0\ell\ge0, denote by

bℓ=#{n∈D/N(O)∣h(n)=ℓ}.b_\ell=\#\left\{n\in\mathcal{D}/N(\mathcal{O})\mid h(n)=\ell\right\}.

Proposition 3.8. For any r∈Z≥0mr\in\mathbb{Z}_{\geq0}^m, we have

∣Sr∣≤∑ℓ=0Rbℓ,R=∑iri.|S_r|\le\sum_{\ell=0}^{R}b_\ell,\qquad R=\sum_i r_i.

Proof. We shall prove that

∣Sr∣≤#{n∈D/N(O)∣h(n)≤R}.|S_r| \le\#\left\{n \in\mathcal{D}/N(\mathcal{O}) \mid h(n) \le R\right\}.

By [CLYY26, Lemma 3.13], there is a bijective map

{x=(xγ)γ∈F+∣qγ(x)∈R for all γ∈Φ+}→∼N/N(O),x↦n(x)N(O).\left\{\mathbf{x}=(x_\gamma)_{\gamma\in F^+} \mid q_\gamma(\mathbf{x}) \in\mathcal{R}\text{ for all }\gamma\in\Phi^+\right\} \xrightarrow{\sim} N/N(\mathcal{O}),\quad\mathbf{x}\mapsto n(\mathbf{x})N(\mathcal{O}).

In particular, the restriction of the above map to SrS_r gives an injection

S_r \lhook\joinrel\longrightarrow\mathcal{D}/N(\mathcal{O}),\quad\mathbf{x}\mapsto n(\mathbf{x})N(\mathcal{O}).

In the definition of SrS_r, we require that v(pij(x))≥−riv(p_{ij}(\mathbf{x})) \ge-r_i for all i,ji,j. We have

Hi(n(x))=max⁡j∣pij(x)∣≤qriH_i(n(\mathbf{x}))=\max_j |p_{ij}(\mathbf{x})| \le q^{r_i}

so that

H(n(x))≤qR,h(n(x))≤R.H(n(\mathbf{x})) \le q^R,\quad h(n(\mathbf{x})) \le R.

Therefore, the restriction of the above map x↦n(x)N(O)\mathbf{x}\mapsto n(\mathbf{x})N(\mathcal{O}) to SrS_r gives an injective map

S_r \lhook\joinrel\longrightarrow\left\{n\in\mathcal{D}/N(\mathcal{O})\mid h(n)\le R\right\}.

The Zeta function

In this section, we still assume that GG is semisimple and simply connected. We normalize the measure on NN such that Vol⁡(N(O))=1\operatorname{Vol}(N(\mathcal{O}))=1.

Definition 4.1 (The Plücker height zeta function). We define the following zeta function for s∈Cs\in\mathbb{C}

ZPl(s)=∫DH(n)−s dn.Z_{\mathrm{Pl}}(s)=\int_{\mathcal{D}}H(n)^{-s}\,dn.

Lemma 4.2. The series defining ZPl(s)Z_{\mathrm{Pl}}(s) is absolutely convergent for Re⁡(s)>1\operatorname{Re}(s)>1. In particular, for any ℓ≥0\ell\ge0, the cardinality bℓb_\ell is finite and for any ss with Re⁡(s)>1\operatorname{Re}(s)>1

ZPl(s)=∑ℓ≥0bℓq−ℓs.Z_{\mathrm{Pl}}(s)=\sum_{\ell\ge0}b_\ell q^{-\ell s}.

Proof. By Proposition 3.4(2), for any s∈Cs\in\mathbb{C} and n∈Nn\in N,

H(n)−s=fs(w˙0n)H(n)^{-s}=f_s(\dot{w}_0n)

with fsf_s the normalized spherical function on the induced representation I(s−1)I(s-1). In particular,

ZPl(s)=∫Dfs(w˙0n) dn.Z_{\mathrm{Pl}}(s)=\int_{\mathcal{D}}f_s(\dot{w}_0n)\,dn.

The absolutely convergence follows from the absolutely convergence of the integral defining the intertwining operator from I(s−1)I(s-1) to I(1−s)I(1-s). This is a special case of [26 Proposition IV.2.1].

As we normalize the measure on NN such that Vol⁡(N(O))=1\operatorname{Vol}(N(\mathcal{O}))=1, for any ss with Re⁡(s)>1\operatorname{Re}(s)>1

ZPl(s)=∑n∈D/N(O)H(n)−s=∑ℓ≥0∑n∈D/N(O)h(n)=ℓq−ℓs=∑ℓ≥0bℓq−ℓs.Z_{\mathrm{Pl}}(s)=\sum_{n\in\mathcal{D}/N(\mathcal{O})}H(n)^{-s}=\sum_{\ell\ge0}\sum_{\substack{n\in\mathcal{D}/N(\mathcal{O})\\h(n)=\ell}}q^{-\ell s}=\sum_{\ell\ge0}b_\ell q^{-\ell s}.

In particular, for each ℓ≥0\ell\ge0, bℓb_\ell is finite.

Before we state the main result for the zeta function, we introduce some notations. Denote by ωi∨∈X∗(A)Q\omega_i^\vee\in X_*(A)_{\mathbb{Q}}, 1≤i≤m1\le i\le m be the fundamental coweights of GG, that is, ⟨αi,ωj∨⟩=δij\langle\alpha_i,\omega_j^\vee\rangle=\delta_{ij}. As we assume that GG is semisimple and simply connected, ωi∨∈X∗(Aad)\omega_i^\vee\in X_*(A_{\mathrm{ad}}). For each z∈Z≥0mz\in\mathbb{Z}_{\ge0}^m, denote by

μz=∑iziωi∨∈X∗(Aad).\mu_z=\sum_i z_i\omega_i^\vee\in X_*(A_{\mathrm{ad}}).

Denote by G^ad\widehat{G}_{\mathrm{ad}} the Langlands dual group of GadG_{\mathrm{ad}}. In particular, μz\mu_z can be viewed as a dominant weight of G^ad\widehat{G}_{\mathrm{ad}}. Let πz\pi_z the irreducible representation on G^ad\widehat{G}_{\mathrm{ad}} with the highest weight μz\mu_z. Denote by

Pz(u)=∑χm(πz,χ)u⟨ρ,μz+χ⟩.P_z(u)=\sum_\chi m(\pi_z,\chi)u^{\langle\rho,\mu_z+\chi\rangle}.

Here, χ∈X∗(Aad)\chi\in X_*(A_{\mathrm{ad}}) and m(πz,χ)m(\pi_z,\chi) is the multiplicity of χ\chi in πz\pi_z.

Lemma 4.3. For each z∈Z≥0mz \in\mathbb{Z}_{\geq0}^{m}, Pz(u)P_z(u) is a polynomial with nonnegative integral coefficients and degree dz=⟨2ρ,μz⟩d_z = \langle2\rho,\mu_z\rangle.

Proof. We shall prove that for any weight χ∈X∗(Aad)\chi\in X_*(A_{\mathrm{ad}}) of πz\pi_z,

0≤⟨ρ,μz+χ⟩≤dz.0 \leq\langle\rho,\mu_z+\chi\rangle\leq d_z.

In fact, for any weight χ\chi of μz\mu_z, we have μz−χ=∑i=1mciαi∨\mu_z-\chi=\sum_{i=1}^{m}c_i\alpha_i^\vee with some ci∈Z≥0c_i \in\mathbb{Z}_{\geq0}. Therefore, ⟨ρ,μz−χ⟩≥0\langle\rho,\mu_z-\chi\rangle\geq0 and ⟨ρ,χ⟩≤⟨ρ,μz⟩\langle\rho,\chi\rangle\leq\langle\rho,\mu_z\rangle. On the other hand, w0χw_0\chi is also a weight of μz\mu_z so that

−⟨ρ,χ⟩=⟨ρ,w0χ⟩≤⟨ρ,μz⟩-\langle\rho,\chi\rangle=\langle\rho,w_0\chi\rangle\leq\langle\rho,\mu_z\rangle

We obtain that −⟨ρ,μz⟩≤⟨ρ,χ⟩≤⟨ρ,μz⟩-\langle\rho,\mu_z\rangle\leq\langle\rho,\chi\rangle\leq\langle\rho,\mu_z\rangle and the Lemma holds.

Theorem 4.4. We have the following.

  1. In the region Re⁡(s)>1\operatorname{Re}(s)>1, we have

ZPl(s)=(1−q−1)mE(q1−s)∑z∈Z≥0mq−ZPz(q1−s).(2)Z_{\mathrm{Pl}}(s)=(1-q^{-1})^{m}\mathcal{E}(q^{1-s})\sum_{z\in\mathbb{Z}_{\geq0}^{m}}q^{-Z}P_z(q^{1-s}). \tag*{(2)}

Here, the polynomial E(u)=∏γ∈Φ+(1−q−1u⟨ρ,γ∨⟩)\mathcal{E}(u)=\prod_{\gamma\in\Phi^{+}}(1-q^{-1}u^{\langle\rho,\gamma^\vee\rangle}) and for each z=(zi)i∈Z≥0mz=(z_i)_i\in\mathbb{Z}_{\geq0}^{m}, we denote by Z=∑iziZ=\sum_i z_i.

  1. Write 2ρ=∑iaiαi2\rho=\sum_i a_i\alpha_i with ai≥1a_i\geq1 and denote by

MG=max⁡iai.M_G=\max_i a_i.

The right hand side of Equation (4.1) is absolutely convergent for Re⁡(s)>1−1MG\operatorname{Re}(s)>1-\frac{1}{M_G}. In particular, this gives the holomorphic continuation of ZPl(s)Z_{\mathrm{Pl}}(s) to Re⁡(s)>1−1MG\operatorname{Re}(s)>1-\frac{1}{M_G}. Moreover, if s=1−1MG+δs=1-\frac{1}{M_G}+\delta for a positive real number δ\delta near to 00, then

ZPl(s)≪δ−(L+m).Z_{\mathrm{Pl}}(s)\ll\delta^{-(L+m)}.

We consider the proof of Theorem 4.4 (1).

For each c∈Omc\in\mathcal{O}^{m}, denote by

Js(c)=∫Nfs(w˙0n)ψc(n)‾ dn.J_s(c)=\int_N f_s(\dot{w}_0n)\overline{\psi_c(n)}\,dn.

Here, fs∈I(s−1)f_s\in I(s-1) is the normalized spherical function in Proposition 3.4(2).

Lemma 4.5 (Fourier expansion of ZPl(s)Z_{\mathrm{Pl}}(s)). For any ss with Re⁡(s)>1\operatorname{Re}(s)>1, we have

ZPl(s)=∫OmJs(c) dc.Z_{\mathrm{Pl}}(s)=\int_{\mathcal{O}^{m}}J_s(c)\,dc.

Here, the measure on O\mathcal{O} is given with Vol⁡(O)=1\operatorname{Vol}(\mathcal{O})=1.

Proof. By the Fubini theorem, we have

∫OmJs(c) dc=∫Nfs(w˙0n)∫Omψc(n)‾ dc dn=∫Nfs(w˙0n)∏i1O(xαi(n)) dn=∫Dfs(w˙0n) dn.\int_{\mathcal{O}^{m}}J_s(c)\,dc=\int_N f_s(\dot{w}_0n)\int_{\mathcal{O}^{m}}\overline{\psi_c(n)}\,dc\,dn=\int_N f_s(\dot{w}_0n)\prod_i1_{\mathcal{O}}(x_{\alpha_i}(n))\,dn=\int_D f_s(\dot{w}_0n)\,dn.

The Lemma now follows from Proposition 3.4(2).

Theorem 4.6 (Casselman-Shalika). Let c=(ci)i∈Omc=(c_i)_i\in\mathcal{O}^{m} with ci=uiϖzic_i=u_i\varpi^{z_i}, ui∈O×u_i\in\mathcal{O}^{\times} and zi≥0z_i\geq0. Denote by z=(zi)i∈Z≥0mz=(z_i)_i\in\mathbb{Z}_{\geq0}^{m}. Then

Js(c)=E(q1−s)Pz(q1−s).J_s(c)=\mathcal{E}(q^{1-s})P_z(q^{1-s}).

Proof. This is in fact the Casselman-Shalika formula for spherical Whittaker functions. Let Ws(g)W_s(g) be the unique spherical Whittaker function of I(s−1)I(s-1)

Ws(g)=∫Nfs(w˙0ng)ψ0(n)‾ dn.W_s(g)=\int_N f_s(\dot{w}_0ng)\overline{\psi_0(n)}\,dn.

Then the Casselman-Shalika formula can be stated as following ([13 Theorem 6.5.1])

Ws(tz)=δ1/2(tz)E(q1−s)tr⁡πz(As),z∈Z≥0m.W_s(t_z)=\delta^{1/2}(t_z)\mathcal{E}(q^{1-s})\operatorname{tr}_{\pi_z}(A_s),\qquad z\in\mathbb{Z}_{\geq0}^{m}.

Here,

  • for each z∈Z≥0mz\in\mathbb{Z}_{\geq0}^{m}, tz=μz(ϖ)∈Aadt_z=\mu_z(\varpi)\in A_{\mathrm{ad}} and tr⁡πz\operatorname{tr}_{\pi_z} is the trace for πz\pi_z.

  • the Satake parameter As∈A^ad(C)A_s \in\widehat{A}_{\mathrm{ad}}(\mathbb{C}) for I(s−1)I(s-1) is characterized by

χ(As)=q−(s−1)⟨ρ,χ⟩,χ∈X∗(A^ad)=X∗(Aad).\chi(A_s)=q^{-(s-1)\langle\rho,\chi\rangle},\qquad\chi\in X^*(\widehat{A}_{\mathrm{ad}})=X^*(A_{\mathrm{ad}}).

We have

tr⁡πz(As)=∑χm(πz,χ)q(1−s)⟨ρ,χ⟩,δ1/2(tz)=q−⟨ρ,μz⟩.\operatorname{tr}_{\pi_z}(A_s)=\sum_{\chi}m(\pi_z,\chi)q^{(1-s)\langle\rho,\chi\rangle},\qquad\delta^{1/2}(t_z)=q^{-\langle\rho,\mu_z\rangle}.

This gives

Ws(tz)=δ1−s/2(tz)E(q1−s)Pz(q1−s).W_s(t_z)=\delta^{1-s/2}(t_z)\mathcal{E}(q^{1-s})P_z(q^{1-s}).

Now we fix c=(ci)c=(c_i) with ci=uiϖzic_i=u_i\varpi^{z_i}, ui∈O×u_i\in\mathcal{O}^{\times} and zi≥0z_i\geq0. Consider z=(zi)i∈Z≥0mz=(z_i)_i\in\mathbb{Z}_{\geq0}^m. Then, we have

αi(tz)=ϖzi,1≤i≤m.\alpha_i(t_z)=\varpi^{z_i},\qquad1\leq i\leq m.

Moreover, as X∗(Aad)X^*(A_{\mathrm{ad}}) is spanned by simple roots αi\alpha_i, 1≤i≤m1\leq i\leq m, there exists tu∈Aad(O)t_u\in A_{\mathrm{ad}}(\mathcal{O}) such that

αi(tu)=ui,1≤i≤m.\alpha_i(t_u)=u_i,\qquad1\leq i\leq m.

Therefore,

Js(c)=∫Nfs(w˙0n)ψc(n)‾ dn=∫Nfs(w˙0n)ψ0(Ad⁡(tutz)n)‾ dn=δ−1(tz)∫Nfs(w˙0Ad⁡(tutz)−1n)ψ0(n)‾ dn=δs/2−1(tz)∫Nfs(w˙0ntz)ψ0(n)‾ dn=δs/2−1(tz)Ws(tz).\begin{aligned} J_s(c)&=\int_N f_s(\dot{w}_0n)\overline{\psi_c(n)}\,dn\\ &=\int_N f_s(\dot{w}_0n)\overline{\psi_0(\operatorname{Ad}(t_ut_z)n)}\,dn\\ &=\delta^{-1}(t_z)\int_N f_s\left(\dot{w}_0\operatorname{Ad}(t_ut_z)^{-1}n\right)\overline{\psi_0(n)}\,dn\\ &=\delta^{s/2-1}(t_z)\int_N f_s(\dot{w}_0nt_z)\overline{\psi_0(n)}\,dn\\ &=\delta^{s/2-1}(t_z)W_s(t_z). \end{aligned}

Now, the result follows from the above Casselman-Shalika formula for Ws(tz)W_s(t_z).

By Lemma 4.5 and Theorem 4.6 with Z=∑iziZ=\sum_i z_i, we have

ZPl(s)=∑z=(zi)i∈Z≥0mq−Z∫(O×)mJs((uiϖzi)i)∏i dui=(1−q−1)m∑z=(zi)i∈Z≥0mq−ZJs((ϖzi)i)=(1−q−1)mE(q1−s)∑z∈Z≥0mq−ZPz(q1−s).\begin{aligned} Z_{\mathrm{Pl}}(s) &=\sum_{z=(z_i)_i\in\mathbb{Z}_{\geq0}^m}q^{-Z}\int_{(\mathcal{O}^{\times})^m}J_s\left((u_i\varpi^{z_i})_i\right)\prod_i\,du_i\\ &=(1-q^{-1})^m\sum_{z=(z_i)_i\in\mathbb{Z}_{\geq0}^m}q^{-Z}J_s\left((\varpi^{z_i})_i\right)\\ &=(1-q^{-1})^m\mathcal{E}(q^{1-s})\sum_{z\in\mathbb{Z}_{\geq0}^m}q^{-Z}P_z(q^{1-s}). \end{aligned}

We obtain Theorem 4.4 (1).

Now, we consider the proof of Theorem 4.4(2).

Lemma 4.7. Let z∈Z≥0mz\in\mathbb{Z}_{\geq0}^m and consider the representation πz\pi_z on G^ad\widehat{G}_{\mathrm{ad}}. Then

Pz(1)=dim⁡πz≤(1+Z)LP_z(1)=\dim\pi_z\leq(1+Z)^L

with Z=∑iziZ=\sum_i z_i and L=∣Φ+∣L=|\Phi^+|.

Proof. By the Weyl dimension formula [10 Corollary 24.6], we have

dim⁡πz=∏γ∈Φ+⟨μz+ρ^,γ⟩⟨ρ^,γ⟩\dim\pi_z=\prod_{\gamma\in\Phi^+}\frac{\langle\mu_z+\widehat{\rho},\gamma\rangle}{\langle\widehat{\rho},\gamma\rangle}

where ρ^=∑iϖi∨∈X∗(Aad)\widehat{\rho}=\sum_i\varpi_i^\vee\in X_*(A_{\mathrm{ad}}) and γ∈Φ+\gamma\in\Phi^+ runs over positive coroots of G^ad\widehat{G}_{\mathrm{ad}}. If γ=∑ikiαi\gamma=\sum_i k_i\alpha_i, then

⟨μz+ρ^,γ⟩⟨ρ^,γ⟩=1+∑ikizi∑iki≤1+Z.\frac{\langle\mu_z+\widehat{\rho},\gamma\rangle}{\langle\widehat{\rho},\gamma\rangle} =1+\frac{\sum_i k_i z_i}{\sum_i k_i}\leq1+Z.

The Lemma now holds.

Write u=q1−su=q^{1-s}. Consider

1−1MG<Re⁡(s)≤1,1≤∣u∣<q1/MG.1-\frac{1}{M_G}<\operatorname{Re}(s)\leq1,\qquad1\leq|u|<q^{1/M_G}.

For each z∈Z≥0mz\in\mathbb{Z}_{\geq0}^{m}, By Lemma 4.3, we can write

Pz(u)=∑k=0dzckuk,ck≥0P_z(u)=\sum_{k=0}^{d_z}c_k u^k,\qquad c_k\geq0

where dz=⟨2ρ,μz⟩=∑iaizid_z=\langle2\rho,\mu_z\rangle=\sum_i a_i z_i.

As ∣u∣≥1|u|\geq1, we have

∣Pz(u)∣≤∑k=0dzck∣u∣k≤∣u∣dz∑kck=∣u∣dzPz(1)≤∣u∣dz(1+Z)L|P_z(u)|\leq\sum_{k=0}^{d_z}c_k|u|^k\leq|u|^{d_z}\sum_k c_k=|u|^{d_z}P_z(1)\leq|u|^{d_z}(1+Z)^L

and

q−Z∣Pz(u)∣≤q−Z∣u∣dz(1+Z)L=∏i(q−1∣u∣ai)zi⋅(1+Z)L.(3)q^{-Z}|P_z(u)|\leq q^{-Z}|u|^{d_z}(1+Z)^L=\prod_i(q^{-1}|u|^{a_i})^{z_i}\cdot(1+Z)^L. \tag*{(3)}

Moreover, as ∣u∣<q1/MG|u|<q^{1/M_G}, there exists a real number x<1x<1 (depending on ∣u∣|u|) such that q−1∣u∣ai<xq^{-1}|u|^{a_i}<x for all ii. Hence we have

∑z∈Z≥0mq−Z∣Pz(u)∣≤∑z∈Z≥0m(1+Z)LxZ<∞.\sum_{z\in\mathbb{Z}_{\geq0}^{m}}q^{-Z}|P_z(u)|\leq\sum_{z\in\mathbb{Z}_{\geq0}^{m}}(1+Z)^Lx^Z<\infty.

In particular, the right hand side of Equation (4.1) is absolutely convergent for any ss with Re⁡(s)>1−1MG\operatorname{Re}(s)>1-\frac{1}{M_G}. Combining with Theorem 4.4(1), this gives the holomorphic continuation of ZPl(s)Z_{\mathrm{Pl}}(s) to Re⁡(s)>1−1MG\operatorname{Re}(s)>1-\frac{1}{M_G}.

We estimate ZPl(s)Z_{\mathrm{Pl}}(s) near its convergence boundary. Consider

s=1−1MG+δ<1s=1-\frac{1}{M_G}+\delta<1

where δ\delta is a positive real number and near to 0.

By (4.1)

ZPl(s)=(1−q−1)mE(q1−s)∑z∈Z≥0mq−ZPz(q1−s),E(q1−s)=∏γ∈Φ+(1−q−1q(1−s)⟨ρ,γ∨⟩).Z_{\mathrm{Pl}}(s)=(1-q^{-1})^m\mathcal{E}(q^{1-s})\sum_{z\in\mathbb{Z}_{\geq0}^{m}}q^{-Z}P_z(q^{1-s}),\qquad\mathcal{E}(q^{1-s})=\prod_{\gamma\in\Phi^+}(1-q^{-1}q^{(1-s)\langle\rho,\gamma^\vee\rangle}).

As 1−s=1MG−δ<1MG1-s=\frac{1}{M_G}-\delta<\frac{1}{M_G}, E(q1−s)\mathcal{E}(q^{1-s}) is bounded by a constant (depending only on FF and GG). By (3), we have

ZPl(s)≪∑z∈Z≥0m(1+Z)Lq−Z∑i[1−(1−s)ai]zi.Z_{\mathrm{Pl}}(s)\ll\sum_{z\in\mathbb{Z}_{\geq0}^{m}}(1+Z)^Lq^{-Z}\sum_i[1-(1-s)a_i]z_i.

As ai≥1a_i\geq1 for all ii, we have

1−(1−s)ai=1−aiMG+δai≥δ1-(1-s)a_i=1-\frac{a_i}{M_G}+\delta a_i\geq\delta

and

ZPl(s)≪∑z∈Z≥0m(1+Z)Lq−δZ=∑Z≥0(Z+m−1m−1)(1+Z)Lq−δZ.Z_{\mathrm{Pl}}(s)\ll\sum_{z\in\mathbb{Z}_{\geq0}^{m}}(1+Z)^Lq^{-\delta Z} =\sum_{Z\geq0}\binom{Z+m-1}{m-1}(1+Z)^Lq^{-\delta Z}.

As

(Z+m−1m−1)≤mm−1(m−1)!(1+Z)m−1,\binom{Z+m-1}{m-1}\leq\frac{m^{m-1}}{(m-1)!}(1+Z)^{m-1},

we obtain

ZPl(s)≪∑Z≥0(1+Z)L+m−1q−δZ≪(1−q−δ)−(L+m)≪δ−(L+m).Z_{\mathrm{Pl}}(s)\ll\sum_{Z\geq0}(1+Z)^{L+m-1}q^{-\delta Z}\ll(1-q^{-\delta})^{-(L+m)}\ll\delta^{-(L+m)}.

In the above, we use

∑n=0∞(n+1)ktn≪k(1−t)−(k+1)\sum_{n=0}^{\infty}(n+1)^k t^n\ll_k(1-t)^{-(k+1)}

and 1−q−δ∼(log⁡q)δ1-q^{-\delta}\sim(\log q)\delta when δ→0+\delta\to0^+. We complete the proof of Theorem 4.4 (2).

Applying the bounding of the zeta function ZPl(s)Z_{\mathrm{Pl}}(s) in Theorem 4.4 (2), we give a bound for Plücker-constrained sets.

Theorem 4.8. For any r∈Z≥0r \in\mathbb{Z}_{\geq0},

∣Sr∣≪q(1−12MG)R,R=∑iri.|S_r| \ll q^{\left(1-\frac{1}{2M_G}\right)R}, \qquad R=\sum_i r_i.

Proof. By Proposition 3.8, we have

∣Sr∣≤∑ℓ=0Rbℓ.|S_r| \leq\sum_{\ell=0}^{R} b_\ell.

Since bℓ≥0b_\ell\geq0, Landau’s theorem on Dirichlet series with nonnegative coefficients [21 Theorem 1], together with Lemma 4.2 and Theorem 4.4, implies that

ZPl(s)=∑ℓ≥0bℓq−ℓsZ_{\mathrm{Pl}}(s)=\sum_{\ell\geq0}b_\ell q^{-\ell s}

with absolute convergence for Re⁡(s)>1−1/MG\operatorname{Re}(s)>1-1/M_G. For any real s>0s>0, since each ℓ≤R\ell\leq R satisfies 1≤qs(R−ℓ)1\leq q^{s(R-\ell)}, we have

∣Sr∣≤qsR∑ℓ=0Rbℓq−ℓs≤qsRZPl(s),s>1−1MG.|S_r|\leq q^{sR}\sum_{\ell=0}^{R}b_\ell q^{-\ell s}\leq q^{sR}Z_{\mathrm{Pl}}(s),\qquad s>1-\frac{1}{M_G}.

By Theorem 4.4 (2), for s=1−1MG+δs=1-\frac{1}{M_G}+\delta with the number δ\delta positive and near to 00, we have

ZPl(s)≪δ−(L+m)Z_{\mathrm{Pl}}(s)\ll\delta^{-(L+m)}

and

∣Sr∣≪q(1−1MG)RqδRδ−(L+m).|S_r|\ll q^{\left(1-\frac{1}{M_G}\right)R}q^{\delta R}\delta^{-(L+m)}.

We may choose δ=12MG(1+R)\delta=\frac{1}{2M_G(1+R)} (so that s<1s<1). Then

∣Sr∣≪(1+R)L+mq(1−1MG)R.|S_r|\ll(1+R)^{L+m}q^{\left(1-\frac{1}{M_G}\right)R}.

For each ε>0\varepsilon>0, (1+R)L+m≤CεqεR(1+R)^{L+m}\leq C_\varepsilon q^{\varepsilon R} for some constant CεC_\varepsilon. Consider ε=1/(2MG)\varepsilon=1/(2M_G). Then

∣Sr∣≪q(1−12MG)R.|S_r|\ll q^{\left(1-\frac{1}{2M_G}\right)R}.

The proof of Theorem 1.1

In this section, we firstly still assume that GG is semisimple and simply connected. Let {Kd}d\{K_d\}_d be the family of principal congruence subgroups of GG, that is, the kernel of the reduction map G(O)→G(O/pd)G(\mathcal{O})\to G(\mathcal{O}/\mathfrak{p}^d). Let {Kl⁡d}d\{\operatorname{Kl}_d\}_d be the corresponding Kloosterman sums, that is,

Kl⁡d(t)=Kl⁡(t;ψ0,w˙0,Kd),t∈A.\operatorname{Kl}_d(t)=\operatorname{Kl}(t;\psi_0,\dot{w}_0,K_d),\qquad t\in A.

We also use Xd(w˙0t)X_d(\dot{w}_0t) to denote the Kloosterman set in this case.

For t∈At\in A, write t=t0λt(ϖ)t=t_0\lambda_t(\varpi) with t0∈A(O)t_0\in A(\mathcal{O}) and λt∈X∗(A)\lambda_t\in X_*(A). By Dąbrowski–Reeder [9 Theorem 0.3], Xd(w˙0t)X_d(\dot{w}_0t) is nonempty only if λt∈Z≥0Φ∨,+\lambda_t\in\mathbb{Z}_{\geq0}\Phi^{\vee,+}. It therefore suffices to consider

t=t0∏i=1mαi∨(ϖri),t0∈A(O),ri∈Z≥0.t=t_0\prod_{i=1}^{m}\alpha_i^\vee(\varpi^{r_i}),\qquad t_0\in A(\mathcal{O}),\quad r_i\in\mathbb{Z}_{\geq0}.

Theorem 5.1 (Theorem 1.3 in [CLYY26]). Let GG be a split semisimple and simply connected group. Assume there exist constants ε>0\varepsilon>0 and C>0C>0 such that for any r∈Z≥0mr\in\mathbb{Z}_{\geq0}^{m}

∣Sr∣≤Cq(1−ε)R,R=∑iri.|S_r|\leq Cq^{(1-\varepsilon)R},\qquad R=\sum_i r_i.

Then for all sufficiently large dd, the Kloosterman sum Kl⁡d\operatorname{Kl}_d has a nontrivial bound.

In fact, there is the condition that p∤det⁡(CG)p\nmid\det(C_G) with CGC_G the Cartan matrix of GG in [CLYY26, Theorem 1.3]. This condition is from the decomposition of the Kloosterman sum on GG to products of Kloosterman sums on GL2\mathrm{GL}_2 in [4 Proposition 5.3]. We remove this condition in the Appendix (Theorem A.6).

Corollary 5.2. Let GG be a split semisimple and simply connected group. Then for all sufficiently large dd, the Kloosterman sum Kl⁡d\operatorname{Kl}_d has a nontrivial bound

Proof. This is an immediate Corollary from Theorem 4.8 and Theorem 5.1. □\square

Now, we let GG be a general split reductive group over FF. As before, we fix a Borel subgroup B=ANB=AN of GG and ψ=ψ0\psi=\psi_0 is the generic character on NN given before [4 Definition 2.1]. Denote by {Gi}1≤i≤s\{G_i\}_{1\leq i\leq s} the set of minimal nontrivial normal smooth connected subgroups of GderG^{\mathrm{der}}. By [14 Theorem in Section 27.5], the groups GiG_i are simple and the multiplication homomorphism

G1×⋯×Gs⟶GderG_1\times\cdots\times G_s\longrightarrow G^{\mathrm{der}}

is an isogeny map. These groups GiG_i are called simple factors of GG. We have the following isogeny map

G~1×⋯×G~s⟶Gder\widetilde{G}_1\times\cdots\times\widetilde{G}_s\longrightarrow G^{\mathrm{der}}

where G~i\widetilde{G}_i, 1≤i≤s1\leq i\leq s is the universal cover of the simple factor GiG_i of GderG^{\mathrm{der}}. For each 1≤i≤s1\leq i\leq s, the group G~i\widetilde{G}_i is simple, simply connected group.

An open compact subgroup K⊂GK\subset G with ψ0∣N∩K=1\left.\psi_0\right|_{N\cap K}=1 is called admissible (with aspect to ψ0\psi_0) if for any z∈Zderz\in Z^{\mathrm{der}} and n∈Nn\in N with zn∈Kzn\in K, we have z=1z=1 and n∈Kn\in K, where ZderZ^{\mathrm{der}} is the center of GderG^{\mathrm{der}} ([4 Definition 2.25]).

Proposition 5.3. Let GG be a split reductive group. Then there exists an admissible group KGK_G such that the Kloosterman sum Kl⁡(⋅,ψ0−1,w˙0,KG)\operatorname{Kl}(\mathord\cdot,\psi_0^{-1},\dot{w}_0,K_G) has a nontrivial bound.

Proof. By Corollary 5.2, for every 1≤i≤s1\leq i\leq s, there exists ϵi>0\epsilon_i>0, 1≤i≤s1\leq i\leq s such that the Kloosterman sum Kl⁡G~i(ti,ψ0−1∣Ni,w˙0,i,KG~i)\operatorname{Kl}_{\widetilde{G}_i}(t_i,\left.\psi_0^{-1}\right|_{N_i},\dot{w}_{0,i},K_{\widetilde{G}_i}) has a nontrivial bound. By Lemma [4 Lemma 3.23 (1)], the product of KG~iK_{\widetilde{G}_i} gives an admissible group KGderK_{G^{\mathrm{der}}} of GderG^{\mathrm{der}}. By Lemma [4 Lemma 3.20] and Lemma [4 Lemma 3.23], for a representative w˙0\dot{w}_0 of the longest Weyl element in WW, the Kloosterman sums Kl⁡Gder(⋅;ψ0−1,w˙0,KGder)\operatorname{Kl}_{G^{\mathrm{der}}}(\mathord\cdot;\psi_0^{-1},\dot{w}_0,K_{G^{\mathrm{der}}}) has a nontrivial bound. Here, we use the fact that the modular characters are compatible under isogeny maps. By Lemma [4 Lemma 3.22], there exists an admissible group KGK_G of GG and a representative w˙0\dot{w}_0 of the longest Weyl element in WW such that the Kloosterman sums Kl⁡G(⋅;ψ0−1,w˙0,KG)\operatorname{Kl}_G(\mathord\cdot;\psi_0^{-1},\dot{w}_0,K_G) has a nontrivial bound. By [4 Lemma 3.21 (2)], the above bound holds for any representative of longest Weyl element w0∈Ww_0\in W. □

The proof of Theorem 1.1. Now let K⊂GK\subset G be a compact open subgroup such that ψ0∣N∩K=1\left.\psi_0\right|_{N\cap K}=1 and w0w^0 is the relevant representative of w0w_0 considered [4]. Consider the Kloosterman integral I(t,1w0K)I(t,1_{w^0K}). By [4 Lemma 3.19], we have

I(t,1w0K)≐Kl⁡((w0)−2t;ψ0−1,w0,K),t∈A.I(t,1_{w^0K})\doteq\operatorname{Kl}((w^0)^{-2}t;\psi_0^{-1},w^0,K),\quad t\in A.

Apply [4 Theorem 2.12], we have the following germ expansion of Kloosterman integrals

I(t,1w0K)=∑J⊂Δ(K∅J∗ωJ)(t),t∈A.I(t,1_{w^0K})=\sum_{J\subset\Delta}(K_\varnothing^J\ast\omega_J)(t),\quad t\in A.

where ωJ∈S(AJ)\omega_J\in\mathcal{S}(A_J) for every J⊂ΔJ\subset\Delta and {K∅J}J⊂J\{K_\varnothing^J\}_{J\subset J} is a system of Shalika germs. By [4 Proposition 2.29], for each JJ, we can choose the germ function K∅JK_\varnothing^J such that

K∅J(t)≐Kl⁡MJ((wJ0)−2t,ψ0−1∣NMJ,wJ0,KMJ),t∈A∅J.K_\varnothing^J(t)\doteq\operatorname{Kl}_{M_J}((w_J^0)^{-2}t,\left.\psi_0^{-1}\right|_{N_{M_J}},w_J^0,K_{M_J}),\quad t\in A_\varnothing^J.

where KMJK_{M_J} is the admissible group in Proposition 5.3. In particular, K∅JK_\varnothing^J is supported on A∅J(rJ)A_\varnothing^J(r_J) for some rJ>0r_J>0.

By the above germ expansion and [4 Proposition 3.16], there exists constant CC such that

∣I(t,1w0K)δ12−ϵ(t)∣≤C,t∈A\left|I(t,1_{w^0K})\delta^{\frac{1}{2}-\epsilon}(t)\right|\leq C,\quad t\in A

holds provided that for every J⊂ΔJ\subset\Delta, for any r>0r>0 small enough, there exist some constants ϵ>0\epsilon>0 and CJ>0C_J>0, such that

∣K∅J(t)δBMJ12−ϵ(t)∣≤CJ,t∈A∅J(r).\left|K_\varnothing^J(t)\delta_{B_{M_J}}^{\frac{1}{2}-\epsilon}(t)\right|\leq C_J,\quad t\in A_\varnothing^J(r).

By Proposition 5.3, such a nontrivial bound holds for K∅J(⋅)=Kl⁡MJ((wJ0)−2⋅,ψ0−1∣NMJ,wJ0,KMJ)K_\varnothing^J(\mathord\cdot)=\operatorname{Kl}_{M_J}((w_J^0)^{-2}\mathord\cdot,\left.\psi_0^{-1}\right|_{N_{M_J}},w_J^0,K_{M_J}). The proof is done. □

Generalized Stevens' trick

We follow the section 3.2 in [CLYY26]. We still consider

t=t0λt(ϖ),t0∈A(O),λt=∑i=1mriαi∨,ri∈Z≥0.t=t_0\lambda_t(\varpi),\qquad t_0\in A(\mathcal{O}),\qquad\lambda_t=\sum_{i=1}^{m}r_i\alpha_i^\vee,\qquad r_i\in\mathbb{Z}_{\geq0}.

Using the fixed ordering and root subgroup parameterizations, write every n∈Nn\in N uniquely as

n(x)=∏γ∈Φ+Xγ(xγ),x=(xγ)γ∈Φ+∈FΦ+.n(\mathbf{x})=\prod_{\gamma\in\Phi^+}X_\gamma(x_\gamma),\qquad\mathbf{x}=(x_\gamma)_{\gamma\in\Phi^+}\in F^{\Phi^+}.

For n∈Nn\in N, write nγn_\gamma for its γ\gamma-coordinate. Recall that Nd=∏γ∈Φ+Xγ(pd)N_d=\prod_{\gamma\in\Phi^+}X_\gamma(\mathfrak{p}^d) for d≥0d\geq0. For a positive integer ℓ\ell, put

N−ℓ:=∏γ∈Φ+Xγ(p−ℓht⁡(γ)),N_{-\ell}:=\prod_{\gamma\in\Phi^+}X_\gamma(\mathfrak{p}^{-\ell\operatorname{ht}(\gamma)}),

where ht⁡(γ)=∑α∈Δκα(γ)\operatorname{ht}(\gamma)=\sum_{\alpha\in\Delta}\kappa_\alpha(\gamma) for γ=∑α∈Δκα(γ)α\gamma=\sum_{\alpha\in\Delta}\kappa_\alpha(\gamma)\alpha.

Put Ad=A∩KdA_d=A\cap K_d, Nd=N∩KdN_d=N\cap K_d, and for d≥1d\geq1 we have

Ad={t∈A:χ(t)∈1+pd for all χ∈X∗(A)}.A_d=\{t\in A:\chi(t)\in1+\mathfrak{p}^d\text{ for all }\chi\in X^*(A)\}.

Moreover, ⋂d≥1Kd={1}\bigcap_{d\geq1}K_d=\{1\} and Nd=∏γ∈Φ+Xγ(pd)N_d=\prod_{\gamma\in\Phi^+}X_\gamma(\mathfrak{p}^d) for every d≥0d\geq0.

Choose ℓ>d\ell>d such that for every x∈Xd(w˙0t)x\in X_d(\dot{w}_0t) we have u(x)∈Nd\N−ℓu(x)\in N_d\backslash N_{-\ell} and u′(x)∈N−ℓ/Ndu'(x)\in N_{-\ell}/N_d. For x=[n1w˙0tn2]∈Xd(w˙0t)x=[n_1\dot{w}_0tn_2]\in X_d(\dot{w}_0t), let

θi(x)=(n1)αi∈p−ℓ/pd,θi′(x)=(n2)αi∈p−ℓ/pd.\theta_i(x)=(n_1)_{\alpha_i}\in\mathfrak{p}^{-\ell}/\mathfrak{p}^d,\qquad\theta_i'(x)=(n_2)_{\alpha_i}\in\mathfrak{p}^{-\ell}/\mathfrak{p}^d.

Write θ′=(θi′)1≤i≤m:Xd(w˙0t)→(p−ℓ/pd)m\theta'=(\theta_i')_{1\leq i\leq m}:X_d(\dot{w}_0t)\to(\mathfrak{p}^{-\ell}/\mathfrak{p}^d)^m and set τ(x)=min⁡{v(x),d}\tau(x)=\min\{v(x),d\}. For y∈[−ℓ,d]my\in[-\ell,d]^m, let Xd(w˙0t)yX_d(\dot{w}_0t)_y be the fibre over yy under the composition

Xd(w˙0t)→θ′(p−ℓ/pd)m→τ[−ℓ,d]m.X_d(\dot{w}_0t)\xrightarrow{\theta'}(\mathfrak{p}^{-\ell}/\mathfrak{p}^d)^m\xrightarrow{\tau}[-\ell,d]^m.

Theorem A.1. Let GG be a split semisimple and simply connected group. Assume there exist constants ε>0\varepsilon>0 and C>0C>0 such that

∣Sr∣≤Cq(1−ε)∑i=1mri,r∈Z≥0m.|S_r|\leq Cq^{(1-\varepsilon)\sum_{i=1}^m r_i},\qquad r\in\mathbb{Z}_{\geq0}^m.

Then for all sufficiently large dd, the Kloosterman sum Kl⁡d\operatorname{Kl}_d has a nontrivial bound.

Following the proof of Theorem 5.1 in [CLYY26], we reduce the above theorem to the following proposition.

Proposition A.2. For d>0d>0 sufficiently large, there exists an integer ℓ>0\ell>0, depending on dd and tt, such that

∣Kl⁡d(t)∣≤Cd∑y∈[−ℓ,d]m∣Xd(w˙0t)y∣q12∑i=1myi,|\operatorname{Kl}_d(t)|\leq C_d\sum_{y\in[-\ell,d]^m}|X_d(\dot{w}_0t)_y|q^{\frac{1}{2}\sum_{i=1}^m y_i},

where CdC_d is a constant depend only on $F,G,d.

Assume throughout that dd is large enough so that the pp-adic exponential and logarithm give isomorphisms between pr\mathfrak{p}^r and 1+pr1+\mathfrak{p}^r for all r≥dr\geq d. Let

f:X∗(A)→Zm,f(λ)=(⟨α1,λ⟩,…,⟨αm,λ⟩).f:X_*(A)\to\mathbb{Z}^m,\qquad f(\lambda)=(\langle\alpha_1,\lambda\rangle,\ldots,\langle\alpha_m,\lambda\rangle).

The size of the finite abelian group coker⁡(f)\operatorname{coker}(f) is the determinant of the Cartan matrix cC:=det⁡(CG)c_C:=\det(C_G) of GG, and set

EC:=v(cC),D:=d+EC.E_C:=v(c_C),\qquad D:=d+E_C.

For r≥1r\geq1, put

Ur,ℓ:=1+pr1+pℓ,Ur,ℓm:=(Ur,ℓ)×m.U_{r,\ell}:=\frac{1+\mathfrak{p}^r}{1+\mathfrak{p}^\ell},\qquad U_{r,\ell}^m:=(U_{r,\ell})^{\times m}.

The image of the simple-root map

ϕd:Ad⟶(1+pd)m,t↦(α1(t),…,αm(t))\phi_d:A_d\longrightarrow(1+\mathfrak{p}^d)^m,\qquad t\mapsto(\alpha_1(t),\ldots,\alpha_m(t))

is

Hd,ℓ:=im⁡(Ad→Ud,ℓm).H_{d,\ell}:=\operatorname{im}(A_d\to U_{d,\ell}^m).

For each x∈Xd(w˙0t)x \in X_d(\dot{w}_0t), define the constrained Kloosterman sum attached to xx by

SH(x):=∑λ=(λ1,…,λm)∈Hd,ℓψF(∑i=1m(θi(x)λi+θσ(i)′(x)λi−1)).S_H(x) := \sum_{\lambda=(\lambda_1,\ldots,\lambda_m)\in H_{d,\ell}} \psi_F\left(\sum_{i=1}^{m}\left(\theta_i(x)\lambda_i+\theta'_{\sigma(i)}(x)\lambda_i^{-1}\right)\right).

where σ\sigma is the permutation of {1,…,m}\{1,\ldots,m\} defined by −w0αi=ασ(i)-w_0\alpha_i=\alpha_{\sigma(i)}.

Proposition A.3. We have,

Kl⁡d(t)=1∣Hd,ℓ∣∑x∈Xd(w˙0t)SH(x).\operatorname{Kl}_d(t)=\frac{1}{|H_{d,\ell}|}\sum_{x\in X_d(\dot{w}_0t)}S_H(x).

Proof. For fixed xx, the function t↦ψ(u(t⋅x))ψ(u′(t⋅x))t\mapsto\psi(u(t\cdot x))\psi(u'(t\cdot x)) is constant on cosets of Ad,ℓ:=ker⁡(Ad→Hd,ℓ)A_{d,\ell}:=\ker(A_d\to H_{d,\ell}). Therefore the orbit integral over AdA_d can be computed by summing over Hd,ℓH_{d,\ell}; the rest of the proof is identical to that of [4 Proposition 5.3]. The details are omitted. □

We now prove that Hd,ℓH_{d,\ell} always contains a product subgroup of depth DD.

Proposition A.4. If ℓ>D\ell>D, then UD,ℓm⊆Hd,ℓU^m_{D,\ell}\subseteq H_{d,\ell}.

Proof. Since cCZm⊆im⁡(f)c_C\mathbb{Z}^m\subseteq\operatorname{im}(f), tensoring with OF\mathcal O_F gives

(pcC)m⊆im⁡(f⊗ZOF).(\mathfrak{p}^{c_C})^m\subseteq\operatorname{im}(f\otimes_{\mathbb Z}\mathcal O_F).

Multiplying by pd\mathfrak{p}^d yields

(pD)m⊆(f⊗id⁡)(X∗(A)⊗pd).(\mathfrak{p}^D)^m\subseteq(f\otimes\operatorname{id})(X_*(A)\otimes\mathfrak{p}^d).

Composing with the pp-adic exponential map, we have UD,ℓm⊆Hd,ℓU^m_{D,\ell}\subseteq H_{d,\ell}. □

Let V:=UD,ℓmV:=U^m_{D,\ell} and choose a set of representatives R\mathcal R for the finite quotient Hd,ℓ/VH_{d,\ell}/V. Then

Hd,ℓ=⨆η∈RηV.H_{d,\ell}=\bigsqcup_{\eta\in\mathcal R}\eta V.

For integers 0<r<ℓ0<r<\ell and a,b∈Oa,b\in\mathcal O, define

Kl⁡(a,b;r,ℓ):=∑λ∈(1+pr)/(1+pℓ)ψF(aλ+bλ−1ϖℓ).\operatorname{Kl}(a,b;r,\ell):=\sum_{\lambda\in(1+\mathfrak{p}^r)/(1+\mathfrak{p}^{\ell})}\psi_F\left(\frac{a\lambda+b\lambda^{-1}}{\varpi^\ell}\right).

Proposition A.5. For every x∈Xd(w˙0t)x\in X_d(\dot{w}_0t),

SH(x)=∑η∈R∏i=1mKl⁡(ϖℓθi(x)ηi,ϖℓθσ(i)′(x)ηi−1;D,ℓ).S_H(x)=\sum_{\eta\in\mathcal R}\prod_{i=1}^{m}\operatorname{Kl}\left(\varpi^\ell\theta_i(x)\eta_i,\varpi^\ell\theta'_{\sigma(i)}(x)\eta_i^{-1};D,\ell\right).

Proof. This follows by substituting λi=ηiμi\lambda_i=\eta_i\mu_i with μi∈UD,ℓm\mu_i\in U^m_{D,\ell} into the definition of SH(x)S_H(x). □

Combining Propositions A.3 and A.5 yields the generalized Stevens’ trick.

Theorem A.6 (Generalized Stevens’ trick). With D=d+v(cC)D=d+v(c_C) and ℓ>D\ell>D,

Kl⁡d(t)=1∣Hd,ℓ∣∑x∈Xd(w˙0t)∑η∈R∏i=1mKl⁡(ϖℓθi(x)ηi,ϖℓθσ(i)′(x)ηi−1;D,ℓ).\operatorname{Kl}_d(t)=\frac{1}{|H_{d,\ell}|}\sum_{x\in X_d(\dot{w}_0t)}\sum_{\eta\in\mathcal R}\prod_{i=1}^{m}\operatorname{Kl}\left(\varpi^\ell\theta_i(x)\eta_i,\varpi^\ell\theta'_{\sigma(i)}(x)\eta_i^{-1};D,\ell\right).

Remark A.7. We record the following estimates for the generalized formula.

  1. Since Hd,ℓH_{d,\ell} contains UD,ℓmU^m_{D,\ell}, we have ∣Hd,ℓ∣≥∣UD,ℓm∣=qm(ℓ−D)|H_{d,\ell}|\geq|U^m_{D,\ell}|=q^{m(\ell-D)}. Hence,

1∣Hd,ℓ∣≤q−m(ℓ−D).\frac{1}{|H_{d,\ell}|}\leq q^{-m(\ell-D)}.
  1. The number of coset representatives satisfies

∣R∣=∣Hd,ℓ∣∣UD,ℓm∣≤qm(D−d)=qm⋅v(cC).|\mathcal R|=\frac{|H_{d,\ell}|}{|U^m_{D,\ell}|}\leq q^{m(D-d)}=q^{m\cdot v(c_C)}.
  1. Each component ηi\eta_i of η∈R⊆Hd,ℓ\eta\in\mathcal R\subseteq H_{d,\ell} belongs to Ud,ℓU_{d,\ell}, hence is invertible.

Fix an integer k≥0k \ge0 such that q−k≤∣2∣2q^{-k} \le|2|^{2}. By [16 Lemmas 3.2–3.4], for all sufficiently large rr and ℓ>r\ell> r, we have

∣Kl⁡(a,b;r,ℓ)∣≤q(min⁡{v(b),ℓ−2r−k}+k+ℓ)/2,a,b∈OF,|\operatorname{Kl}(a,b;r,\ell)| \le q^{\left(\min\{v(b),\ell-2r-k\}+k+\ell\right)/2}, \qquad a,b \in\mathcal{O}_{F},

where v(0)=+∞v(0)=+\infty. Applying this estimate with r=Dr=D and using Remark A.7, we obtain

∣Kl⁡d(t)∣≤q−m(ℓ−D)∑x∈Xd(w˙0t)∑η∈R∏i=1mq(min⁡{v(ϖℓθσ(i)′(x)ηi−1),ℓ−2D−k}+k+ℓ)/2≤qm(D+k/2)∑x∈Xd(w˙0t)∑η∈R∏i=1mqτ(θσ(i)′(x))/2≤qm(D+k/2)qm⋅v(cC)∑x∈Xd(w˙0t)q∑i=1mτ(θσ(i)′(x))/2.\begin{aligned} |\operatorname{Kl}_{d}(t)| &\le q^{-m(\ell-D)}\sum_{x\in X_{d}(\dot{w}_{0}t)}\sum_{\eta\in\mathcal{R}}\prod_{i=1}^{m}q^{\left(\min\{v(\varpi^{\ell}\theta'_{\sigma(i)}(x)\eta_{i}^{-1}),\ell-2D-k\}+k+\ell\right)/2}\\ &\le q^{m(D+k/2)}\sum_{x\in X_{d}(\dot{w}_{0}t)}\sum_{\eta\in\mathcal{R}}\prod_{i=1}^{m}q^{\tau(\theta'_{\sigma(i)}(x))/2}\\ &\le q^{m(D+k/2)}q^{m\cdot v(c_{C})}\sum_{x\in X_{d}(\dot{w}_{0}t)}q^{\sum_{i=1}^{m}\tau(\theta'_{\sigma(i)}(x))/2}. \end{aligned}

Here we used min⁡{v(c),−2D−k}≤τ(c)\min\{v(c),-2D-k\}\le\tau(c) for c∈Fc\in F, and C:=qm(D+v(cC)+k/2)C:=q^{m(D+v(c_{C})+k/2)} depends only on F,G,dF,G,d. Hence,

∣Kl⁡d(t)∣≤C∑y∈[−ℓ,d]m∣Xd(w˙0t)y∣q∑i=1myi/2.|\operatorname{Kl}_{d}(t)|\le C\sum_{y\in[-\ell,d]^{m}}|X_{d}(\dot{w}_{0}t)_{y}|q^{\sum_{i=1}^{m}y_{i}/2}.

References

(L.C.) Academy for Multidisciplinary Studies, Beijing National Center for Applied Mathematics, Capital Normal University, Beijing, 100048, People’s Republic of China Email address: [email protected]

(J.C.) School of Mathematics, Physics and Finance, Anhui Polytechnic University, Wuhu, Anhui, 241000, People’s Republic of China Email address: [email protected]

(Y.L.) Academy for Multidisciplinary Studies, Beijing National Center for Applied Mathematics, Capital Normal University, Beijing, 100048, People’s Republic of China Email address: [email protected]

(M.Y.) School of Mathematics, Nanjing University, Jiangsu, 210093, People’s Republic of China Email address: [email protected]

References

  1. [1][AB24] E. Assing and V. Blomer. The density conjecture for principal congruence subgroups. Duke Math. J. 173(2024), no. 7, 1359-1426.DOI
  2. [2][BFG88] D. Bump, S. Friedberg, D. Goldfeld, Poincaré series and Kloosterman sums for SL(3, ℤ), Acta Arith. 50 (1988), 31–89.DOI
  3. [3][BM24] V. Blomer and S. Man, Bounds for Kloosterman sums on GL(n). Math. Ann. 390 (2024), 1171-1200.
  4. [4][CCL26] L. Cai, J. Chai and Y. Liu, Bessel distributions and Kloosterman sums, arXiv:2606.29852v2 (2026).arxiv.org/abs/2606.29852
  5. [6][CLPS91] J. Cogdell, J.-S. Li, L. Piateski-Shapiro and P. Sarnak. Poincare series for SO(n, 1). Acta Math. 167(1991), 229-285.DOI
  6. [7][SGA3] M. Demazure, A. Grothendieck, Schémas en groupes I, II, III, Lecture Notes in Math 151, 152, 153, Springer-Verlag, New York (1970).
  7. [8][Dab93] R. Dąbrowski, Kloosterman sums for Chevalley groups, Transactions of the American Mathematical Society, 337 (1993), no. 2, 757–769.DOI
  8. [9][DR98] R. Dąbrowski, M. Reeder, Kloosterman sets in reductive groups. Journal of Number Theory 73 (1998), no.2, 228-255.DOI
  9. [10][FH91] W. Fulton and J. Harris, Representation theory: a first course, Graduate Texts in Mathematics, vol. 129, Springer-Verlag, New York, 1991.DOI
  10. [11][Fri87] S. Friedberg, Poincaré Series for GL(n): Fourier Expansion, Kloosterman Sums, and Algebreo- Geometric Estimates., Math. Z. 196 (1987), 165-188.DOI
  11. [12][GSW21] D. Goldfeld, E. Stade, and M. Woodbury, An orthogonality relation for GL(4, ℝ) (with an appendix by Bingrong Huang). arXiv: 1910.13586. Forum of Mathematics, Sigma (2021), 1-83.
  12. [13][HKP10] T. Haines, R. Kottwitz and A. Prasad. Iwahori-Hecke algebras. J. Ramanujan Math.Soc. 25, no.2 (2010), 113-145.
  13. [14][H80] J. E. Humphreys, Linear Algebraic Groups, Graduate Texts in Mathematics, 021, Springer-Verlag, 1980.
  14. [15][Jan03] J. C. Jantzen Representations of algebraic groups. Mathematical Surveys and Monographs, Vol. 107. American Mathematical Society, Providence, RI, second edition.
  15. [16][JY99] H. Jacquet, Y. Ye, Germs of Kloosterman integrals for GL(3). Transactions American Math. Society, 351, no.3, (1999), 1227-1255.DOI
  16. [17][KP23] T. Kaletha, G. Prasad, Bruhat-Tits Theory–A New Approach. New Math. Monogr., 44 (Cambridge University Press, Cambridge, 2023), xxx+718 pp.DOI
  17. [18][Lin26] J. Linn, Bounds for Kloosterman Sums for GLn, Adv. in Math., vol.499, 2026.DOI
  18. [19][Man22] S. Man, Symplectic Kloosterman sums and Poincare series. Ramanujan J. 57 (2022), 707-753.DOI
  19. [20][Man24] S. Man. Fourier coefficients of Sp(4) Eisenstein series. Acta Arith., 213(3):227–271, 2024.
  20. [21][Mau11] B. N. Maurizi, Extending Landau’s theorem on Dirichlet series with non-negative coefficients. Missouri J. Math. Sci. 23 (2011), no. 2, 105–122.arxiv.org/abs/1009.0228
  21. [22][Mia24] Xinchen Miao. Bessel functions and Kloosterman integrals on GL(n). Journal of Functional Analysis. 286 (2024), 110267 University Press, Cambridge, 2017.arxiv.org/abs/2208.01016
  22. [23][Muj24] C. Mujdai. Kloosterman sums on orthogonal groups. arXiv:2410.20569(2024).
  23. [24][Sal32] H. Salie. Über die Kloostermanschen Summen S(u, v : q). Math. Z. 34(1932), 91-109.
  24. [25][Ste87] G. Stevens, Poincaré series on GL(r) and Kloostermann sums. Math. Annalen 277 (1987), 25-51.DOI
  25. [26][Wal03] J.-L. Waldspurger, La formule de Plancherel pour les groupes p-adiques (d’après Harish-Chandra). J. Inst. Math. Jussieu 2 (2003), 235–333.DOI

Paper details

vlnc:W_5yfn7638
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

ChatGPT

View statement
Looking for related papers…
Contents