betaFeedback

WorksMathematics

Maximal Unipotent Monodromy and pp-adic Eigenvalue Formulas

Abstract

We establish p-adic formulas for the eigenvalues of Frobenius in the case of the family of hyperkloosterman sums. The deformation equation in this case has maximal unipotent monodromy, and the interplay of Frobenius and the differential equation plays an important role in establishing p-adic analytic formulas for the eigenvalues.

Introduction

We describe here some of the key properties we use. We consider the hyperkloosterman family of exponential sums. Let Fq\mathbb{F}_q be a finite field of characteristic pp, with q=paq=p^a. Let

Kn(Λ,x):=x1+⋯+xn+Λx1⋯xn∈Z[Λ,x1±1,…,xn±1].K_n(\Lambda,x):=x_1+\cdots+x_n+\frac{\Lambda}{x_1\cdots x_n}\in\mathbb{Z}[\Lambda,x_1^{\pm1},\ldots,x_n^{\pm1}].

Let Θ\Theta be a non-trivial additive character of Fp\mathbb{F}_p and for any positive integer jj let Θj=Θ∘Tr⁡Fqj/Fp\Theta_j=\Theta\circ\operatorname{Tr}_{\mathbb{F}_{q^j}/\mathbb{F}_p} be the additive character of Fqj\mathbb{F}_{q^j} defined by taking the composition of Θ\Theta with the trace map. For λˉ∈Fq\bar{\lambda}\in\mathbb{F}_q, define

Kl⁡n(λˉ/Fqj)=∑x=(x1,…,xn)∈(Fqj×)nΘj(Kn(λˉ,x)).\operatorname{Kl}_n(\bar{\lambda}/\mathbb{F}_{q^j})=\sum_{x=(x_1,\ldots,x_n)\in(\mathbb{F}_{q^j}^{\times})^n}\Theta_j(K_n(\bar{\lambda},x)).

It is well-known that for λˉ∈Fq×\bar{\lambda}\in\mathbb{F}_q^\times the associated LL-function

L(Kl⁡n/Fq,λˉ,T)(−1)n+1=exp⁡(∑j=1∞Kl⁡n(λˉ/Fqj)Tjj).L(\operatorname{Kl}_n/\mathbb{F}_q,\bar{\lambda},T)^{(-1)^{n+1}}=\exp\left(\sum_{j=1}^{\infty}\frac{\operatorname{Kl}_n(\bar{\lambda}/\mathbb{F}_{q^j})T^j}{j}\right).

is a polynomial in TT of degree n+1n+1 (see [1]), with reciprocal roots {αi}i=0n\{\alpha_i\}_{i=0}^n with

αi=qiui(λˉ)(1)\alpha_i=q^i u_i(\bar{\lambda}) \tag*{(1)}

where ui(λˉ)u_i(\bar{\lambda}) is pp-adically a principal unit. Our goal in the present work is to give a pp-adic formula which realizes ui(λˉ)u_i(\bar{\lambda}) for each i=0,1,…,ni=0,1,\ldots,n as a special value of a suitable pp-adic analytic function at λ^\hat{\lambda}, the Teichmüller lifting of λˉ\bar{\lambda}. Such formulas have appeared in the literature for the unique unit root u0(λˉ)u_0(\bar{\lambda}) in the hyperkloosterman family (and other families when such a unique unit root exists). In the generalized Airy case (Zhang [8]), formulas were proven for all reciprocal roots and the methods used there apply here as well. The approach in [8] looks at the Frobenius action on the exterior powers of the deformation equation. A related approach developed by Dwork [3, 4] involves a factorization of normalized local solution matrices to the deformation equation using a global factor.

In [2] Adolphson and Sperber show that equation (1.1) holds in all characteristics. That work required us to work with the splitting function θ∞\theta_\infty. We will prefer here to use the θ1\theta_1 splitting function of Dwork. (The loss that this choice incurs in the restriction of the characteristic pp is necessitated by our need for a compatible pp-adic formulation of the functional equation.) This necessitates that we restrict to the case p>n+1p>n+1 and this hypothesis will hold throughout the paper.

As usual, let Qp\mathbb{Q}_p be the field of pp-adic numbers, Zp\mathbb{Z}_p be the ring of pp-adic integers, Qq\mathbb{Q}_q be the unramified extension of Qp\mathbb{Q}_p having residue class field Fq\mathbb{F}_q, the finite field of qq (=pa=p^a) elements. Let Ω1=Qp(ζp)\Omega_1=\mathbb{Q}_p(\zeta_p), and Ω=Qq(ζp)\Omega=\mathbb{Q}_q(\zeta_p) be the respective cyclotomic extensions of Qp\mathbb{Q}_p and Qq\mathbb{Q}_q and let O\mathcal{O} be the ring of integers of Ω\Omega. The nontrivial additive character Θ\Theta of Fp\mathbb{F}_p determines a uniformizer π\pi for Ω\Omega by the congruence Θ(1)≡1+π(modπ2)\Theta(1)\equiv1+\pi\pmod{\pi^2}. We define a ring of functions over which we will work. As in previous works (Haessig-Sperber [6]), we define a weight function ww with w(λ)=n+1w(\lambda)=n+1 and for any α∈Z≥0\alpha\in\mathbb{Z}_{\geq0} we have w(λα)(=w(α))=α(n+1)w(\lambda^\alpha)(=w(\alpha))=\alpha(n+1). We will work over the ring of functions (where λ\lambda is a variable)

C0={∑α∈Z≥0c(α)πw(α)λα∣c(α)∈O, c(α)→0 as α→∞}.(2)C_0=\left\{\sum_{\alpha\in\mathbb{Z}_{\geq0}}c(\alpha)\pi^{w(\alpha)}\lambda^\alpha\mid c(\alpha)\in\mathcal{O},\ c(\alpha)\to0\ \text{as }\alpha\to\infty\right\}. \tag*{(2)}

In addition, for future use we also define a C0C_0-module of “bounded functions”

C~0={∑α∈Z≥0c(α)πw(α)λα∣c(α)∈O}.\widetilde{C}_0=\left\{\sum_{\alpha\in\mathbb{Z}_{\geq0}}c(\alpha)\pi^{w(\alpha)}\lambda^\alpha\mid c(\alpha)\in\mathcal{O}\right\}.

Of course, C0⊂C~0C_0\subset\widetilde{C}_0 and we give both of these a norm by defining for ξ=∑α∈Z≥0c(α)πw(α)λα∈C~0\xi=\sum_{\alpha\in\mathbb{Z}_{\geq0}}c(\alpha)\pi^{w(\alpha)}\lambda^\alpha\in\widetilde{C}_0, ∣ξ∣=sup⁡α∈Z≥0∣c(α)∣|\xi|=\sup_{\alpha\in\mathbb{Z}_{\geq0}}|c(\alpha)|.

Let M\mathcal{M} denote the monoid generated by ⟨x1,…,xn,λx1x2⋯xn⟩\langle x_1,\ldots,x_n,\frac{\lambda}{x_1x_2\cdots x_n}\rangle. We recall a weight function can be defined on elements of λσxu∈M\lambda^\sigma x^u\in\mathcal{M} via W(σ,u)=σ(n+1)+∑i=1nuiW(\sigma,u)=\sigma(n+1)+\sum_{i=1}^n u_i. If we set m(u)=max⁡{0,−u1,…,−un}m(u)=\max\{0,-u_1,\ldots,-u_n\} for any u∈Znu\in\mathbb{Z}^n, then the usual polyhedral weight of uu is w(u)=W(m(u),u)=(n+1)m(u)+∑i=1nuiw(u)=W(m(u),u)=(n+1)m(u)+\sum_{i=1}^n u_i. We may define closed sectors {Ij}0≤j≤n\{I_j\}_{0\leq j\leq n} of Zn\mathbb{Z}^n as follows: I0=(Z≥0)nI_0=(\mathbb{Z}_{\geq0})^n and for 1≤j≤n1\leq j\leq n, Ij={u∈Zn∣−uj=m(u)}I_j=\{u\in\mathbb{Z}^n\mid-u_j=m(u)\}. With these definitions, W(σ,u)W(\sigma,u) is linear with respect to σ\sigma and uu while w(u+v)≤w(u)+w(v)w(u+v)\leq w(u)+w(v) with equality holding if and only if uu and vv belong to the same closed sectors.

We define a ring of functions L\mathcal{L}, which is a C0C_0 algebra, via

L={∑μ∈ZnC(μ)λm(μ)πw(μ)xμ∣C(μ)∈C0,C(μ)→0 as μ→∞}.\mathcal{L}=\left\{\sum_{\mu\in\mathbb{Z}^n}C(\mu)\lambda^{m(\mu)}\pi^{w(\mu)}x^\mu\mid C(\mu)\in C_0,C(\mu)\to0\text{ as }\mu\to\infty\right\}.

References

  1. [1]A. Adolphson and S. Sperber. Exponential sums and Newton polyhedra: cohomology and estimates. Ann. of Math. (2) 130 (1989), no. 2, 367–406.DOI
  2. [2]A. Adolphson and S. Sperber. Hyperkloosterman sums revisited. J. Number Theory 243 (2023), 328–351.arxiv.org/abs/1911.10639
  3. [3]B. Dwork. Normalized period matrices. I. Plane curves. Ann. of Math. (2) 94 (1971), 337–388.DOI
  4. [4]B. Dwork. Normalized period matrices. II. Ann. of Math. (2) 98 (1973), 1–57.DOI
  5. [5]B. Dwork. On the uniqueness of Frobenius operator on differential equations. Algebraic number theory, 89–96, Adv. Stud. Pure Math., 17, Academic Press, Boston, MA, 1989.
  6. [6]C. D. Haessig and S. Sperber. Symmetric power L-functions for families of generalized Kloosterman sums. Trans. Amer. Math. Soc. 369 (2017), no. 2, 1459–1493.
  7. [7]S. Sperber. Congruence properties of the hyper-Kloosterman sum. Compositio Math. 40 (1980), no. 1, 3–33.
  8. [8]H. Zhang. Formulas for roots of L-functions of the airy exponential sums. Finite Fields Appl. 49 (2018), 29–48.DOI

Paper details

vlnc:W_rcupazar
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC0-1.0
Usage

views · 0

downloads · 0

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

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents