Mahler’s method and Carlitz logarithms

Guillaume Estienne · 2026

Read Mahler’s method and Carlitz logarithms by Guillaume Estienne (2026) 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-2603-18832.txt.

Introduction

Estienne gives a new proof of a 2007 theorem by Papanikolas on Carlitz logarithms. The theorem links linear and algebraic independence for these function-field logarithms. Papanikolas used t-motives in the first proof. Estienne uses Mahler's method instead. The proof revisits and extends Denis's 2006 approach, which handled logarithms of elements of Fq(theta) through a Mahler system. The new route shows Mahler systems can carry the full result. It offers a second road into algebraic independence questions in the Carlitz setting.

Catalog record

CREATORGuillaume Estienne
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE74258
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.18832

Browse the full library · About the Tower of Babel

Full text

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

Mahler’s method and Carlitz logarithms Guillaume Estienne Abstract . In 2007, Papanikolas [ 9 ] established that if Carlitz logarithms of algebraic functions are linearly independent over the rational function field, then they are algebraically independent. The purpose of the present paper is to provide a new proof of this theorem using Mahler’s method instead of the theory of t t -motives. We revisit and extend the approach developed by Denis in [ 6 ] , which enabled him in 2006 to prove this result in the particular case of the logarithm of elements in F q ​ ( θ ) F_{q}(\theta) via a Mahler system. 1. Introduction Throughout this article, we let k=\mathbb{F}_{q}(\theta) denote the function field in the indeterminate \theta over the finite field \mathbb{F}_{q} with q elements, where q is a power of a prime number p . We denote by v_{\infty} the \frac{1}{\theta} -adic valuation on k . We also let \mathbb{C}_{\infty} be a completion of an algebraic closure of k_{\infty}:=\mathbb{F}_{q}(\!(\frac{1}{\theta})\!) , and \overline{k} be the algebraic closure of k contained in \mathbb{C}_{\infty} . For any element z\in\mathbb{C}_{\infty} , the Carlitz exponential is defined by: \exp_{C}(z)=\sum_{k\geq 0}\frac{z^{q^{k}}}{D_{k}}, where D_{k}=\prod_{i=1}^{k}(\theta^{q^{i}}-\theta)^{q^{k-i}} . This function satisfies the functional equation \exp_{C}(\theta z)=\theta\exp_{C}(z)+\exp_{C}(z)^{q} . Its kernel is a free \mathbb{F}_{q}[\theta] -module of rank 1, and we denote a generator by \tilde{\pi} . Explicitly, we have: \tilde{\pi}=\theta\tilde{\theta}\prod_{k=1}^{\infty}\left(1-\theta^{1-q^{k}}\right)^{-1} where \tilde{\theta} denotes a fixed q-1 -th rooth of -\theta . The Carlitz exponential admits an inverse on the disk v_{\infty}(z)>-\frac{q}{q-1} called the Carlitz logarithm, which is given by: \mathrm{Log}_{C}(z):=\sum_{k\geq 0}(-1)^{k}\frac{z^{q^{k}}}{L_{k}} where L_{k}:=\prod_{i=1}^{k}(\theta^{q^{i}}-\theta) . Here we are interested in an algebraic independence result for these logarithms obtained by M. Papanikolas [ 9 ] in 2007: Theorem 1.1 . (Papanikolas 2007.) Let \lambda_{1},...,\lambda_{n}\in\mathbb{C}_{\infty} satisfy \exp_{C}(\lambda_{i})\in\overline{k} for i\in[\![1,n]\!] . If the family (\lambda_{i})_{i=1}^{n} is k -linearly independent, then it is \overline{k} -algebraically independent. This result relies on the theory of t -motives, especially through the use of the ABP criterion [ 4 ] established by G. Anderson, W. Brownawell ad M. Papanikolas in 2004. Our aim here is to provide a new proof of this theorem, using the Mahler’s method. In the general case, to study the algebraic relations among values of functions of arithmetic nature, the commonly employed method is to construct ’good’ functions interpolating these values. One then studies the algebraic relations among these functions, which, under certain technical conditions, yield the set of relations among the values under consideration. Three particular frameworks can be distinguished for this transition to functions: • if our functions are solutions of differential systems (Siegel-Shidlovskii’s theorem [ 13 ] ); • if our functions are solutions of \sigma -difference systems (ABP theorem [ 4 ] ); • if our functions are solutions of a Mahler system (Nishioka’s theorem [ 8 ] ). We focus here on the case of Mahler systems. For a natural number d\geq 2 , we say that a function f(z)\in\overline{k}[\![z]\!] is a d -Mahler function if there exists m\geq 1 and polynomials P_{0}(z),...,P_{m}(z)\in\overline{k}[z] where P_{0}(z)\neq 0 and P_{m}(z)\neq 0 such that P_{0}(z)f(z)+P_{1}(z)f(z^{d})+P_{2}(z)f(z^{d^{2}})+....+P_{m}(z)f(z^{d^{m}})=0. Likewise, a column vector F(z)=(f_{1}(z),...,f_{n}(z))^{T} of functions in \overline{k}[\![z]\!] is said to satisfy a d -mahler system if there exists A(z) an invertible matrix with coefficients in \overline{k}(z) such that F(z^{d})=A(z)F(z). If f is a d -Mahler function, then the vector F(z)=(f(z),...f(z^{d^{n-1}}))^{T} is a solution of a d -mahler system where A(z) is a companion matrix, invertible because P_{0}(z)\neq 0 . The major breakthrough in Mahler’s method came in 1990 with Kumiko Nishioka’s remarkable result in the case of number fields: under certain technical hypotheses, the algebraic relations among the values of Mahler functions at an algebraic point arise solely from the algebraic relations among the functions themselves. A more recent proof of this theorem was given in 2023 by B. Adamczewski and C. Faverjon in [ 3 ] (cf. [ 2 ] for Mahler’s method in several variables). However, these results focus on fields of characteristic zero. F. Pellarin, in his Bourbaki seminar [ 10 ] , presents the principal transcendence results for the positive characteristic case, among which is the following: in 2006, Laurent Denis [ 6 ] established, using a Mahler system, the equivalence between linear and algebraic independence independance of the Carlitz logarithm for elements of \mathbb{F}_{q}(\theta^{1/e}) within its disk of convergence, where e is a positive integer. His idea was based on a deformation of the denominators L_{k}:=\prod_{i=1}^{k}(\theta^{q^{i}}-\theta) into functions L_{k}(z):=\prod_{i=1}^{k}(z^{q^{i}}-\theta). The resulting deformation of the logarithm then satisfies a Mahler system of a particular form, for which Laurent Denis established an analogue of Ku. Nishioka’s theorem in [ 5 ] . Since then, G. Fernandes gave, in 2018, a general analogue to Ku. Nishioka’s theorem for the positive characteristic case in [ 7 ] building on the particular case handled by L. Denis : Theorem 1.2 . (Fernandes 2018.) Let \mathbb{K}\subset\overline{k} be a finite extension of k . Let n\geq 1 , d\geq 2 be two integers and f_{1}(z),...,f_{n}(z)\in\mathbb{K}[\![z]\!] be v_{\infty} -adic analytic functions defined in a neighborhood of the origin satisfying the d -Mahler system: \begin{pmatrix}f_{1}(z^{d})\\ \vdots\\ f_{n}(z^{d})\end{pmatrix}=A(z)\begin{pmatrix}f_{1}(z)\\ \vdots\\ f_{n}(z)\end{pmatrix} where A(z)\in\mathrm{GL}_{n}(\mathbb{K}(z)) . Let \alpha\in\overline{k}\backslash\{0\} be such that v_{\infty}(\alpha)>0 and such that for all k\in\mathbb{N} , the matrix A(\alpha^{d^{k}}) is well defined and invertible. Then the following equality holds: \mathrm{tr.deg}_{k}\big(f_{1}(\alpha),...,f_{n}(\alpha)\big)=\mathrm{tr.deg}_{k(z)}\big(f_{1}(z),...,f_{n}(z)\big). We will use a slightly more general version of this theorem, essentially based on the work of D. Adam and L. Denis in [ 1 ] , a proof of which will be given in the appendix. Theorem 1.3 . Let \mathbb{K}\subset\overline{k} be a finite extension of k . Let n\geq 1 , d\geq 1 be two integers and f_{1}(z),...,f_{n}(z)\in\mathbb{K}[\![z]\!] be v_{\infty} -adic analytic functions defined in a neighborhood of the origin satisfying the q^{d} -Mahler system: \begin{pmatrix}f_{1}(z^{q^{d}})\\ \vdots\\ f_{n}(z^{q^{d}})\end{pmatrix}=A(z)\begin{pmatrix}f_{1}(z)\\ \vdots\\ f_{n}(z)\end{pmatrix} where A(z)\in\mathrm{GL}_{n}(\mathbb{K}(H)) , H=\overline{\mathbb{F}_{q}}[\![Z]\!]\cap\overline{\mathbb{F}_{q}(Z)} . Let \alpha\in\overline{k}\backslash\{0\} be such that v_{\infty}(\alpha)>0 and such that for all k\in\mathbb{N} , the matrix A(\alpha^{q^{kd}}) is well defined and invertible. Then the following equality holds: \mathrm{tr.deg}_{k}\big(f_{1}(\alpha),...,f_{n}(\alpha)\big)=\mathrm{tr.deg}_{k(z)}\big(f_{1}(z),...,f_{n}(z)\big). Recall that \tilde{\theta}\in\mathbb{C}_{\infty} is a q-1 -th rooth of -\theta . We here propose interpolating functions of the Carlitz logarithm generalizing those constructed by L. Denis. To this end, let us write u_{i}=\exp_{C}(\lambda_{i}) and assume that v_{\infty}(u_{i})>-\frac{q}{q-1} for i\in[\![1,n]\!] . Let u\in\overline{k} be a uniformizer such that k_{\infty}(\tilde{\theta},u_{1},...u_{n})=\mathbb{F}_{q^{d}}(\!(u)\!) for an integer d\geq 1 . We first define the functions \mathbf{l}(Z)\in\mathbb{F}_{q^{d}}[\![Z]\!] and \mathbf{l}_{i}(Z)\in\mathbb{F}_{q^{d}}[\![Z]\!] such that \mathbf{l}(u)=\frac{1}{\theta} and \mathbf{l}_{i}(u)=\frac{u_{i}}{\theta\tilde{\theta}} , following the idea of D. Adam et L. Denis in [ 1 ] . We then introduce the following functions: G_{i}(Z)=G_{i}(Z,\beta,s):=\sum_{k\geq 0}\frac{\mathbf{l}_{i}^{q^{k}}(Z)}{\pi_{k}(Z,\beta,s)}\in\mathbb{F}_{q^{d}}(\theta,\beta)[\![Z]\!], whith s a positive integer, \beta\in\overline{k}^{*} , and where \pi_{k}(Z)=\pi_{k}(Z,\beta,s):=\beta^{-k}\prod_{i=1}^{k}(1-\theta\mathbf{l}(Z)^{q^{i}})^{s} ; constructed so that \theta\tilde{\theta}G_{i}(u,1,1)=\mathrm{Log}_{C}(u_{i}) . These functions satisfy the functional equation: G_{i}(Z^{q^{d}})=\pi_{d}(Z)(G_{i}(Z)-M_{i}(Z)) where M_{i}(Z)=\sum_{k=0}^{d-1}\frac{\mathbf{l}_{i}^{q^{k}}(Z)}{\pi_{k}(Z)} .This equation yields, on the one hand, a q^{d} -Mahler system satisfying the conditions of the theorem 1.3 , and on the other hand, the convergence of the sequence \left(\sum_{j=0}^{k-1}\frac{M_{i}(Z^{q^{jd}})}{\pi_{jd}(Z)}\right)_{k\geq 1} to G_{i}(Z) . A key step in the proof will be to also consider an interpolating function of the period \tilde{\pi} , closely linked to the denominators \pi_{k}(Z) . Let us then consider the function: G_{0}(Z)=G_{0}(Z,s):=\prod_{k=1}^{\infty}\left(1-\theta\mathbf{l}(Z)^{q^{k}}\right)^{-s}=\underset{k\to\infty}{\lim}\pi_{k}(Z,1,s)^{-1} where s is a positive integer, constructed in such a way that G_{0}(u,1)=\frac{1}{\theta\tilde{\theta}}\tilde{\pi} . It also satisfies a functional equation: G_{0}(Z^{q^{d}})=\beta^{d}\pi_{d}(Z)G_{0}(Z) which is similar to the one satisfied by the

Read the complete text