\documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage[utf8]{inputenc} \usepackage{lmodern} \usepackage[margin=1.05in]{geometry} \usepackage{amsmath,amssymb,amsthm} \usepackage{aliascnt} \usepackage{microtype} \usepackage{xcolor} \usepackage{hyperref} \usepackage[capitalize,noabbrev]{cleveref} \hypersetup{ colorlinks=true, linkcolor=blue!45!black, citecolor=green!35!black, urlcolor=blue!55!black, pdftitle={Algebraic Exponential Integrals: Transcendence and Linear Relations}, pdfauthor={Christopher D. Long; Antoine-Auguste Le Blanc} } \numberwithin{equation}{section} \setlength{\emergencystretch}{3em} \newtheorem{theorem}{Theorem}[section] \newaliascnt{proposition}{theorem} \newtheorem{proposition}[proposition]{Proposition} \aliascntresetthe{proposition} \newaliascnt{lemma}{theorem} \newtheorem{lemma}[lemma]{Lemma} \aliascntresetthe{lemma} \newaliascnt{corollary}{theorem} \newtheorem{corollary}[corollary]{Corollary} \aliascntresetthe{corollary} \theoremstyle{definition} \newaliascnt{definition}{theorem} \newtheorem{definition}[definition]{Definition} \aliascntresetthe{definition} \theoremstyle{remark} \newaliascnt{remark}{theorem} \newtheorem{remark}[remark]{Remark} \aliascntresetthe{remark} \newcommand{\C}{\mathbb C} \newcommand{\Qbar}{\overline{\mathbb Q}} \newcommand{\Aone}{\mathbb A^1} \newcommand{\Span}{\operatorname{span}} \newcommand{\Spec}{\operatorname{Spec}} \newcommand{\ord}{\operatorname{ord}} \newcommand{\supp}{\operatorname{supp}} \newcommand{\Tr}{\operatorname{Tr}} \newcommand{\Mcal}{\mathcal M} \newcommand{\Ecal}{\mathcal E} \newcommand{\Rcal}{\mathcal R} \newcommand{\Lcal}{\mathcal L} \newcommand{\Dmod}{\mathcal D} \newcommand{\Ocal}{\mathcal O} \title{Algebraic Exponential Integrals:\\ Transcendence and Linear Relations} \author{Christopher D. Long \and Antoine-Auguste Le Blanc\thanks{Le Blanc cryptographic author identifier: \texttt{b6048125\allowbreak{}30b3535b\allowbreak{}23d6dbb0\allowbreak{}f5abe32d\allowbreak{}1ed1a9de\allowbreak{}35fb4f7d\allowbreak{}04d62cff\allowbreak{}08ea19c8}}} \date{August 2026} \begin{document} \maketitle \begin{center} \small \texttt{galizur@gmail.com} \end{center} \begin{abstract} Let $q\in\Qbar[x]$ be nonconstant, and let $\Delta=\sum_\alpha d_\alpha[\alpha]$ be a finite degree-zero zero-cycle on $\Aone(\Qbar)$ with coefficients in $\Qbar$. If $\Gamma$ is any finite piecewise $C^1$ one-chain in $\C$ with $\partial\Gamma=\Delta$, put \[ I_{q,\Delta}(z)=\int_\Gamma e^{zq(x)}\,dx. \] We prove that every nonzero function $I_{q,\Delta}$ is a purely transcendental $E$-function: for every $\xi\in\Qbar^\times$, $I_{q,\Delta}(\xi)$ is transcendental. Equivalently, \[ I_{q,\Delta}(1)\in\Qbar \quad\Longleftrightarrow\quad I_{q,\Delta}\equiv0. \] In particular, $\int_A^B e^{q(x)}dx$ is transcendental for every pair of distinct algebraic endpoints. For a fixed polynomial $q$, we also determine all $\Qbar$-linear relations among finite families of these values: they are exactly the specializations of constant-coefficient functional relations, and the latter are characterized by vanishing of explicit inverse-branch densities on a finite tree in the target of $q$. The proof combines Beukers' lifting theorem with a homogeneous moment connection. A Fourier--Laplace description proves semisimplicity, while an explicit residue calculation shows that every rational horizontal endomorphism is constant. An algebraic value therefore forces an exponential-polynomial identity. Constant-coefficient annihilation and a Cauchy-transform argument then force the original exponential period function to vanish identically. \end{abstract} \medskip \noindent\textit{2020 Mathematics Subject Classification.} Primary 11J81; Secondary 14F10, 34M35. \noindent\textit{Keywords.} $E$-functions, exponential period functions, Siegel--Shidlovskii theorem, Fourier--Laplace transform, polynomial moments, Cauchy transforms, transcendence, linear relations. \section{Introduction and main results} Following Fres\'an and Jossen, we use \emph{exponential period function} for a parameter-dependent integral of the form $\int_\gamma e^{-zf}\omega$ with algebraic data \cite{FresanJossen}. The present functions form a one-dimensional relative family: $X=\Aone$, the regular function is $-q$, the algebraic differential form is $dx$, and the integration chain has finite boundary. This is compatible with the general terminology of exponential periods \cite{CommelinHabeggerHuber,FresanJossenExp}. We refer simply to the polynomial $q$ and to its values and critical values. \begin{definition}[Zero-cycles and exponential period functions] A finite zero-cycle on $\Aone(\Qbar)$ with coefficients in $\Qbar$ is a formal sum \[ \Delta=\sum_{\alpha\in S}d_\alpha[\alpha], \qquad S\subset\Qbar\text{ finite},\quad d_\alpha\in\Qbar. \] Its degree is $\deg\Delta=\sum_{\alpha\in S}d_\alpha$. Suppose $\deg\Delta=0$. Choose a finite piecewise $C^1$ one-chain $\Gamma$ in $\C$, with complex coefficients, such that $\partial\Gamma=\Delta$, and set \begin{equation}\label{eq:I} I_{q,\Delta}(z)=\int_\Gamma e^{zq(x)}\,dx. \end{equation} For every $z$, the one-form $e^{zq(x)}dx$ has an entire primitive in $x$. Thus \eqref{eq:I} is independent of the chosen $\Gamma$, and compactness of $\Gamma$ shows that $I_{q,\Delta}$ is entire in $z$. \end{definition} \begin{theorem}[Algebraic-value rigidity]\label{thm:rigidity} Let $q\in\Qbar[x]$ be nonconstant, and let $\Delta$ be a finite degree-zero zero-cycle on $\Aone(\Qbar)$ with coefficients in $\Qbar$. Then \begin{equation}\label{eq:rigidity} I_{q,\Delta}(1)\in\Qbar \quad\Longleftrightarrow\quad I_{q,\Delta}(z)\equiv0. \end{equation} In the equivalent cases, $I_{q,\Delta}(1)=0$. \end{theorem} The fixed evaluation point $1$ entails no loss: a nonzero algebraic parameter is already part of the polynomial in the exponent. \begin{corollary}[Pure transcendence and algebraic zeros] \label{cor:purely-transcendental} If $I_{q,\Delta}$ is not identically zero, then \[ I_{q,\Delta}(\xi)\notin\Qbar \qquad(\xi\in\Qbar^\times). \] In particular, $I_{q,\Delta}$ has no nonzero algebraic zero. As an $E$-function, it is purely transcendental in the standard sense. \end{corollary} \begin{proof} For $\xi\in\Qbar^\times$, \[ I_{\xi q,\Delta}(z)=I_{q,\Delta}(\xi z). \] The right side is nonzero as an entire function if and only if $I_{q,\Delta}$ is nonzero. Apply \cref{thm:rigidity} to the polynomial $\xi q$. \end{proof} \begin{corollary}[A single algebraic exponential integral] \label{cor:single} Let $q\in\Qbar[x]$ be nonconstant and let $A,B\in\Qbar$ with $A\ne B$. Then, for every $\xi\in\Qbar^\times$, \[ \int_A^B e^{\xi q(x)}\,dx \] is transcendental and nonzero. \end{corollary} \begin{proof} For $\Delta=[B]-[A]$, one has $I_{q,\Delta}(0)=B-A\ne0$, so $I_{q,\Delta}$ is not identically zero. Apply \cref{cor:purely-transcendental}. \end{proof} For a fixed $E$-function, Adamczewski and Rivoal gave an algorithm that computes all nonzero algebraic arguments at which its value is algebraic \cite{AdamczewskiRivoal}; Bostan, Rivoal, and Salvy made the required minimal-equation computations effective in practice \cite{BostanRivoalSalvy}. The point of \cref{thm:rigidity} is different: it gives a uniform answer for the entire family indexed by $(q,\Delta)$ and identifies exactly when the function itself vanishes. Delaygue's Lindemann--Weierstrass theorem gives linear independence under a separation condition on the finite singularity sets of the associated $G$-series \cite{Delaygue}. The functions attached to a fixed $q$ arise from one common moment connection and can have overlapping singularity data; the results below require no such separation and describe the whole relation space. There is classical overlap in special cases. For example, \[ \int_0^A e^{cx^n}\,dx =A\,{}_1F_1\!\left(\frac1n;1+\frac1n;cA^n\right), \] so monomial exponents belong to the hypergeometric $E$-function setting; see, for example, Delaygue's discussion of hypergeometric values \cite{Delaygue}. The no-algebraic-zero consequence also connects with recent work of Fischler and Rivoal on zeros of $E$-functions \cite{FischlerRivoalZeros}. \paragraph{Prior work and the present contribution.} The use of parameter-dependent exponential periods and Siegel--Shidlovskii theory is established in the work of Fres\'an and Jossen. Jossen's 2018 report develops polynomial-phase families with noncompact rapid-decay cycles and states the Andr\'e--Beukers lifting principle for polynomial relations \cite[\S\S2--3]{Jossen2018}. For compact loops on $\mathbb G_m$, Fres\'an and Jossen derive $E$-function systems from linearity, differentiation, and twisted Stokes \cite[\S12.7]{FresanJossenExp}. They also identify the absolute polynomial-phase connection with the Fourier--Laplace transform of the reduced polynomial direct image \cite[Proposition~5.1]{FresanJossen}. In a 2025 lecture reporting joint work in progress with Fres\'an, Jossen presented the quadratic marked-point model, including $\int_0^a e^{-zx^2}\,dx$ at algebraic $a$ and the endpoint exponentials \cite{Jossen2025}. He described its differential Galois group and explained the arithmetic independence obtained from Siegel--Shidlovskii theory, separating the additional $\sqrt\pi$-independence question for the full rapid-decay period field. Our contribution is the uniform rigidity theorem for compact algebraic boundary data and arbitrary polynomial phase, together with the complete fixed-$q$ linear-relation criterion. The argument uses constant-coordinate endomorphism rigidity and inverse-branch densities, without assuming simple critical points or distinct critical values. The proof has three parts. First, moments of \eqref{eq:I} satisfy one explicit inhomogeneous $E$-function system. Second, the homogeneous connection is semisimple and every rational horizontal endomorphism is constant; this converts a polynomial-coefficient relation supplied by Beukers' theorem into a constant-coordinate relation. Third, an exponential-polynomial identity produces vanishing polynomial moments, and a Cauchy transform on a finite tree forces all inverse-branch densities to vanish. \paragraph{Scope of the differential form.} The fixed form $dx$ is essential. For $q(x)=x(x-1)$ and $r(x)=2x^2-x+1$, one has \[ r(x)e^{q(x)}=\frac{d}{dx}\bigl(xe^{q(x)}\bigr), \qquad \int_0^1 r(x)e^{q(x)}\,dx=1. \] Yet the function $\int_0^1 r(x)e^{zq(x)}\,dx$ is not identically zero: its value at $z=0$ is $7/6$. Thus \cref{thm:rigidity} does not extend unchanged to arbitrary polynomial amplitudes. \section{The moment system} We use the standard definition: an entire function $F(z)=\sum_{m\ge0}a_m z^m/m!$ is an $E$-function if its coefficients are algebraic, $h(a_0,\ldots,a_m)=O(m)$, and $F$ satisfies a nonzero linear differential equation over $\Qbar[z]$. Assume throughout this section that $n=\deg q\ge2$, and put \[ V=\C[x]_{