Searcharxiv⌕ Search

arXiv subjects

Michel Waldschmidt

Publications and source records attributed to Michel Waldschmidt.

At least 19 recordsLinked to original sources
arXiv2025

Number of integers represented by families of binary forms III: fewnomials

In a series of papers we investigated the following question: given a family $\calF$ of binary forms having nonzero discriminant and integer coefficients, for each $d\geqslant 3$, we estimate the number of integers $m$ with $|m|\leqslant N$ which are represented by an element in $\calF$ of degree $\geqslant d$. Under suitable assumptions, asymptotically as $N\to\infty$, the main term in the estimate is given by the forms in $\calF$ having degree $d$ (if any), while the forms of degree $>d$ contribute only to the error term. The present text is devoted to fewnomials \[ a_0X^{kr}+a_1X^{k(r-1)}Y^k+\cdots +a_{r-1}X^kY^{k(r-1)}+a_rY^{kr} \] with fixed $r\geqslant 1$ and varying $k,a_0,a_1,\dots,a_r$.

math.NT↗
arXiv2025

Schanuel Property for Elliptic and Quasi--Elliptic Functions

For almost all tuples $(x_1,\dots,x_n)$ of complex numbers, a strong version of Schanuel's Conjecture is true: the $2n$ numbers $x_1,\dots,x_n, {\mathrm e}^{x_1},\dots, {\mathrm e}^{x_n}$ are algebraically independent. Similar statements hold when one replaces the exponential function ${\mathrm e}^z$ with algebraically independent functions. We give examples involving elliptic and quasi--elliptic functions, that we prove to be algebraically independent: $z$, $\wp(z)$, $\zeta(z)$, $\sigma(z)$, exponential functions, and Serre functions related with integrals of the third kind.

math.NT↗
arXiv2025

Variations on Schanuel's Conjecture for elliptic and quasi-elliptic functions I: the split case

It is expected that Schanuel's Conjecture contains all ``reasonable" statements that can be made on the values of {\em the exponential function}. In particular it implies the Lindemann-Weierstrass Theorem and the Conjecture on algebraic independence of logarithms of algebraic numbers. Our goal is to state conjectures {\em \`a la Schanuel}, which imply conjectures {\em \`a la Lindemann-Weierstrass}, for the exponential map of an extension $G$ of an elliptic curve ${\mathcal E}$ by the multiplicative group ${\mathbb G}_m$. In the present paper we assume that the extension is split, that is $G={\mathbb G}_m\times {\mathcal E}$. In a second paper in preparation we will deal with the non-split case, namely when the extension is not a product. Here we propose the {\em split semi-elliptic Conjecture}, which involves the exponential function and the Weierstrass $\wp$ and $\zeta$ functions, related with integrals of the first and second kind. In the second paper, our {\em non-split semi-elliptic Conjecture} will also involve Serre's functions, related with integrals of the third kind. We expect that our conjectures contain all ``reasonable" statements that can be made on the values of these functions. In the present paper we highlight the geometric origin of the split semi-elliptic Conjecture: it is {\em equivalent to} the Grothendieck-Andr\'{e} generalized period Conjecture applied to the 1-motive $M=[u:\mathbb{Z} \rightarrow {\mathbb G}_m^s \times {\mathcal E}^n ]$, which is the Elliptico-Toric Conjecture of the first author. We show that our split semi-elliptic Conjecture implies three theorems of Schneider on elliptic analogs of the Hermite-Lindemann and Gel'fond-Schneider's theorems, as well as a conjecture on the Weierstrass zeta function.

math.NT↗
arXiv2023

Number of integers represented by families of binary forms

We extend our previous results on the number of integers which are values of some cyclotomic form of degree larger than a given value (see \cite{FW1}), to more general families of binary forms with integer coefficients. Our main ingredient is an asymptotic upper bound for the cardinality of the set of values which are common to two non isomorphic binary forms of degree greater than $3$. We apply our results to some typical examples of families of binary forms.

math.NT↗
arXiv2023

Number of integers represented by families of binary forms II: binomial forms

We consider some families of binary binomial forms $aX^d+bY^d$, with $a$ and $b$ integers. Under suitable assumptions, we prove that every rational integer $m$ with $|m|\ge 2$ is only represented by a finite number of the forms of this family (with varying $d,a,b$). Furthermore {the number of such forms of degree $\ge d_0$ representing $m$ is bounded by $O(|m|^{(1/d_0)+ε})$} uniformly for $\vert m \vert \geq 2$. We also prove that the integers in the interval $[-N,N]$ represented by one of the form of the family with degree $d\geq d_0$ are almost all represented by some form of the family with degree $d=d_0$. In a previous {paper} we investigated the particular case where the binary binomial forms are positive definite. We now treat the general case by using a lower bound for linear forms of logarithms.

math.NT↗
arXiv2023

Lidstone interpolation I. One variable

According to Lidstone interpolation theory, an entire function of exponential type $<π$ is determined by it derivatives of even order at $0$ and $1$. This theory can be generalized to several variables. Here we survey the theory for a single variable. Complete proofs are given. This first paper of a trilogy is devoted to Univariate Lidstone interpolation; Bivariate and Multivariate Lidstone interpolation will be the topic of two forthcoming papers.

math.CV↗
arXiv2021

On entire functions of several variables with derivatives of even order taking integer values

We extend to several variables an earlier result of ours, according to which an entire function of one variable of sufficiently small exponential type, having all derivatives of even order taking integer values at two points, is a polynomial. The proof in the one dimensional case relies on Lidstone expansion of the function. For $n$ variables, we need $n+1$ points, having the property that the differences of $n$ of them with the remaining one give a basis of ${\mathbb C}^n$. The proof is by reduction to the one variable situation.

math.CV↗
arXiv2020

On transcendental entire functions with infinitely many derivatives taking integer values at several points

Let $s_0,s_1,\dots,s_{m-1}$ be complex numbers and $r_0,\dots,r_{m-1}$ rational integers in the range $0\le r_j\le m-1$. Our first goal is to prove that if an entire function $f$ of sufficiently small exponential type satisfies $f^{(mn+r_j)}(s_j)\in{\mathbb Z}$ for $0\le j\le m-1$ and all sufficiently large $n$, then $f$ is a polynomial. Under suitable assumptions on $s_0,s_1,\dots,s_{m-1}$ and $r_0,\dots,r_{m-1}$, we introduce interpolation polynomials $Λ_{nj}$, ($n\ge 0$, $0\le j\le m-1$) satisfying $$ Λ_{nj}^{(mk+r_\ell)}(s_\ell)=δ_{j\ell}δ_{nk}, \quad\hbox{for}\quad n, k\ge 0 \quad\hbox{and}\quad 0\le j, \ell\le m-1 $$ and we show that any entire function $f$ of sufficiently small exponential type has a convergent expansion $$ f(z)=\sum_{n\ge 0} \sum_{j=0}^{m-1}f^{(mn+r_j)}(s_j)Λ_{nj}(z). $$ The case $r_j=j$ for $0\le j\le m-1$ involves successive derivatives $f^{(n)}(w_n)$ of $f$ evaluated at points of a periodic sequence ${\mathbf{w}}=(w_n)_{n\ge 0}$ of complex numbers, where $w_{mh+j}=s_j$ ($h\ge 0$, $0\le j\le m$). More generally, given a bounded (not necessarily periodic) sequence ${\mathbf{w}}=(w_n)_{n\ge 0}$ of complex numbers, we consider similar interpolation formulae $$ f(z)=\sum_{n\ge 0}f^{(n)}(w_n)Ω_{\mathbf{w},n}(z) $$ involving polynomials $Ω_{\mathbf{w},n}(z)$ which were introduced by W.~Gontcharoff in 1930. Under suitable assumptions, we show that the hypothesis $f^{(n)}(w_n)\in{\mathbb Z}$ for all sufficiently large $n$ implies that $f$ is a polynomial.

math.NT↗
arXiv2020

Integer--valued functions, Hurwitz functions and related topics: a survey

An integer--valued function is an entire function which maps the nonnegative integers $\mathbb N$ to the integers. An example is $2^z$. A Hurwitz function is an entire function having all derivatives taking integer values at $0$. An example is ${\mathrm e}^z$. Lower bound for the growth order of such functions have a rich history. Many variants have been considered: for instance, assuming that the first $k$ derivatives at the integers are integers, or assuming that the derivatives at $k$ points are integers. These as well as and many other variants have been considered. We survey some of them.

math.NT↗
arXiv2019

On transcendental entire functions with infinitely many derivatives taking integer values at two points

Given a subset $S=\{s_0, s_1\}$ of the complex plane with two points and an infinite subset ${\mathscr S}$ of $S\times {\mathbb N}$, where ${\mathbb N}=\{0,1,2,\dots\}$ is the set of nonnegative integers, we ask for a lower bound for the order of growth of a transcendental entire function $f$ such that $f^{(n)}(s)\in{\mathbb Z}$ for all $(s,n)\in{\mathscr S}$. We first take ${\mathscr S}=\{s_0,s_1\}\times 2{\mathbb N}$, where $2{\mathbb N}=\{0,2,4,\dots\}$ is the set of nonnegative even integers. We prove that an entire function $f$ of sufficiently small exponential type such that $f^{(2n)}(s_0)\in{\mathbb Z}$ and $f^{(2n)}( s_1)\in{\mathbb Z}$ for all sufficiently large $n$ must be a polynomial. The estimate we reach is optimal, as we show by constructing a noncountable set of examples. The main tool, both for the proof of the estimate and for the construction of examples, is Lidstone polynomials. Our second example is $(\{s_0\}\times (2{\mathbb N}+1))\cup( \{ s_1\}\times 2{\mathbb N})$ (odd derivatives at $s_0$ and even derivatives at $ s_1$). We use analogs of Lidstone polynomials which have been introduced by J.M.~Whittaker and studied by I.J.~Schoenberg. Finally, using results of W.~Gontcharoff, A. J.~Macintyre and J.M.~Whittaker, we prove lower bounds for the exponential type of a transcendental entire function $f$ such that, for each sufficiently large $n$, one at least of the two numbers $f^{(n)}(s_0)$, $f^{(n)}(s_1)$ is in ${\mathbb Z}$.

math.NT↗
arXiv2019

Sur la représentation des entiers par les formes cyclotomiques de grand degré

For each integer $d\ge 4$, we study the sequence of positive integers which are represented by one at least of the cyclotomic binary forms $Φ_n(X,Y)$, with $n$ a positive integer satisfying $φ(n)\ge d$. The case $d=2$ was studied in our previous work [FLW]. Our demonstration is based on a variant of a statement of [SX] concerning the common values taken by two binary forms of the same degree and non-zero discriminants. All constants are effectively calculable.

math.NT↗
arXiv2018

Linear recurrence sequences and twisted binary forms

Let $ \prod_{i=1}^d (X-α_i Y) \in{\mathbb C}[X,Y]$ be a binary form and let $ε_1,\dots,ε_d$ be nonzero complex numbers. We consider the family of binary forms $ \prod_{i=1}^d (X-α_i ε_i^aY)$, $a\in {\mathbb Z}$, which we write as $$ X^d-U_1(a)X^{d-1}Y+\cdots+(-1)^{d-1} U_{d-1}(a) XY^{d-1}+(-1)^d U_d(a) Y^d.$$ In this paper we study these sequences $\bigl(U_h(a)\bigr)_{a\in {\mathbb Z}}$ which turn out to be linear recurrence sequences.

math.NT↗
arXiv2017

Representation of integers by cyclotomic binary forms

The homogeneous form $Φ_n(X,Y)$ of degree $φ(n)$ which is associated with the cyclotomic polynomial $ϕ_n(X)$ is dubbed a {\it cyclotomic binary form}. A positive integer $m\ge 1$ is said to be {\it representable by a cyclotomic binary form} if there exist integers $n,x,y$ with $n\ge 3$ and $\max\{|x|, |y|\}\ge 2$ such that $Φ_n(x,y)=m$. We prove that the number $a_m$ of such representations of $m$ by a cyclotomic binary form is finite. More precisely, we have $\,φ(n) \le ({2}/ {\log 3})\log m\, $ and $\, \max\{|x|,|y|\} \le ({2}/{\sqrt{3}})\, m^{1/φ(n)}.\,$ We give a description of the asymptotic cardinality of the set of values taken by the forms for $n\geq 3$. This will imply that the set of integers $m$ such that $a_m\neq 0$ has natural density 0. We will deduce that the average value of the integers $a_m$ among the nonzero values of $a_m$ grows like $\sqrt{\log \, m}$.

math.NT↗
arXiv2017

Parametric geometry of numbers in function fields

Parametric geometry of numbers is a new theory, recently created by Schmidt and Summerer, which unifies and simplifies many aspects of classical Diophantine approximations, providing a handle on problems which previously seemed out of reach. Our goal is to transpose this theory to fields of rational functions in one variable and to analyze in that context the problem of simultaneous approximation to exponential functions.

math.NT↗
arXiv2017

Families of Thue equations associated with a rank one subgroup of the unit group of a number field

Twisting a binary form $F_0(X,Y)\in{\mathbb{Z}}[X,Y]$ of degree $d\ge 3$ by powers $\upsilon^a$ ($a\in{\mathbb{Z}}$) of an algebraic unit $\upsilon$ gives rise to a binary form $F_a(X,Y)\in{\mathbb{Z}}[X,Y]$. More precisely, when $K$ is a number field of degree $d$, $σ_1,σ_2,\dots,σ_d$ the embeddings of $K$ into $\mathbb{C}$, $α$ a nonzero element in $K$, $a_0\in{\mathbb{Z}}$, $a_0>0$ and $$ F_0(X,Y)=a_0\displaystyle\prod_{i=1}^d (X-σ_i(α) Y), $$ then for $a\in{\mathbb{Z}}$ we set $$ F_a(X,Y)=\displaystyle a_0\prod_{i=1}^d (X-σ_i(α\upsilon^a) Y). $$ Given $m\ge 0$, our main result is an effective upper bound for the solutions $(x,y,a)\in{\mathbb{Z}}^3$ of the Diophantine inequalities $$ 0<|F_a(x,y)|\le m $$ for which $xy\not=0$ and ${\mathbb{Q}}(α\upsilon^a)=K$. Our estimate involves an effectively computable constant depending only on $d$; it is explicit in terms of $m$, in terms of the heights of $F_0$ and of $\upsilon$, and in terms of the regulator of the number field $K$.

math.NT↗
arXiv2015

Solving simultaneously Thue equations in the almost totally imaginary case

Let $α$ be an algebraic number of degree $d\ge 3$ having at most one real conjugate and let $K$ be the algebraic number field ${\mathbf Q}(α)$. For any unit $ε$ of $K$ such that ${\mathbf Q}(αε)=K$, we consider the irreducible polynomial $f_ε(X)\in{\mathbf Z}[X]$ such that $f_ε(αε)=0$. Let $F_ε(X,Y)\ = Y^df_ε(X/Y)\in{\mathbf Z}[X,Y]$ be the associated binary form. For each positive integer $m$, we exhibit an effectively computable bound for the solutions $(x,y,ε)$ of the diophantine equation $|F_ε(x,y)|\leq m$.

math.NT↗
arXiv2015

Familles d'équations de Thue associées à un sous-groupe de rang 1 d'unités totalement réelles d'un corps de nombres

Let $F$ be an irreducible binary form attached to a number field $K$ of degree $\geq 3$. Let $ε\not\in \{-1, 1\}$ be a totally real unit of $K$. By twisting $F$ with the powers $ε^a$ of $ε$, ($a\in{\mathbf Z}$), we obtain an infinite family $F_a$ of binary forms. Let $m\in{\mathbf Z}$. We give an effective bound for $\max\{|a|, \log|x|, \log|y|\}$ when $a,x,y$ are rational integers satisfying $F_a(x,y)=m$ with $xy\not=0$.

math.NT↗