\documentclass[11pt,a4paper]{article} \usepackage[T1]{fontenc} \usepackage{lmodern} \usepackage{cmap} \usepackage[utf8]{inputenc} \usepackage{amsmath,amssymb,amsthm} \usepackage[margin=2.8cm]{geometry} \usepackage{hyperref} \theoremstyle{plain} \newtheorem{theorem}{Theorem} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{remark}[theorem]{Remark} \newcommand{\C}{\mathcal{C}} \title{A Chebyshev bound for the constant\\ in the Prouhet--Tarry--Escott problem} \author{David Niedbala Giraudin\\\small Independent researcher, France\quad ORCID 0009-0009-1526-1178} \date{18 November 2026} \begin{document} \maketitle \begin{abstract} Let $(A,B)$ be an ideal solution of size $n$ of the Prouhet--Tarry--Escott problem, with constant $C$ defined by $\prod_i(x-a_i)-\prod_i(x-b_i)=C$, and let $L$ be the diameter of $A\cup B$. We prove \[ |C|\ \le\ 4\left(\frac{L}{4}\right)^{\!n}, \] with equality exactly for the configuration given by the roots of $T_n\mp 1$, where $T_n$ is the Chebyshev polynomial of the first kind. Equality is attained by integer solutions for $n\le 3$ and, by Niven's theorem, is impossible for $n\ge 4$; for $n=4$, $n=5$ and $n=6$, however, we exhibit infinite families of integer solutions, governed by a Pell equation, along which the ratio $4^{n-1}|C|/L^{n}$ tends to $1$, so that the constant $4^{n-1}$ cannot be improved. The bounds on $C$ in the standard references are $p$-adic throughout: they determine which prime powers divide $C$, and the present inequality appears to be the first of archimedean type. Two further consequences are recorded: an unconditional lower bound for the height of any ideal solution, in terms of the known divisors of the universal constant $C_n$, which yields the first such statements for sizes $17\le n\le 20$, where no search has been published; and the observation that the extremal configuration is exactly the trigonometric configuration of Chen Shuwen, for which the analogous extremality statement for the individual variables is an open conjecture. The present result is its proven counterpart for the constant. \end{abstract} \section{Introduction} For $n\ge 2$, two multisets of integers $A=\{a_1,\dots,a_n\}$ and $B=\{b_1,\dots,b_n\}$ with $A\ne B$ form an \emph{ideal solution of size $n$} of the Prouhet--Tarry--Escott (PTE) problem if \begin{equation}\label{eq:pte} \sum_{i=1}^{n}a_i^{\,k}=\sum_{i=1}^{n}b_i^{\,k},\qquad k=1,2,\dots,n-1 . \end{equation} Throughout, $n$ denotes the \emph{size}; the degree is $n-1$. Ideal solutions are known for $n\le 10$ and $n=12$; none is known for $n=11$ or $n\ge 13$. We refer to \cite{Borwein,Caley,CMSV,Shuwen} for background. Writing $F(x)=\prod_i(x-a_i)$ and $G(x)=\prod_i(x-b_i)$, condition \eqref{eq:pte} is equivalent to $F-G$ being a nonzero constant, \begin{equation}\label{eq:C} F(x)-G(x)=C . \end{equation} The arithmetic of $C$ has been studied since Kleiman \cite{Kleiman}, who proved $(n-1)!\mid C$, and Rees and Smyth \cite{ReesSmyth}; the survey of Borwein and Ingalls \cite{BorweinIngalls} and the search of Borwein, Lison\v{e}k and Percival \cite{BLP} record the state of the question at the turn of the century; the current state of the divisibility conditions, for $n\le 20$, is tabulated by Coppersmith, Mossinghoff, Scheinerman and VanderKam \cite[Table~1]{CMSV}, and the analogous questions over imaginary quadratic rings are treated by Caley \cite{Caley}. These references are $p$-adic in nature: they assert that certain prime powers divide $C$. The only archimedean information in them is the trivial bound $|C|\le L^{n}$, immediate from the product formula $|C|=\prod_{i}|b_j-a_i|$, valid for every $j$ \cite[E5]{Borwein}; no bound limiting the \emph{size} of $C$ beyond it appears.\footnote{We have checked this directly in \cite{Borwein}, \cite{BLP}, \cite{Caley}, \cite{CMSV}, \cite{ReesSmyth} and \cite{Shuwen}; Chen Shuwen has also confirmed that no bound of this kind is known to him (personal communication). Theorem~\ref{thm:main} improves the trivial bound by the factor $4^{\,n-1}$, which is $262144$ at $n=10$.} Our result is such a bound. Set \[ L \;=\; \max(A\cup B)-\min(A\cup B), \] the diameter of the solution; note that $L$ is invariant under the translations $x\mapsto x+t$ under which \eqref{eq:pte} is invariant, and scales linearly under $x\mapsto \lambda x$, exactly as $|C|^{1/n}$ does. \begin{theorem}\label{thm:main} Let $(A,B)$ be an ideal solution of size $n$ of the PTE problem, with constant $C$ and diameter $L$. Then \begin{equation}\label{eq:main} |C|\ \le\ 4\left(\frac{L}{4}\right)^{\!n}, \end{equation} or equivalently, in integral form, $4^{n-1}|C|\le L^{n}$. \end{theorem} The bound is sharp. Equality holds precisely when \[ F(x)=\widetilde T(x)+\tfrac{|C|}{2},\qquad G(x)=\widetilde T(x)-\tfrac{|C|}{2} \qquad\text{(or with $F$ and $G$ interchanged),} \] where $\widetilde T$ is the monic Chebyshev polynomial of degree $n$ for the interval $[\min(A\cup B),\max(A\cup B)]$. After normalising that interval to $[0,1]$, the $2n$ elements of $A\cup B$ then take the $n+1$ values \begin{equation}\label{eq:nodes} \cos^{2}\!\left(\frac{j\pi}{2n}\right),\qquad j=0,1,\dots,n , \end{equation} the extreme ones simply and the others doubly. For $n\le 3$ these values are rational and equality is attained by genuine integer solutions: for $n=2$ by $\{0,2\}$ versus $\{1,1\}$ with $|C|=1=4(2/4)^2$, and for $n=3$ by $\{0,3,3\}$ versus $\{1,1,4\}$ with $|C|=4=4(4/4)^3$. For $n\ge 4$ equality is impossible over $\mathbb{Q}$: it would force $\cos(\pi/n)$ to be rational, which by Niven's theorem happens only for $n\in\{1,2,3\}$. Thus \eqref{eq:main} is strict for every ideal solution of size $n\ge4$, although, as the numerical table of Section~\ref{sec:verif} shows, it can be approached closely. Section~\ref{sec:proof} contains the proof, Section~\ref{sec:extremal} discusses the extremal configuration and its relation to a conjecture of Chen Shuwen, Section~\ref{sec:sharp} shows that the bound is approached arbitrarily closely by integer solutions, Section~\ref{sec:height} records the resulting height bounds, and Section~\ref{sec:verif} the numerical verification. \section{Proof of Theorem~\ref{thm:main}}\label{sec:proof} Write $m_0=\min(A\cup B)$ and $M_0=\max(A\cup B)$, so $L=M_0-m_0$, and set \begin{equation}\label{eq:H} H \;=\; \tfrac12\,(F+G),\qquad\text{so that}\qquad F=H+\tfrac{C}{2},\qquad G=H-\tfrac{C}{2} \end{equation} by \eqref{eq:C}. Thus $H$ is monic of degree $n$, and both $H+\frac{C}{2}$ and $H-\frac{C}{2}$ split over $\mathbb{R}$ with all their roots in $[m_0,M_0]$. Also $F'=G'=H'$. \begin{lemma}\label{lem:crit} Let $P$ be a monic polynomial of degree $n$ with all roots real, and let $c_1\le\dots\le c_{n-1}$ be the roots of $P'$. Then $(-1)^{n-j}P(c_j)\ge 0$ for $1\le j\le n-1$. Moreover, if $r$ denotes the largest root of $P$ then $P(x)\ge0$ for $x\ge r$, and if $s$ denotes the smallest root then $(-1)^nP(x)\ge0$ for $x\le s$. \end{lemma} \begin{proof} Let $r_1\le\dots\le r_n$ be the roots of $P$. By Rolle's theorem and a count of multiplicities, $r_j\le c_j\le r_{j+1}$ for $1\le j\le n-1$. On the open interval $(r_j,r_{j+1})$, exactly $n-j$ roots of $P$ lie to the right, so $P$ has constant sign $(-1)^{n-j}$ there; at the endpoints $P$ vanishes. Hence $(-1)^{n-j}P(c_j)\ge0$. The two final assertions are immediate. \end{proof} \begin{lemma}\label{lem:altern} Let $c_1\le\dots\le c_{n-1}$ be the roots of $H'$ and put $x_0=m_0$, $x_j=c_j$ for $1\le j\le n-1$, and $x_n=M_0$. Then $x_0\le x_1\le\dots\le x_n$, all $x_j$ lie in $[m_0,M_0]$, and \begin{equation}\label{eq:altern} (-1)^{n-j}\,H(x_j)\ \ge\ \frac{|C|}{2},\qquad j=0,1,\dots,n . \end{equation} \end{lemma} \begin{proof} The roots of $H'=F'$ lie in the convex hull of the roots of $F$ by the Gauss--Lucas theorem, hence in $[m_0,M_0]$; this gives the ordering. For $1\le j\le n-1$, Lemma~\ref{lem:crit} applied to $F$ and to $G$ --- which have the same derivative, hence the same critical points --- gives, with $\sigma=(-1)^{n-j}$, \[ \sigma F(c_j)\ \ge\ 0,\qquad \sigma G(c_j)\ \ge\ 0 . \] Since $F-G=C$ we have $\sigma F-\sigma G=\sigma C$. If $\sigma C\ge0$ then $\sigma C=|C|$ and $\sigma F(c_j)=\sigma G(c_j)+|C|\ge|C|$; if $\sigma C<0$ then $\sigma G(c_j)=\sigma F(c_j)+|C|\ge |C|$. In either case $\sigma F(c_j)+\sigma G(c_j)\ge |C|$, that is, $\sigma H(c_j)\ge |C|/2$ by \eqref{eq:H}. At $x_n=M_0$ one of $F,G$ vanishes and the other, being monic and evaluated at a point at least as large as all of its roots, is nonnegative by Lemma~\ref{lem:crit}; its value there is $\pm C$, hence equals $|C|$. Therefore $H(M_0)=|C|/2$, which is \eqref{eq:altern} for $j=n$. Symmetrically, at $x_0=m_0$ one of $F,G$ vanishes and the other has sign $(-1)^n$ and absolute value $|C|$, so $H(m_0)=(-1)^n|C|/2$ and $(-1)^{n}H(m_0)=|C|/2$. \end{proof} \begin{proof}[Proof of Theorem~\ref{thm:main}] Let \[ \widetilde T(x)\;=\;2\left(\frac{L}{4}\right)^{\!n} T_n\!\left(\frac{2x-m_0-M_0}{L}\right) \] be the monic Chebyshev polynomial of degree $n$ for $[m_0,M_0]$, where $T_n$ is the Chebyshev polynomial of the first kind; it satisfies \begin{equation}\label{eq:mu} \max_{x\in[m_0,M_0]}|\widetilde T(x)|=\mu:=2\left(\frac{L}{4}\right)^{\!n}. \end{equation} Assume, for contradiction, that $|C|/2>\mu$. Put $Q=H-\widetilde T$; since both $H$ and $\widetilde T$ are monic of degree $n$, $\deg Q\le n-1$. Let $\sigma_j=(-1)^{n-j}$. By Lemma~\ref{lem:altern} and \eqref{eq:mu}, for each $j=0,\dots,n$, \[ \sigma_jQ(x_j)=\sigma_jH(x_j)-\sigma_j\widetilde T(x_j) \ \ge\ \frac{|C|}{2}-\mu\ >\ 0 . \] In particular $Q(x_j)\ne0$ for every $j$, and consecutive values $Q(x_j),Q(x_{j+1})$ have opposite signs; this forces $x_0\max B$, then \begin{equation}\label{eq:shuwen} \frac{\max B}{\max A}\ \ge\ \sin^{2}\!\left(\frac{(n-1)\pi}{2n}\right) \;=\;\cos^{2}\!\left(\frac{\pi}{2n}\right), \end{equation} the right-hand side being exactly the second node in \eqref{eq:nodes}. The statement is restated in \cite[Conjecture~2]{Shuwen} together with an explicit description of the minimising configuration, which is precisely the pattern of double roots of $T_n\mp1$. Inequality \eqref{eq:shuwen} has been verified on millions of numerical solutions; no proof is available. Thus the configuration \eqref{eq:nodes} is conjecturally extremal for the individual variables and, by Theorem~\ref{thm:main}, provably extremal for the constant. The two statements concern different functionals of the same solution, and they are tight at the same place: for the ideal solution of size $5$ \[ \{0,664,701,1787,1788\}\ \text{ versus }\ \{188,189,1275,1312,1976\} \] quoted in \cite[(4.4)]{Shuwen} --- with $A$ the second set, so that $\max A>\max B$ as \eqref{eq:shuwen} requires --- the left-hand side of \eqref{eq:shuwen} exceeds the right-hand side by a factor $1.000387$, while $|C|$ reaches $0.9981$ times the bound \eqref{eq:main}. \input{sec_sharp} \section{Lower bounds for the height}\label{sec:height} Let $C_n$ denote the universal constant of the PTE problem of size $n$, that is, the greatest common divisor of $C$ over all ideal solutions of size $n$, and let $C'_n$ denote the corresponding quantity restricted to symmetric solutions. Divisors of $C_n$ and of $C'_n$ are tabulated in \cite[Table~1]{CMSV} for $n\le20$. Since $C_n\mid C$ for every ideal solution, Theorem~\ref{thm:main} inverts to: \begin{corollary}\label{cor:height} Every ideal solution of size $n$ satisfies \[ L\ \ge\ 4\left(\frac{C_n}{4}\right)^{\!1/n}. \] For a symmetric solution, whose elements may be taken symmetric about $0$ so that $L=2h$ with $h=\max|a_i|$, this reads $h\ \ge\ 2\left(C'_n/4\right)^{1/n}$. \end{corollary} The resulting values are listed in Table~\ref{tab:height}. They are weaker than the exhaustive searches of \cite{CMSV} wherever those have been carried out. For $17\le n\le20$ no search has been published, and Corollary~\ref{cor:height} provides, as far as we are aware, the first unconditional lower bound on the magnitude of a hypothetical ideal solution for those $n$. \input{tab_height} \section{Numerical verification}\label{sec:verif} Table~\ref{tab:verif} lists, for classical ideal solutions of size $n\le12$, the ratio of $|C|$ to the bound \eqref{eq:main}. The defining identities \eqref{eq:pte} are checked exactly for each entry. The computation is reproduced by the ancillary file \texttt{verify\_pte\_chebyshev.py}, which uses exact integer arithmetic and requires only the Python standard library. The same file replays the families of Proposition~\ref{prop:pell}, the extremal family of Theorem~\ref{thm:size4}, and the enumeration of Remark~\ref{rem:enum}. \input{tab_verif} \paragraph{Status of the statements.} Theorem~\ref{thm:main}, its equality case, Proposition~\ref{prop:pell}, Corollary~\ref{cor:sharp}, Lemma~\ref{lem:sym}, Theorem~\ref{thm:size4} and Corollary~\ref{cor:height} are PROVED. The bound \eqref{eq:sharp4} is proved for solutions symmetric about their mean and, for the others, VERIFIED by the enumeration recorded above. The entries of Table~\ref{tab:height} rest on the divisibility data of \cite[Table~1]{CMSV}, which we have VERIFIED independently for $n=7,9,11,13$; the columns $L$ and $h$ there are computed from the exact integer products. Inequality \eqref{eq:shuwen} is OPEN and is quoted here as a conjecture of \cite{Shuwen}. \paragraph{Acknowledgements and declaration.} The integer examples of Table~\ref{tab:shuwen} were provided by Chen Shuwen, who also verified Theorem~\ref{thm:main} on them independently, and he supplied copies of several of the sources consulted for Section~\ref{sec:proof}. Formulas of this kind for ideal solutions of sizes $4$, $5$ and $6$ were obtained by him in 2024 \cite{Shuwen}; the families of Proposition~\ref{prop:pell} were found independently, and I learned of his earlier work by personal communication. I thank him for his generous help and correspondence. Prepared with Claude (Anthropic), which carried out the computations and drafted the proofs under the author's direction; the author is responsible for the content. \begin{thebibliography}{99} \bibitem{Borwein} P.~Borwein, \emph{The Prouhet--Tarry--Escott problem}, Chapter 11 of \emph{Computational Excursions in Analysis and Number Theory}, Springer, New York, 2002, pp.~85--95. \bibitem{BorweinErdelyi} P.~Borwein and T.~Erd\'elyi, \emph{Polynomials and Polynomial Inequalities}, Graduate Texts in Mathematics 161, Springer, 1995. \bibitem{BorweinIngalls} P.~Borwein and C.~Ingalls, \emph{The Prouhet--Tarry--Escott problem revisited}, Enseign. Math. (2) \textbf{40} (1994), 3--27. \bibitem{BLP} P.~Borwein, P.~Lison\v{e}k and C.~Percival, \emph{Computational investigations of the Prouhet--Tarry--Escott problem}, Math. Comp. \textbf{72} (2003), no.~244, 2063--2070. \bibitem{Caley} T.~Caley, \emph{The Prouhet--Tarry--Escott problem}, PhD thesis, University of Waterloo, 2012. \bibitem{CMSV} D.~Coppersmith, M.~J.~Mossinghoff, D.~Scheinerman and J.~M.~VanderKam, \emph{Ideal solutions in the Prouhet--Tarry--Escott problem}, preprint, arXiv:2304.11254. \bibitem{Kleiman} H.~Kleiman, \emph{A note on the Tarry--Escott problem}, J. Reine Angew. Math. \textbf{278/279} (1975), 48--51. \bibitem{ReesSmyth} E.~Rees and C.~J.~Smyth, \emph{On the constant in the Tarry--Escott problem}, in: Cinquante Ans de Polyn\^omes --- Fifty Years of Polynomials, Lecture Notes in Mathematics 1415, Springer, 1990, pp.~196--208. \bibitem{Shuwen} Chen Shuwen, \emph{A survey of the Prouhet--Tarry--Escott problem and its generalizations}, arXiv:2506.11429 (2025). \end{thebibliography} \end{document}