\documentclass[12pt,reqno]{article}
\usepackage[usenames]{color}
\usepackage{graphicx}
\usepackage{amscd}
\usepackage{amsthm}
\usepackage[colorlinks=true,
linkcolor=webgreen,
filecolor=webbrown,
citecolor=webgreen]{hyperref}
\definecolor{webgreen}{rgb}{0,.5,0}
\definecolor{webbrown}{rgb}{.6,0,0}
\usepackage{color}
\usepackage{fullpage}
\usepackage{float}
\usepackage{psfig}
\usepackage{graphics,amsmath,amssymb}
\usepackage{amsfonts}
\usepackage{latexsym}
\usepackage{epsf}
%\usepackage[dvips]{graphicx}
\newtheorem{theorem}{Theorem}[section]
\newtheorem{proposition}{Proposition}[section]
\newtheorem{corollary}{Corollary}[section]
\newtheorem{lemma}{Lemma}[section]
\newtheorem{defin}[theorem]{Definition}
\newenvironment{definition}{\begin{defin}\normalfont\quad}{\end{defin}}
\newtheorem{examp}[theorem]{Example}
\newenvironment{example}{\begin{examp}\normalfont\quad}{\end{examp}}
\setlength{\textwidth}{6.5in}
\setlength{\oddsidemargin}{.1in}
\setlength{\evensidemargin}{.1in}
\setlength{\topmargin}{-.5in}
\setlength{\textheight}{8.9in}
\newcommand{\seqnum}[1]{\href{http://www.research.att.com/cgi-bin/access.cgi/as/~njas/sequences/eisA.cgi?Anum=#1}{\underline{#1}}}
\begin{document}
\begin{center}
\epsfxsize=4in
\leavevmode\epsffile{logo129.eps}
\end{center}
\begin{center}
\vskip 1cm{\LARGE\bf Characterizing the Sum of Two Cubes\\
}
\vskip 1cm
\large
Kevin A. Broughan\\
University of Waikato\\
Hamilton 2001\\
New Zealand\\
\href{mailto:kab@waikato.ac.nz}{\tt kab@waikato.ac.nz} \\
\end{center}
\vskip .2 in
\begin{abstract}
An intrinsic characterization of positive integers which can be
represented as the sum or difference of two cubes is given. Every
integer has a smallest multiple which is a sum of two cubes and
such that the multiple, in the form of an iterated composite
function of the integer, is eventually periodic with period one or
two. The representation of any integer as the sum of two cubes to
a fixed modulus is always possible if and only if the modulus is
not divisible by 7 or 9.
\end{abstract}
% text should be lower case, unless caps are necessary for meaning
\def\a{\alpha}
\def\z{\zeta}
\def\sx{\sum_{n\le \sqrt{x}}}
\section{Introduction}
Consider the beautiful characterization of numbers which are the
sum of two squares, namely \cite[Theorem 366]{hardy} a number $n$
is the sum of two squares if and only if all the prime factors of
$n$ of the form $4m+3$ have even exponent in the standard
factorization of $n$. This is not matched by any known comparable
condition for the sum of two cubes. In the absence of such a
characterization there has been a great deal of interest in
questions related to the sum of two cubes, see for example
\cite{nathanson}, \cite{silverman}.
In Section 2 we give an intrinsic characterization, a property of
$n$ itself, which will determine whether it is representable as
the sum of two cubes or not. The characterization is not so simple
but is complete, and covers both $n=x^3+y^3$ and $n=x^3-y^3$. To
have a representation in either of these forms $n$ must have a
divisor $m$ which satisfies strict conditions: $m^3-n/m$ must be
divisible by 3 with quotient $l$ satisfying $m^2+4l$ is a perfect
square. The applicable range for values of $m$ and sign of $l$
discriminates between the two equations $n=x^3+y^3$ and
$n=x^3-y^3$.
In Section 3 the equation $n=x^3+y^3$ modulo $m$ is considered and
the main result of the paper proved. The divisibility of $m$ by 7
or 9 is definitive, in that it is in these cases, and {\bf only}
in these cases, that the form $n\equiv x^3+y^3 \mod m$ does {\bf
not} have a solution for every $n$.
Every positive integer has a multiple which is the sum of two
cubes. This phenonena is studied in Section 4 where functions,
$\theta(n)$ and $\eta(n)$ giving the ``minimum multiple" of an
integer which represents the sum of two cubes, are defined. These
functions, when iterated, are eventually periodic with period
length one or two.
%====================================================================================
\section{Characterizing the sum of two cubes}
\begin{theorem}
Let $n$ be a positive integer. Then the equation $n=x^3+y^3$ has a
solution in positive integers $x$ and $y$ if and only if the
following three conditions are satisfied:
\begin{itemize}
\item[1a.] There exists a divisor $m\mid n$ with $n^{\frac{1}{3}}\le m\le
2^{\frac{2}{3}} n^{\frac{1}{3}}$ such that
\item[2a.] for some positive integer $l$, $m^2-n/m=3 l$ and such
that
\item[3a.] the integer $m^2-4l$ is a perfect square.
\end{itemize}
The conditions equivalent to the existence of a solution to
$n=x^3-y^3$ in positive integers are as follows:
\begin{itemize}
\item[1b.] There exists a divisor $m\mid n$ with $1\le m<
n^\frac{1}{3}$ such that
\item[2b.] for some positive integer $l$, $n/m-m^2=3 l$ and such
that
\item[3b.] the integer $m^2+4l$ is a perfect square.
\end{itemize}
\end{theorem}
\begin{proof}
First we show that if the equation $n=x^3+y^3$ has a solution then
(1a-3a) must be satisfied.
(1a) Let $n=u^3+v^3=(u+v)(u^2-uv+v^2)$
in positive integers $u,v$ and let $m=u+v$ so $m\mid n$. The form
$$
x^2-xy+y^2=\frac{n}{m}
$$
is the equation of an ellipse, called here E, with major axis the
line $y=x$, and $(u,v)$ is a point on the ellipse in the first
quadrant.
The straight line $m=x+y$ cuts the x-axis at $x=m$ which is equal
or to the right of the point where the ellipse cuts the axis,
namely $x=\sqrt{n/m}$. Hence
$$
\sqrt{\frac{n}{m}}\le m\quad\Rightarrow\quad n^\frac{1}{3}\le m
\quad(1).
$$
The length of the major axis of E is $\sqrt{2n/m}$ and the
distance of the line $x+y=m$ from the origin $m/\sqrt{2}$. Since
the line cuts the ellipse we must have
$$
\frac{m}{\sqrt{2}}\le \sqrt{\frac{2n}{m}}\quad\Rightarrow\quad
m\le 2^\frac{2}{3}n^\frac{1}{3} \quad(2).
$$
By (1) and (2)
$$n^{\frac{1}{3}}\le m\le 2^{\frac{2}{3}}
n^{\frac{1}{3}}.$$
(2a) Substitute $v=m-u$ in $n/m = u^2-uv+v^2$ to obtain the
equation
$$
\frac{n}{m}= 3(u^2-mu)+m^2.
$$
Hence $3\mid m^2-n/m$. Since, by (1a) $n\le m^3$,
$l=(m^2-n/m)/3\ge 0$.
(3a) Now consider the value of $l$: $l=-u^2+mu$. This means $u$ is
an integer root of the quadratic equation $x^2-mx+l=0$ with
integer coefficients, so the discriminant, namely $m^2-4l$, must
be a perfect square.
(1b) If $u\ge 0$ and $v<0$ then the point $(u,v)$ lies on E in the
fourth quadrant so the line $m=x+y$ cuts the x-axis to the left of
$x=\sqrt{n/m}$ leading to the bound $m0$ then
$\theta(n)\leq x+y$. Another problem is to characterize those $n$
such that $\theta(n)=x+y$, this minimum positive value.
Note also that a function like $\theta$ can be defined for forms
with appropriate symmetry properties, e.g. $f(x,y)=x^k+y^k$ for
$k$ odd.
\bigskip
%====================================================================================
{\bf Acknowledgements:} The support of the Department of
Mathematics of the University of Waikato, the discussions held
with Ian Hawthorn and the valuable contributions, especially to
shortening the proof of Theorem 3, of an anonymous referee are
warmly acknowledged.
\begin{thebibliography}{6}
\bibitem{apostol}
T. M. Apostol, \textit{Introduction to Analytic Number Theory},
Springer Verlag, 1976.
\bibitem{hardy}
G. H. Hardy and E. M. Wright, \textit{An Introduction to the
Theory of Numbers}, Fifth Edition, Oxford, 1979.
\bibitem{hua}
K. L. Hua, \textit{Introduction to Number Theory},
Springer-Verlag, 1982.
\bibitem{lang}
S. Lang, Old and new conjectured diophantine inequalities, {\it
Bull. Amer. Math. Soc.}, {\bf 23}, (1990), 37--75.
\bibitem{mordell}
L. J. Mordell, \textit{Diophantine Equations}, Academic Press,
1969.
\bibitem{nathanson}
M. B. Nathanson, \textit{Additive Number Theory: The Classical
Bases}, Springer-Verlag, 1996.
\bibitem{nathanson2}
M. B. Nathanson, \textit{Additive Number Theory: Inverse Problems
and the Geometry of Sumsets}, Springer-Verlag, 1996.
\bibitem{silverman}
J. H. Silverman and J. Tate, \textit{Rational Points on Elliptic
Curves}, Springer-Verlag, 1992.
\bibitem{vosper}
A. G. Vosper, The critical pairs of subsets of a group of prime
order. {\it J. London Math. Soc.}, {\bf 31}, (1956), 200--205,
280--282.
\end{thebibliography}
\bigskip
\hrule
\bigskip
\noindent 2000 {\it Mathematics Subject Classification}:
Primary 11A07; Secondary 11B13, 11B50, 11D25, 11D79, 11P05.
\noindent \emph{Keywords: sum of two cubes; diophantine equation}
\bigskip
\hrule
\bigskip
\noindent (Concerned with sequence
\seqnum{A045980}.)
\bigskip
\hrule
\bigskip
\vspace*{+.1in}
\noindent
Received March 31 2003;
revised version received December 18 2003.
Published in {\it Journal of Integer Sequences}, January 14 2004.
\bigskip
\hrule
\bigskip
\noindent
Return to
\htmladdnormallink{Journal of Integer Sequences home page}{http://www.math.uwaterloo.ca/JIS/}.
\vskip .1in
\end{document}
\end{document}