\section{Sharpness over the integers}\label{sec:sharp} Equality in \eqref{eq:main} requires the configuration \eqref{eq:nodes}, which is irrational for $n\ge4$. It is therefore natural to ask how closely integer solutions can approach the bound. The answer is: arbitrarily closely, at least for $n=4$, $n=5$ and $n=6$. The four examples of Table~\ref{tab:shuwen} were supplied by Chen Shuwen, who verified Theorem~\ref{thm:main} on them independently. \input{tab_shuwen} Their structure is visible once $A\cup B$ is normalised to $[0,1]$: the two simple nodes $0$ and $1$ of \eqref{eq:nodes} are hit exactly, and each of the $n-1$ double nodes is straddled by a pair of elements. For $n=4$ and $n=5$ that pair is consecutive and the node is its exact midpoint, so the deviation from \eqref{eq:nodes} is half a unit of the original scale; for $n=6$ the largest deviation is two units, and for $n=7$ it is close to $46$, which is why that example is markedly less tight. This is not an accident, and the first three cases extend to infinite families. \begin{proposition}\label{prop:pell} \emph{(i)} Let $u,w$ be positive integers with $u^{2}-2w^{2}=1$ and $u>3$, and put $m=(u-1)/2$ and $a=(u-w-1)/2$. Then \[ A=\{0,\,m,\,m+1,\,u\},\qquad B=\{a,\,a+1,\,2m-a,\,2m+1-a\} \] is an ideal solution of size $4$ with diameter $L=u$. \emph{(ii)} Let $v,w$ be integers with $w>v>0$ and $2w^{2}-6vw+2v^{2}-v-w-1=0$, and put $u=2(w-v)$. Then \[ A=\{0,\,u-v-1,\,u-v,\,u+w,\,u+w+1\},\qquad B=\{2u-x:\ x\in A\} \] is an ideal solution of size $5$ with diameter $L=2u$. \emph{(iii)} Let $d,h,e$ be positive integers with $h$ odd, $e\equiv d\pmod2$ and $e^{2}-3h^{2}=1-3d^{2}$, and put $a=(h-1)/2$ and $b=(e-d)/2$. Then \[ A=\{\pm a,\ \pm(a+1),\ \pm h\},\qquad B=\{\pm d,\ \pm b,\ \pm(b+d)\} \] is a symmetric ideal solution of size $6$ with diameter $L=2h$, and \begin{equation}\label{eq:ratio6} \frac{4^{5}|C|}{L^{6}}\;=\;\Bigl(1-\frac{1}{h^{2}}\Bigr)^{\!2} \;-\;\frac{d^{2}\bigl(3h^{2}+1-4d^{2}\bigr)^{2}}{h^{6}} \;=\;1-\frac{2+9d^{2}}{h^{2}}+O\bigl(h^{-4}\bigr). \end{equation} In all three cases the defining equation has infinitely many solutions, and along any of the families \[ \frac{4^{n-1}|C|}{L^{n}}\ \longrightarrow\ 1 . \] \end{proposition} \begin{proof} That each displayed pair satisfies \eqref{eq:pte} is a direct computation. For (i), Pell's equation $u^{2}-2w^{2}=1$ has infinitely many solutions. For (ii), solving the defining conic for $w$ gives discriminant $20v^{2}+20v+9=5(2v+1)^{2}+4$, so its integer points correspond to the solutions of $y^{2}-5X^{2}=4$ with $X=2v+1$; these are indexed by the Fibonacci numbers, $X=F_m$ with $m$ even, and the integrality of $w$ selects $m\equiv2\pmod 6$. There are infinitely many. For (iii), both sides are symmetric about $0$, so the odd power sums vanish identically and only $k=2$ and $k=4$ remain. Writing $S=(3h^{2}+1)/2$, which is an integer because $h$ is odd, those two conditions read \[ b^{2}+(b+d)^{2}=S-d^{2},\qquad b^{4}+(b+d)^{4}=\tfrac{1}{2}S^{2}-d^{4}, \] so that $b^{2}$ and $(b+d)^{2}$ are the roots of $z^{2}-(S-d^{2})z+(\tfrac12 S-d^{2})^{2}$. Its discriminant is $d^{2}(3h^{2}+1-3d^{2})=d^{2}e^{2}$, a square by the defining equation, and substituting $3h^{2}=e^{2}+3d^{2}-1$ turns the two roots into $\bigl((e-d)/2\bigr)^{2}$ and $\bigl((e+d)/2\bigr)^{2}$, which are perfect squares; the only remaining condition is the Pell equation. For its infinitude, the map $(e,h)\mapsto(7e+12h,\,4e+7h)$, multiplication by $(2+\sqrt3)^{2}$ in $\mathbb{Z}[\sqrt3]$, preserves $e^{2}-3h^{2}$ and the parities of $e$ and of $h$; when $d$ is odd the map $(e,h)\mapsto(2e+3h,\,e+2h)$ already serves, since it interchanges those two parities and both are then odd. Admissible seeds are $(e,h)=(1,1)$ for $d=1$, and $(e,h)=(4,3)$ and $(8,5)$ for $d=2$. The seeds themselves give $A=B$ and only start the orbits; for $d=2$ the two orbits under the first map give $h=37,515,7173,\dots$ and $h=67,933,12995,\dots$. Identity \eqref{eq:ratio6} is a direct computation, and it gives the limit for (iii) at once, $d$ being fixed. For the limit, note that $0\in A$ in both cases, so $\prod_i a_i=0$ and \[ \frac{4^{n-1}|C|}{L^{n}}\;=\;4^{n-1}\prod_{i=1}^{n}\frac{b_i}{L}, \] a continuous function of the normalised configuration $(A\cup B-\min)/L$. Along family (i) one has $w/u\to2^{-1/2}$, and along family (ii) the conic gives $w/v\to(3+\sqrt5)/2$; in either case the normalised configuration converges to \eqref{eq:nodes}, with $0$ still in the part corresponding to $A$. The displayed quantity is therefore continuous at the limit and takes there the value $1$, by the equality case of Theorem~\ref{thm:main}. \end{proof} \begin{corollary}\label{cor:sharp} For $n=4$, $n=5$ and $n=6$, the supremum of $4^{n-1}|C|/L^{n}$ over all ideal solutions of that size over $\mathbb{Z}$ equals $1$, and is not attained. In particular the constant $4^{n-1}$ in Theorem~\ref{thm:main} cannot be replaced by any larger one. \end{corollary} \input{tab_families} \begin{lemma}\label{lem:sym} Let $(A,B)$ be an ideal solution of size $4$. If $A$ is symmetric about its mean, then so is $B$, about the same point. \end{lemma} \begin{proof} Translate the mean to $0$, so that $\sum a_i=\sum b_i=0$. For any four quantities of zero sum, \[ \sum_{i} a_i^{3} \;=\; 3\,(a_1+a_2)(a_1+a_3)(a_1+a_4). \] If $A$ is symmetric then $\sum a_i^{3}=0$, hence $\sum b_i^{3}=0$ by \eqref{eq:pte}, so one of the three pairings of $B$ has a vanishing pair sum; the complementary pair then sums to zero as well. \end{proof} Not every ideal solution of size $4$ is symmetric: $\{0,18,18,35\}$ against $\{3,8,30,30\}$ is ideal, and neither side is symmetric about its mean. \begin{theorem}\label{thm:size4} Let $(A,B)$ be an ideal solution of size $4$ which is symmetric about its mean. Translate that mean to $0$ and, if it is a half-integer, scale by $2$, and write $A=\{-p,-q,q,p\}$ and $B=\{-r,-s,s,r\}$ with integers $p>q\ge0$ and $r\ge s\ge1$, the two sides being labelled so that $p=\max(A\cup B)$. Then $r^{2}+s^{2}=p^{2}+q^{2}$, $r^{2}s^{2}>p^{2}q^{2}$, and \[ \frac{4^{3}|C|}{L^{4}}\;=\;\frac{4\,(r^{2}s^{2}-p^{2}q^{2})}{p^{4}} \;\le\;\Bigl(1-\frac{q^{2}}{p^{2}}\Bigr)^{\!2}, \] with equality if and only if $r=s$; when $q=0$ this equality is impossible, and the ratio equals $1-\delta^{2}(2p^{2}-\delta^{2})/p^{4}$ with $\delta=r-s\ge1$. Consequently every ideal solution of size $4$ that is symmetric about its mean, of diameter $L$ and in lowest terms, satisfies \begin{equation}\label{eq:sharp4} \frac{4^{3}|C|}{L^{4}}\;\le\;\Bigl(1-\frac{1}{L^{2}}\Bigr)^{\!2}, \end{equation} and equality holds precisely for \[ A=\Bigl\{0,\tfrac{L-1}{2},\tfrac{L+1}{2},L\Bigr\},\qquad B=\{c,c,d,d\},\qquad c=\tfrac{L-w}{2},\quad d=\tfrac{L+w}{2}, \] where $L^{2}+1=2w^{2}$. For these, $cd=(L^{2}-1)/8$, so that $|C|=\bigl((L^{2}-1)/8\bigr)^{2}$ and the ratio equals $(1-1/L^{2})^{2}$ exactly. \end{theorem} \begin{proof} Both sides being symmetric, the odd power sums vanish and \eqref{eq:pte} reduces to its $k=2$ instance, $r^{2}+s^{2}=p^{2}+q^{2}=:T$. Interchanging $A$ and $B$ changes only the sign of $C$, so the labelling $p=\max(A\cup B)$ costs no generality, and $L=2p$. From $r\ge s$ we get $r^{2}\ge T/2$, and $p^{2}\ge T/2$ because $p>q$. On $[\sqrt{T/2},\sqrt{T}\,]$ the function $r\mapsto r\sqrt{T-r^{2}}$ is strictly decreasing and takes the value $pq$ at $r=p$; since $\{r,s\}\ne\{p,q\}$ forces $r
pq$. Next
$C=p^{2}q^{2}-r^{2}s^{2}$, so $4^{3}|C|/L^{4}=4(r^{2}s^{2}-p^{2}q^{2})/p^{4}$.
Finally $rs\le T/2$, with equality if and only if $r=s$, so
\[
r^{2}s^{2}-p^{2}q^{2}\;\le\;\frac{T^{2}}{4}-p^{2}q^{2}
\;=\;\frac{(p^{2}-q^{2})^{2}}{4},
\]
which gives the bound.
If $q\ge1$, the right-hand side is at most $(1-1/p^{2})^{2}$, with equality
exactly when $q=1$; together with $r=s$ that is $p^{2}+1=2r^{2}$.
If $q=0$, which can only happen when the mean is an integer, then $r=s$ would
give $p^{2}=2r^{2}$, which is impossible, so $\delta=r-s\ge1$; also
$\delta