IMO 2001 · Problem 4

Let \(n\) be an odd integer greater than 1, and let \(k_1,k_2,\ldots,k_n\) be given integers. For each of the \(n!\) permutations \(a=(a_1,a_2,\ldots,a_n)\) of \(1,2,\ldots,n\), let \[ S(a)=\sum_{i=1}^{n}k_i a_i. \] Prove that there are two permutations \(b\) and \(c\), \(b\ne c\), such that \(n!\) is a divisor of \(S(b)-S(c)\).

Official text of the problem. Official problem paper (PDF)

Solution

Write \(N=n!\), and let \(\mathcal P\) be the set of all permutations of \(1,2,\ldots,n\).

Suppose, for a contradiction, that no two distinct permutations have the same value of \(S\) modulo \(N\). There are exactly \(N\) permutations and \(N\) residue classes modulo \(N\), so their values represent every residue class exactly once. Consequently, \[ \sum_{a\in\mathcal P}S(a) \equiv 0+1+\cdots+(N-1) =\frac{N(N-1)}2 \pmod N. \] Since \(n>1\), the integer \(N\) is even. Thus \[ \frac{N(N-1)}2 =N\left(\frac N2-1\right)+\frac N2 \equiv\frac N2\not\equiv0\pmod N. \]

On the other hand, for each fixed position \(i\), every number \(j\in\{1,\ldots,n\}\) occurs as \(a_i\) in exactly \((n-1)!\) permutations, because the remaining entries can be arranged arbitrarily. Hence \[ \begin{aligned} \sum_{a\in\mathcal P}S(a) &=\sum_{i=1}^{n}k_i\sum_{a\in\mathcal P}a_i\\ &=\sum_{i=1}^{n}k_i\,(n-1)!\,\frac{n(n+1)}2\\ &=N\,\frac{n+1}{2}\sum_{i=1}^{n}k_i. \end{aligned} \] Because \(n\) is odd, \((n+1)/2\) is an integer, so this sum is divisible by \(N\), a contradiction.

Therefore two distinct permutations \(b,c\) satisfy \(S(b)\equiv S(c)\pmod{n!}\), as required. \(\square\)

The solution was checked against Evan Chen's published IMO solution notes.