Maths Olympiad Prep

Library / /10 of 18

Algebra Difficulty 8.0 National olympiad, round 2 Prove it Balkan Mathematical Olympiad

Let nNn \in \mathbb{N}, n>2n > 2, and suppose a1,a2,,a2na_1, a_2, \dots, a_{2n} is a permutation of the numbers 1,2,,2n1, 2, \dots, 2n such that a1<a3<<a2n1a_1 < a_3 < \dots < a_{2n-1} and a2>a4>>a2na_2 > a_4 > \dots > a_{2n}. Prove that
(a1a2)2+(a3a4)2++(a2n1a2n)2>n3. (a_1 - a_2)^2 + (a_3 - a_4)^2 + \dots + (a_{2n-1} - a_{2n})^2 > n^3.

Solution

Denote S=(a1a2)2+(a3a4)2++(a2n1a2n)2S = (a_1 - a_2)^2 + (a_3 - a_4)^2 + \dots + (a_{2n-1} - a_{2n})^2. We have
S=i=12ni22(a1a2+a3a4++a2n1a2n)=n(2n+1)(4n+1)32(a1a2+a3a4++a2n1a2n). \begin{aligned} S &= \sum_{i=1}^{2n} i^2 - 2(a_1a_2 + a_3a_4 + \dots + a_{2n-1}a_{2n}) \\ &= \frac{n(2n+1)(4n+1)}{3} - 2(a_1a_2 + a_3a_4 + \dots + a_{2n-1}a_{2n}). \end{aligned}
Next, observe that for each j=1,2,,nj = 1, 2, \dots, n, one of the numbers a2j1,a2ja_{2j-1}, a_{2j} is greater than nn, and the other is at most nn. Indeed, suppose a2j1,a2jna_{2j-1}, a_{2j} \le n. Then a1<a3<<a2j1na_1 < a_3 < \dots < a_{2j-1} \le n and a2n<a2n2<<a2jna_{2n} < a_{2n-2} < \dots < a_{2j} \le n, yielding j+(nj+1)=n+1j + (n - j + 1) = n + 1 distinct positive integers not exceeding nn, a contradiction. The case a2j1,a2j>na_{2j-1}, a_{2j} > n is handled similarly. It follows that a1a2+a3a4++a2n1a2na_1a_2 + a_3a_4 + \dots + a_{2n-1}a_{2n} has form 1b1+2b2++nbn1 \cdot b_1 + 2 \cdot b_2 + \dots + n \cdot b_n, where b1,b2,,bnb_1, b_2, \dots, b_n is some permutation of n+1,n+2,,2nn+1, n+2, \dots, 2n. But it is known that such an expression will be maximal if and only if b1<b2<<bnb_1 < b_2 < \dots < b_n. Therefore,
a1a2+a3a4++a2n1a2n1(n+1)+2(n+2)++n2n=nn(n+1)2+n(n+1)(2n+1)6.(2) a_1a_2+a_3a_4+\dots+a_{2n-1}a_{2n} \le 1(n+1)+2(n+2)+\dots+n \cdot 2n = n \cdot \frac{n(n+1)}{2} + \frac{n(n+1)(2n+1)}{6}. \quad (2)
From (1) and (2), we find
Sn(2n+1)(4n+1)3n2(n+1)n(n+1)(2n+1)3=n3.(3) S \ge \frac{n(2n+1)(4n+1)}{3} - n^2(n+1) - \frac{n(n+1)(2n+1)}{3} = n^3. \quad (3)
By the above arguments, for equality to hold, there would have to exist indices i,j,ki, j, k (since n3n \ge 3) such that {a2i1,a2i}={1,n+1}\{a_{2i-1}, a_{2i}\} = \{1, n+1\}, {a2j1,a2j}={2,n+2}\{a_{2j-1}, a_{2j}\} = \{2, n+2\} and {a2k1,a2k}={3,n+3}\{a_{2k-1}, a_{2k}\} = \{3, n+3\}. It is easy to check that this is impossible, given the assumptions on the permutation a1,a2,,a2na_1, a_2, \dots, a_{2n}. Therefore, equality cannot hold in (3) and S>n3S > n^3.

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement and solution reproduced as published; topic and difficulty added by this site.