Olympiad Maths Prep

Track / Stage 7 / 166 of 300 #1566 of 2000

Problem 1566

National olympiad second round; IMO P1/P4
Combinatorics Difficulty 7.3 Prove it

The National Foundation of Happiness (NFoH) wants to estimate the happiness of people of country. NFoH selected nn random persons, and on every morning asked from each of them whether she is happy or not. On any two distinct days, exactly half of the persons gave the same answer. Show that after kk days, there were at most nnkn-\frac{n}{k} persons whose “yes” answers equals their “no” answers.

This one wants a proof. Work it on paper, read the official solution, then mark yourself honestly — the ladder only means something if the record is true.

Official solution

1. Define the vectors:
Let vi v_i be the vector representing the answers on the i i -th day, where vi=(ai1,ai2,,ain) v_i = (a_{i1}, a_{i2}, \ldots, a_{in}) and aij=1 a_{ij} = 1 if the j j -th person answered "yes" on the i i -th day, and aij=1 a_{ij} = -1 if the j j -th person answered "no".

2. Orthogonality condition:
The problem states that on any two distinct days, exactly half of the persons gave the same answer. This implies that the dot product of any two distinct vectors vi v_i and vj v_j is zero:
vivj=0forij v_i \cdot v_j = 0 \quad \text{for} \quad i \neq j
This orthogonality condition means that the vectors are orthogonal in Rn \mathbb{R}^n .

3. Norm of the vectors:
Each vector vi v_i has a norm squared equal to n n :
vi2=j=1naij2=n v_i^2 = \sum_{j=1}^n a_{ij}^2 = n
because each aij a_{ij} is either 1 1 or 1 -1 .

4. Sum of vectors:
Let xj x_j be the sum of the answers of the j j -th person over k k days:
xj=i=1kaij x_j = \sum_{i=1}^k a_{ij}
We need to show that there are at most nnk n - \frac{n}{k} persons whose "yes" answers equal their "no" answers over k k days.

5. Sum of squares:
Consider the sum of the squares of xj x_j :
j=1nxj2=j=1n(i=1kaij)2 \sum_{j=1}^n x_j^2 = \sum_{j=1}^n \left( \sum_{i=1}^k a_{ij} \right)^2
Using the orthogonality condition, we can write:
j=1nxj2=i=1kvi2=i=1kvi2=kn \sum_{j=1}^n x_j^2 = \left| \sum_{i=1}^k v_i \right|^2 = \sum_{i=1}^k v_i^2 = kn

6. Contradiction:
If there are more than nnk n - \frac{n}{k} persons whose "yes" answers equal their "no" answers, then the number of such persons is greater than nnk n - \frac{n}{k} . Let m m be the number of persons whose "yes" answers equal their "no" answers. Then:
m>nnk m > n - \frac{n}{k}
For these m m persons, xj=0 x_j = 0 . For the remaining nm n - m persons, xj0 x_j \neq 0 . Therefore:
j=1nxj2(nm)12=nm \sum_{j=1}^n x_j^2 \geq (n - m) \cdot 1^2 = n - m
Since m>nnk m > n - \frac{n}{k} , we have:
nm<nk n - m < \frac{n}{k}
Thus:
j=1nxj2<knk=n \sum_{j=1}^n x_j^2 < k \cdot \frac{n}{k} = n
This contradicts the earlier result that j=1nxj2=kn \sum_{j=1}^n x_j^2 = kn .

Therefore, the assumption that there are more than nnk n - \frac{n}{k} persons whose "yes" answers equal their "no" answers leads to a contradiction.

\blacksquare

Source: NuminaMath-1.5, licensed Apache-2.0. Statement and solution reproduced as published; topic, difficulty and ordering added by this site.