Browse · MathNet
PrintEstonian Math Competitions
Estonia counting and probability
Problem
Let be a positive integer and let .
For , we call two -element subsets of neighbours, if they have elements in common (i.e. differ by exactly one element). Let be the size of the largest possible collection of -element subsets of , in which no two subsets are neighbours. Prove that .
For , we call two -element subsets of neighbours, if they have elements in common (i.e. differ by exactly one element). Let be the size of the largest possible collection of -element subsets of , in which no two subsets are neighbours. Prove that .
Solution
For any two subsets belonging to such a collection, the sets of the smallest elements must be different (or else they would be neighbours). There are ways to choose the smallest elements, since the number cannot be one of them. Therefore , as desired.
---
Alternative solution.
For all we have , since any two 1-element sets are neighbours. So for all and the statement holds. We will prove the general statement by induction by , taking the base case to be . We will assume that the statement holds for and consider the -element set . Let . The number of -element subsets containing the number can be at most in a neighbour-free collection, since removing the number from all of those subsets yields a neighbour-free collection of -element subsets of . The number of -element subsets not containing can be at most . Therefore . By the induction assumption and . Therefore by Pascal's rule , as desired.
---
Alternative solution.
For all we have , since any two 1-element sets are neighbours. So for all and the statement holds. We will prove the general statement by induction by , taking the base case to be . We will assume that the statement holds for and consider the -element set . Let . The number of -element subsets containing the number can be at most in a neighbour-free collection, since removing the number from all of those subsets yields a neighbour-free collection of -element subsets of . The number of -element subsets not containing can be at most . Therefore . By the induction assumption and . Therefore by Pascal's rule , as desired.
Techniques
Pigeonhole principleInduction / smoothingRecursion, bijectionColoring schemes, extremal arguments