Maths Olympiad Prep

Library / /43 of 49

, 2022

Number theory Difficulty 7.0 National Olympiad Prove it Bulgaria

Let mm and nn be positive integers, while pp is a prime number. Find the maximal sNs \in \mathbb{N} (as a function of m,nm, n, and pp) such that from an arbitrary group of mnpmnp positive integers, one can only choose snpsnp among them, satisfying the following property: The numbers can be split into ss disjoint subsets of npnp elements, such that the sum of the elements in each of the subsets gives the same residue modulo pp.

Solution

s=m1s = m - 1. Assume s=ms = m and consider a set of mnp1mnp - 1 positive integers, congruent 1(modp)1 \pmod{p}, and pp. Clearly this set does not fulfill the statement, thus sm1s \le m - 1.

Lemma. Among every np+p1np + p - 1 positive integers, there exist npnp with sum, divisible by pp.

Proof. We apply induction to nn. For the base n=1n = 1 we have to show, that among every 2p12p - 1 positive integers, there exist pp with sum, divisible by pp. Consider residues modulo pp, and via induction on kk, prove that for all pk2p \ge k \ge 2 the set of the residues SS of all sums of kk elements of an arbitrary (2k1)(2k-1)-element set that does not contain kk equal elements, satisfies Sk|S| \ge k. For k=2k = 2 the statement is obvious. Assume, we proved it for kp1k \le p-1 and let s1,s2,,sks_1, s_2, \dots, s_k be the different residues. Consider a 2k+12k+1-element set and two different elements in it aa and bb. Apply the inductive hypothesis to the remaining subset (it is clear that we can choose aa and bb such that, the remaining subset be admissible). Consider the sets {s1+a,s2+a,,sk+a}\{s_1 + a, s_2 + a, \dots, s_k + a\} and {s1+b,s2+b,,sk+b}\{s_1 + b, s_2 + b, \dots, s_k + b\}. If they do not coincide, we derive k+1k+1 different residues and the auxiliary induction is completed. Assuming the sets coincide, we derive after summation that ab(modp)a \equiv b \pmod{p} - a contradiction. Hence, we have proven the base n=1n = 1 for our major induction. Finally, if we have constructed our (n1)p(n-1)p-element subset with sum, divisible by pp, then among the remaining 2p12p-1 numbers there are pp with sum, divisible by pp, and we can add them to the subset. The lemma is proved.

Now, applying consecutively the lemma, and extracting new npnp elements with sum 0(modp)0 \pmod{p} at each step, we will end up with m1m-1 disjoint subsets with sum of the elements 0(modp)0 \pmod{p}, so the proof is completed.

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 reproduced verbatim; metadata (topic, difficulty) added by this project.