Let m>1 be an integer. It is known that there exists a prime number in the interval [2m−m+1,2m]. Prove that among any m pairwise distinct positive integers a1,a2,…,am, there exist two numbers ai and aj (1≤i,j≤m) such that (ai,aj)ai≥m, where (ai,aj) denotes the greatest common divisor of the positive integers ai and aj.
Solution
Proof. Without loss of generality, assume that the greatest common divisor of a1,a2,…,am is 1, otherwise divide all a1,a2,…,am by their greatest common divisor, which does not change the conclusion. Let p be a prime number in the interval [2m−m+1,2m]. If p divides any ai, since the greatest common divisor of a1,…,am is 1, there exists some aj not divisible by p, then (ai,aj)ai≥p>m. So, we can assume that a1,a2,…,am are not divisible by p. If there exists ai≡aj(modp), assume ai>aj, then (ai,aj)ai=(ai,aj)ai−aj+(ai,aj)aj≥p+1>m. Therefore, we can assume that a1,a2,…,am have pairwise distinct remainders modulo p. Let m=2p−1+r, then the condition 2m−m+1≤p≤2m is equivalent to r∈[1,2m]. Divide all non-zero residues modulo p into 2p−1 groups, each group containing the residues i and p−i. By the pigeonhole principle, there are r groups, each containing two elements from a1,a2,…,am. Therefore, we can assume a2i−1≡−a2i(modp),a2i−1<a2i,i=1,2,…,r. Notice that for i=1,2,…,r, we have p∣(a2i−1,a2i)a2i−1+a2i. If (a2i−1,a2i)a2i−1+a2i≥2p, then max((a2i−1,a2i)a2i−1,(a2i−1,a2i)a2i)≥p>m. Therefore, we can assume that for i=1,2,…,r, we have (a2i−1,a2i)a2i−1+a2i=p. Let di=(a2i−1,a2i),a2i−1=uidi,a2i=vidi. Then ui<vi and ui+vi=p. Notice that if any vi≥m, the conclusion is already satisfied. Therefore, we can assume vi∈{2p−1+1,2p−1+2,…,2p−1+r−1=m−1}. By the pigeonhole principle, there are two equal numbers among v1,…,vr. Assume v1=v2=v, hence u1=u2=p−v, denoted as u. Clearly, d1=d2 (since a1=a3). Now consider a1,a2,a3,and a4, assuming d1 and d2 are coprime (otherwise divide them by their greatest common divisor). We have (a1,a4)a1=(d1,v)⋅(d2,u)d1u≥(d2,u)(d1,u)⋅u;(a2,a3)a3=(d1,u)⋅(d2,v)d2u≥(d1,u)(d2,u)⋅u;(6) (a2,a3)a2=(d1,u)⋅(d2,v)d1v≥(d2,v)(d1,v)⋅v;(a1,a4)a4=(d1,v)⋅(d2,u)d2v≥(d1,v)(d2,v)⋅v.(7) Notice that the conditions r∈[1,2m] and p−m+1≤u<v≤m−1 give the following inequality: vm−1<um−1≤p−m+1m−1≤m−m+2m−1<m−mm−1=mm+1. If both (ai,aj)ai in (6) are not greater than m−1, then mm+1≥(d2,u)(d1,u),mm+1≥(d2,u)(d1,u). Thus, either (d1,u)=(d2,u) or (d1,u)>m and (d2,u)>m both hold simultaneously. But note that (d1,u)⋅(d2,u)≤u<m, so it must be that (d1,u)=(d2,u), and since d1 and d2 are coprime, they must both be equal to 1. Similarly, if both (ai,aj)ai in (7) are not greater than m−1, then (d1,v)=(d2,v)=1, but then (a1,a4)a4=(d2,u)(d1,v)d2v=d2v≥2v>m, which is a contradiction! Therefore, there must exist ai and aj such that (ai,aj)ai≥m. □
Want a route through all this instead of an archive? The track
puts 2,444 problems in a working order, from Junior Challenge level to the IMO shortlist.
Source: MathNet,
licensed CC-BY-4.0.
Statement reproduced verbatim; metadata (topic, difficulty) added by this project.