CombinatoricsDifficulty 8.5ShortlistProve itUnited States
Prove that there exists a real constant c such that for any pair (x,y) of real numbers, there exist relatively prime integers m and n satisfying the relation (x−m)2+(y−n)2<clog(x2+y2+2). (This problem was suggested by Daniel Kane.)
Solution
(By Adam Hesterberg). Without loss of generality we may consider points (x,y) with x>y>0. For any c, let d=2clog(x2+y2+2). Choose c large enough that c>log(2)2andd≥max{9⋅20⋅21,21+21log(x)}. We claim that (x,y) lies within distance 2d=clog(x2+y2+2) of a lattice point (m,n) with relatively prime coordinates.
If y<1, then (x,y) is at distance at most 2<clog(2)<2d from the point (⌊x⌋,1), which has relatively prime coordinates. Otherwise, consider the points (a,b)∈Z2 with ⌊x⌋≤a<⌊x+d⌋ and ⌊y⌋≤b<⌊y+d⌋, all of which are within distance 2d of (x,y). The number of such pairs with a common factor of k is at most (d/k+1)2, so the number of pairs with a common factor between 2 and d is at most k=2∑⌊d⌋(kd+1)2=d2k=2∑⌊d⌋k21+2dk=2∑⌊d⌋k1+⌊d⌋. We have now the estimates k=2∑⌊d⌋k21≤41+k=3∑⌊d⌋k(k−1)1=41+k=3∑⌊d⌋(k−11−k1)≤41+21−⌊d⌋1≤43(17) and k=2∑⌊d⌋k1≤4+k=10∑⌊d⌋k1≤4+10d.(18) Applying the estimates (17) and (18), we find that the number of pairs with a common factor between 2 and d is at most d2k=2∑⌊d⌋k21+2dk=2∑⌊d⌋k1+⌊d⌋≤43d2+102d2+8d+d=2019d2+9d≤2120d2, where the final inequality holds because we chose d≥9⋅20⋅21.
Therefore, at least 21d2 of the pairs have no common factor between 2 and d. By the pigeonhole principle there exists a with ⌊x⌋≤a<⌊x+d⌋ such that at least 21d of the lattice points (a,b) with ⌊y⌋≤b<⌊y+d⌋ have no common factor at most d. Hence either some (a,b) is the desired point with relatively prime coordinates, or each such b has a prime factor greater than d in common with a. These prime factors must be distinct, since the different values of b differ by at most d. Hence a is divisible by their product, which is at least d21d. But this shows that dx>x+d>a≥d21d≥d1+log(x)>dx, where the first inequality holds because x>y>1, the third because d was chosen so that d≥21+21log(x), and the last because d>e, meaning dlog(x)>elog(x)=e. This is a contradiction. Thus, there must have been some point (a,b) with relatively prime coordinates.
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.