Maths Olympiad Prep

Library / Canadian Mathematical Olympiad /47 of 57

Canadian Mathematical Olympiad

Algebra Difficulty 7.5 National Olympiad, round 2 Prove it Canada

Problem:
A function ff from the positive integers to the positive integers is called Canadian if it satisfies
gcd⁡(f(f(x)),f(x+y))=gcd⁡(x,y) \operatorname{gcd}(f(f(x)), f(x+y)) = \operatorname{gcd}(x, y)
for all pairs of positive integers xx and yy.
Find all positive integers mm such that f(m)=mf(m) = m for all Canadian functions ff.

Solution

Solution:
Define an m∈Nm \in \mathbb{N} to be good if f(m)=mf(m) = m for all such ff. It will be shown that mm is good if and only if mm has two or more distinct prime divisors. Let P(x,y)P(x, y) denote the assertion
gcd⁡(f(f(x)),f(x+y))=gcd⁡(x,y) \operatorname{gcd}(f(f(x)), f(x+y)) = \operatorname{gcd}(x, y)
for a pair x,y∈Nx, y \in \mathbb{N}. Let xx be a positive integer with two or more distinct prime divisors and let pkp^{k} be the largest power of one of these prime divisors such that pk∣xp^{k} \mid x. If x=pk⋅qx = p^{k} \cdot q, then pkp^{k} and qq are relatively prime and x>pkx > p^{k}, q>1q > 1. By P(q,x−q)P(q, x-q),
gcd⁡(f(f(q)),f(x−q+q))=gcd⁡(f(f(q)),f(x))=gcd⁡(q,x−q)=q \operatorname{gcd}(f(f(q)), f(x-q+q)) = \operatorname{gcd}(f(f(q)), f(x)) = \operatorname{gcd}(q, x-q) = q
which implies that q∣f(x)q \mid f(x). By P(pk,x−pk)P\left(p^{k}, x-p^{k}\right),
gcd⁡(f(f(pk)),f(x−pk+pk))=gcd⁡(f(f(pk)),f(x))=gcd⁡(pk,x−pk)=pk \operatorname{gcd}\left(f\left(f\left(p^{k}\right)\right), f\left(x-p^{k}+p^{k}\right)\right) = \operatorname{gcd}\left(f\left(f\left(p^{k}\right)\right), f(x)\right) = \operatorname{gcd}\left(p^{k}, x-p^{k}\right) = p^{k}
which implies that pk∣f(x)p^{k} \mid f(x). Since pkp^{k} and qq are relatively prime, x=pk⋅qx = p^{k} \cdot q divides f(x)f(x), which implies that f(x)≥xf(x) \geq x. Now assume for contradiction that f(x)>xf(x) > x. Let y=f(x)−x>0y = f(x) - x > 0 and note that, by P(x,y)P(x, y), it follows that
f(f(x))=gcd⁡(f(f(x)),f(x+f(x)−x))=gcd⁡(x,f(x)−x)=gcd⁡(x,f(x)). f(f(x)) = \operatorname{gcd}(f(f(x)), f(x+f(x)-x)) = \operatorname{gcd}(x, f(x)-x) = \operatorname{gcd}(x, f(x)).
Therefore f(f(x))∣xf(f(x)) \mid x and f(f(x))∣f(x)f(f(x)) \mid f(x). By P(x,x)P(x, x), it follows that
gcd⁡(f(f(x)),f(2x))=gcd⁡(x,x)=x \operatorname{gcd}(f(f(x)), f(2x)) = \operatorname{gcd}(x, x) = x
This implies that x∣f(f(x))x \mid f(f(x)), which when combined with the above result, yields that f(f(x))=xf(f(x)) = x. Since x∣f(x)x \mid f(x) and xx is divisible by at least two distinct prime numbers, f(x)f(x) is also divisible by at least two distinct prime numbers. As shown previously, this implies that f(x)∣f(f(x))=xf(x) \mid f(f(x)) = x, which is a contradiction since f(x)>xf(x) > x. Therefore f(x)=xf(x) = x for all positive integers xx with two or more distinct prime divisors.

Now it will be shown that all m∈Nm \in \mathbb{N} such that either mm has one prime divisor or m=1m = 1 are not good. In either case, let m=pkm = p^{k} where k≥0k \geq 0 and pp is a prime number and consider the function satisfying that f(pk)=pk+1f\left(p^{k}\right) = p^{k+1}, f(pk+1)=pkf\left(p^{k+1}\right) = p^{k} and f(x)=xf(x) = x for all x≠pk,pk+1x \neq p^{k}, p^{k+1}. Note that this function also satisfies that f(f(x))=xf(f(x)) = x for all positive integers xx. If x+y≠pk,pk+1x+y \neq p^{k}, p^{k+1}, then P(x,y)P(x, y) holds by the Euclidean algorithm since f(f(x))=xf(f(x)) = x and f(x+y)=x+yf(x+y) = x+y. If x+y=pk+1x+y = p^{k+1}, then P(x,y)P(x, y) is equivalent to gcd⁡(x,pk)=gcd⁡(x,pk+1−x)=gcd⁡(x,pk+1)\operatorname{gcd}\left(x, p^{k}\right) = \operatorname{gcd}\left(x, p^{k+1}-x\right) = \operatorname{gcd}\left(x, p^{k+1}\right) for all x<pk+1x < p^{k+1} which holds since the greatest power of pp that can divide xx is pkp^{k}. If x+y=pkx+y = p^{k}, then P(x,y)P(x, y) is equivalent to gcd⁡(x,pk+1)=gcd⁡(x,pk−x)=gcd⁡(x,pk)\operatorname{gcd}\left(x, p^{k+1}\right) = \operatorname{gcd}\left(x, p^{k}-x\right) = \operatorname{gcd}\left(x, p^{k}\right) for all x<pkx < p^{k} which holds as shown above. Note that if m=1m = 1 then this case cannot occur. Since this function satisfies P(x,y)P(x, y), mm is good if and only if mm has two or more distinct prime divisors.

Want a route through all this instead of an archive? The track puts 2,604 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.