Maths Olympiad Prep

Library / /6 of 31

Number theory Difficulty 5.7 AIME, harder Prove it Italy

Problem:

Given the sequence
{x1=2;xn+1=2xn21 for n1 \left\{\begin{array}{l} x_{1}=2 ; \\ x_{n+1}=2 x_{n}^{2}-1 \quad \text{ for } n \geq 1 \end{array}\right.
Prove that nn and xnx_{n} are relatively prime for every n1n \geq 1.

Solution

Solution:

We show that, if pp is a prime number dividing xnx_{n}, then pp does not divide nn.
If p=2p=2, then the claim is trivially true, since all terms xnx_{n} are odd for n>1n>1. Hence 2 divides xnx_{n} if and only if n=1n=1.
Suppose then p>2p>2. Let kk be the smallest positive integer such that pp divides xkx_{k}. The sequence rnr_{n} of the remainders of xnx_{n} upon division by pp takes only finitely many values (between 0 and p1p-1), so there exist two integers i<ji<j such that ri=rjr_{i}=r_{j}. For every m0m \geq 0 we then have ri+m=rj+mr_{i+m}=r_{j+m}, so the sequence rnr_{n} is periodic for nin \geq i; it follows that the value rk=0r_{k}=0 (pp divides xkx_{k}) must be attained before two terms of the sequence rnr_{n} repeat. Moreover, if n<kn<k, then rnp1r_{n} \neq p-1, because otherwise rn+1=rn+2=rn+3==1r_{n+1}=r_{n+2}=r_{n+3}=\cdots=1, and hence rnr_{n} would never be zero. Therefore we must have rk=0r_{k}=0 for some kp1k \leq p-1. But if rk=0r_{k}=0, then rk+1=p1r_{k+1}=p-1 and rn=1r_{n}=1 for every n>k+1n>k+1, and thus xkx_{k} is the only term of the sequence divisible by pp. Since pp does not divide any of the numbers 1,2,,p11,2, \ldots, p-1, pp cannot divide kk.

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 translated into English from it; metadata (topic, difficulty) added by this project.