Maths Olympiad Prep

Library / Brazilian Mathematical Olympiad /103 of 120

Brazilian Mathematical Olympiad

Algebra Difficulty 6.5 National Olympiad Prove it Brazil

Consider the sequence (an)n∈N(a_n)_{n \in \mathbb{N}} with a0=a1=a2=a3=1a_0 = a_1 = a_2 = a_3 = 1 and anan−4=an−1an−3+an−22a_n a_{n-4} = a_{n-1} a_{n-3} + a_{n-2}^2, n≥4n \ge 4. Prove that all the terms of this sequence are integer numbers.

Solution

Let's prove by induction that both ana_n is an integer and gcd⁡(an,an−1)=gcd⁡(an,an−2)=gcd⁡(an,an−3)=1\gcd(a_n, a_{n-1}) = \gcd(a_n, a_{n-2}) = \gcd(a_n, a_{n-3}) = 1. It is certainly true for n=3n = 3 and direct substitutions show that it is true for n≤7n \le 7. Suppose that it's true for n≤kn \le k. First we prove that ak+1a_{k+1} is an integer by showing that ak−3∣akak−2+ak−12a_{k-3} \mid a_k a_{k-2} + a_{k-1}^2.
In fact,
ak−2ak+ak−12=ak−5ak−3+ak−42ak−6⋅ak−3ak−1+ak−22ak−4+(ak−4ak−2+ak−32ak−5)2=ak−52(ak−5ak−3+ak−42)⋅(ak−3ak−1+ak−22)+ak−4ak−6(ak−4ak−2+ak−32)2ak−4ak−6ak−52 \begin{aligned} a_{k-2}a_k + a_{k-1}^2 &= \frac{a_{k-5}a_{k-3} + a_{k-4}^2}{a_{k-6}} \cdot \frac{a_{k-3}a_{k-1} + a_{k-2}^2}{a_{k-4}} + \left( \frac{a_{k-4}a_{k-2} + a_{k-3}^2}{a_{k-5}} \right)^2 \\ &= \frac{a_{k-5}^2(a_{k-5}a_{k-3} + a_{k-4}^2) \cdot (a_{k-3}a_{k-1} + a_{k-2}^2) + a_{k-4}a_{k-6}(a_{k-4}a_{k-2} + a_{k-3}^2)^2}{a_{k-4}a_{k-6}a_{k-5}^2} \end{aligned}
Reducing modulo ak−3a_{k-3} and noting that gcd⁡(ak−3,ak−4)=gcd⁡(ak−3,ak−5)=gcd⁡(ak−3,ak−6)=1\gcd(a_{k-3}, a_{k-4}) = \gcd(a_{k-3}, a_{k-5}) = \gcd(a_{k-3}, a_{k-6}) = 1 by induction hypothesis, this is equivalent to prove that
ak−22ak−42(ak−52+ak−4ak−6)≡0(modak−3) a_{k-2}^2 a_{k-4}^2 (a_{k-5}^2 + a_{k-4} a_{k-6}) \equiv 0 \pmod{a_{k-3}}
But ak−52+ak−4ak−6=ak−3ak−7a_{k-5}^2 + a_{k-4}a_{k-6} = a_{k-3}a_{k-7}, so ak−3∣akak−2+ak−12a_{k-3} \mid a_k a_{k-2} + a_{k-1}^2.
Now the gcd⁡\gcd part. Notice that
gcd⁡(ak+1,ak)=gcd⁡(akak−2+ak−12ak−4,ak)≤gcd⁡(akak−2+ak−12,ak)=gcd⁡(ak−12,ak)=1 \begin{aligned} \gcd(a_{k+1}, a_k) &= \gcd\left(\frac{a_k a_{k-2} + a_{k-1}^2}{a_{k-4}}, a_k\right) \\ &\le \gcd(a_k a_{k-2} + a_{k-1}^2, a_k) = \gcd(a_{k-1}^2, a_k) = 1 \end{aligned}
where the last equality follows from the induction hypothesis. The other equalities follow similarly:
gcd⁡(ak+1,ak−1)≤gcd⁡(akak−2+ak−12,ak−1)=gcd⁡(a2ak−2,ak−1)=1 \gcd(a_{k+1}, a_{k-1}) \le \gcd(a_k a_{k-2} + a_{k-1}^2, a_{k-1}) = \gcd(a_2 a_{k-2}, a_{k-1}) = 1
gcd⁡(ak+1,ak−2)≤gcd⁡(akak−2+ak−12,ak−2)=gcd⁡(ak−12,ak−1)=1 \gcd(a_{k+1}, a_{k-2}) \le \gcd(a_k a_{k-2} + a_{k-1}^2, a_{k-2}) = \gcd(a_{k-1}^2, a_{k-1}) = 1
and the induction step is complete.

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.