Maths Olympiad Prep

Track / Stage 8 / 69 of 180 #1769 of 1964

Problem 1769

IMO Shortlist mid-range; USAMO P2/P5
Number theory Difficulty 8.1 Prove it

Let aa, bb be positive integers such that bn+nb^n+n is a multiple of an+na^n+n for all positive integers nn. Prove that a=ba=b.

Proposed by Mohsen Jamali, Iran

This one wants a proof. Work it on paper, then read the official solution and mark yourself. Be honest about it: the record is only any use to you if it is.

Next problem →

Official solution

1. Assume for contradiction: Suppose a≠b a \neq b . Then, b−a b - a has finitely many prime divisors.
2. Lemma I: There exists an odd prime p>b>a p > b > a such that 4∤ap+p 4 \nmid a^p + p and gcd⁡(ap+p,b−a)≤2 \gcd(a^p + p, b - a) \leq 2 .

Proof of Lemma I:
- Let p p be an odd prime. If a a is even and p p is odd, then 4∤ap+p 4 \nmid a^p + p .
- If a≡±1(mod4) a \equiv \pm 1 \pmod{4} , choose p≡a(mod4) p \equiv a \pmod{4} . Thus, there exists an odd prime p p for which 4∤ap+p 4 \nmid a^p + p .
- Now, we prove the second part of Lemma I. Let q q be an odd prime divisor of b−a b - a . For all q q , if a≡1(modq) a \equiv 1 \pmod{q} , let p≡1(modq) p \equiv 1 \pmod{q} , then ap+p≡2(modq) a^p + p \equiv 2 \pmod{q} .
- Otherwise, if a≢1(modq) a \not\equiv 1 \pmod{q} , let p≡−1(modq) p \equiv -1 \pmod{q} , so ap+p≡ap−1(modq) a^p + p \equiv a^p - 1 \pmod{q} .
- Assume for contradiction that q∣ap−1 q \mid a^p - 1 . Then, either ordq(a)=1 \text{ord}_q(a) = 1 or ordq(a)=p \text{ord}_q(a) = p . In the former case, this means a≡1(modq) a \equiv 1 \pmod{q} , contradiction. In the latter case, this gives us q≡1(modp) q \equiv 1 \pmod{p} , contradiction, since p>q p > q , as p≡−1(modq) p \equiv -1 \pmod{q} .
- Hence, by Dirichlet's Theorem of Arithmetic Progression together with the Chinese Remainder Theorem, we find that there exists an odd prime p>b>a p > b > a such that gcd⁡(ap+p,b−a) \gcd(a^p + p, b - a) has no prime factors other than possibly 2 2 , and since we can choose p p such that 4∤ap+p 4 \nmid a^p + p , it follows that gcd⁡(ap+p,b−a)≤2 \gcd(a^p + p, b - a) \leq 2 , as desired. ■\blacksquare

3. Using Lemma I: Let n=p n = p , where p p is an odd prime which satisfies the conditions in Lemma I, so we have ap+p∣bp−ap a^p + p \mid b^p - a^p .
4. Prime divisor analysis: Let q q be a prime divisor of ap+p a^p + p . If q∣b q \mid b , then q∣a q \mid a , so q∣p q \mid p , which means that q=p q = p . However, p>a p > a , therefore we must have q∤a,b q \nmid a, b .
5. Order analysis: Then, q∣(ba−1)p−1 q \mid (ba^{-1})^p - 1 , so ordq(ba−1)=1 \text{ord}_q(ba^{-1}) = 1 or ordq(ba−1)=p \text{ord}_q(ba^{-1}) = p . In the former case, we obtain q∣b−a q \mid b - a , however, since gcd⁡(ap+p,b−a)≤2 \gcd(a^p + p, b - a) \leq 2 , this implies q=2 q = 2 . In the latter case, this implies q≡1(modp) q \equiv 1 \pmod{p} , so for all primes q∣ap+p q \mid a^p + p , either q=2 q = 2 or q≡1(modp) q \equiv 1 \pmod{p} , and also 4∤ap+p 4 \nmid a^p + p , hence we must have either ap+p≡1(modp) a^p + p \equiv 1 \pmod{p} or ap+p≡2(modp) a^p + p \equiv 2 \pmod{p} , or, equivalently, a≡1(modp) a \equiv 1 \pmod{p} or a≡2(modp) a \equiv 2 \pmod{p} . Since p>b>a p > b > a , this is only possible if a=1 a = 1 or a=2 a = 2 .

6. Case analysis:
- If a=1 a = 1 , then if any prime q∣b q \mid b , setting n=q−1 n = q - 1 , we obtain q∣bq−1+q−1≡−1(modq) q \mid b^{q - 1} + q - 1 \equiv -1 \pmod{q} , contradiction. Hence, b b cannot have any prime divisors, so b=1 b = 1 .
- If a=2 a = 2 , then ap+p a^p + p is odd, so gcd⁡(ap+p,b−a)=gcd⁡(2p+p,b−2)=1 \gcd(a^p + p, b - a) = \gcd(2^p + p, b - 2) = 1 . Then, we have ap+p=2p+p∣bp−ap=bp−2p a^p + p = 2^p + p \mid b^p - a^p = b^p - 2^p . Let q q be a prime divisor of 2p+p 2^p + p . Note that q q must be odd. If q∣b q \mid b , then q∣2 q \mid 2 , so q=2 q = 2 , contradiction. As a result, q∤b q \nmid b , then q∣(b⋅2−1)p−1 q \mid (b \cdot 2^{-1})^p - 1 , therefore ordq(b⋅2−1)=1 \text{ord}_q(b \cdot 2^{-1}) = 1 or ordq(b⋅2−1)=p \text{ord}_q(b \cdot 2^{-1}) = p . In the former case, this gives us q∣b−2 q \mid b - 2 , but this is impossible since gcd⁡(2p+p,b−2)=1 \gcd(2^p + p, b - 2) = 1 . In the latter case, this implies q≡1(modp) q \equiv 1 \pmod{p} , so for all primes q∣2p+p q \mid 2^p + p , q≡1(modp) q \equiv 1 \pmod{p} , so this gives us 2p+p≡1(modp) 2^p + p \equiv 1 \pmod{p} , but we also have 2p+p≡2(modp) 2^p + p \equiv 2 \pmod{p} , contradiction.

7. Conclusion: Then, the initial assumption that b≠a b \neq a must be wrong, so we find that b=a b = a .

■\blacksquare

The final answer is a=b \boxed{ a = b } .

Source: NuminaMath-1.5, licensed Apache-2.0. Statement reproduced verbatim; metadata (topic, difficulty, ordering) added by this project.