Maths Olympiad Prep

Track / Stage 8 / 2 of 180 #1702 of 1964

Problem 1702

IMO Shortlist mid-range; USAMO P2/P5
Number theory Difficulty 8.0 Prove it

Let ff and gg be two nonzero polynomials with integer coefficients and degf>degg\deg f>\deg g. Suppose that for infinitely many primes pp the polynomial pf+gpf+g has a rational root. Prove that ff has a rational root.

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.

Official solution

1. Define the Polynomials and Rational Roots:
Let f(x)=anxn++a0 f(x) = a_n x^n + \cdots + a_0 and g(x)=bmxm++b0 g(x) = b_m x^m + \cdots + b_0 . We are given that for infinitely many primes p p , the polynomial pf+g pf + g has a rational root. Let rp=upvp r_p = \frac{u_p}{v_p} be a rational root of pf+g pf + g for some coprime integers up u_p and vp v_p .

2. Boundedness of Rational Roots:
Since degg<degf\deg g < \deg f, there exists a constant C C such that for xC|x| \geq C, g(x)f(x)<0.01\left| \frac{g(x)}{f(x)} \right| < 0.01. Given pf(rp)+g(rp)=0 pf(r_p) + g(r_p) = 0 , we have g(rp)f(rp)=p\left| \frac{g(r_p)}{f(r_p)} \right| = p . Since p>0.01 p > 0.01 , it follows that rp<C|r_p| < C.

3. Rational Root Theorem Application:
By the Rational Root Theorem, vppan v_p \mid pa_n and uppa0+b0 u_p \mid pa_0 + b_0 . Therefore, for each valid p p , one of the following must hold:
- pvp p \mid v_p (call such primes peasants), or
- vpan v_p \mid a_n (call such primes rebels).

4. Infinitely Many Peasant Primes:
Suppose finitely many valid p p are peasants. Let S S be the set of all valid rebel primes p p . Since vpan v_p \mid a_n , there are finitely many possible values for vp v_p . For each vp v_p , there are finitely many possible values for up u_p because rp=upvp|r_p| = \left| \frac{u_p}{v_p} \right| is bounded. Thus, there are finitely many possible values for rp r_p over all pS p \in S . This contradicts the infinitude of valid p p since p=g(rp)f(rp) p = -\frac{g(r_p)}{f(r_p)} .

5. Case Analysis:
Consider a sufficiently large peasant prime p p such that p p is larger than all ai a_i and bi b_i . Since rp=upvp r_p = \frac{u_p}{v_p} , we have:
(pf+g)(rp)=i=0npai(upvp)i+j=0mbj(upvp)j=0 (pf + g)(r_p) = \sum_{i=0}^n pa_i \left( \frac{u_p}{v_p} \right)^i + \sum_{j=0}^m b_j \left( \frac{u_p}{v_p} \right)^j = 0
Multiplying through by vpn v_p^n gives:
i=0npaiupivpni+j=0mbjupjvpnj=0() \sum_{i=0}^n pa_i u_p^i v_p^{n-i} + \sum_{j=0}^m b_j u_p^j v_p^{n-j} = 0 \quad (\heartsuit)
Taking ()modp2(\heartsuit) \mod p^2, since pvp p \mid v_p , the first term reduces to panupnmodp2 pa_n u_p^n \mod p^2 , and the second term reduces to bmupmvpnmmodp2 b_m u_p^m v_p^{n-m} \mod p^2 .

6. **Case 1: nm+2 n \geq m + 2 :**
bmupmvpnm0(modp2) b_m u_p^m v_p^{n-m} \equiv 0 \pmod{p^2}
Thus, panupn0(modp2) pa_n u_p^n \equiv 0 \pmod{p^2} , implying anupn0(modp) a_n u_p^n \equiv 0 \pmod{p} . Since p>an p > a_n and gcd(up,vp)=1\gcd(u_p, v_p) = 1, this is impossible.

7. **Case 2: n=m+1 n = m + 1 :**
panupn+bn1upn1vp0(modp2) pa_n u_p^n + b_{n-1} u_p^{n-1} v_p \equiv 0 \pmod{p^2}
Since pup p \nmid u_p , we can divide and reduce:
anup+bn1(vpp)0(modp) a_n u_p + b_{n-1} \left( \frac{v_p}{p} \right) \equiv 0 \pmod{p}
Let dp=vpp d_p = \frac{v_p}{p} . Then:
updpbn1an(modp) \frac{u_p}{d_p} \equiv -\frac{b_{n-1}}{a_n} \pmod{p}
Hence, updp=C+pZpan\frac{u_p}{d_p} = C + \frac{pZ_p}{a_n} for some integer Zp Z_p . Since rp=upvp=1pupdp r_p = \frac{u_p}{v_p} = \frac{1}{p} \frac{u_p}{d_p} , we have:
rp=Cp+Zpan r_p = \frac{C}{p} + \frac{Z_p}{a_n}
Since rp|r_p| is bounded, there exists some integer Z Z such that Zp=Z Z_p = Z for infinitely many p p . Thus:
rp=Cp+Zan r_p = \frac{C}{p} + \frac{Z}{a_n}
As p p \to \infty , rpZan r_p \to \frac{Z}{a_n} , a fixed rational number.

8. Contradiction:
Since p=g(rp)f(rp) p = -\frac{g(r_p)}{f(r_p)} and rpZan r_p \to \frac{Z}{a_n} , but f f has no rational roots, we have f(Zan)0 f\left( \frac{Z}{a_n} \right) \neq 0 . Thus:
g(rp)f(rp)g(Zan)f(Zan) -\frac{g(r_p)}{f(r_p)} \to -\frac{g\left( \frac{Z}{a_n} \right)}{f\left( \frac{Z}{a_n} \right)}
This is a fixed number, contradicting the infinitude of valid p p .

Therefore, f f must have a rational root.

\blacksquare

Source: NuminaMath-1.5, licensed Apache-2.0. Statement and solution reproduced as published; topic, difficulty and ordering added by this site.