Prove there are coprime polynomials P(x) and Q(x) with integer coefficients and a real number u>0 such that if ca−12021≤dc1010u,(ca)2020−db≤dc1010u, for some positive integers a,b,c,d, then we have bP(ca)=dQ(ca).
Solution
We shall prove a more general statement in the following lines. Proposition. Let k be a positive integer. Prove that for all pairs (c,d) of integers such that ∣c∣ and ∣d∣ are prime numbers there is a positive integer N such that the number of pairs (a,b) of integers satisfying the following inequalities is at most N, for all sufficiently small positive real numbers u,v. ∣b∣∣a−c∣2021≤u∣c∣1011,∣akb−ckd∣≤v∣c∣k−1010. Let us rewrite the original inequalities as ∣b∣ca−12021≤u∣c∣−1010 and (ca)kb−d≤v∣c∣−1010. Then if k≤1010 then akb=ckd and hence, we only have finitely many solutions for a,b. Hence, assume that k>1010. We then prove the following lemma: Lemma. There are polynomials P(x),Q(x) of degree 1010 with integer coefficients such that the polynomial P(x)−xkQ(x) has a 2021-fold zero at x=1. Proof. We shall prove that for all triples (m,n,k) of natural numbers such that n>k there are polynomials P(x)=P(m,n,k)(x), Q(x)=Q(m,n,k)(x) of degrees m,n with rational coefficients such that P(x)−xkQ(x) is divisible by (x−1)m+n+1. We shall prove this through induction, considering the following expression: x(P(m,n−1,k)(x)−xkQ(m,n−1,k)(x))+r(P(m−1,n,k)(x)−xkQ(m−1,n,k)(x)). Since both brackets are divisible by (x−1)m+n then there would be a rational number r such that the above expression is divisible by (x−1)m+n+1 then defining P(m,n,k)(x)=xP(m,n−1,k)(x)+rP(m−1,n,k)(x) and Q(m,n,k)(x)=xQ(m,n−1,k)(x)+Q(m−1,n,k)(x). Moreover, for the base case, that is when one of m,n are zero, the statement is obvious. This completes our proof.
Comment 1. If P(m,n,k)(0)=0 for some non-zero m,n then the polynomial P(m,n,k)(x)−xkQ(m,n,k)(x) has at most m+n non-zero coefficients, impossible. Lemma.P(m,n,k)(x),Q(m,n,k)(x) have no common factor. Proof. Assume for the contradiction that there is a non constant polynomial D(x) of degree d dividing both of them. Then at least (x−1)m+n+1−d divides D(x)P(m,n,k)(x)−xkD(x)Q(m,n,k)(x). Since the not-zero coefficient of this polynomial is at most m+n−2d+2. But, according to our lemma 2 it must at least have m+n+2−d non-zero coefficients. Therefore, d≤0. This completes our proof.
Now, there are coprime polynomials P(x),Q(x) of degree 1010 with integer coefficients such that P(x)−xkQ(x) is divisible by (x−1)2021 and hence there are polynomials A(x),B(x) with integer coefficients of the same degree that doesn't exceed 1010 such that A(x)P(x)+B(x)Q(x)=R,(1) for some non-zero constant R. Lemma. If a,b,c and d satisfy the statement of our problem then bP(ca)=dQ(ca). Proof. We prove the lemma for polynomial P(x)−xkQ(x), for some positive integer k≥degP(x)=degQ(x)=d. Then, the second inequality becomes ∣akb−ckd∣≤v∣c∣k−1010. Since ∣b∣≥1, take u<1 then ∣a−c∣≤∣c∣ therefore, ∣a∣≤2∣c∣. Moreover, let d=1010, and R(x)=P(x)−xkQ(x), the following quantity is an integer: cd(bP(ca)−dQ(ca))=bcdR(ca)+(akb−ckd)cd−kQ(ca). Now, if bP(ca)=dQ(ca) then, ∣cd(bP(ca)−dQ(ca))∣≥1. Thus, by triangle inequality, we find that ∣bcdR(ca)∣+(akb−ckd)cd−kQ(ca)≥1. Write R(x)=(1−x)2d+1S(x) then, bcdR(ca)=bcd(1−ca)2d+1S(ca)=c−d−1b(c−a)2d+1S(ca). Thus, ∣c−d−1b(c−a)2d+1S(ca)∣+(akb−ckd)cd−kQ(ca)≥1. That is, ∣b∣∣c−a∣2d+1∣c−d−1∣∣S(ca)∣+∣akb−ckd∣∣cd−k∣∣Q(ca)∣≥1. By use of problem's assumption, we find for all u,v: ∣u∣∣S(ca)∣+∣v∣∣Q(ca)∣≥1.
Assume that S(x)=ck−d−1xk−d−1+⋯+c0 be a polynomial with integer coefficients. Then, ∣S(ca)∣≤∑i=0k−d−1∣ci∣2i=C1 which is a constant number. Furthermore, by the same argument we can find that ∣Q(ca)∣≤C2 for some constant C2. This yields to: uC1+vC2≥1. The last inequality would be absurd for all sufficiently small u,v. We are done.
Putting x=ca in (1) and multiply both sides by c2020 it follows that c1010A(ca)c1010P(ca)+c1010B(ca)c1010Q(ca)=c2020R. Now, using lemma 4 yielding to c1010P(ca)(dc1010A(ca)+bc1010B(ca))=c2020dR. Note that if gcd(a,c)>1 then c must divide a. But then, ∣b∣∣a−c∣2021≤u∣c∣1011 becomes ∣b∣ca−1∣c∣2021≤u∣c∣−1010. Hence, it follows that a=c. Thus, assume that gcd(a,c)=1. Hence, c1010P(ca) divides c2020dR. Note that gcd(c1010P(ca),c)=gcd(c,a1010a1010)=gcd(c,a1010)=g≤∣a1010∣, where a1010 is the leading coefficient of P(x). Writing c=gf, gcd(f,a1010)=1 it follows that c1010P(ca) divides g2020dR. Since the number of divisors of Rdg2020 is at most 2∣R∣g2020≤2∣R∣a10102020, hence the total number of possibilities for a is at most 2020∣R∣a10102020. Finally for each certain a,b=P(ca)dQ(ca) would be determined uniquely. Moreover, note that P(ca)=0 because otherwise, P,Q will have a common factor. Since R only depends on the coefficients of P(x),Q(x) we can put N=2020∣R∣a10102020. ■
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 reproduced verbatim; metadata (topic, difficulty) added by this project.