Maths Olympiad Prep

Library / /222 of 462

Number theory Difficulty 5.8 AIME, harder Prove it Ireland

Let aa, bb be two distinct positive integers. Prove that the number
(a+b)3a3a2b+ab2b3 \frac{(a+b)^3}{a^3 - a^2b + ab^2 - b^3}
is not an integer.

Solution

Suppose the given fraction is equal to the integer kk, then
(a+b)3=k(a2+b2)(ab).(14) (a+b)^3 = k(a^2+b^2)(a-b). \qquad (14)
First note that replacing aa by dada and bb by dbdb does not change the value of kk, because numerator and denominator are both homogeneous of degree three. Therefore, we may assume gcd(a,b)=1\gcd(a, b) = 1. Moreover, when we swap aa and bb, the given fraction just changes its sign. Therefore, we may assume a>ba > b. Also note that a3a2b+ab2b3=(a2+b2)(ab)a^3 - a^2b + ab^2 - b^3 = (a^2 + b^2)(a - b). In particular, ab(a+b)3a - b \mid (a + b)^3. Because gcd(ab,a+b)=gcd(ab,2a)=gcd(ab,2b)\gcd(a - b, a + b) = \gcd(a - b, 2a) = \gcd(a - b, 2b) we distinguish two cases according to aba - b being even or odd.

Case 1: If aba - b is odd, gcd(ab,2b)=gcd(ab,b)=gcd(a,b)=1\gcd(a - b, 2b) = \gcd(a - b, b) = \gcd(a, b) = 1, hence gcd(ab,a+b)=1\gcd(a - b, a + b) = 1 and we can only have ab(a+b)3a - b \mid (a + b)^3 if ab=1a - b = 1. Substituting a=b+1a = b + 1 in (14), we obtain
(2b+1)3=k(2b2+2b+1). (2b + 1)^3 = k(2b^2 + 2b + 1).
Because (2b+1)3=8b3+12b2+6b+1=(4b+2)(2b2+2b+1)(2b+1)(2b + 1)^3 = 8b^3 + 12b^2 + 6b + 1 = (4b + 2)(2b^2 + 2b + 1) - (2b + 1), we see that 2b2+2b+12b^2 + 2b + 1 can only divide (2b+1)3(2b + 1)^3 when it also divides 2b+12b + 1, but this is impossible for b>0b > 0 because 2b2+2b+1>2b+12b^2 + 2b + 1 > 2b + 1.

Case 2: If aba - b is even, gcd(ab,2b)=2d\gcd(a - b, 2b) = 2d for some positive integer dd which divides bb and aba - b. Hence dd divides gcd(ab,b)=gcd(a,b)=1\gcd(a - b, b) = \gcd(a, b) = 1, i.e. d=1d = 1. Therefore, gcd(ab,a+b)=2\gcd(a - b, a + b) = 2 in this case. This implies that no prime p>2p > 2 can divide aba - b, since otherwise pp would divide a+ba + b as well according to (14). Hence, ab=2ma - b = 2^m for some m1m \ge 1. It is not hard to show that mm can only be 1, 2 or 3, but we don't need this later and therefore skip the proof.

On the other hand, aa and bb are both odd, because they are coprime and have the same parity in the current case.
When we divide (a+b)3(a+b)^3 by a2+b2a^2+b^2, we find
(a+b)3=a3+3a2b+3ab2+b3=(a+3b)(a2+b2)2(ab)b2. (a+b)^3 = a^3 + 3a^2b + 3ab^2 + b^3 = (a+3b)(a^2+b^2) - 2(a-b)b^2.

Together with equation (14) this implies that a2+b2a^2 + b^2 divides 2(ab)b22(a - b)b^2. If we substitute a=b+2ma = b + 2^m, where m1m \ge 1, we obtain that
a2+b2=2b2+2m+1b+22mdivides2(ab)b2=2m+1b2. a^2 + b^2 = 2b^2 + 2^{m+1}b + 2^{2m} \quad \text{divides} \quad 2(a-b)b^2 = 2^{m+1}b^2.
However, because m1m \ge 1 and bb is odd, the term inside the bracket
2b2+2m+1b+22m=2(b2+2mb+22m1) 2b^2 + 2^{m+1}b + 2^{2m} = 2(b^2 + 2^m b + 2^{2m-1})
is odd, hence must divide b2b^2, which is impossible since b2+2mb+22m1>b2b^2 + 2^m b + 2^{2m-1} > b^2. We have now established in both cases that such positive integers aa, bb do not exist.

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.