Maths Olympiad Prep

Library / /208 of 397

Number theory Difficulty 5.9 AIME, harder Prove it Taiwan

For an integer pair (a,b)(a,b), let P(x)=ax3+bxP(x) = ax^3 + bx. If for some positive integer mm the following statement holds:
If integers x,yx, y satisfy that mm divides P(x)P(y)P(x) - P(y), then mm divides xyx - y.
then the pair (a,b)(a,b) is called "mm-sufficient". If there exist infinitely many positive integers kk such that (a,b)(a,b) is "kk-sufficient", then the pair (a,b)(a,b) is called "very sufficient". Question: does there exist a pair (a,b)(a,b) such that (a,b)(a,b) is "1110-sufficient", but not "very sufficient"?

Solution

The answer is no! We will prove below that if (a,b)(a,b) is "1110-sufficient", then for any k=37rk = 37^r, (a,b)(a,b) is kk-sufficient.

(1) If (a,b)(a,b) is 1110-sufficient, then (a,b)(a,b) is 37-sufficient.
Suppose (a,b)(a,b) is not 37-sufficient, then there exist x,yx, y such that 37P(x)P(y)37 \mid P(x) - P(y) but xyx - y is not a multiple of 37. By the Chinese Remainder Theorem, there exist integers x,yx', y' such that both xx' and yy' are multiples of 30, and 37xx37 \mid x' - x, 37yy37 \mid y' - y. Thus we have 30P(x)P(y)30 \mid P(x') - P(y') and 37P(x)P(y)37 \mid P(x') - P(y'), that is, 1110P(x)P(y)1110 \mid P(x') - P(y'). Hence 1110 does not divide xyx' - y', giving a contradiction.

(2) If (a,b)(a,b) is 37-sufficient, then 37a37 \mid a.
By contradiction, suppose aa is not a multiple of 37. First, if bb is also not a multiple of 37, note that P(x)P(y)=(a(x2+xy+y2)+b)(xy)P(x) - P(y) = (a(x^2 + xy + y^2) + b)(x - y), and 4a(x2+xy+y2)+4b=a(2x+y)2+3ay2+4b4a(x^2 + xy + y^2) + 4b = a(2x + y)^2 + 3a y^2 + 4b. Let the sets
S={as2+b mod 37s is an integer},T={3at2 mod 37t is an integer}, S = \{ a s^2 + b \text{ mod } 37 \mid s \text{ is an integer} \}, \quad T = \{ -3a t^2 \text{ mod } 37 \mid t \text{ is an integer} \},
where the residues here are all taken between 0 and 36. Then both SS and TT have 19 elements, so there must be overlap between them; that is, there exist integers s,ts, t such that 37as2+3at2+b37 \mid a s^2 + 3a t^2 + b.
Since 37 does not divide bb, ss and tt are not both 0, so we may take t=tt' = t or t-t, such that 3ts3t' - s is not a multiple of 37. Take an integer ss' such that 2s+ts(mod37)2s' + t' \equiv s \pmod{37}, then
4a(s2+st+t2)+4bas+3t2+b0(mod37), 4a(s'^2 + s' t' + t'^2) + 4b \equiv a s + 3 t^2 + b \equiv 0 \pmod{37},
and since 3ts3t' - s is not a multiple of 37, 3t(2s+t)3t' - (2s' + t') is not a multiple of 37, that is, ss' and tt' have different residues mod 37. However we have 37(a(s2+st+t2)+b)(st)=P(s)P(t)37 \mid (a(s'^2 + s' t' + t'^2) + b)(s' - t') = P(s') - P(t'), a contradiction.
If bb is a multiple of 37, then P(10)P(1)=999a+9bP(10) - P(1) = 999a + 9b is a multiple of 37, also a contradiction, so (2) is proved.

(3) For any positive integer rr, (a,b)(a, b) is 37r37^r-sufficient.
By (2), 37a37 \mid a. Clearly bb cannot be a multiple of 37, otherwise 37P(1)P(0)37 \mid P(1) - P(0), a contradiction. For any integers x,yx, y, if 37rP(x)P(y)=(a(x2+xy+y2)+b)(xy)37^r \mid P(x) - P(y) = (a(x^2 + xy + y^2) + b)(x - y), then note that a(x2+xy+y2)+bb(mod37)a(x^2 + xy + y^2) + b \equiv b \pmod{37} is not a multiple of 37, hence 37rxy37^r \mid x - y, which completes the proof.

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 translated into English from zh; metadata (topic, difficulty) added by this project.