Let $a, b, c, n$ be four integers, where n$\ge 2$, and let $p$ be a prime dividing both $a^2+ab+b^2$ and $a^n+b^n+c^n$, but not $a+b+c$. for instance, $a \equiv b \equiv -1 (mod \,\, 3), c \equiv 1 (mod \,\, 3), n$ a positive even integer, and $p = 3$ or $a = 4, b = 7, c = -13, n = 5$, and $p = 31$ satisfy these conditions. Show that $n$ and $p - 1$ are not coprime.
Problem
Source: Danube 2013 p2
Tags: number theory, coprime, prime, divides
kaede
28.02.2020 02:57
Lemma
Let $n >1$ be a positive integer relatively prime to $6$.
Then there exists $f( x,y) \in \mathbb{Z}[ x,y]$ such that
$( x+y)^{n} -\left( x^{n} +y^{n}\right) =f( x,y)\left( x^{2} +xy+y^{2}\right)$.
proof of the lemmaLet $g( x) =\left( x^{n} +1\right) -( x+1)^{n}$.
It is sufficient to show that $g( \omega ) =0$ where $\omega $ is a primitive third root of unity.
Note that $g( \omega ) =\left( \omega ^{n} +1\right) -( \omega +1)^{n} =\omega ^{n} +1-\left( -\omega ^{2}\right)^{n} =\omega ^{2n} +\omega ^{n} +1$.
Since $\left\{\omega ^{2n} ,\omega ^{n}\right\} =\left\{\omega ^{2} ,\omega \right\}$, we have $\omega ^{2n} +\omega ^{n} +1=\omega ^{2} +\omega +1=0$.
So we must have $g( \omega ) =0$.
$p\mid a^{n} +b^{n} +c^{n} \cdots ( \star )$
$p\mid a^{2} +ab+b^{2} \cdots ( \triangledown )$
Assume that $p\mid \gcd( a,b)$.
From $( \star )$ and the assumption, we have $p\mid c$, which implies $p\mid a+b+c$.
So we must have $p\nmid \gcd( a,b) \ \cdots ( \clubsuit )$.
From $( \triangledown )$ and $( \clubsuit )$, we have $6\mid p-1\ \cdots ( \spadesuit )$
If $\gcd( n,6) >1$, then $\gcd( p-1,n) >1$ due to $( \spadesuit )$.
Suppose, henceforth, $n$ is relatively prime to $6$.
From the lemma, there exists $f( x,y) \in \mathbb{Z}[ x,y]$ such that
$( a+b)^{n} -\left( a^{n} +b^{n}\right) =f( a,b)\left( a^{2} +ab+b^{2}\right) \cdots ( \neptune )$.
From $( \neptune )$ and $( \triangledown )$, we have $( a+b)^{n} \equiv a^{n} +b^{n}\pmod p\ \cdots ( \heartsuit )$.
From $( \heartsuit )$ and $( \star )$, we have $p\mid ( a+b)^{n} +c^{n} \cdots ( \diamondsuit )$.
From $( \diamondsuit )$, we have $p\mid ( a+b)^{2n} -c^{2n} \cdots ( \twonotes )$.
From $( \diamondsuit )$ and $p\nmid a+b+c$, we have $p\nmid c$ and $p\nmid a+b\ \cdots ( \leftmoon )$.
Assume, for the sake of contradiction, $\gcd( p-1,n) =1$.
Since $\gcd( p-1,n) =1$, we have $\gcd( p-1,2n) =2\ \cdots ( \uranus )$
From $( \leftmoon )$, we have $p\mid ( a+b)^{p-1} -c^{p-1} \cdots ( \earth )$.
From $( \twonotes )$ and $( \earth )$, we have $p\mid \gcd\left(( a+b)^{p-1} -c^{p-1} ,( a+b)^{2n} -c^{2n}\right)$
Note that $\gcd\left(( a+b)^{p-1} -c^{p-1} ,( a+b)^{2n} -c^{2n}\right) =( a+b)^{2} -c^{2}$ due to $( \uranus )$.
So we must have $p\mid ( a+b-c)( a+b+c)$, which implies $a+b\equiv c\pmod p\ \cdots ( \eighthnote )$.
From $( \eighthnote )$ and $( \heartsuit )$, we have $p\mid a^{n} +b^{n} -c^{n} \ \cdots ( \flat )$.
From $( \flat )$ and $( \star )$, we have $p\mid 2c^{n}$, which implise $p\mid c$.
This is clearly impossible due to $( \leftmoon )$.
$\blacksquare $
rocketscience
28.02.2020 06:33
@above thanks for the solution. I have added many latex symbols to my arsenal.