1
$\begingroup$

Are there coprime integer solutions to: $$ \frac{x^n \pm y^n}{x \pm y}=z^m $$ with $n>5 , m>1$ and excluding $z=0$?

I suppose the abc conjecture implies finitely many solutions.

$\endgroup$
1
  • $\begingroup$ I suppose you also want to rule out $x=y=z=1$. $\endgroup$ Jul 7, 2011 at 12:26

1 Answer 1

5
$\begingroup$

I suspect this question is very difficult to answer without additional hypotheses. The case $y=1$ is the classic Nagell-Ljunggren equation, where, unlike the (on the surface) very similar equation of Catalan, we do not even know whether there exist finitely many solutions in the variables $(x,z,n,m)$. If we fix $m=2$ and $n > 5$ prime, say, then it is still a substantial problem to solve the corresponding equation; Ivorra [Dissertationes Math. 444 (2007)] treats the cases $n \in \{ 7, 11, 13, 17 \}$ via elliptic Chabauty.

$\endgroup$
3
  • $\begingroup$ @Mike Thank you. Is the case $n=5, m=2$ clear? It appears to be a quartic model of elliptic curve for fixed y. $\endgroup$
    – joro
    Jul 8, 2011 at 10:05
  • $\begingroup$ This curve has positive rank and (I think) conductor $200$, if memory serves, so there are infinitely many solutions to the corresponding equation.... $\endgroup$ Jul 8, 2011 at 12:29
  • $\begingroup$ For $n=5,m=2$, $\langle x,y,z \rangle =\langle 3k, k, 11k^2\rangle$ is a family of solutions (with both $\pm$ being $-$), but I suppose we are to assume that $\gcd(x,y)=1$. $\endgroup$ Jul 8, 2011 at 23:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.