3
$\begingroup$

Suppose that $(k,l,m) \in \mathbb{N_0}^3$.

If $(k,l,m)=(0,0,0)$ then for $n=1,2$ there is an infinite number of solutions and, by the theorem of of Wiles there are no solutions when $n \geq 3$.

Is it known can there be an infinite number of solutions of $a^{n+k}+b^{n+l}=c^{n+m}$ if $k,l,m$ are not all equal and $n \geq 3$?

What is known of when there is a finite non-zero number of solutions?

Can someone summarize all that is known about this generalized Fermat problem?

We see that a mapping $m: (k,l,m) \to a^{n+k}+b^{n+l}=c^{n+m}$ is a mapping that sends some element of $\mathbb{N_0}^3$ to a family of equations over the integers and because every such equation has its own set of solutions for every $n \in \mathbb N$ we can define some mapping $s$ that sends some element of $\mathbb{N_0}^3$ to a set of all solutions.

A question can then be phrased of when is $s(k,l,m)$ an empty set and when it is not an empty set.

We could probably somehow lower "dimensionality" of this problem by choosing some $(k,l,m)$ and fixing either $a$ or $b$ or $c$ but I am not sure how general results could be obtained in such a way.

Also, I do not know are techniques of Wiles suited for this more general problem, and, if they are not, I do not know why they are not?

It would be nice if someone could explain in non-strict technical sense all that is known about this generalized Fermat problem.

If this is off-topic and written in non-professional way then please vote to close this question of mine.

$\endgroup$
1
  • 3
    $\begingroup$ It is common to insist in this sort of problem that $a,b,c$ be relatively prime (as GH from MO does in his answer). Without that insistence, you can make as many solutions as you want. E.g., from $3^3+2^4=43$, we can get $a^3+b^4=c^5$ with $a=3\times43^8$, $b=2\times43^6$, $c=43^5$. $\endgroup$ Apr 5, 2018 at 3:55

1 Answer 1

5
$\begingroup$

Darmon and Granville proved, using Faltings' Theorem, that your equation has finitely many primitive integer solutions for any fixed exponents which are at least $3$. In fact their result is more general. For several concrete exponents beyond Fermat's Last Theorem, the full set of primitive integer solutions is also known, see e.g. the papers of Siksek-Stoll and Anni-Siksek.

You can find more information in the Wikipedia article on Beal's conjecture and the references therein. See also this survey, especially Section 4.5.

$\endgroup$
2
  • $\begingroup$ I note that OP never specifies primitivity. OP might be interested in knowing that if $a,b,c$ are permitted to have common factors, then there are trivial ways to produce lots of solutions. $\endgroup$ Apr 4, 2018 at 23:31
  • 1
    $\begingroup$ @GerryMyerson: I agree. Please mention this comment under the original post as well (so that the OP is notified about it). $\endgroup$
    – GH from MO
    Apr 5, 2018 at 1:23

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.