28
$\begingroup$

We now that Fermat's last theorem is true so there are not positive integer solutions to $$x^n+y^n=z^n$$ for $n\in\mathbb{N}$ and $n>2$.

But what about if $n\in\mathbb{R}$ or $n\in\mathbb{R}^+$?

$\endgroup$
  • 2
    $\begingroup$ You may find this useful: Fermat’s Last Theorem for Fractional and Irrational Exponents, College Math. J. 41 (2010), 182-185 $\endgroup$ – user940 May 18 '13 at 0:32
  • 1
    $\begingroup$ Also math.williams.edu/fermats-last-theorem-for-fractional-exponents $\endgroup$ – user940 May 18 '13 at 0:33
  • $\begingroup$ Didn't Wiles et. al. solve this in the n∈R domain ? $\endgroup$ – Marcus Anderson Mar 24 '18 at 23:45
52
$\begingroup$

Suppose $z> \max(x,y)$ then $x^0+y^0 = 2 > z^0$ but there exists some $N$ such that $x^N+y^N<z^N$. Therefore there exists some $n\in[0,N]$ satisfying $x^n+y^n=z^n$.

| cite | improve this answer | |
$\endgroup$
  • 3
    $\begingroup$ You can generalise this to show that the set of real n for which there exist integers $x, y, z$ with $x^n + y^n = z^n$ is dense in $[0,\infty)$. $\endgroup$ – Bruno Le Floch May 17 '13 at 18:51
18
$\begingroup$

Take $x=4, y=9$ and $n = 0.5$. You can solve to get $z = 25$. So this works!

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Could you give some insights on how did you find that solution? $\endgroup$ – Ambesh May 17 '13 at 14:52
  • 8
    $\begingroup$ Use any perfect square for x and y if n=0.5 And so you can also use any perfect cube for x and y if you pick n = 1/3, for example x = 27, y = 64, you find z = 343 $\endgroup$ – imranfat May 17 '13 at 15:00
  • 1
    $\begingroup$ Do you have an example for n rational and > 2? $\endgroup$ – Foon May 17 '13 at 16:39
  • $\begingroup$ No, I don't unfortunately $\endgroup$ – imranfat May 17 '13 at 17:07
  • 1
    $\begingroup$ @MarcusAnderson your proof is wrong, and the answer has been deleted. This is because FLT only seeks solutions to $x^n+y^n=z^n$ when $x,y,z\in\mathbb Z$ for $n\in\mathbb Z^+$ and $n\geq3$, and no "constraints" of the type you mention exist. $\endgroup$ – YiFan Feb 4 '19 at 9:20
9
$\begingroup$

$1782^n + 1841^n = 1922^n$ with $n \approx 11.999999995097161$

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Yes. Now I realized I can solve this even with my TI-Nspire; $solve(1782^x+1841^x=1922^x,x)$. $\endgroup$ – Ambesh May 18 '13 at 7:15
  • $\begingroup$ Dan, how is this possible? I thought there were not supposed to be integer solutions with n greater than 2? Isn't that what Fermat is about? I typed it in my calculator and it comes out evenly, is my TI rounding or something? $\endgroup$ – imranfat May 18 '13 at 16:52
  • 3
    $\begingroup$ Ah, there is rounding, never mind. 1782^12 is even, 1841^12 is odd, but 1922^12 is even. Wiles is indeed a clever man... $\endgroup$ – imranfat May 18 '13 at 17:05
  • 1
    $\begingroup$ @imranfat: It's possible because the exponent is a non-integer. It just happens to be really close to one. $\endgroup$ – Dan May 18 '13 at 22:45

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