5
$\begingroup$

I would appreciate if someone could check my attempt in proving the Fermat's Last Theorem for prime exponents greater than $2$.

Firstly, let's prove a couple of lemmas which state that sum or difference of two non-equal odd $p$-th powers cannot be a $p$-th power for prime $p>2$.

Lemma 1:
The difference of two $p$-th powers of non-equal odd positive integers is not a $p$-th power of an integer for prime $p>2$.

Take prime $p>2$, and odd $z,x \in\Bbb Z^+, \; z>x$. Let $a=z-x$. Obviously $a$ is even. $$z=a+x$$ $$z^{\ p}=\sum_{i=0}^p \binom{p}{i}a^{\ p-i}x^{\ i}=x^{\ p}+a\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i}$$ $$z^{\ p}-x^{\ p}=a\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i}$$ Assume that the right side is a $p$-th power of an integer.


**** And here comes my erroneous logic: ****
:(

Then, since $x$ is not divisible by $a$ (because $x$ is odd, and $a$ is even), it must be: $$\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i} \equiv 0\;(mod\;a^{\ p-1})$$


$$\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i} \equiv 0\;(mod\;a)$$ $$px^{\ p-1}+\sum_{i=0}^{p-2} \binom{p}{i}a^{\ p-1-i}x^{\ i} \equiv 0\;(mod\;a)$$ $$px^{\ p-1}+a\sum_{i=0}^{p-2} \binom{p}{i}a^{\ p-2-i}x^{\ i} \equiv 0\;(mod\;a)$$ $$px^{\ p-1} \equiv 0\;(mod\;a)$$ However, since both $p>2$ and $x$ are odd, then $px^{\ p-1}$ is also odd, so it cannot be divisible by $a$ which is even. We reached a contradiction so Lemma 1 is proven.

Lemma 2:
The sum of two $p$-th powers of non-equal odd positive integers is not a $p$-th power of an integer for prime $p>2$.

Proof for the sum of two odd $p$-th powers not being a $p$-th power is almost the same. Start with odd $x,y \in\Bbb Z^+$, set $b=x+y$, and raise $x=b-y$ to the p-th power. If the sum $x^{\ p}+y^{\ p}$ were the p-th power, then $py^{\ p-1}$ would have to be divisible by $b$, which cannot be because $py^{\ p-1}$ is odd, while $b$ is even.

Fermat's Last Theorem: (from Wikipedia)

No three positive integers $x, y$, and $z$ can satisfy the equation $x^n+y^n=z^n$ for any integer value of $n$ greater than two.

Proof of the Fermat's Last Theorem for prime exponents greater than 2:

To prove the Fermat's Last Theorem for prime exponents greater than $2$, we assume that there are such $x,y,z$ and then rewrite all $3$ possible types of sums of odd and even $p$-th powers in such a way to get to a contradiction with the one of the above Lemmas.

Assume that for prime $p>2$ there exist $x,y,z \in\Bbb Z^+$, all three different, for which $$x^{\ p}+y^{\ p}=z^{\ p}$$

A) Assume that $x$ is even and $y$ is odd, both positive integers.

First note that then $z$ must be odd too.

We can rewrite it as $$z^{\ p}-y^{\ p}=x^{\ p}$$ however, since both $z$ and $y$ are odd then this is in contradiction with Lemma 1.

Therefore, combinations where one of $x,y$ is even and the other is odd are not possible.

B) Assume that both $x$ and $y$ are odd positive integers, then $x^{\ p}+y^{\ p}=z^{\ p}$ is in direct contradiction with Lemma 2.

Therefore, combinations where both $x$ and $y$ are odd are not possible.

C) Assume both $x$ and $y$ are even, then we can repeatedly divide them with $2$, until one, or both of them gets odd, which reduce this case to A) or B).

Therefore, combinations where both $x,y$ are even are also not possible.

Since neither of A), B), or C) hold, then there are no such $x,y,z \in Z^+$ to satisfy the equation $x^{\ p}+y^{\ p}=z^{\ p}$ for prime $p>2$ and $x,y,z$ all different.

$\endgroup$
4
  • $\begingroup$ Why would $z^p-x^p\equiv0\bmod a^{p}$ even if it were a $p$-th power? $\endgroup$ – AdLibitum Jul 28 '15 at 15:03
  • $\begingroup$ Even if $z^p-x^p$ is a $p$th power and divisible by $a=z-x$, it does not follow that it would be divisible by $a^p$. For example $41^2-9^2=40^2$ is divisible by $41-9=32$, but it is not divisible by $32^2$. You only used $p>2$ later - I'm using $p=2$ here to show that the logic breaks down here. By Wiles' theorem counterexamples with $p>2$ don't exist so this is just for explaining my point :-) $\endgroup$ – Jyrki Lahtonen Jul 28 '15 at 15:04
  • $\begingroup$ @AdLibitum I think he/she assumed that $a$ is prime. $\endgroup$ – Marcus M Jul 28 '15 at 15:05
  • $\begingroup$ Thanks guys! I appreciate it. :) $\endgroup$ – cod3r Jul 28 '15 at 15:45
9
$\begingroup$

The error is here:

Then, since $x$ is not divisible by $a$ (because $x$ is odd, and $a$ is even), it must be: $$\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i} \equiv 0\;(mod\;a^{\ p-1})$$

If you write $a\sum_{i=0}^{p-1} \binom{p}{i}a^{\ p-1-i}x^{\ i} = y^p$, as you do, then we have that $a | y^p$. This does not imply that $a^p | y^p$, as $a$ is not necessarily prime. For instance $12$ divides $36 = 6^2$ but $12$ does not divide $6$.

$\endgroup$
1
  • $\begingroup$ See my answer to Ramunjndscpl's question on "FLT simple proof". $\endgroup$ – nguyen quang do Dec 27 '15 at 8:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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