# Fermat's Last Theorem - Variation with arithmetically descending exponents

Are there solution(s) to the following variant of Fermat's Last Theorem in the positive integers?

$$a^n + b^{n-i} = c^{n-2i}$$

I haven't been able to identify any trivial solutions. To my (mathematically untrained) eye, it seems obvious that at least $n > 2i$ must hold.

Zev Chonoles noted in early commenting that $n=3$, $i=1$, and $a,b$ arbitrary provides a class of trivial solutions and, indeed, such trivial solutions exist for any $n-2i=1$. So! Are there any solutions to the above equation in the positive integers, where $n-2i>1$?

The above was inspired by this question, and in particular the last portion of Ypnypn's answer to it:

We can also note that $$\sum_{k=1}^{z-1}{a^k} = {a^z-1\over a-1} < a^z$$

Meaning, even if we added all the exponents of $a$ less than $z$, we still wouldn't reach $a^z$.

Essentially, it pointed to the question, "What could one reach with sums of exponents, if not 'all the way to $a^z$'"? Any corresponding variant of the original linked question is just a subset of the problem posed in that original question, so I went straight to the above form.

• There's plenty of trivial solutions with $n=3$, $i=1$, and $a,b$ arbitrary. – Zev Chonoles Sep 3 '15 at 16:26
• @ZevChonoles Ha! Indeed. Are there trivial solutions any time $n-2i=1$? – hBy2Py Sep 3 '15 at 16:29
• Yup, that seems to be the case. Of course, it's your prerogative to disallow such cases. – Zev Chonoles Sep 3 '15 at 16:30
• @ZevChonoles Yep - this is what tends to happen when I post on Math.SE -- it takes several iterations to iron out the trivialities that actual mathematicians see right away. – hBy2Py Sep 3 '15 at 16:32
• @ZevChonoles What think ye now? – hBy2Py Sep 3 '15 at 16:37

Assume $\gcd(n,2i)=1$. Given any $x$, $y$, let $x^n+y^{n-i}=w$. Then for any $r$ we have $$x^nw^{rn(n-i)}+y^{n-i}w^{rn(n-i)}=w^{1+rn(n-i)}$$ which is to say $$(xw^{r(n-i)})^n+(yw^{rn})^{n-i}=w^{1+rn(n-i)}$$ Since $n$ (and, thus, $n-i$) is prime to $n-2i$, we can choose $r$ such that $1+rn(n-i)=(n-2i)s$ for some integer $s$, and so we have a solution of $a^n+b^{n-i}=c^{n-2i}$.
Much more interesting would be to find solutions with $a,b,c$ pairwise coprime. I suspect that all the work that has been done on the general $a^r+b^s=c^t$ since Wiles-Taylor has already settled this problem.
Well, the simplest one I could find in a few minutes was $100^4 + 500^3 = 15000^2$.
• $1^4+2^3=3^2$ is simpler. – Julián Aguirre Sep 3 '15 at 16:56
• Another example: $n=6$, $i=2$, $3^6+6^4=45^2$. – Julián Aguirre Sep 3 '15 at 17:07