HMMT February

Break all possible values of $n$ into the four cases: $n=2, n=4, n>4$ and $n$ odd. By Fermat's theorem, no solutions exist for the $n=4$ case because we may write $y^{4}+\left(2^{25}\right)^{4}=x^{4}$. We show that for $n$ odd, no solutions exist to the more general equation $x^{n}-y^{n}=2^{k}$ where $k$ is a positive integer. Assume otherwise for contradiction's sake, and suppose on the grounds of well ordering that $k$ is the least exponent for which a solution exists. Clearly $x$ and $y$ must both be even or both odd. If both are odd, we have $(x-y)\left(x^{n-1}+\ldots+y^{n-1}\right)$. The right factor of this expression contains an odd number of odd terms whose sum is an odd number greater than 1, impossible. Similarly if $x$ and $y$ are even, write $x=2 u$ and $y=2 v$. The equation becomes $u^{n}-v^{n}=2^{k-n}$. If $k-n$ is greater than 0, then our choice $k$ could not have been minimal. Otherwise, $k-n=0$, so that two consecutive positive integers are perfect $n$th powers, which is also absurd. For the case that $n$ is even and greater than 4, consider the same generalization and hypotheses. Writing $n=2 m$, we find $\left(x^{m}-y^{m}\right)\left(x^{m}+y^{m}\right)=2^{k}$. Then $x^{m}-y^{m}=2^{a}<2^{k}$. By our previous work, we see that $m$ cannot be an odd integer greater than 1. But then $m$ must also be even, contrary to the minimality of $k$. Finally, for $n=2$ we get $x^{2}-y^{2}=2^{100}$. Factoring the left hand side gives $x-y=2^{a}$ and $x+y=2^{b}$, where implicit is $a<b$. Solving, we get $x=2^{b-1}+2^{a-1}$ and $y=2^{b-1}-2^{a-1}$, for a total of 49 solutions. Namely, those corresponding to $(a, b)=(1,99),(2,98), \cdots,(49,51)$.