I am currently reading Ivanovs „Easy as Pi?“ in fact, I am trying to understand a proof in this book. Following statement is to prove:
For $n\geqslant 3$ the curve $x^n+y^n=1$ has no rational parametrization.
The proof goes by contradiction. For this we look at three relatively prime real polynomials $p(t),q(t),r(t)$, not all constant; so that
$(\frac{p(t)}{r(t)})^n+ (\frac{q(t)}{r(t)})^n=1$
this is equivalent to:
$p^n(t)+q^n(t)-r^n(t)=0$
by taking the derivative we derive:
$p‘(t)p^{n-1}(t)+ q‘(t)q^{n-1}(t) \ +$ $r‘(t)r^{n-1}(t)=0 $
If we now look at the system:
$p(t)x+q(t)y+r(t)z=0$ $p‘(t)x+q‘(t)y+r‘(t)z=0$
We see $x= p^{n-1}(t),y= q^{n-1}(t), z= r^{n-1}(t)$ is a solution.
Now comes the part, that I don‘t understand, from now I will write the functions without the argument
Since every solution of this system is proportional to the triple,(which is also a solution):
$(q‘r-qr‘,r‘p-rp‘,p’q-pq‘)$
We find relatively prime polynomials $f,g$, so that
$p^{n-1}=\frac{f}{g}(q‘r-qr‘)$ $q^{n-1}=\frac{f}{g}(r‘p-r‘p)$ $r^{n-1}=\frac{f}{g}(p‘q-pq‘)$
Why is every solution proportional to the triple?
My inital thought was that, if you think about a linear system of equations, the solutions to it create a vector subspace. But this can‘t be the reason, since we have multiplication of polynomials.
The Answer is quite simple, just don‘t overcomplicate it.
Since,
solves the system, and
does aswell, we know that
and
Since for some arbitrary $s\in\mathbb{R}[x]$ the triple
also solves the system (multiply $s\neq0$ on both sides). We know for a fact that there are relatively prime $f,g\in\mathbb{R}[x]$, so that:
$(\frac{f}{g}(q‘r-qr‘),\frac{f}{g}(r‘p-rp‘), \frac{f}{g}(p’q-pq‘)) = (p^{n-1},q^{n-1},r^{n-1})$,
since both of them are solutions, this also works for every other solution.