Let $p$ be a prime number, and let $K = \mathbb{F}_p$. Show that in the field $K (x)$ of rational functions over $K$, the element $x$ has no $p$th root.

I am having trouble understanding what $x$ is. I know that it is just an indeterminate
and that $x$ is transcendental over $K$ but how do I show $x$ has no $p$th root?

I did something similar to this but without the field of fractions. What if I let $f(t) = t^p-x$ where $t$ is just another indeterminate? If $f$ is irreducible then $x$ has no $p$-th root. Am I allowed to use Eisenstein's criterion on $f$? Sadly, I only know the proof of Eisenstein's Criterion over the field of rational numbers. But I've read about a more generalized criterion. Is $x$ a prime element?


It seems you don't have any trouble understanding what $x$ is: you know it's an indeterminate that is transcendental over $\Bbb F_p$. You're correct that Eisenstein's criterion works: since $(x)$ is a prime ideal of the ring $\Bbb F_p[x]$, the polynomial $t^p-x\in \Bbb F_p(x)[t]$ has no roots in $\Bbb F_p(x)$ by the criterion.

Alternately, if $f(x)\in\Bbb F_p(x)$ then $f(x)^p=f(x^p)$ is rational in $x^p$ by little Fermat, whereas clearly $x$ is not rational in $x^p$ so the equation $f(x)^p=x$ is impossible.

  • $\begingroup$ How do you know $(x)$ is a prime ideal? $\endgroup$ – abe May 27 '14 at 4:32
  • 1
    $\begingroup$ @abe: It's one of the easiest ideals to prove prime of all. Have you tried? Do you know what a prime ideal is? The only things not in $(x)$ are constants, and multiplying two constants gives another constant. Equivalently, $F[x]/(x)\cong F$ is a field whenever $F$ is. $\endgroup$ – blue May 27 '14 at 4:33
  • $\begingroup$ Oh Yeah I recall this, Thank You :D $\endgroup$ – abe May 27 '14 at 4:36
  • $\begingroup$ The factor ring $R/I$ is an integral domain if and only if $I$ is a prime ideal of $R$. $\endgroup$ – abe May 27 '14 at 4:38

I think this argument should do it, but maybe I am missing something:

For a contradiction, suppose there are two non-zero polynomials $f,g \in k[x]$ such that $(\frac{f}{g})^p=x$. Then $f^p=g^px$. The degree of the polynomial on the left is $\deg(f)*p$ and the degree of the polynomial on the right is $\deg(g)*p+1$. So these are equal. I.e. $$\deg(f)*p=\deg(g)*p+1$$ But this is outrageous and impossible by divisibility considerations.

  • $\begingroup$ This is exactly what I was thinking, it just seems too easy! What bugs me is that it seems at first glance like it should show $x$ has no $r$-th roots for a lot of other $r$! $\endgroup$ – Robert Lewis May 27 '14 at 4:17
  • 1
    $\begingroup$ @RobertLewis That's correct, $x$ has no $r$th roots except when $r=\pm1$, by the same argument. The interesting thing about $r=p$ is that the characteristic can be used in an argument (as I mention in my answer). $\endgroup$ – blue May 27 '14 at 4:19

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.