6
$\begingroup$

I am bad (but trying to improve!) at very basic number theory and algebra. I'm quite sure this question is easy, but I do not know what fundamentals I am missing. This is from Ireland & Rosen's "Classical Introduction to Modern Number Theory" and is question 10 of Chapter 7. I have copied it exactly.

Let $K\supset F$ be finite fields and $[K:F]=2$. For $\beta\in K$ show that $\beta^{1+q}\in F$ and moreover that every element in $F$ is of the form $\beta^{1+q}$ for some $\beta\in K$.

The question uses the context of the previous one, in which $|F|=q$.

What I've tried so far:

For the first part, it seems there are two cases: i) $\beta\in F$ or ii) $\beta\in K\backslash F$. In i) it is easy, since $\beta^q=\beta$ and the extension is of degree $2$, $\beta^{q+1}=\beta^2\in F$. In ii), I suppose the minimal polynomial of $\beta$ in $K[x]$ is $x^2-\beta^2=x^2-\beta^{q+1}$... but where to go from here?

And I can't even begin to see what to do with the second part of the problem. Is any of this right so far? What do I do next if so? Thanks so much, ya'll. I'll go try to answer a question I can help with in the meantime.

$\endgroup$
2
  • 1
    $\begingroup$ For (i), I'm not sure you need to worry about the degree of the extension — fields are closed under multiplication. For (ii), how do you know that the minimal polynomial has that form? Do I have to have $\beta^2 \in F$? That seems like a strong statement. For the last part, it might help to view this $(q+1)$-th power operation as a homomorphism $K^* \to F^*$. You want this to be surjective. $\endgroup$ – Hoot Jul 4 '16 at 1:57
  • $\begingroup$ I think this is a duplicate of this older question. $\endgroup$ – Jyrki Lahtonen Jul 4 '16 at 4:56
2
$\begingroup$

Here is the first part:

Let $\beta\in K$. We can assume that $\beta\ne0$ because $0^{1+q}=0\in F$.

If $\beta\ne0$, then, by Lagrange's theorem, we have $(\beta^{q+1})^{q-1}=\beta^{q^2-1}=1$.

Now, the roots of $x^{q-1}-1$ in $K$ are exactly the non-zero elements of $F$.

Therefore, $\beta^{q+1} \in F$.

For the second part:

The map $\beta \mapsto \beta^{q+1}$ is a group homomorphism $K^\times \to F^\times$. (The codomain is right by the first part.)

Its kernel are the roots of $x^{q+1}=1$. Since $K^\times$ is cyclic and $q+1$ divides $q^2-1$, the kernel is a subgroup of order $q+1$. Therefore, the image has order $(q^2-1)/(q+1)=q-1$, which is the order of $F^\times$. Thus, the map is surjective.

$\endgroup$
2
  • $\begingroup$ Thanks for this, Ihf. I am reading through it slowly, but I wanted to express my appreciation for the fast response right away! $\endgroup$ – The Count Jul 4 '16 at 2:24
  • $\begingroup$ This is great! I need to be sure I understand each statement exactly, but I get it. Thanks so much! One question: to justify the statement "Now, the roots of $x^{q−1}-1$ in $K$ are exactly the non-zero elements of $F$", I can just say "since $K$ is an extension over $F$, and hence $x^{q-1}-1=\prod\limits_{a\in F, a\neq 0}(x-a)$", right? $\endgroup$ – The Count Jul 4 '16 at 3:52
6
$\begingroup$

We have $x^q=x$ for all $x\in F$ and $x^{q^2}=x$ for all $x\in K$. Since $q$ is a power of $p$ and $p$ is characteristic of $K$, hence $\sigma\colon y\mapsto y^q$ is an automorphism of the field $K$. Note that $\sigma$ fixes all the points of $F$, and it can not fix any point of $K\setminus F$ (otherwise, the polynomial $X^q-X$ will have more than $q$ roots).

Now, $\sigma(\beta^{q+1})=\beta^{q(q+1)}=\beta^{q^2}\beta^{q}=\beta\beta^q=\beta^{q+1}$, thus $\beta^{q+1}$ is a fixed point of $\sigma$, it must be inside $F$, thats what you expected.

$\endgroup$
2
  • 1
    $\begingroup$ Thanks very much for this response. I am not quite used to this level of language on the subject yet, but I'm sure it will be an important thing for me to revisit as I learn more. $\endgroup$ – The Count Jul 5 '16 at 1:07
  • 1
    $\begingroup$ In a word, $x^{q+1}$ is the field-theoretic norm of $x$ from $K$ down to $F$: the product of all ( both ! ) the $K$-conjugates of $x$. $\endgroup$ – Lubin Jul 14 '16 at 4:14

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.