A question inspired by the concept of constructible numbers.

Let $K$ be a field, $\Omega$ some algebraic closure of $K$ and $p$ a prime number. Let $L$ be the set of all elements $\alpha \in \Omega$ for which $\dim_K(K[\alpha])$ is a power of $p$. Is $L$ a subring/subfield of $\Omega$ in general? If not, under what conditions on $p$ or $K$ is it?

The answer is positive for the constructible numbers ($p = 2, K = \mathbb{Q}$).

  • $\begingroup$ Perhaps the constructible field is the compositum of all fields gotten as towers of quadratic extensions. You’ll have to show that if $K$ and $L$ are both at the top of such towers, so is $KL$. It should be a direct verification. $\endgroup$ – Lubin Feb 22 '17 at 22:32

$\newcommand{\Size}[1]{\left\lvert #1 \right\rvert}$$\newcommand{\C}{\mathbb{C}}$$\newcommand{\Q}{\mathbb{Q}}$Let $K = \Q$, $\Omega = \C$, $p = 3$, and $\omega \in \C$ a primitive third root of unity.

Then $\sqrt[3]{2}, \omega \sqrt[3]{2} \in L$, but $\omega = (\sqrt[3]{2})^{-1} \cdot \omega \sqrt[3]{2} \notin L$, as $\Size{\Q[\omega] : \Q} = 2$

So $L$ is not a subfield in this case, nor a subring, as $(\sqrt[3]{2})^{-1} \in \Q[\sqrt[3]{2}]$.

You can do the same, mutatis mutandis, for any odd prime $p$. The reason the argument fails for $p = 2$ is of course that $\Size{\Q[-1] : \Q} = 1 = 2^{0}$.

| cite | improve this answer | |
  • $\begingroup$ Interesting answer, thanks! Could you give a (sketch of a) general proof in the case $p = 2$? All proofs I have found seem to somehow rely on geometric constructions rather than pure algebra. $\endgroup$ – Bib-lost Feb 21 '17 at 17:37
  • 2
    $\begingroup$ @Bib-lost, on re-reading your post, I believe there might be a misconception. The result is not true for $p = 2$ either. Just consider a polynomial $f \in \mathbb{Q}[x]$ of degree $4$, its splitting field $E$, and assume the Galois group is $S_{4}$. Then $E$ is generated by the four roots of $f$, and for each $\alpha$ of them one has $\lvert \mathbb{Q}[\alpha] : \mathbb{Q} \rvert = 4$, but then $E$ contains elements $\beta$ such that $\lvert \mathbb{Q}[\beta] : \mathbb{Q} \rvert = 3$. $\endgroup$ – Andreas Caranti Feb 22 '17 at 12:05
  • $\begingroup$ In that case I need to work on my understanding of the constructible numbers. Thanks for pointing this out. $\endgroup$ – Bib-lost Feb 22 '17 at 15:22

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.