The question is from Marcus' book, "Number Fields" (exercise 12, Chapter 4)

Let $\omega= e^{\frac{2\pi i}{m}}$ and $p$ be a rational prime not dividing $m$. Then how does $p$ split in $\mathbb{Q}[\omega + \omega^{-1}]$?

I know that $\mathbb{Q}[\omega + \omega^{-1}]$ is subfield of $\mathbb{Q}[\omega]$ and that $\text{Gal}(\mathbb{Q}[\omega]/\mathbb{Q})$ is isomorphic to $\mathbb{Z}_m^{\times}$.

The following result is known:

Lemma: Let $H$ be the subgroup of $\mathbb{Z}_m^{\times}$ fixing a subfield $K$ of $\mathbb{Q}[\omega]$ pointwise and $p$ be a rational prime not dividing $m$. Then $f$ is the inertial degree of any prime $\mathfrak{P}$ of $K$ lying over $p\mathbb{Z}$, where $f$ is the least positive integer such that $p^f \equiv x \pmod m$ with $x\in H$.

Since $\mathbb{Q}[\omega]$ has degree 2 over $ \mathbb{Q}[\omega + \omega^{-1}]$, I get $H = \mathbb{Z}_3^{\times}$ and have to find least possible values of $f$ such that $p^f \equiv 1,2 \pmod m$.

What can I conclude from all this, about the splitting of primes?

  • 1
    $\begingroup$ The real subfield $\Bbb{Q}[\omega+\omega^{-1}]$ is the fixed field of complex conjugation. In the identification of the Galois group with $\Bbb{Z}_m^*$ the complex conjugation corresponds to the residue class of $-1$. Therefore $H=\{\overline{1},\overline{-1}\}\le\Bbb{Z}_m^*$. I don't understand where your view that $H=\Bbb{Z}_3^*$ would come from? For one that is not a subgroup of $\Bbb{Z}_m^*$ in any natural way. $\endgroup$ – Jyrki Lahtonen Jul 7 '16 at 18:29
  • $\begingroup$ And because this extension is Galois, the equation $efg=n$ holds. You know $n$, the Lemma gives you a method for finding $f$. Do you know $e$ or $g$? $\endgroup$ – Jyrki Lahtonen Jul 7 '16 at 21:12
  • $\begingroup$ @JyrkiLahtonen I think, $\{\overline{1}, \overline{-1}\}=\{\overline{1}, \overline{2}\} = \mathbb{Z}_3^*$. Also, what I am asking in question is basically that how to figure out $e$. I know that for $\mathbb{Z}[\omega]$, ramification index is $\phi{(p^k)}$ where $k$ is the highest exponent of $p$ dividing $m$. Also, that $n=\frac{\phi{(m)}}{2}$. I don't know how to put the pieces together! $\endgroup$ – rationalbeing Jul 8 '16 at 4:06
  • 1
    $\begingroup$ No, $\overline{2}\neq\overline{-1}$. After all, you are looking at these modulo $m$, and $2\not\equiv-1\pmod m$. You are correct about $n$ and $e$. Did you forget the assumption $p\nmid m$? $\endgroup$ – Jyrki Lahtonen Jul 8 '16 at 5:55
  • 1
    $\begingroup$ Not really. You can describe the prime ideals using a factorization of the minimal polynomial of $\omega+\overline{\omega}$ over $\Bbb{F}_p$, but that is less explicit (and was probably covered in the proof the lemma already). $\endgroup$ – Jyrki Lahtonen Jul 8 '16 at 6:10

Based on hints by Jyrki Lahtonen :

First of all, $H=\{\overline{1}, \overline{-1}\}$ (which is not same as $\mathbb{Z}_3^{\times}$). So, using the Lemma stated in question I can calculate $f$ to be the least positive integer such that $p^f \equiv \pm 1 \pmod m$.

Since $p\not|m$, by a standard theorem for $\mathbb{Q}[\omega]$, we know that its ramification index is 1. Moreover, $\mathbb{Q}[\omega + \omega^{-1}]$ is a subfield of $\mathbb{Q}[\omega]$, hence we conclude that its ramification index is also 1.

We know that $\mathbb{Q}[\omega]$ has degree 2 over $\mathbb{Q}[\omega + \omega^{-1}]$ and degree $\phi(m)$ over $\mathbb{Q}$ hence we conclude that $\mathbb{Q}[\omega + \omega^{-1}]$ and degree $\frac{\phi(m)}{2}$ over $\mathbb{Q}$.

Let $g$ be the number of prime factors in $\mathbb{Q}[\omega + \omega^{-1}]$, then by another standard theorem $fg = \frac{\phi(m)}{2}$ and we get $g = \frac{\phi(m)}{2f}$.

Thus, if $K=\mathbb{Q}[\omega + \omega^{-1}]$, then $$ p\mathcal{O}_K = \prod_{\ell = 1}^{\frac{\phi(m)}{2f}} \mathfrak{P}_{\ell}$$ where $f$ is the smallest positive integer satisfying $p^f \equiv \pm 1 \pmod m$.


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.