12
$\begingroup$

This is the problem I am facing:

Compute the intermediate fields of the extension $K | \mathbb{Q}(i)$ where $K = \mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i)$ and find the intermediate fields $M$ such that $M | \mathbb{Q}(i)$ is a Galois extension.

What I have done:

I start by thinking about the grade of the extension $[\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, : \, \mathbb{Q}(i)]$

$[\mathbb{Q}(\sqrt{3},i) \, : \, \mathbb{Q}(i)] = 2$ because the irreducible polinomial of $\sqrt{3}$ over $\mathbb{Q}(i)$ is $x^2 -3$ (since $\sqrt{3} \notin \mathbb{Q}(i)$

$[\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, : \, \mathbb{Q}(\sqrt{3},i)] = 3$ because the irreducible polinomial of $\sqrt[3]{2}$ over $\mathbb{Q}(\sqrt{3},i)$ is $x^3 -2$ (*)

Hence, $[\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, : \, \mathbb{Q}(i)] =[\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, : \, \mathbb{Q}(\sqrt{3},i)] * [\mathbb{Q}(\sqrt{3},i) \, : \, \mathbb{Q}(i)] = 3 * 2 = 6$

Now, a $\mathbb{Q}(i) $-base for $\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i)$ is $\{ 1, \sqrt[3]{2} \, \xi , \sqrt[3]{2} \, \xi^2, \sqrt{3} ,\sqrt{3} \sqrt[3]{2} \, \xi , \sqrt{3} \sqrt[3]{2} \, \xi^2 \}$

The non-linear elements whose images fix every $\sigma \in \textrm{Gal}(\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, | \, \mathbb{Q}(i))$ are:

$$ \sqrt{3} \to \pm \sqrt{3}$$ $$ \sqrt[3]{2} \to \{\sqrt[3]{2}, \sqrt[3]{2} \, \xi, \sqrt[3]{2} \, \xi ^2 \}$$

The extension $\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, | \, \mathbb{Q}(i)$ is a Galois extension (it is normal(splitting field of $(x^3-2)(x^2-3)$, separable (char($\mathbb{Q}(i)$) = 0) and finite) so $| \textrm{Gal}(\mathbb{Q}(\sqrt[3]{2}, \sqrt{3},i) \, | \, \mathbb{Q}(i)) | = 6$ which means that all the options are valid.

I am stuck here, how can I determine the Galois group, and the normal subgroups (and then, how to extract the matching fields for those subgroups?)

(*) How to show justify this?

$\endgroup$
  • $\begingroup$ Also, $Gal(K/\mathbb{Q}[i]) \cong S_3$. You've listed your automorphisms, and none of them generate the Galois group, so it cannot be cyclic. Remember that the only groups of order $6$ are $\mathbb{Z}_6$ and $S_3$. $\endgroup$ – Kaj Hansen Jul 17 '14 at 21:27
  • 2
    $\begingroup$ $\mathbb{Z}_2 \times \mathbb{Z}_3 \cong \mathbb{Z}_6$ since $gcd(2, 3) = 1$. See here: proofwiki.org/wiki/Group_Direct_Product_of_Cyclic_Groups $\endgroup$ – Kaj Hansen Jul 17 '14 at 21:30
11
$\begingroup$

To address the (*):

To show that $f(x) = x^3 - 2$ is irreducible over $\mathbb{Q}(\sqrt{3}, i)$, it suffices to show that none of its roots are in this field (since $\deg(f) \leq 3$). Indeed, $\sqrt[3]{2}, \ \omega\sqrt[3]{2}$, and $\omega^2\sqrt[3]{2}$ are not elements of $\mathbb{Q}(\sqrt{3}, i)$. If they were, we would have a tower of fields $\mathbb{Q} \subset \mathbb{Q}(\sqrt[3]{2}) \subset \mathbb{Q}(\sqrt{3}, i)$, which would imply $\Big[ \mathbb{Q}(\sqrt{3}, i):\mathbb{Q} \Big] = \Big[ \mathbb{Q}(\sqrt{3}, i): \mathbb{Q}( \sqrt[3]{2}) \Big] \cdot \Big[ \mathbb{Q}(\sqrt[3]{2}):\mathbb{Q} \Big]$. Contradiction! Look at the degrees of the extensions: $3$ does not divide $4$.


To address the crux of the question:

You're close! You've done a good bit of the work already.

Remember that the only groups of order $6$ are $\mathbb{Z}_6$ and $S_3$. However, you've found $6$ automorphisms and none of them generate the entire group. Therefore, it must be the case that $\operatorname{Gal}(K/\mathbb{Q}(i)) \cong S_3$.

Next, the fundamental theorem of Galois theory tells us that there is a one-to-one correspondence between subgroups of $\operatorname{Gal}(K/\mathbb{Q}(i))$ and the intermediate fields between $K$ and $\mathbb{Q}(i)$ that they fix. Let's list the subgroups of $S_3$. They are:

  • $1$ copy of the trivial subgroup
  • $3$ copies of $\mathbb{Z}_2$
  • $1$ copy of $\mathbb{Z}_3$
  • The whole group

You should check this yourself! Hint: Lagrange's theorem tells us that the proper subgroups are of prime order, and hence are cyclic. First, the trivial subgroup corresponds to the entire field $K$, and the whole group corresponds to the base field $\mathbb{Q}(i)$.

Next, let's call the two 'generator' automorphisms that you found $\phi$ and $\tau$ such that $\phi$ sends $\sqrt{3} \mapsto \sqrt{3}$ while fixing the other basis elements, and likewise $\tau$ sends $\sqrt[3]{2} \mapsto \sqrt[3]{2}\xi$.

You should check that the copies of $\mathbb{Z}_2$ are:

  • $\{\text{id}, \phi \}$
  • $\{\text{id}, \tau \phi \}$
  • $ \{\text{id}, \tau^2 \phi \} $

And the copy of $\mathbb{Z}_3$ is $\{\text{id}, \tau, \tau^2\}$.

Some proper intermediate fields between $K$ and $\mathbb{Q}(i)$ are $\mathbb{Q}(i, \sqrt[3]{2})$, $\mathbb{Q}(i, \sqrt{3})$, $\mathbb{Q}(i, \sqrt[3]{2}\xi)$, and $\mathbb{Q}(i, \sqrt[3]{2}\xi^2)$. How do we know this is a complete list? Keeping in mind that $\displaystyle \xi = -\frac{1}{2} + \frac{\sqrt{3}}{2}i$, I'll leave it to you match the subgroups with the intermediary fields they fix.

Finally, to determine which intermediate fields are Galois over $\mathbb{Q}(i)$, simply apply the fact than an intermediate field $E$ is Galois if and only if its corresponding subgroup is a normal subgroup of $\operatorname{Gal}(K/\mathbb{Q(i)})$. What are the normal subgroups of $S_3$?

$\endgroup$

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.