2
$\begingroup$

Let $E/F$ be a finite Galois extension and $\alpha\in E$ and $p(t)=irr(\alpha,F)$, the monic irreducible polynomial defined over $F$ which has a root $\alpha$.

Let $\beta\in E$ be another root of $p(t)$.

My question arises here.

Are the two Galois groups $Gal(E/F(\alpha))$ and $Gal(E/F(\beta))$ isomorphic?

If this is not the case, could you suggest me some counter-example?

If they are always isomorphic, how can we detect the differences between each roots of an irreducible polynomial? Because, I know the all the roots of some irreducible polynomial are not of equal status from the view of total field $E$, not from the base field $F$.

$\endgroup$
1
$\begingroup$

$\newcommand{\Gal}{\mathrm{Gal}}$Yes, the groups will be isomorphic. This can be seen by noting that $E$ is the splitting field over $F$ of some separable polynomial $f \in F[x]$. Now clearly $E$ is also a splitting for $f$ over both $F(\alpha)$ and $F(\beta)$, and there is an isomorphism $\theta$ from $F(\alpha)$ to $F(\beta)$, that takes $\alpha$ to $\beta$. By the uniqueness theorem for the splitting field, $\theta$ can be extended to an isomorphism $\eta$ of $E$.

Now the inner automorphism $g \mapsto \eta^{-1} g \eta$ (left-to-right composition) of $\Gal(E/F)$ will induce an isomorphism $\Gal(E/F(\alpha)) \to \Gal(E/F(\beta))$.

$\endgroup$
  • $\begingroup$ Thank you for your kind answer.But there is some point to correct. You said that by the uniqueness theorem of splitting field,$theta$ can be extended to an isomorphism of $E$. But it is not due to uniqueness of splitting field but to isomorphism extension theorem. $\endgroup$ – user29422 Feb 26 '16 at 8:08

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.