$Proposition$: An extension $K/F$ is normal if and only if $H \triangleleft Gal(E/F)$

Now let $E/F$ be a finite galois extension and let $F=K_0 \leqslant K_1 \leqslant K_2 \leqslant .....\leqslant K_n=E$ consecutive extensions.

Let $H_i$ (for $i=0,1,2,,,n$) subgroup of $Gal(E/F)$ which corresponds to $K_i$ from the Galois correspondence (i.e $K_i \longleftarrow \longrightarrow H_i$)

Prove that $\forall i \in \{1,2,,,n\}$, we have that $K_i / K_{i-1}$ is a galois extension if and only if $H_i \triangleleft H_{i-1}$

I was going to proceed by induction to $n$.

For $n=1$ its valid because of the proposition .

Thus i assumed that the statement is valid $\forall 1<i<n$

But i find a difficulty to proceed because i cannot find the inductive step.

Can someone help me to solve this and complete the proof?

Thank you in advance!


It has nothing to do with induction. Assume $F\subseteq L\subseteq K\subseteq E$ with $E/F$ Galois. Let $H$ be the subgroup fixing $L$ and $N$ the subgroup fixing $K$. The $K/L$ is Galois iff $N \triangleleft H$. Assume first that $K/L$ is Galois let $\tau \in N$ and $\sigma \in H$. Then $\sigma$ maps $K$ into $K$. So if $x\in K$ then $\sigma(x)\in K$ and thus $\sigma^{-1}\tau\sigma(x)=x$, since $\tau$ fixes $K$ and thus $\sigma^{-1}\tau\sigma\in N$.

Assume conversely that $N\triangleleft H$ then with the same notation as above, $\sigma^{-1}\tau\sigma\in N$ and so $\sigma^{-1}\tau\sigma(x)=x$. So $\tau\sigma(x)=\sigma(x)$ and so $\sigma(x)$ is fixed by $H$ and is thus in $K$.


  • $\begingroup$ why is N a subgroup of H? $\endgroup$ – Marios Gretsas Nov 18 '16 at 12:00
  • 1
    $\begingroup$ @capo Because if an automorphism fixes $K$, then it fixes $L$. Fixes means pointwise fixes. $\endgroup$ – Rene Schipperus Nov 18 '16 at 13:40
  • $\begingroup$ how do we conclude from the converse that Κ/L is galois? $\endgroup$ – Marios Gretsas Nov 18 '16 at 15:30
  • $\begingroup$ Every auto of $K$ extends to an auto of $E$. Thus autos of $E$ and $K$ are the same on $K$. $\endgroup$ – Rene Schipperus Nov 18 '16 at 15:33
  • $\begingroup$ can we conclued that s(k)=k for every automorfism? $\endgroup$ – Marios Gretsas Nov 18 '16 at 15:39

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.