I need help understanding what the final sentence below means. Paraphrased from Lang's Algebra, Chapter VI, proof of Theorem 1.2 (page 266, third edition):

Let $K/k$ be Galois and $F$ an arbitrary extension, $K,F$ subfields of some other field. Then $KF/F$ and $K/K\cap F$ are Galois. Let $H=\mathrm{Gal}(KF/F)$, $G=\mathrm{Gal}(K/k)$. If $\sigma \in H$ then the restriction of $\sigma$ to $K$ is in $G$, and the map $\sigma \mapsto \sigma |_K$ gives an injective map of $H$ into the Galois group of $K/K \cap F$. Let $H'$ denote the image of this map.

It's the "and the map..." part I'm not sure about. Reasonably, the map $\sigma \mapsto \sigma|_K$ is from $H$, but into what? I'm guessing into $H$. In the proof it is assumed to be obvious that $\sigma|_K$ fixes $K \cap F$. If $H'$ is the image of $\sigma \mapsto \sigma|_K$ ($H \rightarrow H$) and $\sigma \in H'$, then $\sigma$ is supposed to leave $F$ unchanged. Since $K\cap F \subset F$, that field is also left unchanged. But how does that render the group $H'$ to be $\mathrm{Gal}(K/K\cap F)$? This group does not leave all of $F$ unchanged; nothing is said about $F\setminus (K\cap F)$. I don't get this...

Anyway, the result and its proof are not the issue, but what the codomain (target set?) of $\sigma \mapsto \sigma|_K$ is, and in particular what can be said about the order of $\mathrm{Gal}(K/K\cap F)$ if we know the orders of $G,H$.


It's a map into $G=\mathrm{Gal}(K/k)$ - this is hinted at by the statement

If $\sigma \in H$ then the restriction of $\sigma$ to $K$ is in $G$

An element $\sigma\in H=\mathrm{Gal}(KF/F)$ is an isomorphism $$\sigma\colon KF\to KF$$ such that $\sigma(x)=x$ for all $x\in F$. The restriction of $\sigma$ to $K$ will be an isomorphism $\sigma|_K\colon K\to K$, which is not an isomorphism from $KF$ to $KF$, and therefore can't possibly be interpreted as an element of $H$. It is however an element of $\mathrm{Gal}(K/k)$ because $\sigma|_K(x)=x$ for all $x\in k$.

Note that it makes no sense to ask whether or not an element of $\mathrm{Gal}(K/K\cap F)$ "fixes the rest of $F$" or not, because elements of this group are functions that are not defined on parts of $F$ other than $K\cap F$. For example, does the automorphism $\psi:\mathbb{Q}(\sqrt{2})\to\mathbb{Q}(\sqrt{2})$ that sends $\sqrt{2}$ to $-\sqrt{2}$ "fix" $\sqrt{3}$ or not?

  • $\begingroup$ Alright, makes sense, mostly ;) I'll return to it later, brain's too mushy at 7am. Thanks! $\endgroup$ – Erik Vesterlund Jul 9 '13 at 4:59

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.