I am currently working through Joseph Rotman's book "Galois Theory" and am trying to prove the following theorem.

If $f(x) \in F[x]$ has $n$ distinct roots in its splitting field $E$, then Gal($E/F$) is isomorphic to a subgroup of the symmetric group $S_n$, thus its order is a divisor of $n!$.

I have struggled to understand the proof in the book and have therefore tried to write my own proof of this fact, I have found some questions close to this on this site, but haven't been able to find exactly what I'm looking for.

My proof (so far) is as follows:


Let $X = \{\alpha_1, ... , \alpha_n \}$ be the set of roots of $f(x)$, as $E$ is the splitting field of $f$, it may be written as $E = F(\alpha_1, ... , \alpha_n)$ and noting that $\alpha_i \neq \alpha_j$ for $i\neq j$. Let $\sigma \in$ Gal($E/F$) then $\sigma(X) = X$ (a result from an earlier lemma). Now I would like to construct an injective homomorphism from $Gal(E/F)$ to $S_n$, my guess is the map which sends $\sigma$ to its restriction in $X$, which I shall denote by $\sigma_X$. Let $g$ be this map, then it is clear that $g$ is a homomorphism.

Now I try to show that $g$ is injective, consider $\sigma , \phi \in Gal(E/F)$ where $\sigma \neq \phi$, both of these functions are F-automorphisms of $E/F$, so they "fix" $F$, then we can conclude that $\sigma$ and $\phi$ can be different if and only if they act differently on $X$, which in the language of mathematics is precisely $\sigma \neq \phi \Leftrightarrow \sigma_X \neq \phi_X$. This is the definition of an injective function.

This means that $|Gal(E/F)| \leq |S_n| = n!$, however due to having constructed a homomorphism, we have mapped a group to another group, hence we have mapped $Gal(E/F)$ to a subgroup of $S_n$, whose order divides $|S_n|$ and therefore $|Gal(E/F)|$ divides $n!$, completing the proof.


I am wondering whether my proof for this is correct, I think that it is mostly there, with the statement

we can conclude that $\sigma$ and $\phi$ can be different if and only if they act differently on X

perhaps being the only issue.

  • $\begingroup$ If $ \sigma $ fixes all elements of $ X $, then it fixes the splitting field $ E $ as well. So $ \sigma $ must be the identity and hence your map is injective. $\endgroup$ – hellHound Dec 30 '18 at 17:26
  • $\begingroup$ Use $\operatorname{Gal}(E/F)$ for $\operatorname{Gal}(E/F)$. $\endgroup$ – Shaun Dec 30 '18 at 17:59
  • $\begingroup$ > implying $E/F$ is galois $\endgroup$ – Kenny Lau Dec 31 '18 at 13:54

Assume $\sigma_X = \varphi_X$, i.e. $\forall i, \sigma(\alpha_i) = \varphi(\alpha_i)$. We shall show that $\sigma = \varphi$, i.e. $\forall x \in E, \sigma(x) = \varphi(x)$.

Now note that $E = F(\alpha_1, \cdots, \alpha_n)$, and every $\alpha_i$ is algebraic over $F$, so every $x \in E$ can be expressed as a polynomial in $(\alpha_i)_{i=1}^n$ with coefficients in $F$, say $x = \sum f_j \alpha_{1}^{v_{j1}} \alpha_{2}^{v_{j2}} \cdots \alpha_{n}^{v_{jn}}$. Then:

$$\begin{array}{rcll} \sigma(x) &=& \sigma \left( \sum f_j \alpha_{1}^{v_{j1}} \alpha_{2}^{v_{j2}} \cdots \alpha_{n}^{v_{jn}} \right) \\ &=& \sum \sigma \left( f_j \alpha_{1}^{v_{j1}} \alpha_{2}^{v_{j2}} \cdots \alpha_{n}^{v_{jn}} \right) & \text {$\sigma$ preserves addition} \\ &=& \sum \sigma \left( f_j \right) \sigma \left( \alpha_{1} \right)^{v_{j1}} \sigma \left( \alpha_{2} \right)^{v_{j2}} \cdots \sigma \left( \alpha_{n} \right)^{v_{jn}} & \text {$\sigma$ preserves multiplication} \\ &=& \sum f_j \sigma \left( \alpha_{1} \right)^{v_{j1}} \sigma \left( \alpha_{2} \right)^{v_{j2}} \cdots \sigma \left( \alpha_{n} \right)^{v_{jn}} & \text {$\sigma$ fixes $F$} \\ &=& \sum f_j \varphi \left( \alpha_{1} \right)^{v_{j1}} \varphi \left( \alpha_{2} \right)^{v_{j2}} \cdots \varphi \left( \alpha_{n} \right)^{v_{jn}} & \forall i, \sigma(\alpha_i) = \varphi(\alpha_i) \\ &=& \sum \varphi \left( f_j \right) \varphi \left( \alpha_{1} \right)^{v_{j1}} \varphi \left( \alpha_{2} \right)^{v_{j2}} \cdots \varphi \left( \alpha_{n} \right)^{v_{jn}} & \text {$\varphi$ fixes $F$} \\ &=& \sum \varphi \left( f_j \alpha_{1}^{v_{j1}} \alpha_{2}^{v_{j2}} \cdots \alpha_{n}^{v_{jn}} \right) & \text {$\varphi$ preserves multiplication} \\ &=& \varphi \left( \sum f_j \alpha_{1}^{v_{j1}} \alpha_{2}^{v_{j2}} \cdots \alpha_{n}^{v_{jn}} \right) & \text {$\varphi$ preserves addition} \\ &=& \varphi(x) \end{array}$$

which is what was to be demonstrated.


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.