Let $F$ be a field, and let $f(x) ∈ F[x]$ be a polynomial with only simple roots. Let $E$ be a splitting field for $f(x)$. Consider the Galois group $G_f$ of the polynomial $f(x)$.

Now I need to prove $G_f$ is the set of permutations $σ ∈ S_n$ that take every $F$-algebraic relation between the roots of $f(x)$ to another $F$-algebraic relation between the roots.

That is: I need to show that:

A permutation $σ ∈ S_n$ comes from an automorphism in $Gal(E/F)$ $\Leftrightarrow$ $(\forall g(x_1,...,x_n) ∈ F[x_1,...,x_n] )(g(α_1,...,α_n) = 0\Rightarrow g(α_{σ(1)},...,α_{σ(n)}) = 0)$.

(Here $\alpha_1,...,\alpha_n$ are all the roots of $f(x)$).

I am considering the following track for one direction:

We have a unique $F$-algebra homomorphism $ψ: F[x_1,...,x_n] → E$ defined by: $x_i \mapsto α_i$.

For $σ ∈ S_n$. Let $σ'$ denote the unique automorphism of the $F$-algebra $F[x_1,...,x_n]$ defined by $x_i \mapsto x_{σ(i)}$.

Take the composition $ψ ◦ σ'$. It is a homomorphism $ϕ: F[x_1,...,x_n] → E$. Using the universal property of quotients, may be we can prove that there is an automorphism $σ_0 ∈ Gal(E/F)$ such that $ψ ◦ σ' = σ_0 ◦ ψ$ and we also have that the permutation of the roots of $f(x)$ induced by $σ_0$ is actually the same as our original permutation $σ$.

And then I am a bit confused. So may I please ask for a way if we can acutally do it? And also how to deal with the other direction? Thanks a lot!


Let $K =F(\alpha_1,\ldots,\alpha_n)$ be the splitting field of an irreducible polynomial $f \in F[x]$ whose roots are $\alpha_1,\ldots,\alpha_n$. We have an isomorphism $$\varphi : F[x_1,\ldots,x_n]/I \to K, \qquad \varphi(p(x_1,\ldots,x_n)+I)= p(\alpha_1,\ldots,\alpha_n)$$ where $I$ is the ideal $$I = \{ p \in F[x_1,\ldots,x_n], p(\alpha_1,\ldots,\alpha_n) = 0\}$$

If and only if $\sigma \in Gal(K/F)$, we have that $\sigma$ induces an automorphism of $K$ permutating the $\alpha_i$, so that $\varphi^{-1}\circ\sigma \circ\varphi$ induces an automorphism of $F[x_1,\ldots,x_n]/I$ permuting the $x_i$.

  • $\begingroup$ It makes sense to me... But how does this lead to an anwer to the question? May I please ask for some more explicit explaination to the last sentence? How does it related to the original question? $\endgroup$ – PropositionX May 24 '17 at 5:21
  • $\begingroup$ @PropositionX Your question is about those automorphisms of $F[x_1,\ldots,x_n]$ permuting the $x_i$ and fixing $I$, ie. about the automorphisms of $F[x_1,\ldots,x_n]/I$ permuting the $x_i$. $\endgroup$ – reuns May 24 '17 at 17:06
  • $\begingroup$ MayI I please ask why the map $\phi$ is an isomorphism? $\endgroup$ – PropositionX May 25 '17 at 0:02
  • $\begingroup$ Why is the evaluation map to the splitting field surjective? $\endgroup$ – PropositionX May 25 '17 at 0:10
  • 1
    $\begingroup$ @PropositionX In the same way. $F[\alpha_1,\alpha_2] = F[\alpha_1](\alpha_2) = F(\alpha_1)(\alpha_2) = F(\alpha_1,\alpha_2)$. Otherwise take any $b \in F[\alpha_1,\alpha_2] $ and note there is a $F$-linear dependence between the powers of $b$, ie $b^{-1} = \sum_{m=0}^d c_m b^m \in F[\alpha_1,\alpha_2] $ $\endgroup$ – reuns May 25 '17 at 1:23

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.