2

I am trying to prove:

Let $K$ be the splitting field of $f(x)=x^4+ax^2+b$, whose roots are denoted by $\pm\alpha,\ \pm\beta$. Prove that $Gal(K/\mathbb Q)\cong V_4$ if and only if $\alpha\beta\in\mathbb Q$.

My question is: if we let $$\sigma:\alpha\mapsto-\alpha,\ \beta\mapsto\beta$$ $$\tau:\alpha\mapsto\alpha,\ \beta\mapsto-\beta$$ Then if both are well-defined automorphisms of $K$, the group $$\{1,\sigma,\tau,\sigma\tau\}$$ is isomorphic to $V_4$. However, $\sigma,\tau$ do not fix $\alpha\beta$, which means that this group must not be $Gal(K/\mathbb Q)$ (otherwise $\alpha\beta\notin\mathbb Q$, contradictory to what I am trying to prove) What is the reason?

p.s. I am aware that there questions asking about the same proposition. But they don't mention the problem that I am having here.

trisct
  • 5,221

1 Answers1

0

Your roots have two different values of $\alpha\beta$, depending the signs you choose. Call these possible values $+\gamma,-\gamma$. You thus have a set of two products $\{+\gamma,-\gamma\}$.

When you apply $\sigma$ or $\tau$, then true, you do not map either of the individual elements $+\gamma,-\gamma$ into itself. But you do map the set $\{+\gamma,-\gamma\}$ into itself. Thus the set is invariant, and that is all you really need to identify a symmetry element. The fact that you changed the order within the set does not change the set.

Oscar Lanzi
  • 39,403