I want to prove that given any integer n, we can find a finite Galois extension K over $\mathbb Q$ such Gal$({K}: {\mathbb Q })$ = $S_n$

For prime p, I know finding a polynomial with exactly 2 nonreal roots will have Galois group of splitting field $S_p$.

Can we find infinitely many such polynomials for prime p?

What about composite n?

  • $\begingroup$ What exactly is your second line saying? Are you saying that any irreducible polynomial (over $\mathbb{Q}$) of degree $p$ with exactly two non-real roots will have a galois group equal to $S_p$? $\endgroup$ Apr 26 at 7:32
  • 2
    $\begingroup$ @DionelJaime Yes, that is a known trick. When $p$ is a prime, a transitive subgroup of $S_p$ that contains a 2-cycle must be all of $S_p$. $\endgroup$ Apr 26 at 7:50

One alternative approach (based on Dedekind's theorem, see also here for a local explanation) is the following:

  1. Select two distinct prime numbers $q$ and $r$.
  2. Select an irreducible monic polynomial $g_1(x)\in\Bbb{Z}_q[x]$ of degree $n$.
  3. Select an irreducible monic polynomial $g_2(x)\in\Bbb{Z}_r[x]$ of degree $n-1$.
  4. Solve the instance of Chinese Remainder Theorem and find a monic polynomial $\tilde{g}(x)\in\Bbb{Z}_{qr}[x]$ such that $$\tilde{g}(x)\equiv g_1(x)\pmod q$$ as well as $$\tilde{g}(x)\equiv x g_2(x)\pmod r.$$
  5. Find a "lift" $g(x)\in\Bbb{Z}[x]$ such that $g(x)$ is monic, congruent to $\tilde{g}(x)$ modulo $qr$ and has exactly two non-real zeros.

With this in place the splitting field of $g(x)$ will have Galois group $G=S_n$ over $\Bbb{Q}$:

  • By item 2 $G$ contains an $n$-cycle. In particular $G$ is transitive and $g(x)$ is irreducible over $\Bbb{Q}$.
  • By item 3 $G$ also contains an $(n-1)$-cycle. Together with the previous bullet this implies that $G$ is doubly transitive.
  • By item 5 $G$ contains a $2$-cycle. By the previous bullet $G$ contains all the $2$-cycles, and hence $G=S_n$.

  • If item 5 is not constructive enough, then you can select a third prime $\ell>n$. Select an irreducible monic quadratic $g_3(x)\in\Bbb{Z}_\ell[x]$ and, instead of 5, add the requirement that $$ g(x)\equiv g_3(x)\prod_{i=0}^{n-3}(x-i)\pmod\ell. $$ Dedekind's theorem again implies that $G$ contains a $2$-cycle.
  • See this nice post for an approach to item 5.

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.