Let $K$ be a number field, $v$ a finite place of $K$. Let $K'$ be the maximal extension of $K$ unramified at $v$.

For instance, for $K = \mathbf{Q}$ and $v = p$ a prime, $K' = \mathbf{Q}(\{\zeta_{n}, p\nmid n\})$, $\zeta_n$ primitive $n$-th roots of unity.

Let $\overline{K}$ an algebraic closure of $K$. Is $\text{Gal}(\overline{K}/K')$ an open normal subgroup of $\text{Gal}(\overline{K}/K)$?


For profinite groups $G$, open subgroups $H$ always have finite index, because the map $G \to G/H$ is continuous, $G/H$ has the discrete topology, and $G$ is compact. In your question, this index $[K' :K]$ is infinite, so the answer is no.

  • 1
    $\begingroup$ Is it clear that $[K' : K]$ is always infinite? It certainly "often" is, but is this provably true for all $K$ and $v$? $\endgroup$ – David Loeffler Mar 16 '18 at 10:44
  • $\begingroup$ Good question, I had only given it cursory thought. Can't one just add $n$-th roots of unity for $v(n) = 0$, as in the $\mathbb{Q}$-case? $\endgroup$ – Johann Haas Mar 16 '18 at 11:48
  • $\begingroup$ Then you get an extension ramified at all primes of $K$ dividing $n$ (and there may be more than one). $\endgroup$ – David Loeffler Mar 16 '18 at 15:43
  • $\begingroup$ I am missing something obvious, I'm afraid. Isn't this extension still unramified at $v$, which is what we care about? $\endgroup$ – Johann Haas Mar 16 '18 at 15:46
  • 1
    $\begingroup$ Oops, I had misread the question! I was thinking of the maximal extension ramified only at $v$ in which case the question of whether or not $[K' : K]$ is finite is much more subtle. $\endgroup$ – David Loeffler Mar 16 '18 at 16:33

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.