Let $p$ be a prime number and $H$ be number field. Denote by $\tilde{H}$ the composite of all $\mathbb{Z}_p$ extensions of $H$, so $\operatorname{Gal}(\tilde{H}/H)=\mathbb{Z}_p^{d}$. Write $\varLambda$ for $\mathbb{Z}_p [[\operatorname{Gal}(\tilde{H}/H)]]\simeq\mathbb{Z}_p[[t_1, ... t_d]]$. Let $M$ be the maximal abelian unramified $p$-extension of $\tilde{H}$, as well known $X=\operatorname{Gal}(M/\tilde{H})$ is noetherian torsion module over $\varLambda$.

1) Is $\frac{X}{(t_1,t_2,...,t_n)X}$ of finite cardinal or torsion group?

Consider $L$ the fixed field by $(t_1,t_2,...,t_n)X$, so $\operatorname{Gal}(L/\tilde{H})=\frac{X}{(t_1,t_2,...,t_n)X}$.

2) Is $L$ an abelian extension of $H$.


1 Answer 1


Let $S$ be the subgroup of $\operatorname{Gal}(M/H)$ generated by the inertia groups of the places $v_1,\cdots,v_s$ of $H$ ramifying in $\tilde{H}$. The index of the image of $S$ in $\operatorname{Gal}(\tilde{H}/H)$ is finite. Hence, replacing $H$ by a finite extension, we can assume that this image is isomorphic to $\mathbb Z_{p}^{d}$. Because $M/\tilde{H}$ is unramified, $\operatorname{Gal}(M/H)$ is a semi-direct product of $S$ with $X$.

Let $M_0$ be the maximal abelian extension of $H$ inside $M$ (note that $M_0$ contains $\tilde{H}$) and let $H_0$ be the maximal unramified extension of $H$ contained in $M_0$ (a finite extension).

By definition, $X/(t_1,\cdots,t_d)X$ is the maximal quotient of $X$ on which $\operatorname{Gal}(\tilde{H}/H)$ acts trivially. Because the action of $\operatorname{Gal}(\tilde{H}/H)$ on $X$ is by inner automorphism, this maximal quotient is also the quotient of $\operatorname{Gal}(M/H)$ by its derived subgroup and so the field $L$ fixed by $(t_1,\cdots,t_d)X$ is the maximal abelian extension of $H$ contained in $M$, and thus is equal to $M_0$. The answer to your question 2 is thus positive.

Moreover, as the places of $H$ ramified in $M_0$ are exactly the $v_i$ and as $M_0/\tilde{H}$ is unramified, we see that $S$ surjects on $\operatorname{Gal}(M_0/H_0)$ and hence that this $\mathbb Z_p$-modules has rank at most $d$. Hence, the rank of $\operatorname{Gal}(M_0/H)$ is less than $d$ and so the rank of $\operatorname{Gal}(M_0/\tilde{H})$ is zero. The module $X/(t_1,\cdots,t_d)$ is thus indeed a torsion $\mathbb Z_p$-module.

In fact, the answer above to your question 2) (resp. to your question 1)) exactly recapitulates the proof of the fact that $X$ is a noetherian $\Lambda$-module (resp. a torsion $\Lambda$-module). Good references include J-P.Serre Classes des corps cyclotomiques (d'après K.Iwasawa) (Séminaire Bourbaki, 1958) and K.Iwasawa On $\mathbb Z_\ell$-extensions of Algebraic Number Fields (Annals of Math.,1973).

  • $\begingroup$ if you can just write me a little more detail on the second point. $\endgroup$ Apr 21, 2015 at 11:49
  • $\begingroup$ @AdelLille Done. Have a look at the references as well, you will probably learn a lot. $\endgroup$
    – Olivier
    Apr 21, 2015 at 20:52
  • $\begingroup$ Why the fact M/H̃ is unramified implies that Gal(M/H) is a semi-direct product of S with X? if we have two places of H ramifying in H̃ or more we don't know if the product of inertia group commute, so we don't kwon if $S$ is isomorphism to Gal(H̃/H), but if we have only one place of H ramifying in H̃, the fact that M/H̃ is unramified implies that Gal(M/H) is a semi-direct product of S with X. $\endgroup$ Apr 29, 2015 at 10:33

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.