1
$\begingroup$

I am trying to prove some consequences of Iwasawa's Theorem for CM-fields. There is a sequence of CM-fields $$K=K_0\subseteq K_1 \subseteq \dots \subseteq K_\infty$$ so that $K_\infty/K$ is a $\mathbb{Z}_p$-extension for a prime $p$. So I think that $[K_{n+1}:K_n]$ must be some $p$-power for all $n\geq0$. Since each $K_n$ is a CM-field, there exist subfields $K_n^+$ which are totally real, so that $[K_n:K_n^+]=2$.
What can I say about the maximal unramified abelian extension $K_\infty/K$? How can I prove that $K_\infty$ also is a CM-field? (I need this result to talk about $K_\infty^+/K^+$.)

$\endgroup$
1
$\begingroup$

You can define a CM-field number $K$ as being a totally imaginary quadratic extension of a totally real field $k$. Then it can be shown that complex conjugation on $\mathbf C$ induces an automorphism on $K$ which is independent of the embedding of $K$ into $\mathbf C$ (see e.g. Washington, chap. 4, just before thm. 4.10). By abuse of language, this is called « the » complex conjugation on $K$, and its fixed field $k$ is denoted $K^+$ for obvious reason. Now consider a $\mathbf Z_p$-extension $K_\infty/K$, i.e. an infinite Galois extension whose Galois group is isomorphic to the additive group $\mathbf Z_p$ = inverse limit of $(\mathbf Z /p^n \mathbf Z$ , +). By definition and Galois theory, $K_\infty$ is the direct limit (actually, the union here) of the fields $K_n$ fixed by the subgroups of $\mathbf Z_p$ of index $p^n$ . Note that $K_n/K$ is the unique subextension of degree $p^n$, and $K_{n+1}/K_n$ has degree $p$. Moreover, if all the $K_n$ ‘s are CM, so is obviously the union $K_\infty$.

You ask what can be said of the maximal abelian unramified extension of $K_\infty$. Almost by defintion, it’s the direct limit of the maximal abelian unramified extensions of the $K_n$ ‘s, so by class field theory, its Galois group $X$ over $K_\infty$ is isomorphic to the inverse limit of the ideal class groups $C_n$ of the $K_n$ ‘s. The interesting point would be to describe the action of $\Gamma = Gal(K_\infty/K)$ on $X$. When decomposing $X$ into its (pro)-$l$- Sylow subgroups $X(l)$, a few things can be said if $l \neq p$ (see e.g. Washington, chap. ). But the heart of Iwasawa’s « business » is about the structure of the noetherian compact torsion $\mathbf Z_p[[\Gamma]]$- module $X(p)$. A general algebraic structure theorem is known for such modules (Wash., thm. 13.12), which generalizes in some way the classical description of finite abelian $p$-groups. By descent (i.e. by taking co-invariants under $\Gamma$), one obtains the celebrated Iwasawa formulae giving the asymptotic orders of the $p$-Sylow subgroups of the $C_n$ ‘s ( Wash. , thm. 13. 13).

What we have seen is just the algebraic part of the theory. The most interesting information actually concerns the precise relationship of the module $X(p)$ with the $p$-adic L-functions attached to totally real number fields. This is the so-called Main Conjecture, the « simplest » version of which is the Mazur-Wiles theorem : let $K$ be CM, abelian over $\mathbf Q$, of degree prime to $p$ ; let $K_\infty$ be the cyclotomic $\mathbf Z_p$-extension of $K$ (not just any $\mathbf Z_p$-extension); for any non trivial even character of Gal$(K/\mathbf Q)$, say $\chi$, let $X (\omega \chi^-1)$ (where $\omega$ is the Teichmûller character) be the $\omega \chi^-1$ part of $X(p)$, and let $f_\chi$ (T) be « the » « characteristic power series » associated to $X (\omega \chi^-1)$ via the previous structure theorem ; then $f_\chi ((1+p)s – 1) = L_p (\chi, s)$ for all $s$ in $\mathbf Z_p$ (see e.g. Wash., chapter 16). Note that this descibes only the « minus part » (inverted by complex conjugation) of $X(p)$, the « plus part » (fixed by complex conjugation) being still unknown. For example, if $K$ is the $p$-th cyclotomic field, the famous Vandiver conjecture is equivalent to the vanishing of the plus part of $X(p)$.

$\endgroup$

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.