Genus theory of p-adic pseudo-measures -- Tame kernels and abelian p-ramification
Extracted source text ยท arXiv 2310.10112
\documentclass[11pt]{amsart}
\usepackage{url}
\urlstyle{sf}
\usepackage{hyperref,amssymb}
\hypersetup{colorlinks=true,linkcolor=blue,
urlcolor=cyan,citecolor=magenta}
\usepackage[hyperpageref]{backref}
\usepackage{frcursive,calrsfs,mathrsfs}
\textheight=247mm
\textwidth=115mm
\evensidemargin=15mm
\oddsidemargin=15mm
\hoffset=0mm
\voffset=-22mm
\topmargin=5mm
%
\newtheorem{theorem}{Theorem}[section]
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{remarks}[theorem]{Remarks}
\newtheorem{notations}[theorem]{Notations}
%
\numberwithin{equation}{section}
%
\newcommand{\Q}{{\mathbb{Q}}}
\newcommand{\Z}{{\mathbb{Z}}}
\newcommand{\F}{{\mathbb{F}}}
\newcommand{\C}{{\mathbb{C}}}
\newcommand{\ds}{\displaystyle}
\newcommand{\ov}{\overline}
\newcommand{\wt}{\widetilde}
\newcommand{\ft}{\footnotesize}
\newcommand{\ns}{\normalsize}
\newcommand{\CE}{{\mathcal E}}
\newcommand{\CG}{{\mathcal G}}
\newcommand{\CH}{{\mathcal H}}
\newcommand{\CR}{{\mathcal R}}
\newcommand{\CS}{{\mathcal S}}
\newcommand{\CT}{{\mathcal T}}
\newcommand{\CU}{{\mathcal U}}
\newcommand{\CW}{{\mathcal W}}
\newcommand{\CX}{{\mathcal X}}
\newcommand{\order}{\raise0.8pt \hbox{${\scriptstyle \#}$}}
\newcommand{\lien}{\mathrel{\mkern-4mu}}
\newcommand{\too}{\relbar\lien\rightarrow}
\newcommand{\tooo}{\relbar\lien\relbar\lien\too}
\newcommand{\toooo}{\relbar\lien\relbar\lien\tooo}
\newcommand{\plus}{\ds\mathop{\raise 0.5pt \hbox{$\bigoplus$}}\limits}
\newcommand{\prd}{\ds\mathop{\raise 1.0pt \hbox{$\prod$}}\limits}
\newcommand{\sm}{\ds\mathop{\raise 1.0pt \hbox{$\sum$}}\limits}
\newcommand{\ffrac}[2]{\hbox{\ft $\displaystyle\frac{#1}{#2}$}}
\newcommand{\Gal}{{\rm Gal}}
\newcommand{\ram}{{\rm ram}}
\newcommand{\ab}{{\rm ab}}
\newcommand{\ar}{{\rm ar}}
\newcommand{\nr}{{\rm nr}}
\newcommand{\gen}{{\rm gen}}
\newcommand{\pr}{{\rm pr}}
\newcommand{\bp}{{\rm bp}}
\newcommand{\ta}{{\rm ta}}
\newcommand{\rk}{{\rm rk}}
\newcommand{\1}{{\chi_0^{}}}
\newcommand{\Cha}{{\rm III}}
\newcommand{\Hom}{{\rm H}}
\newcommand{\Ker}{{\rm Ker}}
\newcommand{\Norm}{\hbox{\bf$\textbf{N}$}}
\newcommand{\BC}{\hbox{\bf$\textbf{C}$}}
\newcommand{\BD}{\hbox{\bf$\textbf{D}$}}
\newcommand{\BH}{\hbox{\bf$\textbf{H}$}}
\newcommand{\BJ}{\hbox{\bf$\textbf{J}$}}
\newcommand{\BK}{\hbox{\bf$\textbf{K}$}}
\newcommand{\BL}{\hbox{\bf$\textbf{L}$}}
\newcommand{\BR}{\hbox{\bf$\textbf{R}$}}
\newcommand{\BT}{\hbox{\bf$\textbf{T}$}}
\newcommand{\BV}{\hbox{\bf$\textbf{V}$}}
\newcommand{\BX}{\hbox{\bf$\textbf{X}$}}
\newcommand{\BZ}{\hbox{\bf$\textbf{Z}$}}
\author[Georges Gras]{Georges Gras}
\address{\hspace{-0.5cm}Villa la Gardette, 4 chemin Ch\^ateau Gagni\`ere,
38520 Le Bourg d'Oisans \rm {\url{http://orcid.org/0000-0002-1318-4414} } }
\email{g.mn.gras@wanadoo.fr}
\keywords{abelian fields; $\BK$-theory; Hilbert's kernel; $p$-adic $\zeta$ and
$L$-functions, Stickelberger's element, class field theory, PARI programs}
\subjclass{11R42, 11R70, 11S40, 19F15, 11R37}
\begin{document}
\title[Genus theory of $p$-adic pseudo-measures]
{Genus theory of $p$-adic pseudo-measures \\ Hilbert's kernels \& abelian $p$-ramification}
\date{January 17, 2024}
\begin{abstract} We consider, for real abelian fields $K$, the Birch--Tate
formula linking $\order \BK_2(\BZ_K)$ to $\zeta_K(-1)$;
we compare, for quadratic and cyclic cubic fields with $p \in \{2, 3\}$,
$\order \BK_2(\BZ_K)[p^\infty]$ to the order of the torsion group
$\CT_{K, p}$ of abelian $p$-ramification theory given by the residue
of $\zeta_{K, p}(s)$ at $s=1$. This is done via the ``genus theory'' of
$p$-adic pseudo-measures, inaugurated in the 1970/80's and the fact
that $\CT_{K, p}$ only depends on the $p$-class group and on the
normalized $p$-adic regulator of $K$ (Theorem \ref{main}\,{\bf A}). We
apply this to prove a conjecture of Deng--Li giving the structure of
$\BK_2(\BZ_K)[2^\infty]$ for an interesting family of real quadratic
fields (Theorem \ref{main}\,{\bf B}). Then, for $p \geq 5$, we give a
lower bound of $\rk_p(\BK_2(\BZ_K))$ in cyclic $p$-extensions $K/\Q$
(Theorem \ref{main}\,{\bf C}). Complements, PARI programs and tables
are given in Appendices.
\end{abstract}
\maketitle
\tableofcontents
\section{Introduction and prerequisites}
The aim of this paper is to utilize the link between $p$-adic $\BL$-functions
$\BL_p(s, \chi)$, of even Dirichlet's characters $\chi = \varphi \psi$ ($\varphi$
of prime-to-$p$ order, $\psi$ of {\it non-trivial} $p$-power order), and classical
arithmetic $p$-invariants of the corresponding cyclic number field $K$, then to
show that these invariants have some connections due to the character $\psi$
(genus theory principle). This connection does exist because of reflection
theorems of class field theory and this allows to get additional informations.
\smallskip
Let's give some notations for the description of these methods and the
statement of the results:
\subsection{Notations}\label{notations}
(i) Let $\chi =: \varphi \psi$ be an even character, where $\varphi$ is of
prime-to-$p$ order and $\psi$ of $p$-power order $p^e$, $e \geq 1$.
For $a \in \Z_p^\times$, let $\theta (a)$ be the unique $\xi \in {\rm tor}_{\Z_p}(\Z_p^\times)$
such that $a \equiv \xi \pmod p$ for $p \ne 2$ (resp. $a \equiv \xi \pmod 4$ for $p=2$).
The character $\theta$ will be considered as the cyclotomic character of $\Q(\mu_q)$
($q = p$ or $4$).
\smallskip
Let $K$ be the real cyclic field fixed by $\chi$ and let $M$ be the
subfield of $\Q^\ab$ fixed by the character $\theta^{-1}\,\chi$;
since $K$ is real, $M$ is an imaginary field.
Let $M_0$ be the maximal subfield of $M$ of prime-to-$p$ degree:
\smallskip
\quad $\bullet$ for $p \ne 2$, then $M \subseteq K(\mu_p)$, and $M_0$ is the fixed field of
$\theta^{-1}\,\varphi$ since $\theta$ is of order $p-1$; if $\varphi = \1$ (the unit character),
$M_0=\Q(\mu_p)$;
\smallskip
\quad$\bullet$ for $p = 2$, then $M \subset K(\sqrt {-1})$, and $M_0$ is the fixed field
of $\varphi$ since $\theta$ is of order $2$; if $\varphi = \1$, $M_0=\Q$.
\smallskip
(ii) Let ${\mathfrak m}$ be the maximal ideal of $\Q_p(\mu_{p^e})$ and let
$v_{\mathfrak m}$ be the corresponding valuation with image $\Z$ (so,
$v_{\mathfrak m} = (p-1)\,p^{e-1} \cdot v_p$).
\smallskip
(iii) For any prime number $r$, let $p^{d_r}$ be the degree of the splitting field
of $r$ in $M/M_0$; let $\BD = \sum_{\ell} p^{d_\ell}$, for $\ell \ne p$
ramified in $M/M_0$ and totally split in $M_0/\Q$; whence for $\ell \ne p$
such that:
\smallskip
\quad $\bullet$ for $p \ne 2$: $\psi(\ell) = 0$ and $(\theta^{-1}\,\varphi) (\ell) = 1$;
\smallskip
\quad $\bullet$ for $p = 2$: $\psi(\ell) = 0$ and $\varphi(\ell) = 1$ (let $\widehat K$
with $[K : \widehat K] = 2$; then $M$ is the quadratic extension of $\widehat K$
distinct from $K$ and $\widehat K(\sqrt {-1})$).
\subsection{Overview of the method}\label{method}
Genus theory does exist for $\BL_p$-functions,
as this was initiated in the 1970/80's with the genus theory of $p$-adic abelian
pseudo-measures \cite[Th\'eor\`eme (0.3)]{Gr33}, \cite[Th\'eor\`eme (0.1)]{Gr34}
including the tricky split case $\theta^{-1}\varphi (p) = 1$.
The results of genus theory under consideration claim that:
\begin{equation}\label{either/or}
\left\{\begin{aligned}
\hbox{either}\ \ \ \ v_{\mathfrak m} \big(\hbox{$\frac{1}{2}$}\,\BL_p(s, \chi) \big)
& > \BC(s),\,\forall s \in \Z_p, \\
\hbox{or}\ \ \ \ v_{\mathfrak m} \big(\hbox{$\frac{1}{2}$}\,\BL_p(s, \chi) \big)
& = \BC(s),\,\forall s \in \Z_p,
\end{aligned} \right.
\end{equation}
%
for an explicit $\BC(s)$, which is a constant $\BC$ as soon as
$\theta^{-1}\varphi (p) \ne1$ (see Theorem \ref{thmfond} defining
$\BC(s)$, essentially depending on $\BD$).
\smallskip
This allows to assign, for $p \in \{2, 3\}$, similar properties to the following
arithmetic $p$-invariants depending on values of $\BL_p$-functions:
\smallskip
({\bf a}) We will consider, at first (Section \ref{torsion}), the torsion group $\CT_{K, p}$
of the Galois group of the maximal abelian $p$-ramified (non-complexi\-fied)
pro-$p$-extension $H_{K, p}^\pr$ of $K$.
In the real abelian case, $\CT_{K, p} = \Gal(H_{K, p}^\pr/K_\infty)$, where $K_\infty$
is the cyclotomic $\Z_p$-exten\-sion of $K$.
\smallskip
Analytically (where $\sim$ means equality ``up to a $p$-adic unit factor''):
$$\order \CT_{K, p} \sim [K \cap\,\Q_\infty : \Q] \times \prd_{\chi \ne 1}
\hbox{$\frac{1}{2}$}\,\BL_p (1, \chi), $$
where $\chi$ runs trough the set of primitive Dirichlet's characters of $K$
\cite[Appendix]{C}, \cite{Se}, while class field theory elucidates it completely
with Formula \eqref{T=HRW} \cite{Gra8}, which explains that $\CT_{K, p}$
plays a central role in this study.
\smallskip
({\bf b}) Then we will focus (Section \ref{hilbert}) on the the Hilbert tame kernel $\BK_2(\BZ_K)$
of the ring of integers $\BZ_K$ of $K$ and, for $p=2$, we will replace it by the regular
kernel $\BR_2(\BZ_K)$ in the ordinary sense, linked to the Hilbert one via
the exact sequence \cite{Gar}:
\begin{equation}\label{regular}
1 \to \BR_2(\BZ_K) \too \BK_2(\BZ_K) \too (\Z/2\Z)^{[K\,:\,\Q]} \to 1.
\end{equation}
The class field theory information about the regular kernel comes in general
from reflection theorems giving formulas \eqref{rankR} and \eqref{rankT}; we have
refrained from citing all the articles re-proving well-known reflection theorems between
$\BK_2(\BZ_K)[p^\infty]$, $\CT_{K,p}$, $\CH_{K,p}$, published in the 1980--90's
by several authors.
\smallskip
({\bf c}) Then $\order \BK_2(\BZ_K)$, being in relation with the complex $\BL(-1, \chi)$'s,
we use (Section \ref{comparison}), when it is possible, the classical link between
$p$-adic and complex $\BL$-functions for some comparison of the two $p$-adic
invariants $\BK_2(\BZ_K)[p^\infty]$ and $\CT_{K, p}$\,\footnote{\,A $p$-invariant
may be denoted $\BX_K[p^\infty]$ when it is of the form $\BX_K \otimes \Z_p$, for a
{\it finite global invariant $\BX_K$}. Otherwise it is denoted $\CX_{K, p}$;
it is only conjectural \cite[Section 8]{Gr55} that $\CT_{K, p}$ can be written
$\BT_K \otimes \Z_p$; this conjecture is out of reach but may be coherent
with the finiteness of $\BH_K$ (class group) and $\BK_2(\BZ_K)$,
and is supported by many heuristics. For convenience, we
will also use the notation $\CX_{K, p}$ when $\BX_K$ does exist (e.g.,
$\CH_{K, p} = \BH_K[p^\infty]$).},
with the use of property \eqref{either/or} when it applies, thus giving new informations.
\subsection{Torsion group of abelian \texorpdfstring{$p$}{Lg}-ramification}
\label{torsion}
Class field theory is especially convenient to state the arithmetic properties of
$\CT_{K, p}$ (e.g., \cite[\S\,III.2.c, \S\,III.4.b, \S\,IV.3]{Gr5}, after some pioneering
works \cite{BP, Gr3, Jau1, Ng1, GJ} and many others).
\smallskip
Numerical aspects are given in \cite{Gr6} with PARI \cite{PARI} programs and
the order of magnitude of $\order \CT_{K, p}$ is studied in \cite{Gra9}.
\smallskip
We refer to Appendix \ref{A1} for the program computing the structure of $\CT_{K, p}$
in complete generality.
The following diagram is valid for any number field $K$ fulfilling Leopoldt's conjecture,
replacing $K_\infty$ by the compositum $\wt K$ of the $\Z_p$-extensions of $K$:
%
\unitlength=0.98cm
\[\vbox{\hbox{\hspace{-2.8cm}
\begin{picture}(11.5, 5.4)
% horizontales
\put(6.1, 4.50){\line(1, 0){1.6}}
\put(8.4, 4.50){\line(1, 0){2.3}}
\put(3.8, 4.50){\line(1, 0){1.2}}
\put(9.1, 4.2){\ft$\simeq {\mathcal W}_{K, p}$}
\put(4.2, 2.50){\line(1, 0){1.25}}
\bezier{350}(3.8, 4.8)(7.6, 5.5)(11.0, 4.8)
\put(7.2, 5.25){\ft$\CT_{K, p}$}
\bezier{350}(3.7, 4.2)(5.85, 3.4)(8.0, 4.2)
\put(6.0, 3.5){\ft$\CT_{K, p}^\bp$}
% vertical
\put(3.50, 2.9){\line(0, 1){1.25}}
\put(3.50, 0.9){\line(0, 1){1.25}}
\put(5.7, 2.9){\line(0, 1){1.25}}
\bezier{300}(3.9, 0.5)(4.7, 0.5)(5.6, 2.3)
\put(5.2, 1.3){\ft$\simeq \CH_{K, p}$}
\bezier{300}(6.0, 2.5)(8.5, 2.6)(10.8, 4.3)
\put(8.2, 2.7){\ft$\simeq \CU_{K, p}/\CE_{K, p}$}
\put(10.85, 4.4){\ft$H_{K, p}^\pr$}
\put(5.2, 4.4){\ft$\wt K H_{K, p}^\nr$}
\put(7.8, 4.4){\ft$H_{K, p}^\bp$}
\put(6.4, 4.2){\ft$\simeq {\mathcal R}_{K, p}$}
\put(3.4, 4.4){\ft$\wt K$}
\put(5.5, 2.4){\ft$H_{K, p}^\nr$}
\put(2.8, 2.4){\ft$H_{K, p}^\nr\! \cap\! \wt K$}
\put(3.4, 0.40){\ft$K$}
\end{picture}}}\]
Definitions and notations are the following:
\smallskip
(i) $\CH_{K, p}$ is the $p$-class group and $H_{K, p}^\nr$ the $p$-Hilbert class field;
\smallskip
(ii) $\CR_{K, p}$ is the {\it normalized} $p$-adic regulator, defined as a $\Z_p$-module
in \cite[Section 5, Proposition 5.2]{Gra8}; its order is the classical $p$-adic regulator,
up to an explicit factor;
\smallskip
(iii) $\CU_{K, p}$ is the group of principal local units at $p$, $\CE_{K, p}$ the closure in $\CU_{K, p}$
of the group of units $E_K$ and $\CW_{K, p} := {\rm tor}_{\Z_p}(\CU_{K, p})/\mu_p(K)$;
\smallskip
(iv) $H_{K, p}^\bp$ (resp. $\CT_{K, p}^\bp$), named in \cite{Ng1} the Bertrandias--Payan
field (resp. module), is issued from \cite{BP}; $H_{K, p}^\bp$ is the compositum of the
$p$-cyclic extensions of $K$ embeddable in $p$-cyclic extensions of arbitrary large
degree; see \cite{GJN} for complements. Whence:
%
\begin{equation}\label{T=HRW}
\order \CT_{K, p} = \ffrac{\order \CH_{K, p}}{[H_{K, p}^\nr \cap \wt K : K]}
\times \order \CR_{K, p} \times \order \CW_{K, p}.
\end{equation}
In the real case, $K_\infty/K$ is in general totally ramified,
so that the formula becomes $\order \CT_{K, p} = \order \CH_{K, p} \times
\order \CR_{K, p} \times \order \CW_{K, p}$.
\begin{remark}
From \cite[Proposition 5.2]{Gra8}, we can give the following examples:
\smallskip
(i) For $p=2$ and $K = \Q(\sqrt m)$ ($m>0$ square-free), this formula
becomes $\order \CT_{K, 2} \sim h_K \times \ffrac{\log_2(\varepsilon_K)}{2 \sqrt m}$,
where $h_K$ is the class number and $\varepsilon_K$ the fundamental unit.
\smallskip
(ii) For $p=3$ and a cyclic cubic field $K$, then $\order \CT_{K, 3} \sim h_K \times
\ffrac{R_{K, 3}}{3^u}$ with $u=2$ (resp. $u=1$) if $3$ is unramified (resp. ramified)
and where $R_{K, 3}$ is the usual $3$-adic regulator
\end{remark}
\subsection{Tame and regular Hilbert's kernels}\label{hilbert}
Let $K$ be any number field, let $p \geq 2$ be a prime number and let $S_p(K)$ be the set of
$p$-places of $K$; put $K' := K(\mu_p)$ and let $S_p(K')$ be the set of $p$-places of $K'$.
\smallskip
Recall that $\omega : \Gal(\Q(\mu_p)/\Q) \to \mu_{p-1}$, called the cyclotomic (or Teichm\"uller)
character, is defined by the Galois action on $\mu_p$ (conductor $p$, order $p-1$ for
$p \ne 2$); it is the unit character $\1$ for $p=2$ since $\mu_2 \subset \Q$. For $p \ne 2$,
we have in some sense $\omega = \theta$ (Notations \ref{notations}\,(i)), but for $p=2$,
do not confuse $\omega$ with $\theta$ as cyclotomic character of $\Gal(\Q(\mu_4)/\Q)$
(conductor $4$, order$2$).
\subsubsection{First reflection theorem}
The regular kernel $\BR_2(\BZ_K)$ (see \eqref{regular}) fulfills, from Tate's results \cite{Ta1, Ta2}
and reflection theorem \cite[Th\'eor\`eme 11.1]{Gr4} or \cite[Theorem II.7.7.3.1]{Gr5},
the following $p$-rank formula, where $\BH_{K'}^{S_p(K') {\rm res}}$ is the
$S_p(K')$-class group in the restricted sense, $\delta =1$ or$0$, according as
$\mu_p \subset K$ or not:
\begin{equation}\label{rankR}
\left\{ \begin{aligned}
&\rk_p(\BR_2(\BZ_K)) = \rk_p \big((\BH_{K'}^{S_p(K') {\rm res}})_{\omega^{-1}} \big) \\
& \hspace{1cm}+\order \{v \in S_p(K),\ \hbox{$v$ totally split in $K'/K$} \}-\delta, \\
& \rk_2(\BR_2(\BZ_K)) = \rk_2 \big(\BH_{K}^{S_2 {\rm res}} \big)
+\order S_2-1.
\end{aligned}\right.
\end{equation}
\subsubsection{Second reflection theorem}
Due to the reflection theorem between $p$-class groups and $p$-torsion groups
of abelian $p$-ramification theory, we get similarly \cite[Proposition III.4.2.2]{Gr5}:
\begin{equation}\label{rankT}
\left\{\begin{aligned}
& \rk_p(\CT_{K, p}) = \rk_p \big((\BH_{K'}^{S_p(K') {\rm res}})_\omega \big) \\
& \hspace{1cm}+\order \{v \in S_p(K),\,\hbox{$v$ totally split in $K'/K$} \}-\delta \\
& \rk_2(\CT_{K, 2}) = \rk_2 \big(\BH_{K}^{S_2 {\rm res}} \big)
+\order S_2-1.
\end{aligned}\right.
\end{equation}
From the two previous relations, we get:
\begin{equation}\label{equalranks}
\rk_p (\BR_2(\BZ_K)) = \rk_p (\CT_{K, p}),
\end{equation}
as soon as $\omega^2=1$, that is to say, if $K$ contains the maximal real
subfield of $\Q(\mu_p)$, a framework widely developed in \cite{GJ};
these techniques yielding generalizations as that of Keune \cite{K}.
\smallskip
The relation was proven in \cite[Theorems 1, 2]{Gr3} (in which $\BR_2(\BZ_K)$
was denoted $H_2^0K$), to characterize, for $p \in \{2, 3\}$, the abelian $p$-extensions
$K/\Q$ such that $\BR_2(\BZ_K)[p^\infty] = 1$ ($p$-regular fields studied in
\cite{GJ, BGr, RO}).
\smallskip
We note that $\Q(\mu_p)^+ \subseteq K$ is always fulfilled for $p \in \{2, 3\}$.
More generally, \cite[Remarque 11.5]{Gr4} gives, when $K$ contains $\mu_p$,
the corresponding relation \eqref {equalranks} with characters:
$$\rk_\chi (\BR_2(\BZ_K)) = \rk_{\omega^2 \chi^{-1}} (\CT_{K, p}). $$
Class field theory approach of the structures of $\BK_2(\BZ_K)[p^\infty]$ and
$\CT_{K, p}$ has given a huge literature, especially for rank
computations, often restricted to $p \in \{2, 3\}$, probably with
\eqref{equalranks} when $\omega^2 = \1$, to get practical results (e.g.,
\cite{BS, Gr3, K, Ber, Br1, Q1, YF, KM, Br2, Yue2, Q2, DL}).
\smallskip
For results in the case $p \geq 5$, see Section \ref{p>3}.
\subsection{Arithmetic v.s. analytic properties}\label{comparison}
To complete these class field theory aspects, we will use the following
($p$-adic and complex) analytic formulas and the correspondence, for
$p \in \{2, 3\}$, between $p$-adic and complex $\BL$-functions
of characters $\chi$ of $K$:
\begin{equation}\label{analytic}
\left\{\begin{aligned}
\order \CT_{K, p} & \sim [K \cap\,\Q_\infty : \Q] \times \prd_{\chi \ne 1}
\ffrac{1}{2}\,\BL_p (1, \chi), \\
\order \BK_2(\BZ_K) & = 2^{[K : \Q]}\times \Big( \ffrac{1}{6} \prd_{[F:K]=2}
\ffrac{\order \mu(F)}{2} \Big) \times \prd_{\chi \ne 1}\ffrac{1}{2}\,\BL (-1, \chi),
\end{aligned}\right.
\end{equation}
\noindent
where $\mu(F)$ is the group of roots of unity of the field $F$.
\smallskip
For more history and theoretical contributions, one may refer to
\cite{Ba, Ta1, Gar, Ta2, Jau1, Ng1, Hur, GJ, K, Ko2, Ng,
HK, RW, Ko1, Yue1, Gr5, Q3}, among many others, dealing with the more
concrete results we have cited previously; for some generalizations to the
wide \'etale kernels ${\bf WK}_{2i}(K)$, see for instance \cite{Ng2, JS, JM, AAM}.
\smallskip
But our goal is to return to very simple effective $p$-adic methods allowing
computations, by means of pseudo-measures. We intend to provide, simultaneously,
survey and history parts about these various questions.
\smallskip
These pseudo-measures are the Mellin transform (via a reflection principle) of
Stickelberger elements, and give rise, after twisting, to $p$-adic measures of the
form $\big(\CS^*_{\!L_n}(c)(s) \big)_n$, $s \in \Z_p$, in the group algebras
$\Z_p/qp^n \Z_p[\Gal(L_n/\Q)]$ of the layers $L_n := K(\mu_{qp^n})$ of the cyclotomic
$\Z_p$-extension of $K(\mu_q)$ ($q=4$ or $p$) \cite[Section II]{Gr1}.
\smallskip
This was done after the main pioneering works of Kubota--Leopoldt \cite{KL},
then Fresnel \cite{Fre}, Amice--Fresnel \cite{AF}, Iwasawa \cite{Iw}, Coates \cite{C},
Serre \cite{Se}, giving many results, as the value of the residue of Dedekind
$p$-adic $\zeta$-functions at $s=1$ and some annihilation theorems for $p$-class
groups, torsion groups $\CT_{K, p}$ and similar invariants \cite{Gr1, Gr2, Jau2, BN,
Jau3, Jau4}; this was considerably generalized to a totally real base field by
Colmez in \cite{Col} and Deligne--Ribet in \cite{DR}.
\smallskip
The Iwasawa framework of these constructions is very similar and
is detailed in \cite[\S\,7.2]{W}.
\subsection{Main results of the article}\label{main}
We describe three results about the study and the comparison of the
modules $\BK_2(\BZ_K)[p^\infty]$ and $\CT_{K, p}$:
\medskip
{\bf (a)} From the properties of the constant $\BC$ described in \S\,\ref{method}
and stated in Theorem \ref{thmfond}, we obtain the following results for
quadratic fields ($p=2$) and cyclic cubic ones ($p=3$):
\subsubsection*{\bf Theorem A} \label{A} (see Theorem \ref{thmfinal} for the proof).
\smallskip
(i) {\it Let $K = \Q(\sqrt m)$ be a real quadratic field of conductor $\ne 8$.}
\smallskip
{\it Assume that $v_2(\order \CT_{K, 2}) = \BC$
(equivalent to $v_2 \big(\frac{1}{2} \BL_2(1, \chi) \big) = \BC$
from formula \eqref{analytic}); then}
$\order \BK_2(\BZ_K)[2^\infty] = 2^{\hbox{\tiny $\BC$}+2}$.
\smallskip
(ii) {\it Let $K$ be a cyclic cubic field of conductor $\ne 9$.}
{\it Assume that $v_3(\order \CT_{K, 3}) = \BC$
(so, $v_3 \big(\BL_3(1, \chi) \times\BL_3(1, \chi^2) \big) = \BC$);
then} $\order \BK_2(\BZ_K)[3^\infty] = 3^{\hbox{\tiny $\BC$}}$.
\smallskip
(iii) {\it If $v_2 (\CT_{K, 2} ) > \BC$ (resp. $v_3 (\CT_{K, 3} ) > \BC$), }
$\order \BK_2(\BZ_K)[2^\infty] > 2^{\hbox{\tiny $\BC$}+2}$ {\it (resp.}
$\order \BK_2(\BZ_K)[3^\infty] > 3^{\hbox{\tiny $\BC$}}$).
\medskip
We compute, in Appendix \ref{B1}, $\order \BR_2(\BZ_K) [2^\infty]$ for quadratic
fields by means of explicit pseudo-measures yielding expressions of
Theorem \ref{Lp} and Corollary \ref{s=-1}; this allows safe verifications.
\medskip
{\bf (b)} We apply the case of equality to a family of real quadratic fields introduced
in \cite[Theorem 1.2]{DL} and prove the Deng--Li conjecture on
$\BK_2(\BZ_K)[2^\infty]$ in the following Theorem\,{\bf B} (Section \ref{DengLi}):
\subsubsection*{\bf Theorem B}\label{B} {\it
Let $K = \Q(\sqrt m)$, $m= \ell_1 \ell_2 \cdots \ell_n$, $n$ even, with:
\smallskip
$\bullet$ $\ell_1 \equiv 3 \pmod 8$,\ \ $\ell_i \equiv 5 \pmod 8$ for $i \geq 2$,
\smallskip
$\bullet$ $\big( \frac{\ell_1}{\ell_2} \big) = -1$, $\big( \frac{\ell_1}{\ell_j} \big) = 1$ for $j \geq 3$,
\ \ $\big( \frac{\ell_i}{\ell_j} \big) = -1$ for $2 \leq i < j \leq n$.
\smallskip
Then $\CH_{K, 2} \simeq (\Z/2\Z)^{n-1}$, $\CT_{K, 2} \simeq (\Z/2\Z)^{n-2} \times \Z/4\Z$, and:}
$$\BK_2(\BZ_K)[2^\infty] \simeq (\Z/2\Z)^{n-1} \times \Z/2^3 \Z\ \hbox{(Deng--Li conjecture).}$$
\medskip
{\bf (c)} The cases $p \in \{2, 3\}$ being very specific (thus well studied in the literature)
because of the relation \eqref{flp} for $m=2$ between $p$-adic and complex
$\BL$-functions, we have obtained the following Theorem\,{\bf C} when $p$ is
distinct from $2$ and $3$ (see Theorem \ref{pdiv} for a much more general
statement over a real base field $k$ of prime-to-$p$ degree):
\subsubsection*{\bf Theorem C} \label{C}
{\it Let $K/\Q$ be a cyclic $p$-extension for $p \geq 5$ and let $S_\ta$ be the
set of primes $\ell \ne p$ ramified in $K/\Q$.
\smallskip\noindent
Then $\rk_p(\BK_2(\BZ_K)) \geq
\order S_\ta$. In particular, $\BK_2(\BZ_K)[p^\infty] = 1$ if and only if
$K$ is contained in the cyclotomic $\Z_p$-extension $\Q_\infty$ of $\Q$.}
\section{Computation of \texorpdfstring{$\order \BK_2 (\BZ_K)[p^\infty]$,
$p \in \{2, 3\}$}{Lg}}
\subsection{Birch-Tate formula and \texorpdfstring{$\BL_p$}{Lg}-functions}
Let $\zeta_K(s)$ be the De\-dekind zeta-function of the real abelian number field $K$.
\smallskip
Then $\zeta_K(s) = \prd_{\chi} \BL(s, \chi)$, as product of the complex $\BL$-functions,
where $\chi$ runs trough the set of primitive Dirichlet's characters of $K$, for
which $\BL(s, \1) = \zeta_\Q^{}(s)$ for the unit character.
%
In the $p$-adic context, definition of $\BL_p$-functions is as follows
(e.g., \cite[Section 5\,(a), Remarque]{Fre}, Amice--Fresnel \cite[Section 1]{AF}):
\begin{equation}\label{flp}
\left\{\begin{aligned}
& \BL_p(1-m, \chi) = (1- p^{m-1}\,\chi(p))\,\BL(1-m, \chi),
\hbox{\ for:} \\
&\hspace{0.6cm} m > 1\ \&\ \,m \equiv 0 \pmod {(p-1)}\ \hbox{if $p > 2$}, \\
&\hspace{0.6cm} m > 1\ \&\ \,m \equiv 0 \pmod 2\ \hbox{if $p = 2$}.
\end{aligned}\right .
\end{equation}
This allows to compute for instance $\BL(-1, \chi)$ in terms of
$\BL_p$-func\-tions for $p=2$ and $p = 3$ since $m=2$ fulfills the congruent
conditions required in \eqref{flp}.
\smallskip
The Birch-Tate formula is the following equality (proved as consequence
of the Mazur--Wiles ``Main Theorem'' in abelian theory and complements
in the case $p=2$ \cite{Ko2}):
\begin{equation}\label{BT}
\left\{\begin{aligned}
\order \BK_2 (\BZ_K) & = w_2(K)\,\zeta_K(-1) = w_2(K)\,\zeta_\Q(-1)\!\!
\prd_{\chi \ne \1} \BL(-1, \chi), \\
w_2(K) & = 4 \cdot \prd_{[F:K]=2} \ffrac{\order \mu(F)}{2}\,\ \
\hbox{(see \eqref{analytic})}.
\end{aligned}\right .
\end{equation}
For instance, $w_2(\Q) = 24$ and $\zeta_\Q^{}(-1) = \ffrac{1}{12}$; then
$\order \BK_2 (\Z) = 2$, giving $\order \BR_2 (\Z) = 1$ as expected since
$\CT_{\Q, p} = 1$ for all $p$.
\smallskip
Thus, the computation of the $p$-Sylow subgroup
$\order \BK_2(\BZ_K) [p^\infty]$ of $\BK_2 (\BZ_K)$ is possible for $p=2$
and $p=3$ with the following formulas (using $\zeta_\Q^{}(-1) = \ffrac{1}{12}$
and \eqref{flp} for $m=2$):
%
\begin{equation*}
\left\{\begin{aligned}
\order \BK_2 (\BZ_K)[2^\infty] & \sim \ffrac{w_2(K)}{12} \times\!\!
\prd_{\chi \ne \1} \ffrac{1}{1-2 \chi(2)}\,\BL_2(-1, \chi), \\
\order \BK_2 (\BZ_K) [3^\infty] & \sim \ffrac{w_2(K)}{12} \times\!\!
\prd_{\chi \ne \1} \ffrac{1}{1-3 \chi(3)}\,\BL_3(-1, \chi),
\end{aligned}\right .
\end{equation*}
where $\sim$ means equality up to a $p$-adic unit factor for the considered$p$.
Since the denominators $1-p\,\chi(p)$ are invertible in $\Z_p$, the formulas
become with expression \eqref{BT} of $w_2(K)$:
\begin{equation*}
\left\{\begin{aligned}
\order \BK_2 (\BZ_K) [2^\infty] & \sim \prd_{[F:K]=2} \ffrac{ \order \mu_2(F) }{2}
\times\!\! \prd_{\chi \ne \1}\,\BL_2(-1, \chi), \\
\order \BK_2(\BZ_K) [3^\infty] & \sim \ffrac{1}{3} \prd_{[F:K]=2} \order \mu_3(F)
\times\!\! \prd_{\chi \ne \1}\,\BL_3(-1, \chi).
\end{aligned}\right .
\end{equation*}
\subsection{Formulas for quadratic and cubic fields}\label{formulas}
We deduce the following expressions with $p \in \{2, 3\}$;
due to exact sequence \eqref{regular} when $p=2$, we do not write the
corresponding results for $\BR_2(\BZ_K)$.
\smallskip
({\bf a}) Real quadratic fields.
\smallskip
\quad (i) For real quadratic fields $K = \Q(\sqrt m)$, $m \ne 2, 3$, $p=2$:
\begin{equation*}
\left\{\begin{aligned}
\prd_{[F:K] = 2} \ffrac{1}{2} \cdot \order \mu_2(F) & = 2
\hbox{ ($F=K(\mu_4)$, $F=K(\mu_6)$)}; \\
\order \BK_2(\BZ_K) [2^\infty] & \sim 2 \cdot \BL_2(-1, \chi).
\end{aligned}\right .
\end{equation*}
\quad (ii) For $\Q(\sqrt 2)$, $\prd_{[F:K]=2} \ffrac{\order \mu_2(F)}{2} = 4$
($F=K(\mu_8)$, $F=K(\mu_6)$); so,
$\order \BK_2\BZ_{\Q(\sqrt 2)} [2^\infty] \sim 4 \cdot \BL_2(-1, \chi) \sim 4$.
\smallskip\smallskip
\quad (iii) For $\Q(\sqrt 3)$, $\prd_{[F:K]=2} \ffrac{\order \mu_2(F)}{2} = 4$
($F=K(\mu_4)$, $F=K(\mu_6) = K(\mu_4)$); so,
$\order \BK_2\BZ_{\Q(\sqrt 3)} [2^\infty] \sim 4 \cdot \BL_2(-1, \chi) \sim 8$.
\smallskip
\quad (iv) For $p=3$, $K \ne \Q(\sqrt 3)$, the formulas are:
\begin{equation*}
\left\{\begin{aligned}
\ffrac{1}{3} \prd_{[F:K] = 2} \order \mu_3(F) & = 1 \hbox{ ($F=K(\mu_6)$)}; \\
\order \BK_2(\BZ_K) [3^\infty] & \sim \BL_3(-1, \chi).
\end{aligned}\right .
\end{equation*}
\smallskip
\quad (v) For $\Q(\sqrt 3)$, $\ffrac{1}{3} \prd_{[F:K]=2} \order \mu_3(F) = 3$
($F=K(\mu_3)$, $F=K(\mu_4) = K(\mu_6)$); so,
$\order \BK_2\BZ_{\Q(\sqrt 3)} [3^\infty] \sim 3 \cdot \BL_3(-1, \chi) \sim 3$.
\medskip
({\bf b}) Cyclic cubic fields.
\smallskip
\quad (i) For cubic fields $K$ of conductor distinct from $9$, $p=3$:
\begin{equation*}
\left\{\begin{aligned}
\ffrac{1}{3} \prd_{[F:K]=2} \order \mu_3(F) & = 1
\hbox{ ($F=K(\mu_6)$)}; \\
\order \BK_2 (\BZ_K) [3^\infty] & \sim \BL_3(-1, \chi) \times \BL_3(-1, \chi^2),
\end{aligned}\right .
\end{equation*}
for the two conjugate characters of order $3$ of $K$.
\smallskip
\quad (ii) For the cubic field of conductor $9$,
$\ffrac{1}{3} \prd_{[F:K]=2} \order \mu_3(F) = 3$ ($F=K(\mu_9)$);
then $\order \BK_2(\BZ_K) [3^\infty] = 3 \cdot \BL_2(-1, \chi) \sim 1$.
\smallskip
\quad (iii) For $p=2$, the formulas are, for the two conjugate characters of order
$3$ of $K$:
\begin{equation*}
\left\{\begin{aligned}
\prd_{[F:K]=2} \ffrac{1}{2}\cdot \order \mu_2(F) & = 2
\hbox{ ($F=K(\mu_4)$, $F=K(\mu_6)$)}; \\
\order \BK_2 (\BZ_K) [2^\infty] & \sim 2 \cdot \BL_2(-1, \chi) \times \BL_2(-1, \chi^2).
\end{aligned}\right .
\end{equation*}
\section{Definition of a \texorpdfstring{$p$}{Lg}-adic pseudo-measure
\texorpdfstring{$(\CS_{\!L_n})_n$}{Lg}}
Let $K$ be a real abelian field and put $L_n := K(\mu_{qp^n})$, $q \in \{p, 4\}$,
as usual, and $n \geq 0$.
\subsection{The Stickelberger elements}
The conductor of $L_n$ is of the form $f_{L_n} = q p^n f$, for a prime-to-$p$
integer $f$, taking $n$ large enough if $p$ ramifies in $K$ (otherwise, $f$ is
the conductor of $K$) (in formulas we shall abbreviate $f_{L_n}$ by $f_n$).
\smallskip
Then put, where all Artin symbols are taken over $\Q$:
\centerline{$\CS_{\!L_n} := -\sm_{a=1}^{f_n} \Big(\ffrac{a}{f_n}-\ffrac{1}{2} \Big)
\,\Big(\ffrac{L_n}{a} \Big)^{-1}$, }
\noindent
as restriction to $L_n$ of $\CS_{\Q(\mu_{f_n}^{})}$,
where $a$ runs trough the prime-to-$f_n$ integers $a \in [1, f_n]$.
\subsection{Norms of the Stickelberger elements}
Let $f$ and $m$ be such that $m \mid f$; consider $\Q(\mu_f^{})$ and $\Q(\mu_m)$
and let $\Norm_{\Q(\mu_f^{})/\Q(\mu_m)}$ be the restriction map
$\Q[\Gal(\Q(\mu_f^{}) / \Q)] \to \Q[\Gal(\Q(\mu_m)/\Q)]$.
We have:
\begin{equation}\label{norms}
\Norm_{\Q(\mu_f^{})/\Q(\mu_m)} (\CS_{\Q(\mu_f^{})}) =
\prd_{\ell \mid f,\,\ell \nmid m} \!\! \Big(1-\Big(\ffrac{\Q(\mu_m)}{\ell} \Big)^{-1}\Big)
\cdot \CS_{\Q(\mu_m)}.
\end{equation}
Let $L/K$ be an extension of abelian fields of conductors $m$ and $f$;
we define $\CS_L:= \Norm_{\Q(\mu_f^{})/L}(\CS_{\Q(\mu_f^{})})$ and
$\CS_K:= \Norm_{\Q(\mu_m^{})/K}(\CS_{\Q(\mu_m^{})})$, respectively;
then, $\Norm_{L/K}(\CS_{\!L}) = \prd_{\ell \mid f,\ \ell \nmid m}\!\!
\Big(1-\Big(\ffrac{K}{\ell} \Big)^{-1}\Big) \cdot \CS_K$. This implies
$\Norm_{L/K}(\CS_L) = 0$ as soon as a prime $\ell \mid f$ totally splits in $K$.
\subsection{Twists of the Stickelberger elements}
Let $c$ be an integer prime to $2 p f$; for $L_n = K(\mu_{qp^n}^{})$, put:
$\CS_{\!L_n}(c) := \Big(1-c\Big(\ffrac{L_n}{c} \Big)^{-1} \Big)\cdot \CS_{\!L_n}$.
%
Then $\CS_{\!L_n}(c) \in \Z[\Gal(L_n/\Q)]$. Indeed, we have:
$$\CS_{\!L_n} (c)=\ffrac{-1}{f_n} \sm_a
\Big[a \Big(\ffrac{L_n}{a} \Big)^{-1}-a c \Big(\ffrac{L_n}{a} \Big)^{-1}\Big(\ffrac{L_n}{c} \Big)^{-1}\Big]
+ \ffrac{1-c}{2} \sm_a \Big(\ffrac{L_n}{a} \Big)^{-1}; $$
%
let $a'_{c} \in [1, f_n]$ be the unique integer such that
$a'_{c} \cdot c \equiv a \pmod {f_n}$;
put $a'_{c} \cdot c-a = \lambda^n_a(c) f_n$,
$\lambda^n_a(c) \in \Z$; then, using the bijection
$a \mapsto a'_{c}$ in the second summation and the fact that
$\Big(\ffrac{L_n}{a'_{c}} \Big) \Big(\ffrac{L_n}{c} \Big) = \Big(\ffrac{L_n}{a} \Big)$:
\begin{equation}\label{twist}
\left\{\begin{aligned}
\CS_{\!L_n}(c) & = \ffrac{-1}{f_n} \Big[ \sm_a a \Big(\ffrac{L_n}{a} \Big)^{-1}\!\!-
\sm_a a'_{c} c \Big(\ffrac{L_n}{a'_{c}} \Big)^{-1} \! \Big(\ffrac{L_n}{c} \Big)^{-1}\Big] \\
&\hspace{4.1cm}+\ffrac{1-c}{2} \sm_a \Big(\ffrac{L_n}{a} \Big)^{-1} \\
& = \sm_a \Big( \lambda^n_a(c)+\ffrac{1-c}{2} \Big) \Big(\ffrac{L_n}{a} \Big)^{-1}
\!\! \in \Z[\Gal(L_n/\Q)].
\end{aligned}\right.
\end{equation}
This twisted form gives the Stickelberger $p$-adic measure used to generate the
$\BL_p$-functions of $K$ as explained in the next Section.
\section{The measure \texorpdfstring{$(\CS^*_{\!L_n}(c))_n$}{Lg}
defining \texorpdfstring{$\BL_p(s, \chi)$}{Lg}}
Consider the algebras $A_n := \Z_p / qp^n\Z_p[\Gal(L_n/\Q)],\ n \geq 0$.
The Mellin transform (e.g., \cite[\S\,II.1]{Gr1}) is defined, on $A_n$,
by the following image of any $\sigma \in G$, where $a$,
defined modulo $f_n = qp^n f$, represents $\sigma$ as Artin symbol:
\begin{equation*}
\left \{ \begin{aligned}
\sigma & = \Big(\ffrac{L_n}{a} \Big) \mapsto
\theta(a) \langle a \rangle^s \Big(\ffrac{L_n}{a} \Big)^{-1}
= a \langle a \rangle^{s-1} \Big(\ffrac{L_n}{a} \Big)^{-1} \\
& = a \langle a \rangle^{s-1} \sigma^{-1} \pmod {qp^n f},\ \,s \in \Z_p,
\end{aligned} \right.
\end{equation*}
and the image, in $A_n$, of the expression \eqref{twist} of $\CS_{\!L_n}(c)$
by this transform
yields the $p$-adic measure in $\ds \varprojlim_n A_n$ giving annihilation
theorems for real invariants and $p$-adic functions $\BL_p(s, \chi)$ for $n \to \infty$;
indeed, since $L_n = K(\mu_{qp^n})$, one gets, from the norm relations \eqref{norms},
$$\Norm_{L_{n+1}/L_n} (\CS^*_{\!L_{n+1}}(c)(s)) \equiv \CS^*_{\!L_n}(c)(s)
\pmod{qp^n},\ \,n \geq 0. $$
So, we will use the following approximations:
\begin{equation*}
\CS^*_{\!L_n}(c)(s) \equiv \sm_{a=1}^{f_n} \Big(\lambda^n_a(c)
+ \ffrac{1-c}{2} \Big)\,a^{-1} \langle a \rangle^{1-s} \Big(\ffrac{L_n}{a} \Big)\!\!
\pmod{qp^n}.
\end{equation*}
\subsection{Approximations modulo \texorpdfstring{$q p^n$}{Lg}
of \texorpdfstring{$\BL_p(s, \chi)$}{Lg}}
The $\BL_p$-functions, of even Dirichlet's characters, are obtained by
means of the Mellin transform of the previous twisted pseudo-measures,
so that the Mellin transform of $\Big(1-c\Big(\ffrac{L_n}{c} \Big)^{-1} \Big)$ is
$\Big(1- \langle c \rangle^{1-s} \Big(\ffrac{L_n}{c} \Big)\Big)$ and is a
factor of the $p$-adic measure $\big(\CS^*_{\!L_n}(c)(s) \big)_n$ that must be dropped
when this makes sense (thus in any case, except $\chi = \1$ and $s=1$):
\begin{theorem}\label{Lp}
Let $c$ be any odd integer, prime to $p$ and to the conductor $f$
of $K$. For all $n$ large enough, let $f_n$ be the conductor of $L_n = K(\mu_{qp^n})$,
and for all $a \in [1, f_n]$, prime to $f_n$, let $a'_{c}$ be the unique integer in $[1, f_n]$
such that $a'_{c} \cdot c-a = \lambda^n_a(c) f_n$, $\lambda^n_a(c) \in \Z$.
The twisted $p$-adic measure generating the $\BL_p$-functions is given by
the restriction $\CS^*_{K, n}(c)(s)$ of $\CS^*_{\!L_n}(c)(s)$ to $K$, giving, for
all $n \geq 0$:
\centerline{$\CS^*_{K, n}(c)(s) \equiv \sm_{a=1}^{f_n} \Big[ \lambda^n_a(c)+\ffrac{1-c}{2}\Big]
\,a^{-1}\langle a \rangle^{1-s} \Big(\ffrac{K}{a} \Big) \pmod{qp^n}$.}
\smallskip\noindent
Whence the $\BL_p$-functions of Dirichlet's character $\chi$ of $K$:
\centerline{$\BL_p(s, \chi) = \ffrac{1}{1-\chi(c) \langle c \rangle^{1-s}} \times {\ds \lim_{n \to \infty}}
\sm_{a = 1}^{f_n} \big [\lambda_a^n(c) +\ffrac{1-c}{2} \big]\,a^{-1} \langle a \rangle^{1-s} \chi(a)$.}
\end{theorem}
This formula is used in Appendices \ref{B1}, \ref{B2}, \ref{C1}, \ref{C2}, \ref{D3},
for explicit numerical computations.
\begin{corollary}\label{s=-1}
Assume that $p \in \{2, 3\}$; in this context, $\theta^2 = \1$.
Since for $s=-1$, $a^{-1} \langle a \rangle^{1-s} = a^{-1} \langle a \rangle^2 = a$,
we get:
\centerline{$\BL_p(-1, \chi) = \ffrac{1}{1-\chi(c) \langle c \rangle^2} \times {\ds \lim_{n \to \infty}}
\sm_{a = 1}^{f_n} \big [\lambda_a^n(c) +\ffrac{1-c}{2} \big]\,a \chi(a)$.}
\end{corollary}
\begin{corollary}\label{formulasbis}
Assume $\chi \ne \1$ and $\chi(c) \ne 1$. For $s=-1$, $\langle c \rangle
\equiv 1 \pmod q$ and $\chi(c) \in \mu_{[K:\Q]}^{} \setminus \{1\}$;
so, in the quadratic case with $p=3$ and in the cubic case with $p=2$,
$1-\chi(c) \langle c \rangle^2$ is invertible. Then:
\smallskip
$\bullet$ $1-\chi(c) \langle c \rangle^2 \sim 2$ for the quadratic case and $p=2$,
\smallskip
$\bullet$ $1-\chi(c) \langle c \rangle^2\sim 1-\zeta_3$ for the cubic case and $p=3$.
\smallskip\noindent
From formulas of Section \ref{formulas} one gets:
\smallskip
(i) For $p=2$ in the quadratic case $K=\Q(\sqrt m)$, $m \ne 2, 3$ and a half summation
giving $\frac{1}{2}\,\BL_2(s, \chi)$, for
$\order \BR_2(\BZ_K)[2^\infty] \!=\! \ffrac{1}{4} \order \BK_2(\BZ_K)[2^\infty]$
one obtains (since $1-\chi(c) \langle c \rangle^2 \sim 2$):
\begin{equation*}
\left \{\begin{aligned}
\order \BR_2(\BZ_K)[2^\infty] & \sim \ffrac{1}{2}\,\BL_2(-1, \chi) \\
& \equiv \ffrac{1}{2}\! \sm_{a = 1}^{f_n/2} \big [\lambda_a^n(c)
+ \ffrac{1-c}{2} \big] a \chi(a) \pmod {2^n}.
\end{aligned}\right.
\end{equation*}
\smallskip
(ii) For the cubic case with $p=3$,
$(1-\chi(c) \langle c \rangle^2)(1-\chi^2 (c) \langle c \rangle^2)
\sim (1 - \zeta_3)( 1 - \zeta_3^2) \sim 3$, giving, with $\Norm = \Norm_{\Q(\mu_3)/\Q}$:
\begin{equation*}
\left \{\begin{aligned}
\order \BK_2(\BZ_K)[3^\infty] & \sim \BL_3(-1, \chi) \cdot \BL_3(-1, \chi^2) \\
& \equiv \ffrac{1}{3}\,\Norm \Big( \sm_{a = 1}^{f_n/2} \big [\lambda_a^n(c)
+ \ffrac{1-c}{2} \big] a \chi(a) \Big) \!\! \pmod {3^n}.
\end{aligned}\right.
\end{equation*}
\end{corollary}
\subsection{Genus theory of \texorpdfstring{$p$}{Lg}-adic pseudo-measures}
We have obtained in \cite[Th\'eor\`eme 0.2]{Gr34}, the following results
which does not seem very known but states the property of ``stability''
mentioned in the Introduction and may be linked to the existence (or not) of
zeroes of $\BL_p$-functions (for examples about zeroes of $\BL_p$-functions,
one may refer to \cite{W1, W2, W3, W4, W} and see Appendix \ref{C3} for some
illustrations of this influence):
\begin{theorem} \label{thmfond}
Under Notations \ref{notations}, where $\chi = \varphi \psi$ is even, $\varphi$
of prime-to-$p$ order and $\psi$ of order $p^e$, $e \geq 1$, we have, with
$\BD = \sum_{\ell} p^{d_\ell}$ and $\epsilon = 0$ (resp.
$\epsilon = 1$) when $ \varphi \ne \1$ (resp. $\varphi = \1$):
\smallskip
{\bf (a)} $v_{\mathfrak m} \Big(\frac{1}{2} \BL_p(s, \chi) \Big) \geq \BC(s),\,\forall\,s \in \Z_p$,
where $\BC(s)$ is as follows:
\smallskip
\quad ($i_p$) $p \ne 2$ and $\theta^{-1} \varphi (p) \ne 1$ (i.e., $p$ not totally split in $M_0$); then:
$$\BC(s) = \BC = \BD-\epsilon. $$
\smallskip
\quad ($i_2$) $p = 2$ and $ \varphi(2) \ne 1$ (i.e., $2$ not totally split in $M_0$); then:
$$\BC(s) = \BC=\BD. $$
\smallskip
\quad ($ii_p$) $p \ne 2$, $\theta^{-1} \varphi (p) = 1$ and
$\theta^{-1} \chi (p) \ne 1$ (i.e., $p$ totally split in $M_0$ and not totally split in $M$); then:
$$\BC(s) = \BC=\BD+p^{d_p}. $$
\smallskip
\quad ($ii_2$) $p = 2$, $ \varphi(2) = 1$ and $\theta^{-1} \chi (2) \ne 1$ (i.e., $2$
totally split in $M_0$ and not totally split in $M$); then:
$$\BC(s) = \BC=\BD+2^{d_2}-2 \epsilon. $$
\smallskip
\quad ($iii_p$) $p \ne 2$, $\theta^{-1} \chi (p) = 1$ (i.e., $p$ totally split in $M$); then:
$$\BC(s) :=\BD+ v_{\mathfrak m}(p\,s). $$
\smallskip
\quad ($iii_2$) $p = 2$, $\theta^{-1} \chi (2) = 1$ (i.e., $2$ totally split in $M$); then:
$$\BC(s) :=\BD+ v_{\mathfrak m}(q\,s)-2 \epsilon. $$
\smallskip
{\bf (b)} We have either the equality:
$$v_{\mathfrak m} \big(\ffrac{1}{2} \BL_p(s, \chi) \big) = \BC(s),\,\forall\,s \in \Z_p, $$
or the strict inequality:
$$v_{\mathfrak m} \big(\ffrac{1}{2} \BL_p(s, \chi) \big) >
\BC(s),\,\forall\,s \in \Z_p, $$
(with $s \ne 0$ in cases ($iii_p$) and ($iii_2$)).
\end{theorem}
So, under the case of stability $v_{\mathfrak m} \big(\frac{1}{2} \BL_p(s, \chi) \big)
= \BC(s),\,\forall\,s \in \Z_p$, a computation may be done at a suitable value
$s$ for $\BL_p(s, \chi)$, e.g., $s=1$ giving $\order \CT_{K, p} \sim
[K \cap\,\Q_\infty : \Q] \times \prd_{\chi \ne 1} \ffrac{1}{2}\,\BL_p (1, \chi)$;
thus $\order \CT_{K, 2} \sim \ffrac{1}{2}\,\BL_2 (1, \chi)$ for $K$ real quadratic,
except $K$ of conductor $8$, or $\order \CT_{K, 3} \sim \BL_3 (1, \chi)
\times \BL_3 (1, \chi^2)$ for $K$ cyclic cubic, except $K$ of conductor $9$.
\smallskip
Which gives the main process and Theorem \ref{main}\,{\bf A},
where $\BC(s) = \BC$, as soon as $s \in \Z_p^\times$,
since $v_{\mathfrak m}(s) = 0$ in cases (iii) (e.g., $s = \pm 1$):
\begin{theorem} \label{thmfinal}
We obtain the following results:
\smallskip
(i) Let $K$ be a real quadratic field of character $\chi$ of conductor $\ne 8$, and
set $p=2$. Assume that the knowledge of $\order \CT_{K, 2}$ (e.g., using formula
\eqref{T=HRW}) implies $v_2 \big(\frac{1}{2} \BL_2(1, \chi) \big) = \BC$; then
$v_2 \big(\frac{1}{2} \BL_2(s, \chi) \big) = \BC,\,\forall\,s \in \Z_2$, whence
in particular, $\order \BK_2(\BZ_K)[2^\infty] = 2^{\hbox{\tiny $\BC$}+2}$.
\smallskip
We have $\BC = \BD$ (resp. $\BC = \BD-1$) if $2$ splits (resp. does not split)
in $M = \Q(\sqrt {-m})$.
\smallskip
(ii) Let $K$ be a cyclic cubic field of character $\chi$ of conductor $\ne 9$,
and set $p=3$. Assume that the knowledge of $\order \CT_{K, 3}$ implies
$v_{\mathfrak m} \big(\BL_3(1, \chi) \big) = \BC$; then $v_{\mathfrak m}
\big(\BL_3(s, \chi) \big) = \BC,\,\forall\,s \in \Z_3$, then
in particular, $\order \BK_2(\BZ_K)[3^\infty] = 3^{\hbox{\tiny $\BC$}}$.
We have $\BC = \BD-1$.
\smallskip
(iii) If $v_2 \big(\frac{1}{2} \BL_2(1, \chi) \big) > \BC$
(resp. $v_{\mathfrak m} \big(\BL_3(1, \chi) \big) > \BC$), then one obtains
$\order \BK_2(\BZ_K)[2^\infty] > 2^{\hbox{\tiny $\BC$}+2}$ (resp.
$\order \BK_2(\BZ_K)[3^\infty] > 3^{\hbox{\tiny $\BC$}}$).
\end{theorem}
Considering for instance \cite[Theorem 1.2]{DL} giving, for this specific family of
real quadratic fields:
$$\BK_2(\BZ_K)[2^\infty] \simeq (\Z/2\Z)^{n-1} \times \Z/2^\delta\Z,\ \,\delta \geq 3, $$
the Deng--Li conjecture is $\delta = 3$; we intend to prove it by means of
the previous process from showing that $\order \CT_{K, 2} = 2^{n}$:
\begin{corollary}\label{remfond}
Under the conditions: $\order \CH_{K, 2} = 2^{n-1}$, $\CR_{K, 2}=1$
and $\order \CW_{K, 2} = 2$, then $\order \CT_{K, 2} = 2^n$,
giving the valuation $n$ for $\frac{1}{2} \BL_2(1, \chi)$; so, Theorem \ref{thmfinal}\,(i)
yields $\BC = \BD = n$, since $2$ splits in $\Q(\sqrt {-m})$, and the conjecture
$\delta = 3$ follows.
\end{corollary}
\subsection{Modulus of continuity of \texorpdfstring{$\BL_p(s, \chi)$}{Lg}}
Whatever $p$, if $\chi$ is of prime-to-$p$ order, genus
theory is empty and this raise the question of the ``independence''
(or not) of $\order \BK_2 (\BZ_K)[p^\infty]$ and $\order \CT_{K, p}$.
%
This is related to the rank formula
\eqref{equalranks} when $K$ contains the maximal real subfield of
$\Q(\mu_p)$, but is also a consequence of the existence of a
non-trivial modulus of continuity for $\BL_p(s, \chi)$, whatever
$\chi = \varphi \psi$, as follows \cite[Th\'eor\`eme 0.3]{Gr34}:
\begin{theorem}
With Notations \ref{notations}, we have, for all $s, t \in \Z_p$:
$$\ffrac{1}{2}\BL_p(t, \chi)-\ffrac{1}{2}\BL_p(s, \chi) \equiv a\,(t-s)
\pmod {q\,{\mathfrak m}^{\!\hbox{\tiny $\BV$}} (t-s)}, $$
with a computable constant $a$ ($a = 0$ when $\varphi \ne \1$) and $\BV$ of
the form $\BV = \ds \BD-\max_\ell (p^{d_\ell}-p^{d_p}+\epsilon, \epsilon)$
($\epsilon = 1$ if $\varphi = \1$, $0$ otherwise),
where $\ell$ runs trough the set of primes ramified in $M/M_0$, totally
split in $M_0/\Q$ and such that $\frac{1}{q} \log_p(\ell) \not\equiv 0 \pmod p$.
\end{theorem}
In the case $\psi = \1$, $\varphi \ne \1$, the above formula becomes:
$$\ffrac{1}{2}\BL_p(t, \varphi) \equiv \ffrac{1}{2}\BL_p(s, \varphi) \pmod {q\,(t-s)}. $$
Thus, even if the genus principle is empty, there is a non-trivial congruence
between the orders of the two corresponding invariants.
\smallskip
For instance, we get $\ffrac{1}{2}\BL_p(1, \varphi) \equiv
\ffrac{1}{2}\BL_p(-1, \varphi) \pmod {2q}$, whence:
\begin{equation*}
\left\{\begin{aligned}
\order \BK_2(\BZ_K)[3^\infty] & \equiv \order \CT_{K, 3} \pmod 3,\ \
\hbox{for quadratic fields}, \\
\order \BR_2(\BZ_K)[2^\infty] & \equiv \order \CT_{K, 2} \pmod 8,\ \
\hbox{for cyclic cubic fields}.
\end{aligned}\right.
\end{equation*}
In the case of real quadratic fields for $p=3$ (resp. of cyclic cubic fields for $p=2$), see
the numerical results given in Appendix \ref{B2} (resp. Appendix \ref{C2}).
\section{Proof of the Deng--Li conjecture \texorpdfstring{$\delta = 3$}{Lg}}
\label{DengLi}
We consider the family defined in \cite[Theorem 1.2]{DL}. We will prove the
conjecture $\delta = 3$ in the writing:
$$\BK_2(\BZ_K)[2^\infty] \simeq (\Z/2\Z)^{n-1}
\times \Z/2^\delta \Z$$
and some other properties of this family of real quadratic fields.
\smallskip
Recall that $K = \Q(\sqrt m)$, $m= \ell_1 \ell_2 \cdots \ell_n$, $n$ even, with:
\smallskip
(i) $\ell_1 \equiv 3 \pmod 8$, $\ell_i \equiv 5 \pmod 8$ for $i \geq 2$,
\smallskip
(ii) $\big( \frac{\ell_1}{\ell_2} \big) = -1$, $\big( \frac{\ell_1}{\ell_j} \big) = 1$ for $j \geq 3$,
\smallskip
(iii) $\big( \frac{\ell_i}{\ell_j} \big) = -1$ for $2 \leq i < j \leq n$.
\smallskip
Condition (i) implies $m \equiv -1 \pmod 8$ and $K$ of discriminant $4m$; then
$\big( \frac{\ell_i}{\ell_j} \big) = \big( \frac{\ell_j}{\ell_i} \big)$ for all $i, j$ since
for $i \ne j$, one of the two primes is congruent to $1$ modulo $4$.
It implies also that the fundamental unit $\varepsilon$ is of norm $1$ since $-1$
is not norm in $K/\Q$; then, if $\varepsilon = a+b \sqrt m$, one has $a^2+b^2
\equiv 1 \pmod 8$, whence, either $a = 4 a'$ with $b$ odd, or $b = 4b'$ with $a$ odd.
\smallskip
Let ${\mathfrak p}$ be the prime ideal above $(2)$ and let $K_ {\mathfrak p}$
be the completion of $K$ at ${\mathfrak p}$; thus $K_ {\mathfrak p} = \Q_2(\sqrt {-1})$
proving that $2$ is local norm at $2$ in $K/\Q$ and that the norm
group of local units is equal to $1+4 \Z_2$.
\smallskip
The Hasse norm residue
symbols, of the form $\Big(\frac{\ell_i,\ K/\Q}{\ell_j} \Big)$, characterizing
the property ``$\ell_i$ local norm at $\ell_j$ in $K/\Q$'', are given by the
quadratic residue symbols $\big(\frac{\ell_i}{\ell_j} \big)$; indeed, $\ell_i$
norm in $\Q_{\ell_j}(\sqrt m)/\Q_{\ell_j}$ is equivalent (for $i \ne j$) to
$\ell_i$ square in $\Q_{\ell_j}^\times$ since the norm group of local
units of $\Q_{\ell_j}(\sqrt m)$ is of index $2$ in $\mu_{\ell_j-1}^{}
\oplus (1+\Z_{\ell_j}^{})$.
\smallskip
We can add the properties:
(iv) $\big( \frac{2}{\ell_i} \big) = \big(-1\big)^{\frac{\ell_i^2-1}{8}} = -1$,
for all $i$, $\big( \frac{\ell_1}{2} \big) = -1$, $\big( \frac{\ell_i}{2} \big) = 1$
for $i \geq 2$ (in the meaning $\ell_i$ norm (or not) in $\Q_2(\sqrt{-1})/\Q_2$).
\smallskip
(v) $\big( \frac{\ell_i}{\ell_i} \big) = 1$ for all $i \ne 2$ and $\big( \frac{\ell_2}{\ell_2} \big) = -1$,
obtained from the product formula of the Hasse norm residue symbols of fixed $\ell_i$ with
$n$ even.
\medskip
Put $G := \Gal(K/\Q) =: \langle \sigma \rangle$ and:
$$\Omega_K := \{(s_1, \ldots, s_{n+1})
\in G^{n+1},\ s_1 \cdots s_{n+1} =1 \} \simeq G^n; $$
let $h_K$ be the map $\Q^\times \to \Omega_K$ defined, by
means of the Hasse norm residue symbols, by
$h_K(x) = \Big(\Big(\frac{x,\ K/\Q}{v} \Big) \Big)_{\! v}$,
where $v$ runs trough the $n+1$ places ramified in $K/\Q$,
and for which the product formula:
$$\prd_{v\ {\rm ramified}}\Big(\ffrac{x,\ K/\Q}{v} \Big) = 1$$
holds as soon as $(x)$ is the norm of an ideal in $K/\Q$, hence local
norm at every non-ramified place (i.e., $x \in \{-1, 1\} \cdot \Q^{\times 2}$).
Whence the matrix of symbols, product formula taken on each line:
%
\begin{equation}\label{array}
\begin{tabular}{|l|l|l|l|c|c|c|c|c|c|c}
\hline
\ft${}$\ns & \ft$2$\ns & \ft$\ell_1$\ns & \ft$\ell_2$\ns & \ft$\ell_3$\ns & \ft$\ldots$\ns
& \ft$ \ell_i$\ns & \ft$\ldots$\ns & \ft$\ell_n$\ns \\
\hline
\ft$2$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$\ldots$\ns
& \ft$-1$\ns & \ft$\ldots$\ns & \ft$-1$\ns \\
\hline
\ft$\ell_1$\ns & \ft$-1$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$ 1$\ns & \ft$\ldots$\ns
& \ft$1$\ns & \ft$\ldots$\ns & \ft$1$\ns \\
\hline
\ft$\ell_2$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$\ldots$\ns
& \ft$-1$\ns & \ft$\ldots$\ns & \ft$-1$\ns \\
\hline
\ft$\ell_3$\ns & \ft$1$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$1$\ns & \ft$\ldots$\ns
& \ft$-1$\ns & \ft$\ldots$\ns & \ft$-1$\ns \\
\hline
\ft${\vdots}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns
& \ft${}$\ns & \ft${}$\ns & \ft${}$\ns \\
\hline
\ft$\ell_i$\ns & \ft$1$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$\ldots$\ns
& \ft$1$\ns & \ft$\ldots$\ns & \ft$-1$\ns \\
\hline
\ft${\vdots}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns & \ft${}$\ns
& \ft${}$\ns & \ft${}$\ns & \ft${}$\ns \\
\hline
\ft$\ell_n$\ns & \ft$1$\ns & \ft$1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$\ldots$\ns
& \ft$-1$\ns & \ft$\ldots$\ns & \ft$1$\ns \\
\hline
\ft$-1$\ns & \ft$-1$\ns & \ft$-1$\ns & \ft$1$\ns & \ft$1$\ns & \ft$\ldots$\ns
& \ft$1$\ns & \ft$\ldots$\ns & \ft$1$\ns \\
\hline
\end{tabular}
\end{equation}
\subsection{Proof of \texorpdfstring{$\CH_{K, 2} \simeq (\Z/2\Z)^{n-1}$}{Lg}}
This is proven in \cite{DL}, but we can bring more informations and remarks.
The first one is given by the Chevalley--Herbrand formula \cite[pp. 402-406]{Che}
in $K/\Q$:
$$\order \CH_{K, 2}^G = \frac{2^{n+1}}
{[K : \Q] (E_\Q : E_\Q \cap \Norm_{K/\Q} (K^\times))}, $$
\noindent
since $n+1$ primes ramify, with $E_\Q = \{\pm 1\}$; we shall write instead:
\begin{equation}
\order \CH_{K, 2}^G = \frac{\order \Omega_K}{\order h_K (E_\Q)} =
\frac{2^n}{\order h_K (\{\pm 1\})} = 2^{n-1},
\end{equation}
\noindent
since $-1 \not\in \Norm_{K/\Q} (K^\times)$.
\begin{lemma}\label{lem1}
The subgroup $\CH_{K, 2}^G$ is elementary of $2$-rank $n-1$; it is generated
by the classes of the prime ideals ${\mathfrak l}_i \mid \ell_i$, $i = 1, \ldots , n$,
and by the class of ${\mathfrak p} \mid (2)$. There are two independent
relations of principality of the form ${\mathfrak p}^{a_0} \prod_{i=1}^n
{\mathfrak l}_i^{a_i} = (\alpha)$ between the ramified primes (where the
exponents are $0$ or $1$), an obvious one being
$\prod_{i=1}^n {\mathfrak l}_i = (\sqrt m)$.
\end{lemma}
\begin{proof}
We have the classical exact sequence, where $E_K =
\langle -1, \varepsilon \rangle$ is the group of units of $K$ and
where $\CH_{K, 2}^\ram$ is the subgroup of $\CH_{K, 2}^G$ generated
by the classes of the ramified primes:
$$1 \to \CH_{K, 2}^\ram \too \CH_{K, 2}^G \too
\{\pm 1\} \cap \Norm_{K/\Q} (K^\times)/ \Norm_{K/\Q}(E_K) \to 1, $$
giving here $\CH_{K, 2}^G = \CH_{K, 2}^\ram \simeq (\Z/2\Z)^{n-1}$, the
right term being trivial; so, there are exactly two independent relations of
principality between the ramified primes.
\smallskip\noindent
Then, from \cite[\S\,4.4]{Gr7} generalizing our old papers in ``Annales
de l'Institut Fourier'', we have for the second element of the filtration:
$$\order \Big( \big (\CH_{K, 2}/\CH_{K, 2}^G \big)^G \Big) = \frac{\order \Omega_K}
{\order h_K (\Lambda)},\ \,\Lambda = \langle -1, 2, \ell_1, \ldots, \ell_n \rangle_{\Z}^{}; $$
then, $\order h_K (\Lambda) = (\Lambda : \Lambda\cap \Norm_{K/\Q} (K^\times))
= 2^n$, by computing norm residue symbols with the R\'edei matrix \eqref{array}
which is of rank $n$; whence $\CH_{K, 2} = \CH_{K, 2}^G = \CH_{K, 2}^\ram
\simeq (\Z/2\Z)^{n-1}$.
\end{proof}
A program computing these relations is given Appendix \ref{A2}.
\subsection{The non-trivial relation for \texorpdfstring{$n$}{Lg} even}
Consider the relations:
\begin{equation}\label{eq1}
\left\{\begin{aligned}
{\mathfrak p}^{a_0} \prd_{i = 1}^n {\mathfrak l}_i^{a_i} & =:
{\mathfrak p}^{a_0} \prd_{i \in I} {\mathfrak l}_i = (\alpha),\ \alpha \in K^\times, \\
\Norm_{K/\Q}(\alpha) & = s\,2^{a_0} \prd_{i \in I} \ell_i,
\ s \in \{\pm 1\},
\end{aligned}\right.
\end{equation}
%
where $I$ is a subset of $[1, n]$ and $s = \pm1$;
from the trivial relation $(\sqrt m) =
\prod_{i=1}^n {\mathfrak l}_i$ and $\Norm_{K/\Q}(\sqrt m)
=-\prod_{i=1}^n \ell_i$ we deduce $(\alpha \sqrt m) =
(c)\,{\mathfrak p}^{a_0}\,\prod_{i=1}^n {\mathfrak l}_i^{\ov a_i}$,
where $\ov a_i = 1-a_i$ and $c = \prod_{i \in I} \ell_i$;
taking $\ov \alpha := \frac{\alpha \sqrt m}{c}$, one gets the
equivalent complementary relations, where $\ov I = [1, n] \setminus I$:
\begin{equation}\label{eq2}
\left\{\begin{aligned}
{\mathfrak p}^{a_0} \prd_{i \in \ov I} {\mathfrak l}_i & = (\ov \alpha),\ \\
\Norm_{K/\Q}(\ov \alpha) &= -s\,2^{a_0}\prd_{i \in \ov I} \ell_i.
\end{aligned}\right.
\end{equation}
\begin{lemma}\label{lem2}
(i) In the relations \eqref{eq1} and \eqref{eq2}, one has $a_0=1$
and the two non-trivial equivalent relations may be written:
\smallskip
$\bullet$ ${\mathfrak p}\cdot \prod_{j \in J} {\mathfrak l}_j = (\beta)$,
$J \subseteq \{2, \ldots, n\}$, $\order J \equiv 1 \pmod 2$.
\smallskip
$\bullet$ ${\mathfrak p}\cdot {\mathfrak l}_1 \prod_{j \in \ov J} {\mathfrak l}_j = (\ov \beta)$,
$\ov J \subseteq \{2, \ldots, n\}$, $\order \ov J \equiv 0 \pmod 2$,
\smallskip
The relation without ${\mathfrak l}_1$ is ${\mathfrak p}\cdot {\mathfrak l}_2 = (\beta)$.
\medskip
(ii) These relations are given by the ideals $(\varepsilon+1)$ and $(\varepsilon-1)$
for which $\ffrac{\varepsilon+1}{\varepsilon-1} = C \cdot \sqrt m$, where $C$
is an odd rational number.
\medskip
(iii) There exists a sign $s_0$ such that $(\varepsilon+s_0) =
(c)\,{\mathfrak p}\cdot {\mathfrak l}_2$.
\end{lemma}
\begin{proof}
(i) Assume that $a_0 = 0$; let $\prod_{j \in J} {\mathfrak l}_j = (\beta)$
be the non-trivial relation such that $J \subseteq [2, n]$ (so, $J \ne \emptyset$).
Then $\prod_{j \in J} \ell_j = \pm \Norm_{K/\Q} (\beta)$; since
$\Norm_{K/\Q} (\beta) \equiv 1 \pmod 4$ as well as the $\ell_j$'s, we get
$\prod_{j \in J} \ell_j = \Norm_{K/\Q} (\beta)$.
%
Consider the symbols $\Big(\hbox{$\ffrac{\prod_{j \in J} \ell_j }{\ell_i}$}\Big)$,
$i \in [1, n]$, using \eqref{array}:
\smallskip
$\bullet$ For $i=2$, one gets $\Big(\hbox{$\ffrac{\prod_{j \in J} \ell_j }{\ell_2}$}\Big)=1$
if and only if $\order J$ is even (which solves the case $n=2$ since $J = \{2\}$).
$\bullet$ for $i =1$, one obtains $\Big(\hbox{$\ffrac{\prod_{j \in J} \ell_j }{\ell_1}$}\Big)=1$
if and only if $j=2 \not\in J$;
$\bullet$ for any $i \geq 3$, $\Big(\hbox{$\ffrac{\prod_{j \in J} \ell_j }{\ell_i}$}\Big)=1$
if and only if $j = i \not\in J$.
\smallskip
So we obtain $J = \emptyset$ (absurd); whence the relation
${\mathfrak p}\cdot \prod_{j \in J} {\mathfrak l}_j = (\beta)$
for $J \subseteq [2, n]$.
\smallskip
Then
$\Big(\hbox{$\ffrac{2\,\prod_{j \in J} \ell_j }{\ell_i}$}\Big) = 1$ for all $i \in [1, n]$
since $\Norm_{K/\Q} (\beta) = -2 \prod_{j \in J} \ell_j$ is not possible ($-1$ is not
local norm at $2$); so this is equivalent to $\Big(\hbox{$\ffrac{\prod_{j \in J} \ell_j }
{\ell_i}$}\Big) = -1$ for all $i \in [1, n]$.
The case $i = 2$ gives $\order J$ odd; the case $i \in J$, when $j=2 \not \in J$,
gives a contradiction, as well as $2 \in J$ and $i \not\in J$, so that one verifies
that only the relation ${\mathfrak p}\cdot {\mathfrak l}_2 = (\beta)$ holds (the
$(n+1) \times (n+1)$ matrix \eqref{array} is of rank $n$ and the sum of the two
lines corresponding to $2$ and $\ell_2$ gives again the norm relation).
\smallskip
(ii) We have $(\varepsilon+s)^{1-\sigma} = s \varepsilon$; so $(\varepsilon+s)$
is a principal invariant ideal necessarily of the form $(c_s) {\mathfrak p}{\mathfrak m}_s$,
$c_s \in \Q^\times$, ${\mathfrak p} \mid (2)$, ${\mathfrak m}_s \mid (\sqrt m)$.
Then, $\ffrac{\varepsilon+1}{\varepsilon-1} = C \cdot \sqrt m$ since
$\Big(\ffrac{\varepsilon+1}{\varepsilon-1}\cdot \ffrac{1}{\sqrt m}\Big)^{1-\sigma} = 1$.
\end{proof}
\begin{lemma}\label{lem3}
The case $b$ even does not occur.
\end{lemma}
\begin{proof}
Assume that $b$ is even; then $b = 4b'$; there exists a sign $s = \pm 1$
such that $a+s \equiv 2 \pmod 4$, so that $\varepsilon+s = a+s+4b' \sqrt m$
and the ideal $(\varepsilon+s) = (a+s+4b' \sqrt m)$ is of the form
$(2) (a''+2 b' \sqrt m)$ with $a''$ odd, which can not give the non-trivial relations
of Lemma \ref{lem2} since $a''+2 b' \sqrt m$ is ``odd'' (absurd).
\end{proof}
\begin{lemma}\label{lem4}
In the writing $\varepsilon = a+b \sqrt m$, $a = 4a'$ with $a'$ odd.
\end{lemma}
\begin{proof}
Assume that $a' = 2 a''$ and put $\theta_{s'} := \varepsilon+s' =
8 a''+s'+b \sqrt m$, $s' \in \{\pm1\}$. We have, using $mb^2 = a^2-1$,
$\Norm_{K/\Q}(\theta_{s'}) = 2 (8 a''s'+1)$.
From Lemma \ref{lem2}\,(iii) the non-trivial relation may be written
(where $c$ is an odd rational),
$(\theta_{s'}) = (c)\,{\mathfrak p}\,{\mathfrak l}_2$
giving by taking the norm:
$$8 a''s' +1 = s\,c^2\,{\mathfrak l}_2 \equiv 5s \pmod 8,\ \,s \in \{\pm 1\} ; $$
which is absurd.
\end{proof}
\subsection{Structure of \texorpdfstring{$\CT_{K, 2}$}{Lg} and triviality
of \texorpdfstring{$\CR_{K, 2}$}{Lg}}
Let's give the Kummer generators of the maximal sub-extension, of
$H_{K, 2}^\pr/K$, of exponent $2$.
\begin{lemma}\label{lem5}
The module $\CT_{K, 2}$ is of $\Z_2$-rank $n-1$ and the maximal sub-extension
of $H_{K, 2}^\pr/K_\infty$, of exponent $2$, is the Kummer extension
$K_\infty(\sqrt {\varepsilon}, \sqrt{\ell_1}, \ldots, \sqrt{\ell_n})$, under
the existence of the two independent relations $\sqrt\ell_1 \cdots \sqrt\ell_n
= \sqrt m \in K^\times$ and $\sqrt 2 \sqrt{\ell_2} =
\alpha \sqrt \varepsilon$, $\alpha \in K^\times$.
\end{lemma}
\begin{proof}
The maximal sub-extension of $H_{K, 2}^\pr/K$, of exponent $2$, is the Kummer
extension $K(\sqrt{2}, \sqrt{\varepsilon}, \sqrt{\ell_1}, \ldots, \sqrt{\ell_n})$
(all the radicals $a$ in the $\sqrt a$'s are squares of ideals of $K$ and are totally positive;
there is no other radicals since $\CH_{K, 2} = \CH_{K, 2}^\ram$).
We utilize the fact that $\sqrt 2 \in K_\infty$, then the relations
$\ell_1 \cdots \ell_n = m \in K^{\times 2}$ and
${\mathfrak p}\,{\mathfrak l}_2 = (\alpha)$, $\alpha \in K^\times$,
giving, by squaring, $2 \ell_2 = \alpha^2 \eta > 0$, $\eta \in E_K$,
$\eta \ne 1$; then we may assume that $\eta = \varepsilon$,
whence $2 \ell_2 = \alpha^2 \varepsilon$.
\smallskip
Whence the result since $\CT_{K, 2}$ is a direct factor in $\Gal(H_{K, 2}^\pr/K)$.
\end{proof}
\begin{theorem}\label{regulator}
The normalized $2$-adic regulator $\CR_{K, 2}$ is trivial.
\end{theorem}
\begin{proof}
We have $\varepsilon = a+ b \sqrt m = 4 a'+b \sqrt m$ with $a^2-mb^2=1$; then:
$$\varepsilon^2 = a^2+2ab \sqrt m+mb^2 =1+2mb^2+8 a' b \sqrt m .$$
Recall that $a'$ is odd (Lemma \ref{lem4}), that $b$ is odd and $m\equiv -1 \pmod 8$; thus
$\varepsilon^2 \equiv 1-2 +8 \sqrt m \pmod {16}$; whence $\log_2(\varepsilon) \sim 4$.
\smallskip
From \cite[Proposition 5.2]{Gra8}, we have, in our context:
$$\order \CR_{K, 2} \sim \frac{1}{2} \cdot
\frac{\big(\Z_2 : \log(\Norm_{K/\Q}(\CU_{K, 2})) \big)}
{ \order \CW_{K, 2} \cdot \prod_{{\mathfrak p} \mid 2} \Norm {\mathfrak p}}
\cdot \frac {R_K}{\sqrt {D_K}}, $$
\noindent
where $R_K = \log_2(\varepsilon)$ is the usual $2$-adic regulator \cite[\S\,5.5]{W} and
$D_K$ the discriminant of $K$.
%
Since we have $\big(\Z_2 : \log(\Norm_{K/\Q}(\CU_{K, 2})) \big) =
\big(\Z_2 : \log(1+4\,\Z_2) \big) = 4$, $R_K \sim 4$,
$\order \CW_{K, 2} = 2$, $\prod_{{\mathfrak p} \mid 2}\Norm {\mathfrak p} = 2$,
$\sqrt {D_K} \sim 2$, this yields $\order \CR_{K, 2} \sim 1$.
\smallskip
Formula \eqref{T=HRW} gives $\order \CT_{K, 2} = 2^{n-1} \times 1 \times 2 =2^n$, and
Corollary \ref{remfond} ends the proof of the Deng--Li conjecture $\delta = 3$.
\end{proof}
\section{\texorpdfstring{The \texorpdfstring{$p$}{Lg}-rank of $\BK_2(\BZ_K)$}{Lg}
when \texorpdfstring{$p \geq 5$}{Lg}}\label{p>3}
We consider a prime number $p \geq 5$ and a cyclic $p$-extension
$K/k$ fulfilling some conditions:
\begin{theorem} \label{pdiv} Let $p$ be a prime $\geq 5$.
Let $k$ be a real abelian number field of prime-to-$p$ degree and
let $K$ be any cyclic $p$-extension of $k$, abelian over $\Q$.
We assume that $K \cap \Q(\mu_p) = \Q$ and that $p$
does not totally split in $K(\mu_p)/K$.
Let $S_\ta$ be the set of primes $\ell \ne p$, ramified in $K/k$, and
let $t_\ell \geq 1$ be the number of prime ideals above $\ell$ in $k/\Q$. Then
$\rk_p \big(\BK_2(\BZ_K) \big) \geq \sum_{\ell \in S_\ta} t_\ell$.
\smallskip\noindent
In particular, for all primes $\ell \equiv 1 \pmod p$ and for any non-trivial $p$-extension
$K/\Q$, contained in $\Q(\mu_\ell^{})/\Q$, we have $\BK_2(\BZ_K)[p^\infty] \ne 1$.
\end{theorem}
\begin{proof}
Let $\zeta_p$ be a generator of $\mu_p^{}$ and let $\Q' := \Q(\zeta_p)$,
$k' := k(\zeta_p)$ and $K' := K(\zeta_p)$; put $G := \Gal(K'/k')$ and $g := \Gal(K'/K)$.
Recall that $\omega : g \to \mu_{p-1}^{}(\Z_p)$ is the Teichm\"uller character of $g$
(such that $\zeta_p^s =: \zeta_p^{\omega(s)}$, for all $s \in g$) and that any
$\Z_p[g]$-module $X$ is the sum $\bigoplus_{j=1}^{p-1} X_{\omega^j}$ in an
obvious meaning. Since $p \geq 5$, $\omega \ne \omega^{-1}$ and the
comparison of $\BK_2(\BZ_K)[p^\infty] = \BK_2(\BR_K)[p^\infty]$ with
$\CT_{K, p}$ made in \cite{GJ} does not hold.
\smallskip
The primes of $S_p(K)$ being not totally split in $K'/K$, the $\omega^{-1}$-component
of the $\Z_p[G]$-module $\langle S_p(K') \rangle \otimes \Z_p$ generated by $S_p(K')$
above $S_p(K)$ is trivial; thus, formula \eqref{rankR} becomes:
$$ \rk_p(\BK_2(\BZ_K)) = \rk_p\big((\CH_{K', p})_{\omega^{-1}} \big). $$
We note that a prime number $\ell$, ramified in $K/\Q$ with ramification
index divisible by $p$, fulfills the condition $\ell \equiv 1 \pmod p$
(for generalizations to the non-Galois case, see \cite[II, \S\,(d)]{Gr5}); whence
$\ell$ totally splits in $\Q'/\Q$ and, for ${\mathfrak l} \mid \ell$ in $k$,
the $p-1$ prime ideals ${\mathfrak L}$ above ${\mathfrak l}$ in $k'$ ramify
in $K'/k'$ with the same ramification index; this gives $(p-1) t_\ell$ ramified
ideals in $K'/k'$, whence $\order S_\ta(k') = (p-1)\sm_{\ell \in S_\ta} t_\ell$.
Let $G_v \subseteq G$ be the inertia group in $K'/k'$ of a ramified place
$v$ of $k'$ (so $v$ is a prime ideal above $\ell \ne p$ or a prime ideal above
$p$, but the $\omega^{-1}$-component of $\langle S_p(k') \rangle \otimes \Z_p$
is trivial), and put $S(k') = S_\ta(k')\,\cup\,S_p(k')$.
Let $E_{k'}$ be the group of units of $k'$:
\smallskip
The Chevalley--Herbrand formula in $K'/k'$ writes:
\begin{equation*}
\order \CH_{K', p}^G = \order \CH_{k', p}^\nr \times \frac{\order \Omega_{k'}}
{\order h_{k'}(E_{k'})},
\end{equation*}
where $\CH_{k', p}^\nr \subseteq \CH_{k', p}$ corresponds to
$\Gal(H_{k'}^\nr/K' \cap H_{k'}^\nr)$ and
$\Omega_{k'} =\Big \{(s_v)_v \in \bigoplus_{v \in S(k')}
G_v,\ \ \,\prod_v s_v = 1 \Big\}$,
$h_{k'} : E_{k'} \to \Omega_{k'}$ being the map defined by the
family of Hasse norm residue symbols $\Big(\big(\ffrac{x, K'/k'}{v}\big)\Big)_v$
which fulfill the ``product formula'' on units.
This formula is associated to the following exact sequences:
\begin{equation} \label{suites}
\left\{\begin{aligned}
&1 \to \BJ_{K'/k'}(\CH_{k', p}) \cdot \CH_{K', p}^\ram \too \CH_{K', p}^G \\
&\hspace{1.7cm} \too E_{k'} \cap \Norm_{K'/k'}(K'^\times)/\Norm_{K'/k'}(E_{K'}) \to 1, \\
&1 \to E_{k'} / E_{k'} \cap \Norm_{K'/k'}(K'^\times) \\
&\hspace{1.7cm}\too \Omega_{k'} \too \Gal(H_{K'/k'}^\gen/K' H_{k'}^\nr) \to 1,
\end{aligned}\right.
\end{equation}
where $\BJ_{K'/k'}$ is the extension of classes, $\CH_{K', p}^\ram$ the
subgroup of $\CH_{K', p}$ generated by the ramified primes, $H_{K'/k'}^\gen
\subseteq H_{K'}^\nr$ the genus field, fixed by $\CH_{K', p}^{1-\sigma}$,
where $\sigma$ generates $G$ \cite[Proposition IV.4.5]{Gr5}.
\smallskip
The link between the two aspects (fixed points and genus
exact sequences) is given by $\Gal(H_{K'/k'}^\gen/K') \simeq
\CH_{K', p}/ \CH_{K', p}^{1-\sigma}$ and the exact sequence
$1 \to \CH_{K', p}^G \to \CH_{K', p} \to \CH_{K', p}^{1-\sigma} \to 1$.
\smallskip
More precisely, we need the $\omega^{-1}$-component of $\CH_{K', p}^G$
and of the terms of the exact sequences. For $\CH_{K', p}^G$, this computation
has been done in Jaulent's Thesis \cite[Chapitre III]{Jau1} and reproduced in
\cite[\S\,2.2, Theorem 2.1\,(ii)]{Gra10}; the difficulty comes from the fact that
$\CH_{k', p}$ is not necessary isomorphic to a sub-module of $\CH_{K', p}$
(indeed, only $\BJ_{K'/k'}(\CH_{k', p})$ makes sense as sub-module of
$\CH_{K', p}$ and $\BJ_{K'/k'}$ is not necessarily injective).
It writes under our context:
\begin{equation}\label{omega}
\order (\CH_{K', p}^G)_{\omega^{-1}} = \order (\CH_{k', p}^\nr)_{\omega^{-1}} \times
\frac{\order (\Omega_{k'})_{\omega^{-1}}}{\order (h_{k'}(E_{k'}))_{\omega^{-1}}} .
\end{equation}
Since $E_{k'} = E_{k'^+} \oplus \mu_p$ (up to a prime-to-$p$ index), where
the subgroup $E_{k'^+}$ of real units of $k'$ is of even character, and since
$\mu_p$ is of character $\omega \ne \omega^{-1}$, the $\omega^{-1}$-component
of $h_{k'}(E_{k'})$ is trivial and exact sequences \eqref{suites} and formula \eqref{omega}
reduce to:
\begin{equation}\label{isomorphisms}
\left\{\begin{aligned}
(\CH_{K', p}^G)_{\omega^{-1}} & \simeq \big (\BJ_{K'/k'}(\CH_{k', p}) \cdot
\CH_{K', p}^\ram \big)_{\omega^{-1}}, \\
(\Omega_{k'})_{\omega^{-1}} & \simeq (\Gal(H_{K'/k'}^\gen/K' H_{k'}^\nr))_{\omega^{-1}}, \\
\order (\CH_{K', p}^G)_{\omega^{-1}} & = \order (\CH_{k', p}^\nr)_{\omega^{-1}} \times
\order (\Omega_{k'})_{\omega^{-1}} . \\
\end{aligned}\right.
\end{equation}
Whence:
\begin{equation}\label{fin}
\left\{\begin{aligned}
\rk_p \big(\BK_2(\BZ_K)\big) & = \rk_p\big((\CH_{K', p})_{\omega^{-1}} \big) \\
& \geq \rk_p(\CH_{K', p} / \CH_{K', p}^{1-\sigma})_{\omega^{-1}}
\geq \rk_p(\Omega_{k'})_{\omega^{-1}}.
\end{aligned}\right.
\end{equation}
Let $G' := \langle G_v \rangle_v \simeq \Gal(K'/K' \cap H_{k'}^\nr)$; the product
formula may be interpreted by means of the exact sequence:
$$1 \to \Omega_{k'} \too \plus_{v \in S(k')} G_v \too G' \subseteq G \to 1, $$
in which the character of $G'$ and that of $\bigoplus_{v \in S_p(k')} G_v$
is the unit character. Whence $(\Omega_{k'})_{\omega^{-1}} \simeq
\plus_{v \in S_\ta(k')} G_v$.
For $\ell \in S_\ta$, let $p^{e_\ell}$, $e_\ell \geq 1$, be the ramification index
of $\ell$ in $K'/k'$. The inertia groups $G_v$, for $v \mid \ell$
in $k'$, does not depend on $v$ and by abuse may be denoted $G_\ell$.
Since each of the $t_\ell$ primes ${\mathfrak l} \mid \ell$ in $k$, totally splits in
$k'/k$ into $p-1$ primes ${\mathfrak L} \mid {\mathfrak l}$ of $k'$, one may write:
$$\plus_{{\mathfrak L}} G_{\mathfrak L} = \plus_{\ell \in S_\ta} \
\plus_{{\mathfrak l} \mid \ell} \
\plus_{{\mathfrak L} \mid {\mathfrak l}} G_\ell \simeq
\plus_{\ell \in S_\ta} \Big(\big (\Z/p^{e_\ell} \Z \big)^{p-1} \Big)^{t_\ell}, $$
in which, each $\ell$-component is, as Galois module, isomorphic
to $t_\ell$ copies of the regular representation
$\bigoplus_{j=1}^{p-1}\big (\Z[G_\ell]/
p^{e_\ell} \Z[G_\ell] \big)_{\omega^j}$
whose $\omega^{-1}$-component is of order $p^{e_\ell}$. Whence
$(\Omega_{k'})_{\omega^{-1}} \!\simeq \bigoplus_{\ell \in S_\ta}
(\Z/p^{e_\ell} \Z)^{t_\ell}$.
Thus, one obtains, from \eqref{fin}, $\rk_p(\BK_2(\BZ_K)) \geq \sum_{\ell \in S_\ta} t_\ell$.
\end{proof}
\begin{corollary}
We obtain the following consequences:
\smallskip
(i) $\order (\CH_{k', p}^\nr)_{\omega^{-1}} \times
\order (\Omega_{k'})_{\omega^{-1}} = \order \Big (\BJ_{K'/k'}(\CH_{k', p}) \cdot
\CH_{K', p}^\ram \Big)_{\omega^{-1}}$.
\smallskip
(ii) If $\BK_2(\BZ_k)[p^\infty] = 1$, then $(\CH_{K', p}^\ram)_{\omega^{-1}}
\simeq \bigoplus_{\ell \in S_\ta} (\Z/p^{e_\ell} \Z)^{t_\ell}$.
\smallskip
(iii) If $k = \Q$, then:
\smallskip
\qquad $\bullet$ $\rk_p(\BK_2(\BZ_K)) \geq \order S_\ta$;
\smallskip
\qquad $\bullet$ $(\CH_{K', p}^\ram)_{\omega^{-1}} \simeq
\bigoplus_{\ell \in S_\ta} \Z/p^{e_\ell} \Z$.
\end{corollary}
\begin{proof}
Cases (i), (ii) come from the fact that $\BK_2(\BZ_k)[p^\infty] = 1$ is equivalent to
$(\CH_{k', p})_{\omega^{-1}} = 1$, then to \eqref{isomorphisms}. If $k = \Q$,
$\sum_{\ell \in S_\ta} t_\ell = \order S_\ta$; then $(\CH_{\Q', p})_{\omega^{-1}} = 1$
from Ribet's reciprocal of Herbrand's criterion, since the Bernoulli number
$B_2 = \frac{1}{6}$ is prime to $p$ \cite[Theorem 5, p.\,45]{Rib}. Whence (iii).
\end{proof}
\begin{remarks}
(i) For $p=3$, $\omega^{-1} = \omega$, so that $(\CH_{K', p}^G)_{\omega^{-1}}$
depends on $\ds \frac{\order (\Omega_{\Q'})_{\omega^{-1}}}{\order h_{\Q'}(\mu_3)}$
which may be trivial, when $S_\ta = \{\ell\}$ and $[K : \Q] = 3$.
\smallskip
(ii) Replacing $\BK_2(\BZ_K)[p^\infty]$ by $\CT_{K, p}$ (whatever the prime $p \geq 3$)
and $\omega^{-1}$ by $\omega$, $h_{\Q'}(\mu_p)$ may be non-trivial, being of
character $\omega$. So, if $K/\Q$ is of prime conductor $\ell$, $\CH_{K, p} = 1$
and we may have $\CT_{K, p}=1$ but $\BK_2(\BZ_K)[p^\infty] \ne 1$.
\smallskip
(iii) For a table of numerical examples, see Appendix \ref{D1} for the
computation of $\order\BK_2(\BZ_K)[p^\infty]$, $p \geq 3$, and
Appendix \ref{D2} for the computation of $\rk_5(\BK_2(\BZ_K)[5^\infty]$.
Numerical tables are given in \cite{Br2} for cyclic cubic fields; more on
computations of $p$-ranks are given in Qin's papers \cite{Q1, Q2, Q3}.
\end{remarks}
\section{Conclusion}
These explicit calculations lead to a confirmation of the properties of the
$\BL_p$-functions when $s$ varies in $\Z_p$ and takes the values $s \in \{-1, 1\}$,
giving annihilation theorems, orders of isotypic components of classical $p$-invariants.
\smallskip
Other invariants than class groups, torsion groups and regular kernels exist, as for
instance the Jaulent logarithmic class group $\wt \CH_{K, p}$\,\footnote{\,Also denoted,
most often in the literature, $\wt {\rm C\ell}_{K,p}$, $\wt \CT_{K,p}$ as quotient of $\CT_{K,p}$;
but it looks more like the $p$-class group $\CH_{K, p}$ and there exists a formula of the
form $\order \CT_{K,p} = \wt \CH_{K, p} \cdot \wt \CR_{K, p} \cdot \wt \CW_{K, p}$,
for a suitable ``logarithmic regulator'', see \cite[Schema \S\,2.3]{Jau24}. Moreover
it may capitulates in real $p$-extensions of $K$ (the key for Greenberg's conjecture),
while $\CT_{K,p}$ never capitulates under Leopoldt's conjecture.} \cite{Jau23, Jau3, Jau4}
linked to the wild Hilbert kernel \cite[Th\'eor\`eme 9]{Jau22} and in relation with Greenberg's
conjecture \cite{Jau24, Jau25} (see Appendices \ref{clog2}, \ref{clog3} for some computations),
and invariants of the higher $K$-theory (giving conjectural expressions suggested
in \cite[Section 12, Conjecture 12.2]{Gr4} and some results of \cite{JM}).
\smallskip
An important problem is to find the structure of these invariants, it being understood
that they fulfill, once genus part has been taken into consideration, standard densities
in the spirit of the Cohen--Lenstra--Martinet--Malle distributions. A specific technique,
of genus theory type, does exist in cyclic $p$-extensions, especially for $p$-class
groups with determining the canonical filtration by means of a natural algorithm
generalizing Chevalley--Herbrand fixed points formula (large bibliography,
synthesized in \cite{Gr7}).
\smallskip
A main question being to find such algorithms
for the other $p$-invariants, which unfortunately does not exist to our knowledge;
the case of the most important groups $\CT_{K, p}$ being very interesting; indeed,
class field theory (by means of computation of suitable ray class groups)
gives the group structure, but this computational aspect does not give a method
to determine the filtration in cyclic $p$-extensions; only the fixed points formula,
is known \cite[Theo\-rem IV.3.3]{Gr5}.
\smallskip
In a cohomological viewpoint, $\CT_{K, p}$ is closely related to the
Tate--Chafarevich group
$\Cha_{K, p}^2 := {\rm Ker} \big [{\Hom}^2 (\CG_{K, S}, \F_p) \rightarrow
\oplus_{{\mathfrak p} \mid p}\,{\Hom}^2 (\CG_{K_{\mathfrak p}}, \F_p) \big]$,
and that $\CT_{K, p} \simeq \Hom^2(\CG_{K, S}, \Z_p)^\ast$ \cite[Th\'eor\`eme1.1]{Ng1},
where $\CG_{K, S}$ is the Galois group of the maximal $p$-ramified pro-$p$-extension
of $K$ and $\CG_{K_{\mathfrak p}}$ the local analogue over $K_{\mathfrak p}$;
generalizations of the same kind are done in the literature, as in
\cite[Theorem 0.4, 0.7, 0.11,\,\ldots]{Ko1}, \cite[Theorem 3.5, Example 4.3]{AAM},
with analogous cohomological framework, without any workable algorithm.
\smallskip
A major difficulty is to interpret, in terms of ideals, global units and so on, such
cohomology groups; the case of $\CT_{K, p} \simeq \Hom^2(\CG_{K, S}, \Z_p)^\ast$
is edifying since $\CG_{K, S}$ is inaccessible, while $\CT_{K, p}$ depends on the
$p$-class group and on the normalized $p$-adic regulator, of $K$. Of course, cohomology
gives important and non-obvious relations but, in the previous example, it is the knowledge
of $\CT_{K, p}$ which yields important information on $\CG_{K, S}$ (e.g., case of the simplest
structure giving the notion of $p$-rational fields issued from \cite{Gr3, Mov} and largely
developed in a lot of papers; see \cite[Appendix]{Gra88} for a survey
about this notion and general abelian $p$-ramification theory).
\smallskip
Maybe the link we made in \cite[\S\,III.2, Theorem III.2.6, Theorem III.3.3]{Gr5} and
\cite[Appendice]{GJ}, using logarithms of ideals for an effectiveness of the arithmetic,
may be generalized.
\smallskip
As a final remark, we point out the fact that these analytic genus theory aspects of $p$-adic
pseudo-measures are stated in \cite{Gr33} for any totally real base field $k$ replacing $\Q$
(under Leopoldt's conjecture for $p$, to get the important result of deployment \cite[Theorem
III.4.1.5]{Gr5}), and allowing similar technics with genus fields frome Deligne--Ribet
pseudo-measures; these results being completed with for instance the article by Maire
\cite{Mai} on the class field theory aspects and governing fields about the {\it existence}
of prescribed $p$-cyclic ramified extensions for which an analog of Theorem \ref{thmfond}
should apply.
\subsection*{Acknowledgements} I thank with pleasure Litong Deng and Yong\-xiong Li
for interesting exchanges about their paper, giving to me the idea of taking again into
account my old papers on genus theory of $p$-adic pseudo-measures, hoping that
arithmetic invariants of new families of real abelian fields can be elucidated via this method.
\begin{appendix}
\section{Data relating to the Deng--Li family}
This section is only concerned by the family of quadratic fields described
and studied in Section \ref{DengLi}; we illustrate and confirm various numerical aspects.
\subsection{Computation of \texorpdfstring{$\CH_{K, p}$}{Lg},
\texorpdfstring{$\CT_{K, p}$}{Lg} and \texorpdfstring{$\wt \CH_{K, p}$}{Lg}}\label{A1}
We compute the three fundamental invariants for this family, it being understood that
the structure of $\CH_{K, p}$ and complex analytic properties are given in \cite{DL}
in order to study $\BK_2(\BZ_K)[2^\infty]$.
\subsubsection{General program for \texorpdfstring{$\CH_{K, p}$}{Lg},
\texorpdfstring{$\CT_{K, p}$}{Lg}}
It may be convenient to recall the program computing the structures of $\CH_{K, p}$ and $\CT_{K, p}$
for a number field $K$, for any ${\sf p \in [bp, Bp]}$, as soon as a defining polynomial $P \in \Z[X]$
(monic, irreducible) is given (in ${\sf P}$); for the computation of $\CT_{K, p}$, the parameter
${\sf nu}$ must be chosen such that ${\sf p^{nu}}$ be larger than the exponent of the result
${\sf T}$ \cite[\S\,2.1]{Gr6}; the number ${\sf r}$ is the number $r_2+1$
of independent $\Z_p$-extensions of $K$:
\smallskip
\ft\begin{verbatim}
{P=x^3-7*x+1;bp=2;Bp=5*10^5;K=bnfinit(P,1);r=K.sign[2]+1;
print("P=",P," H=",K.cyc);forprime(p=bp,Bp,nu=6;
Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;T=List;e=matsize(HKn)[2];
R=0;for(k=1,e-r,c=HKn[e-k+1];v=valuation(c,p);if(v>0,R=R+1;
listinsert(T,p^v,1)));if(R>0,print("p=",p," rk(T)=",R," T=",T)))}
P=x^3-7*x+1 H=[] P=x^8-8*x+1 H=[4]
p=7 rk(T)=1 T=[7] p=2 rk(T)=1 T=[4]
p=701 rk(T)=1 T=[701] p=3 rk(T)=1 T=[3]
\end{verbatim}\ns
\subsubsection{Case of the Deng--Li family}
The following program verifies the structure of $\CH_{K, 2}$ and
$\CT_{K, 2}$ for the Deng--Li family of quadratic fields, for $n=4$:
\smallskip
\ft\begin{verbatim}
{L=List
([7215,26455,77415,119535,142935,153735,166335,171015,196359,
226655,241215,243295,257335,283855,311415,315055,420135,430495,
447135,473415,475215,490295,504295,545415,550615,552695,553335,
563695,568815,592215,603655,606615,633399,657735,665223,673215,
685815,687895,727935,751335,755495,757055,790495,798135,803751,
807455,818935,833199,849615,878415,884455,886015,886335,896415,
905255,911495,934935,961935,973655,981695,990015]);
p=2;nu=9;for(j=1,61,P=x^2-L[j];K=bnfinit(P,1);HK=K.cyc;
Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;T=List;e=matsize(HKn)[2];
for(k=1,e-1,c=HKn[e-k+1];v=valuation(c,p);if(v>0,listput(T,p^v)));
print("P=",P," T=",T," H=",HK))}
P=x^2-7215 T=[2,2,4] H=[2,2,2] P=x^2-981695 T=[2,2,4] H=[10,2,2]
P=x^2-26455 T=[2,2,4] H=[2,2,2] P=x^2-990015 T=[2,2,4] H=[18,2,2]
\end{verbatim}\ns
\smallskip
From \eqref{equalranks}, the $2$-ranks of $\BR_2(\BZ_K)$ and $\CT_{K, 2}$ are equal to $3$,
this implies, since $\BK_2(\BZ_K)[2^\infty] \simeq (\Z/2\Z)^3 \times \Z/2^3\Z$, that:
$$\BR_2(\BZ_K)[2^\infty] \simeq (\Z/2\Z)^2 \times \Z/2^2\Z \simeq \CT_{K, 2}. $$
More generally, for the Deng--Li family, we have:
$$\BR_2(\BZ_K)[2^\infty] \simeq \CT_{K, 2} \simeq
(\Z/2\Z)^{n-2} \!\times \Z/2^2\Z,\ \CH_{K, 2} \simeq (\Z/2\Z)^{n-1}. $$
From this, we verify that $\CW_{K, 2} \simeq \Z/2\Z$, which
is coherent with the formula $\order \CT_{K, 2} = \order \CH_{K, 2} \times
\order \CR_{K, 2} \times \order \CW_{K, 2}$ since the normalized $2$-adic regulator
is trivial from Theorem \ref{regulator}.
\subsubsection{Logarithmic class group}\label{classlog}
The computation of $\wt \CH_{K, 2}$ (instruction ${\sf CLog=bnflog(K, 2)}$)
for the Deng--Li family, gives always the structure $ (\Z/2\Z)^2$ for $n=4$;
but we do not know if this invariant is given by the $\BL_2(s, \chi)$'s at some
$s \in \Z_2$ (see \S\,\ref{clog2} for the computation of $\wt \CH_{K, 2}$ for
quadratic fields and \S\,\ref{clog3} for the cubic case):
\smallskip
\ft\begin{verbatim}
P=x^2-7215 CLog=[[2,2],[],[2,2]] T=[2,2,4] H=[2,2,2]
P=x^2-26455 CLog=[[2,2],[],[2,2]] T=[2,2,4] H=[2,2,2]
(...)
P=x^2-981695 CLog=[[2,2],[],[2,2]] T=[2,2,4] H=[10,2,2]
P=x^2-990015 CLog=[[2,2],[],[2,2]] T=[2,2,4] H=[18,2,2]
\end{verbatim}\ns
\smallskip\noindent
Moreover, the third component of ${\sf bnflog(K, 2)}$ gives the $S_2$-class group
$\CH_{K, 2}/{\rm cl}({\mathfrak p}) \simeq \Z/2\Z \times \Z/2\Z$, where
${\mathfrak p} \mid 2$ is ramified; this result confirms that necessarily
${\mathfrak p}$ is non-principal of order $2$.
\subsection{The canonical non-trivial relation}\label{A2}
The program gives the non-trivial relation whose writing does not contain ${\mathfrak l}_1$
(with $\ell_1 \equiv 3 \pmod 8$), then that obtained multiplying by $(\sqrt m)$.
We have proven that this relation is always
${\mathfrak p} {\mathfrak l}_2 = (\alpha)$ (Lemma \ref{lem2}).
The list $L$ gives the exponents $a_0$, $a_1$, $a_2,\,\ldots$,
of ideals ${\mathfrak p}$, ${\mathfrak l}_1$, ${\mathfrak l}_2,\,\ldots$, in this order
especially for the particular primes ${\mathfrak l}_1$, ${\mathfrak l}_2$ (which
is not so immediate since the PARI factorization of $m$ gives primes
in the ascending order). The generator $\alpha$ of the relation of principality is
$\alpha = u+v \sqrt m$; for checking, its norm $N$ and the sign
$S$ of $N$ are computed:
\smallskip
\ft\begin{verbatim}
{m=List
([7215,26455,77415,119535,142935,153735,166335,171015,196359,
226655,241215,243295,257335,283855,311415,315055,420135,430495,
447135,473415,475215,490295,504295,545415,550615,552695,553335,
563695,568815,592215,603655,606615,633399,657735,665223,673215,
685815,687895,727935,751335,755495,757055,790495,798135,803751,
807455,818935,833199,849615,878415,884455,886015,886335,896415,
905255,911495,934935,961935,973655,981695,990015]);p=2;
for(j=1,61,M=m[j];D=quaddisc(M);r=omega(D);L0=List;
for(i=1,r,listput(L0,0));P=x^2-M;K=bnfinit(P,1);print();Lel=List;
Div=component(factor(D),1);listput(Lel,2,1);h=3;for(i=2,r,c=Div[i];
if(Mod(c,4)==-1,el1=c;listput(Lel,c,2)));for(i=2,r,c=Div[i];
if(c!=el1 & kronecker(c,el1)==-1,el2=c;listput(Lel,c,3)));
for(i=2,r,c=Div[i];if(c!=el2 & Mod(c,4)==1,h=h+1;listput(Lel,c,h)));
print("M=",M," D=",Lel);for(k=1,2^r-1,B=binary(k);t=#B;
LB=L0;for(i=1,t,listput(LB,B[i],r-t+i));F=idealfactor(K,D);
Fel=List;CFel=component(F,1);listput(Fel,CFel[1],1);h=3;
for(i=2,r,c=CFel[i][1];if(Mod(c,4)==-1,el1=c;
listput(Fel,CFel[i],2)));
for(i=2,r,c=CFel[i][1];if(c!=el1 & kronecker(c,el1)==-1,el2=c;
listput(Fel,CFel[i],3)));
for(i=2,r,c=CFel[i][1];if(c!=el2 & Mod(c,4)==1,h=h+1;
listput(Fel,CFel[i],h)));
A=1;for(j=1,r,e=LB[j];if(e==0,next);A=idealmul(K,A,Fel[j]);
Q=bnfisprincipal(K,A);u=Q[2][1];v=Q[2][2];
if(Q[1]==0 & u!=0,print("L=",LB);N=u^2-M*v^2;S=sign(N);
print(component(factor(abs(N)),1));
print("u=",u," v=",v," S=",sign(S)))))}
\end{verbatim}\ns
\smallskip
Let's give some excerpts showing all possible cases:
\smallskip
\ft\begin{verbatim}
M=7215 D=[2,3,5,13,37] M=26455 D=[2,11,13,5,37]
L=[1,0,1,0,0] L=[1,0,1,0,0]
[2,5] [2,13]
u=-85 v=1 S=1 u=86654841 v=-532769 S=1
L=[1,1,0,1,1] L=[1,1,0,1,1]
[2,3,13,37] [2,5,11,37]
u=1443 v=17 S=-1 u=1084184915 v=-6665757 S=-1
M=504295 D=[2,11,173,5,53] M=665223 D=[2,3,461,13,37]
L=[1,0,1,0,0] L=[1,0,1,0,0]
[2,173] [2,461]
u=7970980451749 v=11224562253 S=1 u=10603 v=13 S=1
L=[1,1,0,1,1] L=[1,1,0,1,1]
[-1,2,5,11,53] [2,3,13,37]
u=-32719598967495 v=46075031513 S=-1 u=18759 v=23 S=-1
\end{verbatim}\ns
\section{Quadratic fields}
\subsection{Computation of
\texorpdfstring{$\order \BK_2(\BZ_K)[2^\infty]$, $\CT_{K, 2}$ and $\BC$}{Lg}} \label{B1}
The following program computes, instead, $\order \BR_2(\BZ_K)[2^\infty]$ for any real
quadratic field (so $f$ is $m$ or $4m$) and $\CT_{K, 2}$ for
checking of Theorem \ref{thmfond} (cases (ii${}_2$) and (iii${}_2$) and
property {\bf (b)} about equality v.s. inequality, for all $s \in \Z_p$).
In that cases, $\epsilon = 1$, $\BD$ is the number of odd ramified primes,
and $\BC = \BD$ (resp. $\BC = \BD-1$) if $\theta^{-1}\chi(2) = 1$
(resp. $\theta^{-1}\chi(2) \ne 1$).
\smallskip
Recall that ${\sf nu}$ and ${\sf n}$ must be chosen large enough (below we take
$n=10$ for $m >100$, $n=13$ for $m > 10^3$).
\smallskip
We do not write the trivial cases $\CT_{K, 2} = \BR_2(\BZ_K)[2^\infty] = 1$.
\smallskip
\ft\begin{verbatim}
{p=2;q=4;nu=10;n=6;for(m=5,2000,if(core(m)!=m,next);f=quaddisc(m);
P=x^2-m;K=bnfinit(P,1);Kpn=bnrinit(K,p^nu);H=Kpn.cyc;T=List;
e=matsize(H)[2];valT=0;for(k=1,e-1,h=H[e-k+1];v=valuation(h,p);
if(v>0,valT=valT+v;listput(T,p^v)));f=quaddisc(m);fn=q*p^n*f;c=1;
while(gcd(c,2*m)!=1 || kronecker(f,c)!=-1,c=c+1);S=0;
forstep(a=1,fn/2,2,if(gcd(a,m)!=1,next);aa=lift(Mod(a/c,fn));
la=(aa*c-a)/fn;eps=kronecker(m,a);S=S+a*eps*(la+(1-c)/2));
valR=valuation(S,2)-1;D=omega(m);if(Mod(m,2)==0,D=D-1);
C=D;if(Mod(m,8)!=-1,C=D-1);
if(valT==C,print("m=",m," c=",c," T=",T," v_2(T)=",valT,
" val(R_2Z)=",valR," C=",C," Equality"));
if(valT>C,print("m=",m," c=",c," T=",T," v_2(T)=",valT,
" v_2(R_2Z)=",valR," C=",C," Inequality")))}
m=7 c=5 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=1 Inequality
m=14 c=3 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=0 Inequality
m=15 c=13 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=2 Equality
m=17 c=3 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=0 Inequality
m=21 c=11 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=1 Equality
m=23 c=3 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=1 Inequality
m=30 c=11 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=1 Equality
m=31 c=7 T=[8] v_2(T)=3 v_2(R_2Z)=3 C=1 Inequality
m=33 c=5 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=1 Equality
m=34 c=7 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=0 Inequality
m=35 c=3 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=1 Equality
m=39 c=11 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=2 Equality
m=41 c=3 T=[16] v_2(T)=4 v_2(R_2Z)=3 C=0 Inequality
(...)
m=1001 c=3 T=[2,2] v_2(T)=2 v_2(R_2Z)=2 C=2 Equality
m=1002 c=5 T=[16] v_2(T)=4 v_2(R_2Z)=5 C=1 Inequality
m=1003 c=5 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=1 Inequality
m=1005 c=29 T=[2,8] v_2(T)=4 v_2(R_2Z)=5 C=2 Inequality
m=1006 c=7 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=0 Inequality
m=1007 c=3 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=2 Equality
m=1009 c=11 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=0 Inequality
m=1010 c=3 T=[2] v_2(T)=1 v_2(R_2Z)=1 C=1 Equality
m=1011 c=7 T=[4] v_2(T)=2 v_2(R_2Z)=2 C=1 Inequality
m=1015 c=17 T=[2,4] v_2(T)=3 v_2(R_2Z)=3 C=3 Equality
m=1022 c=3 T=[2,16] v_2(T)=5 v_2(R_2Z)=9 C=1 Inequality
m=1023 c=5 T=[2,64] v_2(T)=7 v_2(R_2Z)=7 C=3 Inequality
\end{verbatim}\ns
\smallskip
The equality $v_2(\order \CT_{K, 2}) = \BC$ often occur and when
$v_2(\order \CT_{K, 2}) > \BC$, then $v_2(\order \BR_2(\BZ_K)[2^\infty] > \BC$,
with some cases of distinct valuations ($m=41, 65, 66, 114,\,\ldots$).
\smallskip
Numerical examples given in \cite[Table]{DL}, yield equalities with
$\BC = \BD +v_2(4 s) -2 \epsilon = 4$ for $s=-1$, since $2$ splits in
$M = \Q(\sqrt{-m})$ (Theorem \ref{thmfond}\,(iii$_2$) and Corollary \ref{remfond}).
\subsection{Computation of \texorpdfstring{$\wt \CH_{K, 2}$}{Lg}}\label{clog2}
We give a program of computation of the logarithmic class group to compare with
the invariants $\CH_{K, 2}$, $\CT_{K, 2}$. There are many structures, but the
relation between $\CH_{K, 2}$ and $\CT_{K, 2}$ being well-known, we give only
an excerpt of interesting examples concerning $\wt \CH_{K, 2}$. For $p=2$, the
first component of ${\sf bnflog(K,2)}$ gives $\wt \CH_{K, 2}^0$ of index
$1$ or $2$ in $\wt \CH_{K, 2}$ \cite[\S\,3, Remark 3.1]{BeJa}:
\smallskip
\ft\begin{verbatim}
{bf=10^4;Bf=10^4+10^3;for(m=bf,Bf,if(core(m)!=m,next);PK=x^2-m;
F=component(factor(m),1);nu=12;K=bnfinit(PK,1);HK=K.clgp;
Kpn=bnrinit(K,2^nu);HKn=Kpn.cyc;T=List;w=0;e=matsize(HKn)[2];
for(k=1,e-1,CK=HKn[e-k+1];v=valuation(CK,2);if(v>0,w=w+v;
listput(T,2^v)));K=bnfinit(PK,1);HK=K.clgp;Kpn=bnrinit(K,2^nu);
HKn=Kpn.cyc;T=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,CK=HKn[k+1];
v=valuation(CK,2);if(v>0,w=w+v;listput(T,2^v)));CLK=bnflog(K,2);
print("m=",F," P=",PK," H=",HK[2]," T=",T," CLog=",CLK))}
m=[3,5,23,29] x^2-10005 H=[2,2] T=[4,2,2] CLog=[[2,2],[],[2,2]]
m=[3,47,71] x^2-10011 H=[4,2] T=[4,2] CLog=[[4],[],[4]]
m=[3,5,11,61] x^2-10065 H=[2,2] T=[16,2,2]CLog=[[4],[2],[2]]
m=[2,3,23,73] x^2-10074 H=[8,2] T=[32,4] CLog=[[8],[],[8]]
m=[2,3,19,89] x^2-10146 H=[4,2] T=[8,2] CLog=[[4],[],[4,2]]
m=[7,31,47] x^2-10199 H=[4,2] T=[4,4,2] CLog=[[4,2],[],[4,2]]
m=[10337] x^2-10337 H=[] T=[8] CLog=[[4],[4],[]]
m=[2,5297] x^2-10594 H=[8] T=[32] CLog=[[4],[],[8]]
m=[3,11,17,19]x^2-10659 H=[2,2,2]T=[2,2,2] CLog=[[2,2,2],[],[2,2,2]]
m=[17,641] x^2-10897 H=[2] T=[64,2] CLog=[[2,2],[2],[2]]
m=[3,5,17,47] x^2-11985 H=[2,2] T=[1024,2,2]
CLog=[[128],[64],[2]]
m=[12161] x^2-12161 H=[11] T=[512] CLog=[[16],[16],[]]
m=[2,3,13,167]x^2-13026 H=[2,2] T=[128,2] CLog=[[],[],[2]]
\end{verbatim}\ns
\subsection{Computation of \texorpdfstring{$\CT_{K, 3}$ and
$\order \BK_2(\BZ_K)[3^\infty]$}{Lg}} \label{B2}
For $p=3$, genus theory does not apply; we only give the valuation
of $\order \BK_2(\BZ_K)[3^\infty]$ and the structure of $\CT_{K, 3}$; trivial
cases are not written:
\smallskip
\ft\begin{verbatim}
{p=3;nu=5;n=6;for(m=2,10^2,if(core(m)!=m,next);P=x^2-m;
K=bnfinit(P,1);Kpn=bnrinit(K,p^nu);H=Kpn.cyc;T=List;e=matsize(H)[2];
for(k=1,e-1,C=H[e-k+1];v=valuation(C,p);if(v>0,listput(T,p^v)));
f=quaddisc(m);fn=p^(n+1)*f;c=1;while(Mod(c,p)==0 ||
kronecker(f,c)!=-1,c=c+1);S=0;forstep(a=1,fn/2,[1,2],
if(gcd(a,f)!=1,next);aa=lift(Mod(a/c,fn));la=(aa*c-a)/fn;
eps=kronecker(f,a);S=S+a*eps*(la+(1-c)/2));print
("m=",m," v_3(K_2Z)=",valuation(S,3)," T=",T))}
m=6 v_3(K_2Z)=1 T=[3] m=69 v_3(K_2Z)=1 T=[3]
m=15 v_3(K_2Z)=1 T=[3] m=74 v_3(K_2Z)=1 T=[9]
m=29 v_3(K_2Z)=1 T=[9] m=77 v_3(K_2Z)=1 T=[3]
m=33 v_3(K_2Z)=1 T=[3] m=78 v_3(K_2Z)=1 T=[3]
m=42 v_3(K_2Z)=3 T=[9] m=79 v_3(K_2Z)=1 T=[9]
m=43 v_3(K_2Z)=2 T=[3] m=82 v_3(K_2Z)=4 T=[3]
m=51 v_3(K_2Z)=1 T=[3] m=83 v_3(K_2Z)=1 T=[3]
(...)
m=10187 v_3(K_2Z)=2 T=[81] m=10673 v_3(K_2Z)=4 T=[27]
m=10239 v_3(K_2Z)=2 T=[3,9] m=10771 v_3(K_2Z)=6 T=[9]
m=10281 v_3(K_2Z)=1 T=[27] m=10842 v_3(K_2Z)=4 T=[3,3]
m=10297 v_3(K_2Z)=4 T=[3] m=10942 v_3(K_2Z)=2 T=[3,243]
m=10351 v_3(K_2Z)=1 T=[243] m=11062 v_3(K_2Z)=3 T=[27]
\end{verbatim}\ns
\section{Cyclic cubic fields}
\subsection{Computation of \texorpdfstring{$\order \BK_2(\BZ_K)[3^\infty]$, }{Lg}
\texorpdfstring{$\CT_{K, 3}$}{Lg} and \texorpdfstring{$\BC$}{Lg}}\label{C1}
We consider all cyclic cubic fields, which requires
calculating the Artin group giving $\Gal(\Q(\mu_f^{})/K)$; indeed, a
composite conductor $f$ gives rise to several fields, so that we must work from
the defining polynomial depending on suitable integers $a$, $b$ such that
$f = \frac{a^2+27\,b^2}{4}$ (another method would be to work in the group
$(\Z/f\Z)^\times$, giving other difficulties). We compare
$v_{\mathfrak m} \big(\order \BK_2(\BZ_K)[3^\infty] \big)$ to
$\BC$ to see if we are, or not, in case of equality $v_{\mathfrak m}
\big(\BL_3(s, \chi) \big) = \BC,\,\forall\,s \in \Z_3$.
\subsubsection{General program}
The program gives at first the list of all cyclic cubic fields, of
conductor $f$ (in ${\sf f}$, ${\sf f \in [bf, Bf]}$), with a defining
polynomial $P_K$ (in ${\sf PK}$).
We compute for $p=3$ the structures of the $3$-class group (in ${\sf H}$)
that of the $3$-torsion group (in ${\sf T}$) and $\order \BK_2(\BZ_K)[3^\infty]$
(in ${\sf \order K{\_{}\!\_{}} 2 Z}$).
\smallskip
Except if the first layer of the
cyclotomic $\Z_3$-extension of $K$ is non-ramified (e.g., ${\sf f=657,\
P=x^3-219*x+730,\ H_K=[3, 3],\ T=[3]}$), then $\CT_{K, 3}=1$ implies
$\CH_{K, 3} = \CR_{K, 3} = 1$ since $\CW_{K, 3}=1$:
\smallskip
\ft\begin{verbatim}
{bf=7;Bf=10^3;Q=y^2+y+1;Y=Mod(y,Q);for(f=bf,Bf,
if(isprime(f)==1,next);h=valuation(f,3);if(h!=0 & h!=2,next);
F=f/3^h;if(core(F)!=F,next);F=factor(F);Div=component(F,1);
d=matsize(F)[1];for(j=1,d,D=Div[j];if(Mod(D,3)!=1,break));
Df=component(factor(f),1);for(b=1,sqrt(4*f/27),
if(h==2 & Mod(b,3)==0,next);A=4*f-27*b^2;if(issquare(A,&a)==1,
if(h==0,if(Mod(a,3)==1,a=-a);PK=x^3+x^2+(1-f)/3*x+(f*(a-3)+1)/27);
if(h==2,if(Mod(a,9)==3,a=-a);PK=x^3-f/3*x-f*a/27);
nu=8;K=bnfinit(PK,1);HK=K.clgp;Kpn=bnrinit(K,3^nu);HKn=Kpn.cyc;
T=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,CK=HKn[e-k+1];
v=valuation(CK,3);if(v>0,w=w+v;listput(T,3^v)));
C=omega(f)-1;if(Mod(f,3)==0,C=C-1);forprime(q=2,10^2*f,
if(Mod(3*f,q)!=0 & polisirreducible(PK+O(q))==1,q0=q;break));
A0=List;A1=List;A2=List;c=q0;n=3;fn=3^n*f;N=eulerphi(fn)/(2*3);
for(s=1,fn/2,if(gcd(s,f*3)!=1,next);for(k=0,fn,r=s+k*fn;
if(isprime(r)==0,next);if(polisirreducible(PK+O(r))==0,
listput(A0,s);break)));S0=0;S1=0;S2=0;g=(1-c)/2;
for(j=1,N,s=A0[j];listput(A1,lift(Mod(s*q0,fn))));
for(j=1,N,s=A1[j];listput(A2,lift(Mod(s*q0,fn))));
for(j=1,N,s=A0[j];t=lift(Mod(s/c,fn));ls=(t*c-s)/fn;S0=S0+s*(ls+g));
for(j=1,N,s=A1[j];t=lift(Mod(s/c,fn));ls=(t*c-s)/fn;S1=S1+s*(ls+g));
for(j=1,N,s=A2[j];t=lift(Mod(s/c,fn));ls=(t*c-s)/fn;S2=S2+s*(ls+g));
S=S0+Y*S1+Y^2*S2;K2=norm(S)/3;v=valuation(K2,3);
if(v==C,print("f=",Df," P=",PK," H=",HK[2],
" T=",T," #K_2Z=",3^v," C(s)=",C," Equality"));
if(v>C,print("f=",Df," P=",PK," H=",HK[2],
" T=",T," #K_2Z=",3^v," C(s)=",C," Inequality")))))}
\end{verbatim}\ns
\subsubsection{Numerical results} We only write an excerpt of the results:
\smallskip
\ft\begin{verbatim}
f=[9,7] P=x^3-21*x-35 H=[3] T=[] #K_2Z=1 C(s)=0 Equality
f=[9,7] P=x^3-21*x+28 H=[3] T=[] #K_2Z=1 C(s)=0 Equality
f=[7,13] P=x^3+x^2-30*x-64 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[7,13] P=x^3+x^2-30*x+27 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[9,19] P=x^3-57*x-152 H=[3] T=[3] #K_2Z=3 C(s)=0 Inequality
f=[9,19] P=x^3-57*x+19 H=[3] T=[3] #K_2Z=3 C(s)=0 Inequality
f=[7,31] P=x^3+x^2-72*x+209 H=[3] T=[3,3]#K_2Z=9 C(s)=1 Inequality
f=[7,31] P=x^3+x^2-72*x-225 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[9,37] P=x^3-111*x+370 H=[3] T=[3] #K_2Z=3 C(s)=0 Inequality
f=[9,37] P=x^3-111*x+37 H=[3] T=[3] #K_2Z=3 C(s)=0 Inequality
f=[13,31]P=x^3+x^2-134*x-597 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[13,31]P=x^3+x^2-134*x+209 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[7,61] P=x^3+x^2-142*x+601 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
f=[7,61] P=x^3+x^2-142*x-680 H=[3] T=[9,9]#K_2Z=9 C(s)=1 Inequality
f=[7,67] P=x^3+x^2-156*x-799 H=[3] T=[3,3]#K_2Z=27 C(s)=1 Inequality
f=[7,67] P=x^3+x^2-156*x+608 H=[3] T=[3] #K_2Z=3 C(s)=1 Equality
(...)
f=[9,73] P=x^3-219*x-1241 H=[3,3] T=[3] #K_2Z=3 C(s)=0 Inequality
\end{verbatim}\ns
\subsection{Computation of \texorpdfstring{$\wt \CH_{K, 3}$}{Lg}}\label{clog3}
We use a part of the previous program, adding the computation of the logarithmic
class group and we write some data obtained up to conductors $\leq 10^3$, then
for larger conductors. The first component of ${\sf bnflog(K, 3)}$ gives
$\wt \CH_{K, 3}$, the second one (often trivial) is the sub-group
$\wt {\rm cl} (\langle S_3\rangle)$ generated by $S_3$ and the third component
is $\CH_{K, 3}^{S_3} = \CH_{K, 3}/{\rm cl} (\langle S_3\rangle)$; we have
$\wt \CH_{K, 3}/\wt {\rm cl} (\langle S_3\rangle) \simeq \CH_{K, 3}^{S_3}$ \cite{BeJa}.
\smallskip
\ft\begin{verbatim}
f=[19] x^3+x^2-6*x-7 H=[] T=[3] CLog=[[],[],[]]
f=[7,13] x^3+x^2-30*x-64 H=[3] T=[3] CLog=[[3],[],[3]]
f=[9,61] x^3-183*x-854 H=[3] T=[] CLog=[[],[],[3]]
f=[19,37] x^3+x^2-234*x-729 H=[2,6] T=[3,3] CLog=[[3],[3],[]]
f=[7,181] x^3+x^2-422*x+3191 H=[3,3] T=[3,3] CLog=[[3,3],[],[3,3]]
(...)
f=[7,19,73,103] P=x^3+x^2-333342*x+73964960
H=[3,3,3,3] T=[3,3,3,3] CLog=[[3,3,3,3],[],[3,3,3,3]]
f=[7,19,73,103] P=x^3+x^2-333342*x-69038901
H=[6,6,3] T=[3,3,3,3] CLog=[[3,3,3],[3],[3,3]]
f=[7,19,73,103] P=x^3+x^2-333342*x-49038361
H=[3,3,3] T=[3,3,3] CLog=[[3,3,3],[],[3,3,3]]
f=[397,251893] P=x^3+x^2-33333840*x+41296924413
H=[3] T=[9,3,3] CLog=[[3,3],[3,3],[]]
f=[7,14286007] P=x^3+x^2-33334016*x-56482638787
H=[54,18] T=[27,27] CLog=[[27,9],[],[27,9]]
f=[7,163,87643] P=x^3+x^2-33333554*x-214816239
H=[3,3] T=List([3,3,3]) CLog=[[3,3],[3],[3]]
f=[100001053] P=x^3+x^2-33333684*x-17437220649
H=[] T=List([9,3]) CLog=[[9,3],[9,3],[]]
f=[31,103,31321] P=x^3+x^2-33335984*x+68190607953
H=[6,6,3] T=List([9,9,3]) CLog=[[9,3],[3],[3,3]]
\end{verbatim}\ns
\subsection{Computation of \texorpdfstring{$\order \BK_2(\BZ_K)[2^\infty]$
and $\CT_{K, 2}$}{Lg}}\label{C2}
We still compute $\order \BR_2 (\BZ_K) [2^\infty] \sim \frac{1}{2} \cdot \BL_2(-1, \chi)
\times \frac{1}{2} \cdot \BL_2(-1, \chi^2)$, for prime conductors, noting that genus
theory does not apply for cyclic cubic fields with $p=2$. We compute $\CT_{K, 2}$,
at first with the classical program using ray class groups, then we use formula of
Theorem \ref{Lp} in the following complete program (we only print an example
of the six cases of structures that occur up to $f=3500$):
\smallskip
\ft\begin{verbatim}
{p=2;Q=y^2+y+1;Y=Mod(y,Q);forprime(f=7,5*10^5,
if(Mod(f,3)!=1,next);for(b=1,sqrt(4*f/27),A=4*f-27*b^2;
if(issquare(A,&a)==1,if(Mod(a,3)==1,a=-a);
P=x^3+x^2+(1-f)/3*x+(f*(a-3)+1)/27;K=bnfinit(P,1);nu=8;
Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;T=List;e=matsize(HKn)[2];
for(k=1,e-1,CK=HKn[e-k+1];v=valuation(CK,p);if(v>0,listput(T,p^v)));
g=znprimroot(f);c=lift(g);if(Mod(c,p)==0,c=c+f);n=8;ex=(f-1)/3;
cex=Mod(c,f)^ex;R0=0;R1=0;R2=0;T0=0;T1=0;T2=0;fn=p^(n+2)*f;
forstep(a=1,fn/2,2,if(Mod(a,f)==0,next);aa=lift(Mod(a/c,fn));
am=lift(Mod(a,fn)^-1);la=(aa*c-a)/fn;XR=a*(la+(1-c)/2);XT=am*la;
aex=Mod(a,f)^ex;if(aex==1,R0=R0+XR;T0=T0+XT);if(aex==cex,
R1=R1+XR;T1=T1+XT);if(aex==cex^2,R2=R2+XR;T2=T2+XT));
AR=lift(Mod(R0-R2,fn));BR=lift(Mod(R1-R2,fn));
AT=lift(Mod(T0-T2,fn));BT=lift(Mod(T1-T2,fn));
NR=norm(AR+Y*BR);wR=valuation(NR,p);
NT=norm(AT+Y*BT);wT=valuation(NT,p);print();print("f=",f," P=",P);
print("AR=",AR," BR=",BR," #R_2Z=",p^wR);print("T=",T);
print("AT=",AT," BT=",BT," #T=",p^wT))))}
f=7 P=x^3+x^2-2*x-1 f=739 P=x^3+x^2-246*x-520
AR=114683 BR=6 #R_2Z=1 AR=7232 BR=4072 #R_2Z=64
T=[] T=[8,8]
AT=45707 BT=57070 #T=1 AT=2395776 BT=8911992 #T=64
f=31 P=x^3+x^2-10*x-8 f=2689 P=x^3+x^2-896*x+5876
AR=24 BR=507870 #R_2Z=4 AR=1733440 BR=461096 #R_2Z=64
T=[2,2] T=[16,16]
AT=82976 BT=73118 #T=4 AT=29791312 BT=37866256 #T=256
f=277 P=x^3+x^2-92*x+236 f=3163 P=x^3+x^2-1054*x-13472
AR=3256 BR=4537692 #R_2Z=16 AR=432 BR=51769200 #R_2Z=256
T=[4,4] T=[16,16]
AT=718480 BT=1119412 #T=16 AT=11946416 BT=40520656 #T=256
f=3457 P=x^3+x^2-1152*x+13700 f=6163 P=x^3+x^2-2054*x+17576
AR=286536 BR=3460600 #R_2Z=64 AR=6140160 BR=6195120 #R_2Z=256
T=List([16,16]) T=List([8,8])
AT=2216784 BT=1012224 #T=256 AT=2754032 BT=3116408 #T=64
\end{verbatim}\ns
\smallskip
We know from \eqref{equalranks} that $\BR_2 (\BZ_K) [2^\infty]$ and $\CT_{K, 2}$
have same $2$-rank, but we see that most often the whole structures coincide
(first exceptions $f = 2689, 3457$ and $f=6163$ giving an interesting case).
\subsection{Zeroes of \texorpdfstring{$p$}{Lg}-adic \texorpdfstring{$\BL$}{Lg}-functions
and \texorpdfstring{$\BC(s)$}{Lg}}\label{C3}
A method, to study the influence of the zeroes, close to $s=1$, of $p$-adic
$\BL$-functions, is to construct families of degree-$p$ cyclic fields $K_n$ such that
the $p$-adic regulator tends to $0$ as $n \to \infty$, so $\BL_p(1, \chi_n) \to 0$
and $\order \CT_{K_n, p} \to 0$ from \eqref{T=HRW}; thus, invariants given by
some $\BL_p(s_0^{}, \chi_n)$, $s_0^{} \in \Z_p^\times$, have the same behavior
because of the role of $\BC(s) = \BC$ when $s \in \Z_p^\times$ (Theorem\ref{thmfond}).
\subsubsection{The Lecacheux--Washington cyclic cubic fields}
The following family of cyclic cubic fields $K$ illustrates, for $p=3$, the fact that,
in a genus theory context, the strict inequality (Theorem \ref{thmfond}\,{\bf (b)}):
$$v_{\mathfrak m} \big(\ffrac{1}{2} \BL_3(s, \chi) \big) > \BC,\,\forall\,s \in \Z_3, $$
can be largely exceeded because of $\BL_3(1, \chi)$ very close to $0$:
\smallskip
Using the family of cubic polynomials \cite{Lec, W4}:
$$P_n=x^3-(n^3-2n^2+3n-3) x^2-n^2 x-1, $$
for some values of $n \to 1$ in $\Z_3$, one obtains $3$-adic $\BL$-functions having
a zero near $s=1$, which of course gives large modules $\CT_{K_n, 3}$.
The required conditions are given in \cite[Theorem 3]{W4}.
\subsubsection{Numerical examples}
The constant $\BC$ associated to these fields is always $2$, so that we must obtain
$v_3(\CT_{K, 3}) > 2$ as soon as $n$ is close to $1$ in $\Z_3$.
We give the program and some examples suggesting that $\order \CT_{K_n, 3}$ is
unbounded in such families, as $n \to \infty$; but these fields have an huge discriminant
$D_n$ and the function $\mathcal C_p(K_n) := \ds \frac{\log(\order \CT_{K_n, p})}
{\log(\sqrt {D_n})}$, that we have introduced in \cite{Gra9}, is rather small and fulfills
the various conjectures given in this article:
\smallskip
\ft\begin{verbatim}
{M=3^5;forstep(n=1,10^6,M,if(isprime(n^2+3)==0,next);
A=n^2-3*n+3;if(numdiv(A)!=4,next);F=factor(A);
p=component(F,1)[1];q=component(F,1)[2];
if(Mod(p^2,9)==1 || Mod(q^2,9)==1,next);
P=x^3-(n^3-2*n^2+3*n-3)*x^2-n^2*x-1;K=bnfinit(P,1);
nu=18;HK=K.clgp;Kpn=bnrinit(K,3^nu);HKn=Kpn.cyc;
T=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,CK=HKn[e-k+1];
v=valuation(CK,3);if(v>0,w=w+v;listput(T,3^v)));
print("n=",n," p=",p," q=",q," T=",T," v_3(T)=",w))}
n=2674 p=43 q=166099 T=[3,81,243] v_3(T)=10
n=40096 p=3271 q=491461 T=[3,729,729] v_3(T)=13
n=43498 p=6547 q=288979 T=[3,81,243] v_3(T)=10
n=50788 p=28111 q=91753 T=[3,243,243] v_3(T)=11
n=56134 p=3613 q=872089 T=[3,729,2187] v_3(T)=14
n=76546 p=31 q=189001951 T=[3,2187,2187] v_3(T)=15
n=78490 p=7 q=880063519 T=[3,81,243] v_3(T)=10
n=96958 p=79 q=118994467 T=[3,9,243,729] v_3(T)=14
n=124660 p=7 q=2219963089 T=[3,6561,6561] v_3(T)=17
n=158194 p=135301 q=184957 T=[3,243,729] v_3(T)=12
(...)
n=570808 p=643 q=506718601 T=[3,6561,19683] v_3(T)=18
n=649540 p=229 q=1842359227 T=[3,3,19683,59049]v_3(T)=21
\end{verbatim}\ns
\noindent
Note that $n=570808 \equiv 1 \! \pmod {3^9}$ and $n=649540 \equiv 1\! \pmod {3^{10}}$.
\medskip
For $M = 3$, only giving $n \equiv 1 \pmod 3$, we obtain many equalities:
\smallskip
\ft\begin{verbatim}
n=340 p=7 q=16369 T=[3,3] v_3(T)=2
n=736 p=79 q=6829 T=[3,3] v_3(T)=2
n=970 p=7 q=133999 T=[3,3] v_3(T)=2
(...)
\end{verbatim}\ns
\smallskip
For $M = 9$, giving $n \equiv 1 \pmod 9$, we never obtain
a valuation $2$, but minimal valuations are $4$:
\smallskip
\ft\begin{verbatim}
n=874 p=7 q=108751 T=[3,3,9] v_3(T)=4
n=2926 p=1777 q=4813 T=[3,3,9] v_3(T)=4
n=4042 p=3067 q=5323 T=[3,3,9] v_3(T)=4
(...)
\end{verbatim}\ns
More generally, the minimal valuations are $2\,v_3(M)$.
\section{Numerical results for \texorpdfstring{$p \geq 5$}{Lg}}
When $p \geq 5$ and for a cyclic $p$-extension $K/\Q$, the trick used
for $p \in \{2, 3\}$ does not exist and we only have available the complex
analytic formulas of $\order \BK_2(\BZ_K)$ and the rank formula
\eqref{rankR} giving Theorem \ref{pdiv}.
\subsection{Computation of \texorpdfstring{$\order \BK_2(\BZ_K)[p^\infty]$}{Lg}}
\label{D1}
The computation of the order of $\BK_2(\BZ_K)$ uses Bernoulli
polynomials \cite[Theorem 4.2 using Proposition 4.1]{W}.
\smallskip
The following program computes the $p$-valuation of the product
$\prod_{\chi \ne \1} \BL(-1, \chi)$ for the $p-1$ characters
$\chi$ of order $p$ and prime conductor$\ell$ with the formula
$\BL(-1, \chi) \sim \sum_{a=1}^{\ell-1} \chi(a)\,(a^2-\ell a)$,
where $\chi(a)$ is computed from $\chi(g) := \zeta_p$, where $g$ is a
primitive root modulo $\ell$ and from the writing $a \equiv g^k \pmod \ell$
for $k \in [1, \ell-1]$; by comparison we compute the structure of $\CT_{K, p}$.
The case $p=3$ is included to recall that it is particular and may give trivial
modules; moreover, for $p>3$, equality \eqref{equalranks} does not apply
in general:
\smallskip
\ft\begin{verbatim}
{forprime(p=3,100,Q=polcyclo(p);X=Mod(x,Q);print();
forprime(ell=1,10^4,if(Mod(ell,p)!=1,next);g=znprimroot(ell);
L=0;for(k=1,ell-1,a=lift(g^k);E=a^2-a*ell;L=L+E*X^k);
w=valuation(norm(L),p);P=polsubcyclo(ell,p);K=bnfinit(P,1);
nu=8;Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;T=List;e=matsize(HKn)[2];
for(k=1,e-1,c=HKn[e-k+1];v=valuation(c,p);if(v>0,listput(T,p^v)));
print("p=",p," ell=",ell," T=",T," v(K_2Z)=",w)))}
p=3 ell=7 T=[] v(K_2Z)=0 p=11 ell=23 T=[] v(K_2Z)=1
p=3 ell=19 T=[3] v(K_2Z)=1 p=11 ell=727 T=[11] v(K_2Z)=1
p=3 ell=199 T=[3,3] v(K_2Z)=2 p=11 ell=1321 T=[] v(K_2Z)=3
p=3 ell=4177 T=[9,9] v(K_2Z)=3 p=11 ell=1453 T=[11] v(K_2Z)=1
p=3 ell=2593 T=[3,9] v(K_2Z)=6 p=11 ell=3631 T=[11] v(K_2Z)=2
p=3 ell=21997 T=[27,81]v(K_2Z)=7 p=11 ell=4357 T=[11,11]v(K_2Z)=1
p=5 ell=11 T=[] v(K_2Z)=1 p=13 ell=53 T=[] v(K_2Z)=1
p=5 ell=101 T=[5,5] v(K_2Z)=1 p=13 ell=677 T=[13] v(K_2Z)=1
p=5 ell=181 T=[] v(K_2Z)=3 p=13 ell=1483 T=[] v(K_2Z)=2
p=5 ell=401 T=[5,5] v(K_2Z)=2 p=13 ell=2029 T=[13] v(K_2Z)=1
p=5 ell=3001 T=[5] v(K_2Z)=7 p=13 ell=6761 T=[13] v(K_2Z)=4
p=5 ell=5351 T=[5,5,5]v(K_2Z)=1 p=13 ell=11831T=[13] v(K_2Z)=2
p=7 ell=29 T=[] v(K_2Z)=1 p=17 ell=103 T=[] v(K_2Z)=2
p=7 ell=127 T=[] v(K_2Z)=2 p=17 ell=137 T=[] v(K_2Z)=1
p=7 ell=197 T=[7,7] v(K_2Z)=1 p=17 ell=3469 T=[17] v(K_2Z)=1
p=7 ell=491 T=[7] v(K_2Z)=1 p=17 ell=3571 T=[] v(K_2Z)=1
p=7 ell=4159 T=[] v(K_2Z)=3 p=17 ell=3673 T=[] v(K_2Z)=1
p=7 ell=4229 T=[] v(K_2Z)=4 p=17 ell=3911 T=[] v(K_2Z)=1
\end{verbatim}\ns
\smallskip
For $p$ large, the $p$-valuation is almost often $1$ with a trivial $\CT_{K, p}$.
\subsection{Computations of \texorpdfstring{$\rk_5 (\BK_2(\BZ_K))$}{Lg}}
\label{D2}
The next programs compute the $\omega^{-1}$-component of the $5$-class
group $\CH_{K_2}$ of $K_2 := K(\mu_5^{})$, for degree-$5$ cyclic
fields $K$, of prime conductor in the first section, then with $f=11 \cdot 31$
in the second section, thus giving the $5$-rank
of $\BK_2\BZ_K[5^\infty]$ (formula \eqref{rankR}, where $\langle S_5(K') \rangle$
is of character $\chi_0^{}$).
\smallskip
The program computes the $5$-class group $\CH_{K_1}$ of
$K_1 := K(\sqrt 5$), so that
$\CH_{K_2} = \CH_{K_1} \plus\ (\CH_{K_2})_{\omega^{-1}} \plus\
(\CH_{K_2})_{\omega}$,
where $\rk_5 (\CH_{K_2})_{\omega} = \rk_5(\CT_{K, 5})$ (formula \eqref{rankT}),
where the structure of $\CT_{K, 5}$ is computed as usual via the program
of Section \ref{A1}. Whence:
$$\rk_5 (\BK_2(\BZ_K)) = \rk_5 (\CH_{K_2})-\rk_5 (\CH_{K_1})
- \rk_5 (\CT_{K, 5}). $$
\subsubsection{Case of prime conductors}
So, $f$ is a prime number $\ell \equiv 1 \pmod 5$; thus, $\CH_K = 1$.
We write an excerpt of each structures obtained:
\smallskip
\ft\begin{verbatim}
{p=5;forprime(ell=11,10^5,if(Mod(ell,p)!=1,next);
P=polsubcyclo(ell,p);print("conductor ell=",ell," P=",P);nu=8;
K=bnfinit(P,1);HK=K.clgp[2];Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;
TK=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,Ck=HKn[e-k+1];
v=valuation(Ck,p);if(v>0,listput(TK,p^v)));
R1=polcompositum(P,polsubcyclo(p,2))[1];LK1=bnfinit(R1,1);
R2=polcompositum(P,polcyclo(p))[1];LK2=bnfinit(R2,1);
print("TK=",TK," HK1=",LK1.clgp[2]," HK2=",LK2.clgp[2])}
conductor ell=11 P=x^5+x^4-4*x^3-3*x^2+3*x+1
TK=[] HK1=[] HK2=[5]
conductor ell=31 P=x^5+x^4-12*x^3-21*x^2+x+5
TK=[] HK1=[] HK2=[10,10,2,2]
conductor ell=101 P=x^5+x^4-40*x^3+93*x^2-21*x-17
TK=[5,5] HK1=[] HK2=[2005,5,5]
conductor ell=151 P=x^5+x^4-60*x^3-12*x^2+784*x+128
TK=[5] HK1=[] HK2=[5305,5]
conductor ell=251 P=x^5+x^4-100*x^3-20*x^2+1504*x+1024
TK=[5] HK1=[] HK2=[11810,10,2,2]
conductor ell=281 P=x^5+x^4-112*x^3-191*x^2+2257*x+967
TK=[] HK1=[5] HK2=[261005,5,5]
conductor ell=401 P=x^5+x^4-160*x^3+369*x^2+879*x-29
TK=[5,5] HK1=[] HK2=[42505,5,5,5]
conductor ell=421 P=x^5+x^4-168*x^3+219*x^2+3853*x-3517
TK=[] HK1=[5] HK2=[225915,15,3,3]
\end{verbatim}\ns
\smallskip
The rank $2$ is obtained for $\ell = 31, 281, 401$.
\subsubsection{An example of composite conductor}
The next program computes the $\omega^{-1}$-component
for the four degree-$5$ cyclic fields $K$ of conductor $f=11 \cdot 31$
(for which $\CT_{K, 5} = 1$). One may replace ${\sf f=11*31}$ by any
product of primes $\ell_1 \equiv \ell_2 \equiv 1 \pmod 5$):
\smallskip
\ft\begin{verbatim}
{p=5;f=11*31;LP=polsubcyclo(f,p);d=matsize(LP)[2];
for(i=1,d,P=LP[i];D=nfdisc(P);if(D!=f^(p-1),next);
print();print(P);K=bnfinit(P,1);HK=K.clgp[2];
nu=8;Kpn=bnrinit(K,p^nu);HKn=Kpn.cyc;
TK=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,Ck=HKn[e-k+1];
v=valuation(Ck,p);if(v>0,listput(TK,p^v)));
R1=polcompositum(P,polsubcyclo(p,2))[1];LK1=bnfinit(R1,1);
R2=polcompositum(P,polcyclo(p))[1];LK2=bnfinit(R2,1);
print("HK=",HK," TK=",TK," HK1=",LK1.clgp[2]," HK2=",LK2.clgp[2])}
x^5-x^4-136*x^3+641*x^2-371*x-67
HK=[5] TK=[] HK1=[5,5] HK2=[218305,5,5,5,5]
x^5-x^4-136*x^3-723*x^2-1053*x-67
HK=[5] TK=[] HK1=[5,5] HK2=[145505,5,5,5,5]
x^5-x^4-136*x^3-41*x^2+3039*x-1431
HK=[5] TK=[] HK1=[5,5] HK2=[3805,5,5,5,5,5,5]
x^5-x^4-136*x^3+300*x^2+2016*x-3136
HK=[5] TK=[] HK1=[5,5] HK2=[12605,5,5,5,5,5]
\end{verbatim}\ns
\smallskip
The third case gives $\rk_5((\BK_2\BZ_K[5^\infty]_{\omega^{-1}}) = 4$
and the last one gives the rank $3$.
With $f=11 \cdot 101$, one obtains interesting structures, giving
the ranks $2, 2, 4, 3$, respectively:
\smallskip
\ft\begin{verbatim}
x^5-x^4-444*x^3+3644*x^2+80*x-32000
HK=[5] TK=[5] HK1=[5,5] HK2=[1196410,10,10,10,5]
x^5-x^4-444*x^3+311*x^2+32299*x+26883
HK=[5] TK=[5] HK1=[5,5] HK2=[15020305,5,5,5,5]
x^5-x^4-444*x^3-1911*x^2+12301*x-8669
HK=[5] TK=[5] HK1=[5,5] HK2=[1405,1405,5,5,5,5,5]
x^5-x^4-444*x^3+311*x^2+43409*x+4663
HK=[5] TK=[5] HK1=[5,5] HK2=[5559605,5,5,5,5,5]
\end{verbatim}\ns
\smallskip
These ranks may indeed be larger than the number of tame ramified primes;
this illustrates Theorem \ref{pdiv}.
\subsection{Computation of \texorpdfstring{$\BL(-1-(p-3)\,p^n, \chi)$}{Lg}}\label{D3}
Let $K/\Q$ be a degree-$p$ cyclic extension. For $p \geq 5$ the conditions
\eqref{flp} ($m>1$ and $m \equiv 0 \pmod{p-1}$) are not
fulfilled for $m=2$, which does not allow to express $\BL(-1, \chi)$ by means of
$\BL_p$-function at $s = -1$. But we may compute, with:
$$m := 2+(p-3) p^n > 1\ \,{\rm and}\ \,(-1)_n := 1- m = -1-(p-3)p^n, $$
the following expression:
\begin{equation*}
\left \{\begin{aligned}
\BL_p(-1-(p-3)\,p^n, \chi) & = (1-p^{1+(p-3)\,p^n})\,\BL((-1)_n, \chi)) \\
& \sim \BL_p(-1, \chi),\ \,\hbox{for $n$ large enough}.
\end{aligned}\right.
\end{equation*}
\noindent
This does not give $\order \BK_2(\BZ_K)[p^\infty]$ but is related to
$\order \BK_{2m-2}(\BZ_K)[p^\infty]$ of higher $\BK$-theory, from
Quillen--Lichtenbaum conjecture, via the $\BL((-1)_n, \chi)$'s \cite{Ko1, Ng2}:
$$\zeta_K^{} ((-1)_n) = \zeta_K^{} (1-m) = \pm \frac{\order \BK_{2m-2}(\BZ_K)}{w_m(K)}, $$
where $w_m(K)$ is the largest integer $N$ such that $[K(\zeta_N^{}) : K] \mid m$.
One computes that for a degree-$p$ cyclic extension $K/\Q$ for $p \ne 2$, $w_m(K) \sim p$,
which will give $v_p(\order \BK_2(\BZ_K)[p^\infty] = v_p(\zeta_K^{} (1-m))+1$.
\smallskip
Indeed, the condition $m \equiv 0 \pmod {(p-1)}$ of definition \eqref{flp} is fulfilled
for $m = 2+(p-3) p^n$, so that relation between $p$-adic and complex
$\BL$-functions does exist and leads, for $n$ large enough, by continuity
of $\BL_p(s, \chi)$, to
$\prod_{\chi \ne \1}\,\BL((-1)_n, \chi) \sim \prod_{\chi \ne \1}\,\BL_p(-1, \chi)$,
since the Euler factors $1-p^{m-1} \chi(p)$ are $p$-adic units.
\smallskip
Thus, formula of Theorem \ref{Lp} for $s = (-1)_n$ makes sense for
computing $\BL((-1)_n, \chi)$, of constant valuation for $n \geq n_0$,
and only depending on the computation of $\BL_p(-1, \chi)$ related to
$\order \BK_2(\BZ_K)[p^\infty]$ and, in some sense, to $\order \CT_{K, p}$.
\smallskip
The complex analytic computation of $\BL((-1)_n, \chi)$ is done by the
following program, with $m = 2+(p-3)p^n$, but with less immediate
Bernoulli polynomials; so we use their definition with series;
we give also the computation of the structure of $\CT_{K, p}$:
\smallskip
\ft\begin{verbatim}
{n=2;forprime(p=5,50,m=2+(p-3)*p^n;print();forprime(ell=1,60000,
if(Mod(ell,p)!=1,next);Q=polcyclo(p);X=Mod(x,Q);g=znprimroot(ell);
P=polsubcyclo(ell,p);K=bnfinit(P,1);nu=8;Kpn=bnrinit(K,p^nu);
HKn=Kpn.cyc;TK=List;w=0;e=matsize(HKn)[2];for(k=1,e-1,
Ck=HKn[e-k+1];v=valuation(Ck,p);if(v>0,listput(TK,p^v)));
Factm=1;for(i=1,m,Factm=i*Factm);
B=(x+O(x^(m+2)))*exp(x*y/ell+O(x^(m+2)))/(exp(x+O(x^(m+2)))-1);
Bm=ell^(m-1)/m*Factm*polcoeff(B,m);L=0;dm=poldegree(Bm);
for(k=1,ell-1,a=lift(g^k);E=0;for(j=0,dm,cm=polcoeff(Bm,j);
E=E+cm*a^j);L=L+E*X^k);vm=valuation(norm(L),p);
print("p=",p," ell=",ell," T=",TK," v(K_(2m-2)Z)=",vm+1)))}
\end{verbatim}\ns
\smallskip
For instance, for $p=5$ and cyclic quintic fields of prime conductor$\ell$,
this gives the following examples where $m=52$:
\smallskip
\ft\begin{verbatim}
ell=101 T=[5,5] v(K_(2m-2)Z)=3 ell=401 T=[5,5] v(K_(2m-2)Z)=3
ell=151 T=[5] v(K_(2m-2)Z)=2 ell=5351 T=[5,5,5] v(K_(2m-2)Z)=4
ell=251 T=[5] v(K_(2m-2)Z)=2 ell=29251T=[5,5,5,5] v(K_(2m-2)Z)=5
\end{verbatim}\ns
\smallskip
In the selected interval, $\order \CT_{K, 5}$ and $\prod_{\chi \ne \1} \BL_5(-1, \chi)$
have same valuation, giving $v_5(\order \BK_2(\BZ_K)) = v_5(\order \CT_{K, 5})+1$,
whereas $\order \CT_{K, 5}$ is also obtained $p$-adically,
replacing $\langle a \rangle \cdot \theta^{-1}\chi(a)$ by $a^{-1}\,\chi(a) =
\langle a \rangle^{-1} \theta^{-1}\chi(a)$ in formula of Theorem \ref{Lp}.
However, the minimal $\ell$ giving distinct $5$-valuations is given by the following data:
\smallskip
\ft\begin{verbatim}
ell=56401 T=[5,5,5,25] v(T)=5 v(K_(2m-2)Z)=7
S0=173625530 S1=5807338145*Y S2=-7744983390*Y^2 S3=-4753111575/2*Y^3
S4=-8281263805/2*Y^3-8281263805/2*Y^2-8281263805/2*Y-8281263805/2
S=-6517187690*Y^3-23771230585/2*Y^2+3333412485/2*Y-7934012745/2
=5*(-1303437538*Y^3-4754246117/2*Y^2+666682497/2*Y-1586802549/2)
norm(S)=5^4*539252420555567630791570203161974917625/16=5^7*u
R=norm(S)/p=5^6*u
\end{verbatim}\ns
\smallskip
Computing $\BL_5((-1)_n, \chi)$ by means of the
formula of Theorem \ref{Lp} gives again the results:
\smallskip
\ft\begin{verbatim}
{n=2;p=5;m=2+(p-3)*p^n;Q=polcyclo(p);X=Mod(x,Q);
forprime(ell=1,5351,if(Mod(ell,p)!=1,next);fn=p*p^n*ell;
g=znprimroot(ell);c=lift(g);if(Mod(c,p)==0,c=c+ell);S=0;
for(a=1,fn/2,if(gcd(a,p*ell)!=1,next);aa=lift(Mod(a/c,fn));
la=(aa*c-a)/fn;A=lift(Mod(a,fn)^(m-1));u=znlog(Mod(a,ell),g);
S=S+(la+(1-c)/2)*A*X^u);NS=norm(S);vm=valuation(NS,p)-1;
print("p=",p," ell=",ell," v(K_(2m-2)Z)=",vm+1))}
ell=101 v(K_(2m-2)Z)=3 ell=401 v(K_(2m-2)Z)=3
ell=151 v(K_(2m-2)Z)=2 ell=601 v(K_(2m-2)Z)=2
ell=251 v(K_(2m-2)Z)=2 ell=701 v(K_(2m-2)Z)=3
\end{verbatim}\ns
\end{appendix}
\begin{thebibliography}{xxxxxxxx}
\bibitem[AAM2021]{AAM} J. Assim, H. Asensouyis, Y. Mazigh {\it A genus formula for
the positive \'etale wild kernel}, J. Number Theory {\bf 218} (2021), 161--179. \\
\url{https://doi.org/10.1016/j.jnt.2020.07.013}
\bibitem[AF1972]{AF} Y. Amice, J. Fresnel, {\it Fonctions z\^eta $p$-adiques des corps de
nombres ab\'eliens r\'eels}, Acta Arith. {\bf 20} (1972), 353--384. \\
\url{https://doi.org/10.24033/msmf.37}
\bibitem[Ba1968]{Ba} H. Bass, {\it Algebraic ${\rm K}$-theory}, Mathematics Lecture Note Series,
Benjamin, New York, Amsterdam 1968. \\
\url{https://www.maths.ed.ac.uk/v1ranick/papers/bass.pdf}
\bibitem[BeJa2016]{BeJa} K. Belabas, J-F. Jaulent, {\it The logarithmic class group
package in PARI/GP}, Pub. Math. Besan\c con (2016), 5--18. \\
\url{https://doi.org/10.5802/pmb.o-1}
\bibitem [Ber1990]{Ber} R.I. Berger, {\it Quadratic extensions of number fields with
elementary abelian $2$-prim ${\rm K}_2(O_F)$ of smallest rank},
Jour. Number Theory {\bf 34} (1990) 284--292.
\url{https://doi.org/10.1016/0022-314X(90)90138-H}
\bibitem[BGr1992]{BGr} R.I. Berger, G. Gras, {\it Regular fields: normic criteria in
$p$-extensions}, Publ. Math. Fac. Sci. Besan\c con (Th\'eorie des Nombres),
Ann\'ees 1991/92 (1992), 1--5..
\url{https://doi.org/10.5802/pmb.a-67}
\bibitem[BN2005]{BN} J-R. Belliard, T. Nguyen Quang Do, {\it On modified circular
units and annihilation of real classes}, Nagoya Math. J. {\bf 177} (2005), 77--115.\\
\url{https://doi.org/10.1017/S0027763000009065}
\bibitem[BP1972]{BP} F. Bertrandias, J-J. Payan, {\it $\Gamma$-extensions et invariants
cyclotomiques}, Ann. Sci. Ec. Norm. Sup. {\bf 4}(5) (1972), 517--548. \\
\url{https://doi.org/10.24033/asens.1236}
\bibitem[Br1992]{Br1} J. Browkin, {\it On the p-rank of the tame kernel of algebraic
number fields}, J. Reine Angew. Math. {\bf 432} (1992), 135--149. \\
\url{https://doi.org/10.1515/crll.1992.432.135}
\bibitem[Br2005]{Br2} J. Browkin, {\it Tame kernels of cubic cyclic fields},
Math. Comp. {\bf 74} (2005), 967--999. \\
\url{https://doi.org/10.1090/S0025-5718-04-01726-0}
\bibitem[BS1982]{BS} J. Browkin, A. Schinzel, {\it On Sylow $2$-subgroups of $K_2O_F$
for quadratic number fields $F$}, J. Reine Angew. Math. {\bf 331} (1982), 104--113.\\
\url{https://eudml.org/doc/152417}
\bibitem[Che1933]{Che} C. Chevalley, {\it Sur la th\'eorie du corps de classes
dans les corps finis et les corps locaux}, Th\`ese no. 155, Jour. of the Faculty
of Sciences Tokyo, \textbf{2} (1933), 365--476. \\
\url{http://archive.numdam.org/item/THESE_1934__155__365_0/}
\bibitem[Co1975]{C} J. Coates, {\it $p$-adic $L$-functions and Iwasawa's theory},
Algebraic number fields: $L$-functions and Galois properties (Proc. Sympos.,
Univ. Durham, Durham, 1975), Academic Press, London (1977), 269--353.\\
\url{https://doi.org/10.17863/CAM.72005}
\bibitem[Col1988]{Col} P. Colmez, {\it R\'esidu en $s = 1$ des fonctions z\^eta
$p$-adiques}, Invent. Math. {\bf 91} (1988), 371--389.
\url{https://eudml.org/doc/143545}
\bibitem[DeLi2023]{DL} L-T. Deng, Y-X. Li, {\it An application of Birch--Tate formula
to tame kernels of real quadratic number fields}. \\
\url{https://doi.org/10.48550/arXiv.2307.05895}
\bibitem[DR1980]{DR} P. Deligne, K.A. Ribet, {\it Values of Abelian $L$-functions
at Negative Integers over Totally Real Fields}, Invent. math. {\bf 59} (1980), 227--286. \\
\url{https://eudml.org/doc/142740}
\bibitem[Fr1965]{Fre} J. Fresnel, {\it Nombres de Bernoulli et fonctions $L$ $p$-adiques},
Ann. Sci. Inst. Fourier {\bf 17}(2) (1967), 281--333.
\url{https://doi.org/10.5802/aif.271} \\
S\'eminaire Delange--Pisot--Poitou (Th\'eorie des nombres), Tome {\bf 7},
(1965-1966), no. 2, Expos\'e no. 14, 1--15. \\
\url{http://www.numdam.org/item?id=SDPP_1965-1966__7_2_A3_0}
\bibitem[Gar1971]{Gar} H. Garland, {\it A finiteness theorem for $K_{2}$ of a
number field}, Ann. of Math. (2) {\bf 94} (1971), 534--548.
\url{https://doi.org/10.2307/1970769}
\bibitem[Gra1978]{Gr1} G. Gras, {\it Sur la construction des fonctions $L$ $p$-adiques
ab\'eliennes}, S\'eminaire Delange--Pisot--Poitou (Th\'eorie des nombres), Tome {\bf 20}
(1978--1979), no. 2 , Expos\'e no. 22, 1--20. \\
\url{http://www.numdam.org/item?id=SDPP_1978-1979__20_2_A1_0}
\bibitem[Gra1979]{Gr2} G. Gras, {\it Annulation du groupe des $\ell$-classes
g\'en\'eralis\'ees d'une extension ab\'elienne r\'eelle de degr\'e premier \`a $\ell$},
Ann. Sci. Inst. Fourier {\bf 29}(1) (1979), 15--32.
\url{https://doi.org/10.5802/aif.725}
\bibitem[Gra1986${}^a$]{Gr3} G. Gras, {\it Remarks on $K_2$ of number fields}, Jour.
Number Theory {\bf 23} (1986), 322--335. \\
\url{https://doi.org/10.1016/0022-314X(86)90077-6}
\bibitem[Gra1986${}^b$]{Gr33} G. Gras, {\it Th\'eorie des genres analytique des
fonctions $L$ $p$-adiques des corps totalement r\'eels}, Invent. math. {\bf 86}
(1986), 1--17. \\
\url{https://doi.org/10.1007/BF01391492}
\bibitem[Gra1987]{Gr34} G. Gras, {\it Pseudo-mesures associ\'ees aux fonctions
$L$ de $\Q$}, manuscripta math. {\bf 57} (1987), 373--415.\\
\url{https://doi.org/10.1007/BF01168668}
\bibitem[Gra1998]{Gr4} G. Gras, {\it Th\'eor\`emes de r\'eflexion}, J. Th\'eorie des
Nombres de Bordeaux {\bf 10}(2) (1998), 399--499.
\url{http://eudml.org/doc/248168}
\bibitem[Gra2005]{Gr5} G. Gras, {\it Class Field Theory: from theory to practice},
corr. 2nd ed., Springer Monographs in Mathematics, Springer, 2005, xiii+507 pages. \\
\url{https://doi.org/10.1007/978-3-662-11323-3}
\bibitem[Gra2016]{Gr55} G. Gras, {\it Les $\theta$-r\'egulateurs locaux d'un nombre alg\'ebrique :
Conjectures $p$-adiques}, Canadian Journal of Mathematics {\bf 68}(3) (2016), 571--624.
\url{http://dx.doi.org/10.4153/CJM-2015-026-3} \\
English translation: \url{https://arxiv.org/pdf/1701.02618.pdf}
\bibitem[Gra2017${}^a$]{Gr6} G. Gras, {\it On $p$-rationality of number fields.
Applications -- PARI/GP programs}, Publ. Math. Fac. Sci. Besan\c con
(Th\'eorie des Nombres), Ann\'ees 2017/2018 (2019), 29--51.
\url{https://doi.org/10.5802/pmb.35}
\bibitem[Gra2017${}^b$]{Gr7} G. Gras, {\it Invariant generalized ideal
classes--Structure theorems for $p$-class groups in $p$-extensions},
Proc. Indian Acad. Sci. (Math. Sci.), {\bf 127}(1) ( 2017), 1--34. \\
\url{https://doi.org/10.1007/s12044-016-0324-1}
\bibitem[Gra2018${}^b$]{Gra8} G. Gras, {\it The $p$-adic Kummer--Leopoldt Constant:
Normalized $p$-adic Regulator}, Int. J. Number Theory {\bf 14}(2) (2018), 329--337. \\
\url{https://doi.org/10.1142/S1793042118500203}
\bibitem[Gra2019$^a$]{Gra88} G. Gras, {\it Practice of the Incomplete $p$-Ramification
Over a Number Field -- History of Abelian $p$-Ramification},
Communications in Advanced Mathematical Sciences {\bf 2}(4) (2019), 251--280.\\
\url{https://doi.org/10.33434/cams.573729}
\bibitem[Gra2019$^b$]{Gra9} G. Gras, {\it Heuristics and conjectures in direction of
a $p$-adic Brauer--Siegel theorem}, Math. Comp. {\bf 88}(318) (2019), 1929--1965. \\
\url{https://doi.org/10.1090/mcom/3395}
\bibitem[Gra2023]{Gra10} G. Gras, {\it The Chevalley--Herbrand formula and the real
abelian Main Conjecture (New criterion using capitulation of the class group)},
\emph {J. Number Theory} {\bf 248} (2023), 78--119. \\
\url{https://doi.org/10.1016/j.jnt.2023.01.002}
\bibitem[GJ1989]{GJ} G. Gras, J-F. Jaulent, {\it Sur les corps de nombres r\'eguliers},
Math. Z. {\bf 202} (1989), 343--365.
\url{http://eudml.org/doc/174095}
\bibitem[GJN2016]{GJN} {\it Sur la capitulation du module de Bertrandias--Payan},
In: Publ. Math. Fac. Sci. Besan\c con (Alg\`ebre et Th\'eorie des Nombres) (2016): \par
G. Gras, {\it Sur le module de Bertrandias--Payan dans une $p$-extension--Noyau de
capitulation}, 25--44. \url{https://doi.org/10.5802/pmb.o-3}\\
J-F. Jaulent, {\it Sur la capitulation pour le module de Bertrandias--Payan}, 45--58.
\url{https://doi.org/10.5802/pmb.o-4} \\
T. Nguyen Quang Do, {\it Descente galoisienne et capitulation entre modules de
Bertrandias--Payan}, 59--79. \url{https://doi.org/10.5802/pmb.o-5}
\bibitem[HuK1998]{HK} J. Hurrelbrink, M. Kolster, {\it Tame kernels under relative
qua\-dratic extensions and Hilbert symbols}, J. Reine Angew. Math. {\bf 499} (1998), 145--188.
\url{https://doi.org/10.1515/crll.1998.056}
\bibitem[Hur1987]{Hur} J. Hurrelbrink, {\it Class numbers, units and $K_2$}, Algebraic
$K$-theory: connections with geometry and topology (Lake Louise, AB, 1987), 87--102,
NATO Adv. Sci. Inst. Ser. C: Math. Phys. Sci., 279, Kluwer Acad. Publ., Dordrecht, 1989.\\
\url{https://doi.org/10.1007/978-94-009-2399-7_4}
\bibitem[Iw1972]{Iw} K. Iwasawa, {\it Lectures on $p$-adic $L$-functions}, Annals of
Mathematics Studies {\bf 74}, Princeton Univ. Press 1972, 112 pp. \\
\url{https://doi.org/10.1515/9781400881703}
\bibitem[Jau1986] {Jau1} J-F. Jaulent, {\it L'arithm\'etique des ${\ell}$-extensions
(Th\`ese d'\'etat)}, Publications Math\'ematiques de Besan\c con {\bf 1}(1) (1986), 1--357. \\
\url{https://doi.org/10.5802/pmb.a-42}
\bibitem[Jau1990] {Jau2} J-F. Jaulent, {\it La th\'eorie de Kummer et le $K_2$
des corps de nombres}, J. Th\'eorie des Nombres de Bordeaux {\bf 2} (1990), 377--411.\\
\url{http://www.numdam.org/item/JTNB_1990__2_2_377_0/}
\bibitem[Jau1994${}^a$] {Jau22} J-F. Jaulent, {\it Sur le noyau sauvage des corps de nombres},
Acta Arith. {\bf 67} (1994), 335--348.
\url{http://eudml.org/doc/206636}
\bibitem[Jau1994${}^b$]{Jau23} J-F. Jaulent, {\it Classes logarithmiques des corps de nombres},
J. Th\'eorie Nombres Bordeaux {\bf 6}(2) (1994), 301--325.\\
\url{https://jtnb.centre-mersenne.org/article/JTNB_1994__6_2_301_0.pdf}
\bibitem[Jau2019]{Jau24} J-F. Jaulent, {\it Note sur la conjecture de Greenberg},
J. Ramanujan Math. Soc. {\bf 34}(1) (2019), 59--80. \\
\url{https://www.i-scholar.in/index.php/rms/article/view/181346}
\bibitem[Jau2020]{Jau25} J-F. Jaulent, {\it Normes universelles et conjecture de
Greenberg}, Acta Arithmetica 194 (2020), 99--109. \\
\url{https://doi.org/10.4064/aa190623-27-9}
\bibitem[Jau2021]{Jau3} J-F. Jaulent, {\it Annulateurs de Stickelberger des groupes de
classes logarithmiques}, Acta Arith. {\bf 203} (2021), 241--253. \\
\url{https://doi.org/10.4064/aa201127-22-6}
\bibitem[Jau2023]{Jau4} J-F. Jaulent, {\it Annulateurs circulaires des groupes de classes
logarithmiques} (2023), Ann. Sci. Inst. Fourier (to appear). \\
\url{https://doi.org/10.48550/arXiv.2003.12301}
\bibitem[JaMi2006]{JM} J-F. Jaulent, A. Michel, {\it Approche logarithmique des noyaux
\'etales sauvages des corps de nombres}, Jour. Number Theory {\bf 120} (2006), 72--91.
\url{https://doi.org/10.1016/j.jnt.2005.11.011}
\bibitem[JaSo2001]{JS} J-F. Jaulent, F. Soriano, {\it Sur le noyau sauvage des corps de nombres
et le groupe des classes logarithmiques}, Math. Z. {\bf 238}(2) (2001), 335--354.
\url{https://doi.org/10.1007/s002090100256}
\bibitem[Keu1989]{K} F. Keune, {\it On the structure of the ${\rm K}_2$ of the ring of integers
in a number field}, $K$-Theory {\bf 2} (1989), 625--645. \\
\url{https://doi.org/10.1007/BF00535049}
\bibitem[KL1964]{KL} T. Kubota, H.W. Leopoldt, {\it Eine $p$-adische Theorie der Zetawerte. I.
Einf\"uhrung der $p$-adischen Dirichletschen $L$-Funktionen}, J. Reine Angew. Math.,
{\bf 214/215} (1964), 328--339. \\
\url{https://eudml.org/doc/150624}
\bibitem[Kol2002]{Ko1} M. Kolster, {\it $K$-theory and arithmetic}, In: Contemporary developments
in algebraic $K$-theory 2002 (Karoubi, M., Ed.), ICTP Lecture Notes {\bf 15} (2003), 191--258.\\
\url{https://www.osti.gov/etdeweb/servlets/purl/20945047}
\bibitem[Kol1989]{Ko2} M. Kolster, {\it A relation between the $2$-primary parts of the main
conjecture and the Birch--Tate conjecture}, Canad. Math. Bull. {\bf 32}(2) (1989), 248--251.
\url{https://doi.org/10.4153/CMB-1989-036-8}
\bibitem[KM2003]{KM} M. Kolster, A. Movahhedi, {\it Bi-quadratic number fields with trivial
2-primary Hilbert kernels}, Proc. London Math. Soc. {\bf 87}(3) (2003), 109--136.
\url{https://doi.org/10.1112/S0024611502014028}
\bibitem[Lec1993]{Lec} O. Lecacheux, {\it Units in Number Fields and Elliptic Curves},
Advances in number theory (Kingston, ON, 1991), Oxford Sci. Publ., Oxford Univ. Press,
New York, 1993, pp. 293--301. \\
\url{https://doi.org/10.1093/oso/9780198536680.003.0022}
\bibitem[Mai2018]{Mai} C. Maire, {\it Genus theory and governing fields},
New York J. Math. {\bf 24} (2018) 1056--1067.
\url{http://nyjm.albany.edu/j/2018/24-50v.pdf}
\bibitem[Mov1988]{Mov} A. Movahhedi, {\it Sur les $p$-extensions des corps $p$-rationnels},
Th\`ese, Universit\'e Paris VII, 1988.
\url{http://www.unilim.fr/pages_perso/chazad.movahhedi/These_1988.pdf}
\bibitem[Ng1990]{Ng} T. Nguyen Quang Do, {\it Une \'etude cohomologique de la partie 2-primaire
de $K_2{\mathcal O}$}, $K$-Theory {\bf 3} (1990), 523--542.\\
\url{http://dx.doi.org/10.1007/BF01054449}
\bibitem[Ng1986]{Ng1} T. Nguyen Quang Do, {\it Sur la $\Z_p$-torsion de certains modules
galoisiens}, Ann. Sci. Inst. Fourier {\bf 36}(2) (1986), 27--46.\\
\url{https://doi.org/10.5802/aif.1045}
\bibitem[Ng1992]{Ng2} T. Nguyen Quang Do, {\it Analogues sup\'erieurs du noyau
sauvage}, J. Th\'eorie des Nombres de Bordeaux {\bf 4} (1992), 263--271.\\
\url{http://www.numdam.org/item/JTNB_1992__4_2_263_0/}\\
\url{http://www.numdam.org/item/?id=JTNB_1993__5_1_217_0}
\bibitem[PARI]{PARI} The PARI Group, {\it PARI/GP, version \texttt{2.5.3} (2013)},
Univ. Bordeaux.\\ \url{http://pari.math.u-bordeaux.fr/download.html}
\bibitem[Qin1995]{Q1} H-R. Qin, {\it The $4$-rank of $K_2O_F$ for real quadratic fields $F$},
Acta Arith. {\bf 72}(4) (1995), 323--333.
\url{http://eudml.org/doc/206799}
\bibitem[Qin2005]{Q2} H-R. Qin, {\it The $2$-Sylow subgroup of $K_2O_F$ for number fields $F$},
J. Algebra {\bf 284}(2) (2005), 494--519. \\
\url{https://doi.org/10.1016/J.JALGEBRA.2004.10.024}
\bibitem[Qin2010]{Q3} H-R. Qin, {\it Reflection theorems and the $p$-Sylow subgroup
of $K_2O_F$ for a number field $F$}, Journal of Pure and Applied Algebra
{\bf 214}(7) (2010), 1181--1192.
\url{https://doi.org/10.1016/j.jpaa.2009.10.005}
\bibitem[Rib2008] {Rib} K.A. Ribet, {\it Bernoulli numbers and ideal classes}, SMF,
Gazette \textbf{118} (2008), 42--49. \\
\url{https://smf.emath.fr/system/files/filepdf/gaz-118.pdf}
\bibitem[R{\O}2000]{RO} J. Rognes, P.A. {\O}stv\ae r, {\it Two-primary algebraic
${\rm K}$-theory of two-regular number fields}, Math. Z. {\bf 233} (2000), 251--263.\\
\url{https://doi.org/10.1007/PL00004795}
\bibitem[RWK2000]{RW} J. Rognes, C. Weibel, {\it Two-primary algebraic $K$-theory
of rings of integers in number fields. Appendix A by Manfred Kolster}, J. Amer. Math.
Soc. {\bf 13}(1) (2000), 1--54. \\
\url{https://doi.org/10.1090/S0894-0347-99-00317-3}
\bibitem[Se1978]{Se} J-P. Serre, {\it Sur le r\'esidu de la fonction z\^eta $p$-adique d'un corps
de nombres}, C.R. Acad. Sci. Paris 287, S\'erie I (1978), 183--188. \\
\url{https://doi.org/10.1007/978-3-642-39816-2_116}
\bibitem[Tate1970]{Ta1} J. Tate, {\it Symbols in arithmetic}, In: Congr\`es
International de Math\'e\-matiques 1970, Tome 1, 201--211. \\
\url{https://books.google.fr/books?id=AykSHAAACAAJ}
\bibitem[Tate1976]{Ta2} J. Tate, {\it Relations between $K_2$ and Galois cohomology},
Invent. Math. {\bf 36} (1976), 257--274.
\url{https://eudml.org/doc/142421}
\bibitem[YF2000]{YF} Q. Yue, K. Feng, {\it The $4$-rank of the tame kernel
versus the $4$-rank of the narrow class group in quadratic number fields},
Acta Arith. {\bf 96}(2) (2000), 155--165.
\url{https://doi.org/10.4064/aa96-2-4}
\bibitem[Yue2002]{Yue1} Q. Yue, {\it On tame kernel and class group in terms of quadratic
forms}, Jour. Number Theory {\bf 96}(2) (2002), 373--387. \\
\url{https://doi.org/10.1006/jnth.2002.2808}
\bibitem[Yue2005]{Yue2} Q. Yue, {\it The formula of 8-ranks of tame kernels},
J. Algebra {\bf 286}(1) (2005), 1--25.
\url{https://doi.org/10.1016/j.jalgebra.2004.10.030}
\bibitem[Wa1980]{W1} L.C. Washington, {\it Zeroes of $p$-adic $L$-functions},
S\'eminaire de th\'eorie des nombres, Paris, 1980-81 (S\'em. Delange--Pisot--Poitou),
Birkh\"au\-ser, Boston, 1982, 337--357.\\
\url{https://books.google.fr/books?id=pjoBwwEACAAJ}
\bibitem[Wa1981]{W2} L.C. Washington, {\it Zeroes of $p$-adic $L$-functions},
S\'eminaire de Th\'eorie des Nombres de Bordeaux (1980-1981), 1--4.\\
\url{http://www.jstor.org/stable/44166382}
\bibitem[Wa1995]{W3} L.C. Washington, {\it Siegel zeros for $2$-adic $L$-functions},
Number Theory, Halifax, NS (1994), CMS Conf. Proc., American Mathematical
Society, Providence, RI, 1995, 393--396. \\
\url{https://books.google.fr/books?id=gzRti1o1bXkC}
\bibitem[Wa1996]{W4} L.C. Washington, {\it A Family of Cubic Fields and Zeros
of $3$-adic $L$-Functions}, Journal of Number Theory {\bf 63} (1996), 408--417.\\
\url{https://doi.org/10.1006/jnth.1997.2096}
\bibitem[Wa1997]{W} L.C. Washington, {\it Introduction to cyclotomic fields},
Graduate Texts in Math. {\bf 83}, Springer enlarged second edition 1997.\\
\url{https://doi.org/10.1007/978-1-4612-1934-7}
\end{thebibliography}
\end{document}