Skip to content

Commit

Permalink
add more LaTeX for Fujisaki
Browse files Browse the repository at this point in the history
  • Loading branch information
kbuzzard committed Dec 2, 2024
1 parent ff78769 commit 1930772
Showing 1 changed file with 94 additions and 3 deletions.
97 changes: 94 additions & 3 deletions blueprint/src/chapter/QuaternionAlgebraProject.tex
Original file line number Diff line number Diff line change
Expand Up @@ -194,9 +194,100 @@ \section{Statement of the main result of the miniproject}
and because $f_1(g_i)=f_2(g_i)$ by assumption, we deduce $f_1(g)=f_2(g)$ as required.
\end{proof}

It thus remains to prove Fujisaki's lemma. Note that we will be assuming the results
from the adele mini-project~\ref{Adele_miniproject}.
It thus remains to prove Fujisaki's lemma, which is what we do in the rest of this
miniproject. Note that we will be assuming the results
from the adele mini-project, chapter~\ref{Adele_miniproject} of this blueprint.

\section{The ``Haar character'' on a ring.}

Say $A$ is a topological ring, which is locally compact (i.e. there's a compact
subset of $A$ which contains a nonempty open subset of $A$). Then the additive
abelian group $(A,+)$ has a Haar measure. If $u\in A^\times$ is a unit of $A$,
then multiplication by $u$ is an additive isomorphism $(A,+)\to(A,+)$, and thus
pulls back a Haar measure $\mu$ to a positive real scalar multiple of $\mu$. This
scalar $h(u)\in\R_{>0}$ is independent of the choice of Haar measure $\mu$,
and gives us a canonical group homomorphism $h:A^\times\to\R_{>0}$, which is
presumably continuous although I didn't check this and I don't think we need it.
This initially looks like the ``modular character'' of a group but it isn't.
The modular character of a locally compact topological group is a character on
the group which measures the difference between the left and right Haar measure.
Here the Haar measure is additive, and addition is commutative, so there is
no difference between left and right Haar measure. To put it another way,
the crucial axiom which makes this construction work is distributivity $u(a+b)=ua+ub$,
and the crucial axiom which makes the modular character work is associativity $g(hk)=(gh)k$.

We need a name for this character. Right now I propose calling it the Haar character,
and in mathlib perhaps {\tt distribHaarCharacter} is an appropriate name.

\subsection{Examples}

1) If $A=\R$ then $h(r)=|r|$.

2) If $A=\mathbb{C}$ then $h(r)=|r|^2$ (for example, multiplication by 2 changes the area
of a shape in the complexes by a factor of 4).

3) If $A=\Q_p$ then $h(r)=|r|_p$, the usual $p$-adic norm with $|p|_p=p^{-1}$.
This is because $p\Z_p$ has index $p$ in $\Z_p$, so $\Z_p$ is the disjoint union
of $p$ additive cosets of $p\Z_p$ and each of these has the same additive Haar measure.
As a consequence, the measure of $p\Z_p$ must be $1/p$ times the measure of $\Z_p$.

4) More generally if $A$ is a nonarchimedean local field with integers $R$,
maximal ideal $m=(\varpi)$ and finite residue field of size $q$, then $h(\varpi)=1/q$
for the same reason.

Note that these examples give us a ``canonical'' norm on any local field (set $h(0)=0$).

5) If $A=B\times C$ is a product of two locally compact topological rings, then $A$
is locally compact, and $h_A=h_B\times h_C$ in the sense that $h_A(b,c)=h_B(b)h_C(c)$.

6) Now say $A$ is a restricted product of topological rings $A_i$ with respect to compact
open subrings $R_i$ (note: the existence of such an $R_i$ implies that $A_i$ is
locally compact). If $u_i\in R_i^\times$ then multiplication by $u_i$ sends $R_i$
bijectively to itself, so $h_i(u_i)=1$, with $h_i$ denoting the Haar character of $A_i$.
A general element $u\in A^\times$ with inverse $v$
has local components $u_i$ satisfying $u_iv_i=1$, and $u_i,v_i\in R_i$ for all but
finitely many $i$, meaning $u_i\in R_i^\times$ for all but finitely many $i$, and thus
$h_i(u_i)=1$ for these $i$. Thus the product $\prod_i h_i(u_i)$ is a finite product
(in the sense that all but finitely many terms in it are 1), and one can check
that this is $h(u)$.

7) If $A$ is the adeles of a number field, then $A$ is the product of the finite adeles
and the infinite adeles. Example (6) applies to the finite adeles, and using example (5)
on the infinite adeles we deduce a formula $h(a)=\prod_v h_v(a_v)$ relating the global
Haar character to the product of local Haar chacters.

8) Let $K$ be a number field, and let $B$ be a finite-dimensional $K$-algebra containing
an $\mathcal{O}_K$-order $R\subseteq B$, that is, a module-finite $\mathcal{O}_K$-subalgebra
$R$ of $B$ such that the canonical map $R\otimes_{\mathcal{O}_K}K\to B$ is an isomorphism.
Set $A=B\otimes_K\mathbb{A}_K$. Then $A$ is a product
$(B\otimes_K\mathbb{A}_K^\infty)\times(B\otimes_KK_\infty)$, and
$B\otimes_K\mathbb{A}_K^\infty=\prod'_vB\otimes_KK_v$, with the restricted product over
the finite places of $v$ being with respect to the compact open subrings
$R\otimes_{\mathcal{O}_K}\mathcal{O}_{K_v}$. All orders are commensurable, so the choice
of order $R$ here doesn't really matter: it only changes the subring at finitely many places.
So (5) and (6) apply again, giving us $h(b)=\prod_v h_v(b_v)$, a finite product.

\section{The product formula}

Let $K$ be a number field, and let $B$ be a finite-dimensional $K$-algebra (not necessarily
commutative), containing an $\mathcal{O}_K$-order. Set $A=B\otimes_K\mathcal{A}_K$.
Then $A$ is a locally compact topological ring (give it the module topology coming from
$\mathcal{A}_K$, which is just the product topology if one chooses a $K$-basis for $B$).

What I believe to be true is

\begin{theorem} Write $h_A$ for the Haar character of $A$.
If $b\in B^\times$, regarded as an element of $A^\times$, then $h_A(b)=1.$
\end{theorem}
\begin{proof} WLOG $K=\Q$, as $B\otimes_K\mathcal{A}_K=B\otimes_{\Q}\mathcal{A}_{\Q}$.
There must now be some way to reduce it to the case $B=\Q$. And then it's just
the product formula.
\end{proof}

This proof needs some work.

\section{Proof of Fujisaki's lemma}

See Main Theorem 27.6.14 of Voight. TODO: make these into issues.
TODO.

See Main Theorem 27.6.14 of Voight.

0 comments on commit 1930772

Please sign in to comment.