chevalley weil for de Rham

This commit is contained in:
jgarnek 2024-12-10 18:48:14 +01:00
parent 0cdda6cee7
commit d9c360a8f5
2 changed files with 50 additions and 28 deletions

Binary file not shown.

View File

@ -231,20 +231,37 @@ $k[C]$-module. It turns out that the map
is a bijection (cf. \cite[p. 35--37, 42 -- 43]{Alperin_local_rep}). We write
$\mc V(M, i)$ for the $k[G]$-module corresponding to a pair $(M, i) \in \Indec(k[C]) \times \{ 1, \ldots, p^n \}$.
Finally, we recall the classical Chevalley-Weil formula. Keep the above notation and assume that $p \nmid \# G$. For any $e \in \NN$, denote by $\chi_e$ the primitive character of a cyclic group of order $e$.
Then:
Finally, we recall the classical Chevalley-Weil formula. For any $e \in \NN$, denote by $\chi_e$ the primitive character of a cyclic group of order $e$.
%
\begin{equation}
H^0(X, \Omega_X) \cong \bigoplus_{M \in \Indec(k[G])} M^{\oplus a_M},
\begin{Proposition} \label{prop:chevalley_weil}
Keep the above notation and assume that $p \nmid \# G$. Then:
%
\begin{equation} \label{eqn:cw}
H^0(X, \Omega_X) \cong \bigoplus_{W \in \Indec(k[G])} W^{\oplus a_W},
\end{equation}
%
where:
%
\begin{align*}
a_M := - \dim_k M + \sum_{Q \in Y(k)} \sum_{i = 0}^{e_{X/Y, Q} - 1} \frac{e_{X/Y, Q} - i}{e_{X/Y, Q}} \cdot N_{P, i}(M),
a_W := (g_Y - 1) \cdot \dim_k W + \sum_{Q \in Y(k)} \sum_{i = 1}^{e_{X/Y, Q} - 1} \frac{e_{X/Y, Q} - i}{e_{X/Y, Q}} \cdot N_{Q, i}(W) + \llbracket W \cong k \rrbracket,
\end{align*}
%
and $N_{P, i}(M)$ is the multiplicity of the character $\chi_{e_Q}^i$ in the $k[G_Q]$-module $M \otimes_{k[G_Q]} \theta_{X/Y, Q}$.
and $N_{Q, i}(W)$ is the multiplicity of the character $\chi_{e_Q}^i$ in the $k[G_Q]$-module $W \otimes_{k[G_Q]} \theta_{X/Y, Q}$.
\end{Proposition}
%
\begin{Corollary}
Keep the notation of Proposition~\ref{prop:chevalley_weil}. Then:
\begin{equation} \label{eqn:cw_dR}
H^1_{dR}(X) \cong k[G]^{\oplus 2g_X - 2} \oplus k^{\oplus 2}.
\end{equation}
%
where:
%
\begin{align*}
a_W^{dR} := 2 (g_Y - 1) \cdot \dim_k W + \sum_{Q \in Y(k)} (e_{X/Y, Q} - 1) \cdot \dim_k W + 2 \cdot \llbracket W \cong k \rrbracket,
\end{align*}
\end{Corollary}
\section{Cyclic covers}
%
@ -263,9 +280,9 @@ and $N_{P, i}(M)$ is the multiplicity of the character $\chi_{e_Q}^i$ in the $k[
\end{Theorem}
%
\begin{Remark}
Note that for $g_Y = 0$ the first exponent is negative. However, since $m_{X/Y} = n$ (as $\PP^1$ doesn't have any \'{e}tale covers), the first two summands in~\eqref{eqn:HdR_formula} cancel out. Thus also in this case the module~\eqref{eqn:HdR_formula} is well-defined.
Note that this formula is well-defined for $g_Y = 0$, even though the first exponent is negative. Indeed, since $m_{X/Y} = n$ (as $\PP^1$ doesn't have any \'{e}tale covers), the first two summands in~\eqref{eqn:HdR_formula} cancel out.
\end{Remark}
%
Write $H := \langle \sigma \rangle \cong \ZZ/p^n$.
For any $k[H]$-module $M$ denote:
%
@ -491,8 +508,8 @@ shows that $m_{\sigma - 1}$ is well-defined and injective.
\begin{proof}[Proof of Theorem~\ref{thm:cyclic_de_rham}]
We use the following notation: $H' := \langle \sigma^p \rangle \cong \ZZ/p^{n-1}$,
$H'' := H/\langle \sigma^{p^{n-1}} \rangle \cong \ZZ/p^{n-1}$, $Y' := X/H'$, $X'' := X/\langle \sigma^{p^{n-1}} \rangle$. Note that $H''$ naturally acts on $X''$.
{\color{red}
$H'' := H/\langle \sigma^{p^{n-1}} \rangle \cong \ZZ/p^{n-1}$, $Y' := X/H'$ and $X'' := X/\langle \sigma^{p^{n-1}} \rangle$ {\color{red}, see the diagram below.}
\[
\xymatrix{
& X \ar[rd]^{\langle \sigma^{p^{n-1}} \rangle} \ar[ld]_{\langle \sigma^p \rangle =H'} \ar[dd]^{\pi}& \\
@ -500,7 +517,7 @@ shows that $m_{\sigma - 1}$ is well-defined and injective.
& Y &
}
\]
}
{\color{red} Note that $H''$ naturally acts on $X''$ and $X''/H'' \cong Y$.}
Let also $\mc M := H^1_{dR}(X)$ and write $\mc M_0$ for the module~\eqref{eqn:HdR_formula}.
We consider now two cases. If the cover $X \to Y$ is \'{e}tale, then by induction assumption, since $2(g_{Y'} - 1) = p \cdot 2 \cdot (g_Y - 1)$:
%
@ -740,8 +757,9 @@ Let $X$ be a curve with an action of $G$ and write $Y := X/H$. For any $k[C]$-mo
\end{proof}
\begin{proof}[Proof of Main Theorem]
As explained at the beginning of this section, it suffices to show this in the case when $G = H \rtimes_{\chi} C = \langle \sigma \rangle \rtimes_{\chi} \langle \rho \rangle \cong \ZZ/p^n \rtimes_{\chi} \ZZ/c$ and $k = \ol k$ by Lemma~\ref{lem:reductions}.
We prove this by induction on~$n$. If $n = 0$, then it follows by Chevalley--Weil theorem.
As explained at the beginning of this section, it suffices to show this in the case when $G = H \rtimes_{\chi} C = \langle \sigma \rangle \rtimes_{\chi} \langle \rho \rangle \cong \ZZ/p^n \rtimes_{\chi} \ZZ/c$ and $k = \ol k$ by Lemma~\ref{lem:reductions}. {\color{red} Let $Y := X/H$. Similarly as in the proof of Theorem~\ref{thm:cyclic_de_rham}, we write $H' := \langle \sigma^p \rangle \cong \ZZ/p^{n-1}$,
$H'' := H/\langle \sigma^{p^{n-1}} \rangle \cong \ZZ/p^{n-1}$, $Y' := X/H'$ and $X'' := X/\langle \sigma^{p^{n-1}} \rangle$. Observe that the ramification datum of the covers $X'' \to Y$ and $X \to Y'$ depends only on the ramification data of $X \to Y$.
We prove the result by induction on~$n$.} If $n = 0$, then it follows by Chevalley--Weil theorem.
Consider now two cases. Firstly, we assume that $X \to Y$ is \'{e}tale. Then by Lemma~\ref{lem:G_invariants_\'{e}tale} and \cite[Corollary~2.4]{Garnek_equivariant} we have $\dim_k H^1_{dR}(X)^H = 2g_Y = \dim_k H^0(X, \Omega_X)^H + \dim_k H^1(X, \mc O_X)^H$. Therefore the Hodge--de Rham exact sequence splits by \cite[Lemma~5.3]{Garnek_equivariant} and
%
\begin{align*}
@ -782,28 +800,32 @@ Let $X$ be a curve with an action of $G$ and write $Y := X/H$. For any $k[C]$-mo
%
\section{Examples}
%
Assume that $G$ is a group with a normal $p$-Sylow subgroup $H$ of order $p$. Let $C := G/H$. Then $G = H \rtimes_{\chi} C$
Assume that $G$ is a group with a normal $p$-Sylow subgroup $H$ of order~$p$. Let $C := G/H$. Then $G = H \rtimes_{\chi} C$
for a homomorphism $\chi : C \to \FF_p^{\times}$.
%
\begin{Proposition}
Keep the above notation. If $G$ acts on a curve $X$ and the cover $X \to X/H$ is not \'{e}tale, then:
Keep the above notation. {\color{red} Assume that $k$ is algebraically closed.} If $G$ acts on a curve $X$ and the cover $X \to X/H$ is not \'{e}tale, then:
%
\[
H^1_{dR}(X) \cong \bigoplus_{M \in \Indec(C)} \mc V(M, p-1)^{\oplus b_M} \oplus \mc V(M, p)^{\oplus c_M},
H^1_{dR}(X) \cong \bigoplus_{W \in \Indec(C)} \mc V(W, p)^{\oplus a_W'} \oplus \mc V(W, p-1)^{\oplus b_W},
\]
%
where
where for any $W \in \Indec(k[C])$ the number $a_W$ is as in the equality~\eqref{eqn:cw} for the action of $C$ on $X$,
$a_W'$ is as in the equality~\eqref{eqn:cw} for the action of $C$ on $Y := X/H$ and
%
\begin{align*}
b_M &:= (1 - \frac 1p) \cdot \dim_k M + \sum_{Q \in (X/G)(k)} \sum_{i < e_{X/Y, Q}??}
b_W &:= \frac 1p \left( (p-1) \cdot a_W - \sum_{i = 1}^{p-2} a_{W \otimes \chi^i} \right) - a'_{W \otimes \chi}.
\end{align*}
%
$b_M := \ldots$, $c_M := \ldots$.
\end{Proposition}
\begin{proof}
???
Theorem~\ref{thm:cyclic_de_rham} easily implies that
%
\[
H^1_{dR}(X) \cong \bigoplus_{W \in \Indec(C)} \mc V(W, p)^{\oplus A_W} \oplus \mc V(W, p-1)^{\oplus B_W}
\]
%
for some $A_W, B_W \in \ZZ$. ??
\end{proof}
\noindent Let $p > 2$. Consider the Mumford curve