Lemma 59.84.1. Let $\Lambda$ be a Noetherian ring, let $M$ be a finite $\Lambda$-module which is annihilated by an integer $n > 0$, let $k$ be an algebraically closed field, and let $X$ be a separated, finite type scheme of dimension $\leq 1$ over $k$. Then

1. $H^ q_{\acute{e}tale}(X, \underline{M})$ is a finite $\Lambda$-module if $n$ is prime to $\text{char}(k)$,

2. $H^ q_{\acute{e}tale}(X, \underline{M})$ is a finite $\Lambda$-module if $X$ is proper.

Proof. If $n = \ell n'$ for some prime number $\ell$, then we get a short exact sequence $0 \to M[\ell ] \to M \to M' \to 0$ of finite $\Lambda$-modules and $M'$ is annihilated by $n'$. This produces a corresponding short exact sequence of constant sheaves, which in turn gives rise to an exact sequence of cohomology modules

$H^ q_{\acute{e}tale}(X, \underline{M[n]}) \to H^ q_{\acute{e}tale}(X, \underline{M}) \to H^ q_{\acute{e}tale}(X, \underline{M'})$

Thus, if we can show the result in case $M$ is annihilated by a prime number, then by induction on $n$ we win.

Let $\ell$ be a prime number such that $\ell$ annihilates $M$. Then we can replace $\Lambda$ by the $\mathbf{F}_\ell$-algebra $\Lambda /\ell \Lambda$. Namely, the cohomology of $\mathcal{F}$ as a sheaf of $\Lambda$-modules is the same as the cohomology of $\mathcal{F}$ as a sheaf of $\Lambda /\ell \Lambda$-modules, for example by Cohomology on Sites, Lemma 21.12.4.

Assume $\ell$ be a prime number such that $\ell$ annihilates $M$ and $\Lambda$. Let us reduce to the case where $M$ is a finite free $\Lambda$-module. Namely, choose a short exact sequence

$0 \to N \to \Lambda ^{\oplus m} \to M \to 0$

This determines an exact sequence

$H^ q_{\acute{e}tale}(X, \underline{\Lambda ^{\oplus m}}) \to H^ q_{\acute{e}tale}(X, \underline{M}) \to H^{q + 1}_{\acute{e}tale}(X, \underline{N})$

By descending induction on $q$ we get the result for $M$ if we know the result for $\Lambda ^{\oplus m}$. Here we use that we know that our cohomology groups vanish in degrees $> 2$ by Theorem 59.83.10.

Let $\ell$ be a prime number and assume that $\ell$ annihilates $\Lambda$. It remains to show that the cohomology groups $H^ q_{\acute{e}tale}(X, \underline{\Lambda })$ are finite $\Lambda$-modules. We will use a trick to show this; the “correct” argument uses a coefficient theorem which we will show later. Choose a basis $\Lambda = \bigoplus _{i \in I} \mathbf{F}_\ell e_ i$ such that $e_0 = 1$ for some $0 \in I$. The choice of this basis determines an isomorphism

$\underline{\Lambda } = \bigoplus \underline{\mathbf{F}_\ell } e_ i$

of sheaves on $X_{\acute{e}tale}$. Thus we see that

$H^ q_{\acute{e}tale}(X, \underline{\Lambda }) = H^ q_{\acute{e}tale}(X, \bigoplus \underline{\mathbf{F}_\ell } e_ i) = \bigoplus H^ q_{\acute{e}tale}(X, \underline{\mathbf{F}_\ell })e_ i$

since taking cohomology over $X$ commutes with direct sums by Theorem 59.51.3 (or Lemma 59.51.4 or Lemma 59.52.2). Since we already know that $H^ q_{\acute{e}tale}(X, \underline{\mathbf{F}_\ell })$ is a finite dimensional $\mathbf{F}_\ell$-vector space (by Theorem 59.83.10), we see that $H^ q_{\acute{e}tale}(X, \underline{\Lambda })$ is free over $\Lambda$ of the same rank. Namely, given a basis $\xi _1, \ldots , \xi _ m$ of $H^ q_{\acute{e}tale}(X, \underline{\mathbf{F}_\ell })$ we see that $\xi _1 e_0, \ldots , \xi _ m e_0$ form a $\Lambda$-basis for $H^ q_{\acute{e}tale}(X, \underline{\Lambda })$. $\square$

## Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.

In order to prevent bots from posting comments, we would like you to prove that you are human. You can do this by filling in the name of the current tag in the following input field. As a reminder, this is tag 0GJC. Beware of the difference between the letter 'O' and the digit '0'.