6
$\begingroup$

Let $X$ be a smooth, projective and connected curve over an algebraically closed field, and let $\eta \rightarrow X$ be its generic point (we also call the inclusion as $\eta$). I want to understand the proof of the computation of the étale cohomology of $\Bbb G_{m, \eta}$ on $X$. Reading SGA 4.5, by P. Deligne, I got stacked at the lemma 3.3 of the section "Cohomolgie des courbes" of . It says the following:

$H^q(X_{ét}, \eta_* \Bbb G_{m, \eta})=0$ for $q >0$.

In the proof, they consider the exact sequence arising from the spectral sequence given by the functors $\eta_*: Ab(\eta_{ét}) \rightarrow Ab(X_{ét})$ and $\Gamma(X_{ét}, -): Ab(X_{ét}) \rightarrow Ab$ (I will denote this functor as $\Gamma$), where Ab(-) are the sheaves of abelian groups. More precisely, we obtain the exact sequence $$ 0 \rightarrow R^1\Gamma(\eta_* \Bbb G_{m, \eta}) \rightarrow R^1(\Gamma \circ \eta_*)(\Bbb G_{m, \eta}) \rightarrow \Gamma \circ R^1(\eta_* \Bbb G_{m, \eta}) \rightarrow \cdots.$$

Since the $R^q(\eta_* \Bbb G_{m, \eta})$ are zero, we get isomorphisms $H^q(X_{ét}, \eta_* \Bbb G_{m, \eta}) \cong R^q(\Gamma \circ \eta_*)(\Bbb G_{m, \eta})$.

So the question is, why $R^q(\Gamma \circ \eta_{*}) (\Bbb G_{m, \eta}) = H^q(\eta_{ét}, \Bbb G_{m, \eta})$? Once this equality is stablished, a previous result concludes the proof.

$\endgroup$
4
$\begingroup$

While typing the question, a friend of mine solved it, so I just share his explanation, this might be helpful for someone.

We do this by steps using the same thing: for any $f:X \rightarrow Y$ we have that $\Gamma(X_{ét},-) = \Gamma(Y_{ét}, f_*-)$. Let $\pi: X \rightarrow Spec(k)$ be the structure morphism. In the formulae, we denote $\Gamma_k := \Gamma((Spec k)_{ét}, -)$ and $\Gamma := \Gamma(X_{\acute{e}t},-)$.

Step 1: Here, $\Gamma (X_{ét}, -) = \Gamma ((Spec \: k)_{ét}, \pi_*-)$ imples $$ R^q(\Gamma \circ \eta_*)(\Bbb G_{m, \eta}) = R^q(\Gamma_k \circ \pi_* \circ \eta_*)(\mathbb{G}_{m, \eta}).$$ Step 2: Here, $ \Gamma(\eta_{ét}, -) = \Gamma (X_{ét}, \eta_*-)$ implies $$ R^q(\Gamma_k \circ (\pi \circ \eta)_* \Bbb G_{m, \eta}) = R^q\Gamma (\eta_{ét}, \Bbb G_{m, \eta}).$$

and we are done, since the last term is by definition $H^q(\eta_{\acute{e}t}, \Bbb G_{m, \eta})$. Note that in the rhs, $\Gamma$ is not the same as in the previous equality, sorry for the notation!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.