The Stacks project

Proposition 100.7.4. Summary of results on locally quasi-coherent modules having the flat base change property.

  1. Let $\mathcal{X}$ be an algebraic stack. If $\mathcal{F}$ is an object of $\textit{Mod}(\mathcal{X}_{\acute{e}tale}, \mathcal{O}_\mathcal {X})$ which is locally quasi-coherent and has the flat base change property, then $\mathcal{F}$ is a sheaf for the fppf topology, i.e., it is an object of $\textit{Mod}(\mathcal{O}_\mathcal {X})$.

  2. The category of modules which are locally quasi-coherent and have the flat base change property is a weak Serre subcategory $\mathcal{M}_\mathcal {X}$ of both $\textit{Mod}(\mathcal{O}_\mathcal {X})$ and $\textit{Mod}(\mathcal{X}_{\acute{e}tale}, \mathcal{O}_\mathcal {X})$.

  3. Pullback $f^*$ along any morphism of algebraic stacks $f : \mathcal{X} \to \mathcal{Y}$ induces a functor $f^* : \mathcal{M}_\mathcal {Y} \to \mathcal{M}_\mathcal {X}$.

  4. If $f : \mathcal{X} \to \mathcal{Y}$ is a quasi-compact and quasi-separated morphism of algebraic stacks and $\mathcal{F}$ is an object of $\mathcal{M}_\mathcal {X}$, then

    1. the derived direct image $Rf_*\mathcal{F}$ and the higher direct images $R^ if_*\mathcal{F}$ can be computed in either the étale or the fppf topology with the same result, and

    2. each $R^ if_*\mathcal{F}$ is an object of $\mathcal{M}_\mathcal {Y}$.

  5. The category $\mathcal{M}_\mathcal {X}$ has colimits and they agree with colimits in $\textit{Mod}(\mathcal{X}_{\acute{e}tale}, \mathcal{O}_\mathcal {X})$ as well as in $\textit{Mod}(\mathcal{O}_\mathcal {X})$.

Proof. Part (1) is Sheaves on Stacks, Lemma 93.22.1.

Part (2) for the embedding $\mathcal{M}_\mathcal {X} \subset \textit{Mod}(\mathcal{X}_{\acute{e}tale}, \mathcal{O}_\mathcal {X})$ we have seen in the proof of Lemma 100.7.3. Let us prove (2) for the embedding $\mathcal{M}_\mathcal {X} \subset \textit{Mod}(\mathcal{O}_\mathcal {X})$. Let $\varphi : \mathcal{F} \to \mathcal{G}$ be a morphism between objects of $\mathcal{M}_\mathcal {X}$. Since $\mathop{\mathrm{Ker}}(\varphi )$ is the same whether computed in the étale or the fppf topology, we see that $\mathop{\mathrm{Ker}}(\varphi )$ is in $\mathcal{M}_\mathcal {X}$ by the étale case. On the other hand, the cokernel computed in the fppf topology is the fppf sheafification of the cokernel computed in the étale topology. However, this étale cokernel is in $\mathcal{M}_\mathcal {X}$ hence an fppf sheaf by (1) and we see that the cokernel is in $\mathcal{M}_\mathcal {X}$. Finally, suppose that

\[ 0 \to \mathcal{F}_1 \to \mathcal{F}_2 \to \mathcal{F}_3 \to 0 \]

is an exact sequence in $\textit{Mod}(\mathcal{O}_\mathcal {X})$ (i.e., using the fppf topology) with $\mathcal{F}_1$, $\mathcal{F}_2$ in $\mathcal{M}_\mathcal {X}$. In order to show that $\mathcal{F}_2$ is an object of $\mathcal{M}_\mathcal {X}$ it suffices to show that the sequence is also exact in the étale topology. To do this it suffices to show that any element of $H^1_{fppf}(x, \mathcal{F}_1)$ becomes zero on the members of an étale covering of $x$ (for any object $x$ of $\mathcal{X}$). This is true because $H^1_{fppf}(x, \mathcal{F}_1) = H^1_{\acute{e}tale}(x, \mathcal{F}_1)$ by Sheaves on Stacks, Lemma 93.22.2 and because of locality of cohomology, see Cohomology on Sites, Lemma 21.7.3. This proves (2).

Part (3) follows from Lemma 100.7.2 and Sheaves on Stacks, Lemma 93.11.7.

Part (4)(b) for $R^ if_*\mathcal{F}$ computed in the étale cohomology follows from Lemma 100.7.3. Whereupon part (4)(a) follows from Sheaves on Stacks, Lemma 93.22.2 combined with (1) above.

Part (5) for the étale topology follows from Sheaves on Stacks, Lemma 93.11.8 and Lemma 100.7.2. The fppf version then follows as the colimit in the étale topology is already an fppf sheaf by part (1). $\square$


Comments (2)

Comment #3800 by on

In the proof of (what is now) part (4)(b), you want to write "in the étale topology".

Comment #3918 by on

@#3800: Do you mean that in (4)(b) we are computing the pushforward in the \'etale topology? But by part (4)(a) this is the same as for the fppf topology which is the "default" topology, so I think no confusion can arise from the statement of the lemma as given. OK?


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 0771. Beware of the difference between the letter 'O' and the digit '0'.