8

When I am studying stacky stuffs, I am always confused by the notion of étale abelian sheaves on $X$, because conceivably there might be three different meanings of that:

  1. Take the global étale site (the category of scheme with topology defined by étale coverings) and consider an abelian sheaf $\mathcal F$ on the global site, together with a map $\mathcal F\rightarrow h_X$ to the sheaf represented by $X$.

  2. Take the big étale site on $X$ (the category of schemes over X with topology defined by étale covering) and consider an abelian sheaf on this site.

  3. Take the small étale site on $X$ (the category of schemes étale over X with topology defined by étale covering) and consider an abelian sheaf on this site.

From the context of the books that I am reading, (1) doesn't seem like the right thing to look at. But is (2) or (3) the right thing to look at? Are they the same?

Zhen Lin
  • 97,105
waikit
  • 290
  • Actually, (1) and (2) are the same (modulo a subtlety about "abelian sheaf"). But (3) is different in general. – Zhen Lin Nov 07 '14 at 20:09
  • @ZhenLin Can you expand a bit on how (1) and (2) are the same? Thanks. – waikit Nov 07 '14 at 23:01
  • For any site $\mathcal{C}$ and any object $X$ in $\mathcal{C}$, the topos $\mathbf{Sh}(\mathcal{C}){/ h_X}$ is equivalent to the topos $\mathbf{Sh}(\mathcal{C}{/X})$. Ergo the category of abelian group objects in either topos is equivalent. – Zhen Lin Nov 07 '14 at 23:06
  • I suppose the functor is $\textbf{Sh}(\mathcal C){h_X}\rightarrow \textbf{Sh}(\mathcal C{/X})$ by restriction?

    But then if I have the sheaf $\mathcal O_X \in \mathcal C_{/X}$ where $\mathcal C$ is the global étale site, (i.e., The sheaf represented by $\mathbb A^1_X$ over $X$), then how do I realize that as (1)?

     – waikit Nov 07 '14 at 23:10
  • The functor is not so easy to describe. See the very last part of this answer. – Zhen Lin Nov 07 '14 at 23:18

1 Answers1

4

I don't think 1) is correct. In the notes that I have on stacks (some notes by Martin Olsson) we normally stick to 2) sheaves on the big etale site, I guess because its slightly more general and because:

a) you can prove that $X$-schemes give you sheaves on the big Etale site (or more generally on the fppf site).

b) Algebraic spaces are defined as sheaves on the big etale site satisfying some additional properties.

But I think that your definition 3) would be equally acceptable in other contexts.

Daniel Mckenzie
  • 1,004