6
$\begingroup$

Let $\{X_1 \rightrightarrows X_0\}$ be a smooth groupoid object in the category of affine schemes ($X_0 \to X_1$, $X_1 \to X_1$ and $X_1 \times_{X_0} X_1 \to X_1$ also belong to the datum). Equivalently, we have a smooth commutative Hopf algebroid. Let $X$ be the associated algebraic stack. Thus, we have a presentation $X_0 \twoheadrightarrow X$ (which is smooth, surjective and affine) with $X_0 \times_X X_0 = X_1$, and $X$ is geometric. Every geometric stack arises this way.

Question. If $Y$ is a scheme, how can we describe $X(Y)$ explicitly in terms of $Y$ and the $X_i$?

This should be somewhere in the literature? For example, if $X_0 = \mathrm{Spec}(\mathbb{Z})$, then $X_1$ is a group scheme, $X$ is its classifying stack, so that $X(Y)$ consists of $X_1$-torsors on $Y$, right?

In general, I expect that the answer will be some kind of "torsors under the groupoid". A morphism $Y \to X$ may be pulled back to $Y_0 \to X_0$, where $Y_0 \to Y$ is smooth, surjective and affine. Conversely, if $Y_0 \to Y$ is smooth, surjective and affine, and $Y_0 \to X_0$ is a morphism, then an extension to $Y \to X$ corresponds to a descent datum of $Y_0 \to X_0 \to X$ with respect to the fpqc cover $Y_0 \to Y$. But this still involves $X$. How can we get rid of $X$ in the description of morphisms $Y \to X$?

$\endgroup$

2 Answers 2

3
$\begingroup$

Niels already links to references that answer the question, but I'll briefly summarize them.

If $X$ is a stack presented by a groupoid $X_1 \rightrightarrows X_0$ then a map from a scheme $S$ into $X$ induces a groupoid presentation $S \mathop{\times}_X X_1 \rightrightarrows S \mathop{\times}_X X_0$ of $S$. Conversely, suppose that we have a groupoid presentation $S_1 \rightrightarrows S_0$ of $S$ and compatible maps $S_i \rightarrow X_i$ (such that the two maps $S_1 \rightarrow S_0$ are induced by pullback from the two maps $X_1 \rightarrow X_0$) then there is an induced map $S \rightarrow X$.

The associated stack of the groupoid $X_\bullet$ is therefore the fibered category $X$ for which $X(S)$ is the category of groupoid presentations of $S$ with a map to $X_\bullet$ satisfying the above parenthetical condition.

$\endgroup$
4
  • $\begingroup$ Thanks. Going through the construction of stackification in the book by Laumon and Moret-Bailly, it follows that $X(S)$ consists of covers $S' \to S$ and a morphism of groupoid schemes $(S' \times_S S' \rightrightarrows S') \to (X_1 \rightrightarrows X_0)$. No cartesian condition. What am I missing? Also, is there some more global description? If $X_0$ is the base scheme, we get the local description of $X_1$-torsors via cocycles $S' \times_S S' \to X_1$. But how do we get the global description as special morphisms over $S$ (with an action and local triviality)? $\endgroup$ Oct 17, 2013 at 12:54
  • $\begingroup$ The reason for the cartesian condition is to make the morphisms come out right. If you just look at maps from a groupoid presentation to $X_\bullet$ you can get two representations of the same map to $X$ coming from different presentations of $S$, without an isomorphism between them. You can get an example of this with $X_\bullet = [ (\text{point}) \rightrightarrows (\text{point}) ]$. $\endgroup$ Oct 17, 2013 at 23:23
  • $\begingroup$ Does this mean that the construction of the associated stack in LMB is incomplete? Sorry, I'm confused. $\endgroup$ Oct 18, 2013 at 13:16
  • $\begingroup$ The definition in LMB also works, but you have to take more care in the definition of morphisms in the stack. (What is a morphism between two maps $S \rightarrow X$ that are given by different presentations of $S$?) $\endgroup$ Oct 18, 2013 at 20:16
3
$\begingroup$

This is called "Stackification via torsors" in the book by Kai Behrend, Brian Conrad, Dan Edidin, William Fulton, Barbara Fantechi, Lothar Göttsche and Andrew Kresch. See Andrew Kresch's homepage:

http://www.math.uzh.ch/index.php?pr_vo_det&key1=1287&key2=580&no_cache=1

more specifically chapter 4, \S 4.

http://www.math.uzh.ch/index.php?file&key1=5161

For sure, this is done in the stacks project

http://stacks.math.columbia.edu/

but I couldn't find the exact reference (probably the keyword is: quotient stack).

For a more historical point of view, see also the related question Algebraic stacks as (etale) groupoid alg.spaces/schemes .

$\endgroup$
1
  • $\begingroup$ Thank you! It's a pity that the book project won't be continued, as far as I know. $\endgroup$ Oct 17, 2013 at 12:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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