Sign up ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

When we study sheaves of sets (on a space X or a site C) we are often interested in the stalks of the sheaf (at either a point $p:1\to X$ or a left exact, cover-preserving functor $a:C\to Sets$). I wonder how one generalizes this notion to stacks?

Since we can always pass from a stack to an equivalent sheaf, one option might be to say the stalk of a stack is (any category equivalent to) the stalk of the associated sheaf of categories. This seems a little roundabout, though? Is there a more intrinsic definition of these categories?

share|cite|improve this question
I presume by stack you mean a fibred category (satisfying the usual properties), and by a sheaf of categories you mean a presheaf of categories satisfying the appropriate coherence (weaker than that for a sheaf of sets). The main problem is perhaps what you consider to be the 2-category of stacks, and what you want to take for 1-arrows, in order to define a point and the pullback along the inclusion of that point. –  David Roberts Mar 31 '11 at 6:36
Related question: Is there a nice construction of an étale space associated to a stack? –  Martin Brandenburg Mar 31 '11 at 6:46
I'm confused by the question. Are you trying to understand "the stalk of a sheaf on a stack" or "the stalk of a stack"? It sounds like it's the latter, but then you should consider this question: what is the stalk of a space over $X$? After all, a space over $X$ corresponds to a sheaf over $X$ (in most reasonable categories of spaces). This strikes me as a pretty weird thing to think about. Is this something anybody would want to generalize to stacks? –  Anton Geraschenko Mar 31 '11 at 9:48
@David: maybe I'm confused, but I think even in the case of sheaves of modules over a scheme when you take the pullback along the inclusion of a point you don't get the stalk of the sheaf, but its fiber –  Qfwfq Mar 31 '11 at 19:21
@Martin: Yes. I explain this in this preprint: –  David Carchedi Apr 3 '11 at 20:50

1 Answer 1

Here is the categorical way to think about the stalk:

Let $X$ be space, and $x \in X$ a point. Regard $x$ as a map $$x:pt \to X$$.

Then $x$ induces a geometric morphism $$x_*:Sh(pt) \to Sh(X)$$ $$Sh(pt) \stackrel{}{\longleftarrow} Sh(X):x^*$$

(so $x^*$ is left-adjoint to $x_*$ and left-exact).

Now notice that $Sh(pt) \cong Set$. Under this identification, the stalk of $F \in Sh(X)$ at $x$ is the set $x^*\left(F\right)$.

This can be done in the $2$-categorical setting with stacks with no adjustment:

$x$ induces an adjoint pair of $2$-functors

$$x_*:St(pt) \to St(X)$$ $$St(pt) \stackrel{}{\longleftarrow} St(X):x^*$$

and under the identification $St(pt) \cong Gpd,$ the stalk of $\mathscr{X} \in St(X)$ at $x$ is the groupoid $x^*\left(\mathscr{X}\right)$.

A way of viewing this in a geometric way is by using the etale-realization of the stack $\mathscr{X}$, (explained in ) which is a topological stack $$L\left(\mathscr{X}\right)$$ equipped with a local homoemosphism to $X$. The stalk can then be viewed as the fiber of this map over $x$.

share|cite|improve this answer
My comment comes too late, but I have the impression that your construction actually gives the fiber, not the stalk, of whatever thing you are considering (sheaf, space, stack,...). –  Qfwfq Mar 18 at 17:21
The fiber of the etale-space $L(F) \to X$ of a sheaf $F$ over $X$ at a point $x$ IS precisely the stalk $F_x.$ These are one and the same. –  David Carchedi Mar 19 at 9:53
Oh I see what you mean. (On the other hand, if $F$ is a sheaf of modules on a $k$-scheme $X$ and $x:\mathrm{Spec}(k)\to X$ is a (closed) point, the $k$- vector space $x^{*} F$ is the fiber at $x$, not the stalk). –  Qfwfq Mar 19 at 12:35
Yes, but the question was about generalizing from stalks of sheaves of sets to stalks of stacks so, although I agree that what I said may not make sense for sheaves of more interesting objects than sets (like modules), it does answer the question asked. –  David Carchedi Mar 19 at 21:09

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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