3
$\begingroup$

I am looking for some examples of gerbes over stacks (as defined in Understanding definition of gerbe over a stack) that comes from manifolds.

Let $M$ be a manifold then $\underline{M}$ is a stack associated to $M$ (I can give details if any one needs).

We know when to call a map of stacks $\mathcal{D}\rightarrow \mathcal{C}$ a gerbe over stack. Suppose $\mathcal{D}$ is coming from a manifold $M$ and $\mathcal{C}$ is coming from a manifold $N$ i.e., we have $\underline{M}\rightarrow \underline{N}$.

Given a map of stacks $F:\underline{M}\rightarrow \underline{N}$, we have unique map $F(id):M\rightarrow N$ associated to $F$.

What can we say about this map? Any comments are welcome.

First condition says that $F:\underline{M}\rightarrow \underline{N}$ is an epimorphism. Suppose $f:M\rightarrow N$ is a surjective submersion, then it follows that $F:\underline{M}\rightarrow \underline{N}$ is an epimorphism.

Second condition says that the diagonal map $\underline{M}\rightarrow \underline{M}\times_{\underline{N}}\underline{M}$ is an epimorphism. Suppose diagonal map $M\rightarrow M\times_N M$ is a surjective submersion, then $\underline{M}\rightarrow \underline{M}\times_{\underline{N}}\underline{M}$ is an epimorphism.

But then, $M\rightarrow M\times_N M$ is surjective implies that $f$ is injective.

An injective, surjectve submersion is a diffeomorphism. So, I landed up with a diffeomorphism conidtion on $M\rightarrow N$.

Do we have other non trivial examples?

Edit: Ycor has added algebraic geometry tag. I am also ok with algebraic geometry examples. I do not know much (not even the precise definition in that set up). Smooth manifolds are replaced by schemes (I guess). Given a scheme, there is an associated stack (algebraic stack??). Gerbe over stack in case of algebraic geometry of the form $\underline{X}\rightarrow \underline{Y}$ should correspond to some morphism of schemes $X\rightarrow Y$. It will be good if you can define gerbe over stack in algebraic geometry set up (loosely atleast) and the give examples.

$\endgroup$
  • $\begingroup$ @YCor: Please see the edit. $\endgroup$ – Praphulla Koushik Nov 28 '18 at 15:06
  • 3
    $\begingroup$ Short answer: no (if I understand the question correctly: when is $X \to Y$ a gerbe, for representable stacks $X$ and $Y$). Roughly speaking, the fibres of a gerbe are transitive (Lie) groupoids. So if the fibres are in fact manifolds, they are single points. Gerbes are a truly stacky phenomenon and do not appear until nontrivial stacks are around. $\endgroup$ – David Roberts Nov 28 '18 at 15:25
  • $\begingroup$ You understood the question correctly. Yes. Question is : "When is a map of stacks $\underline{M}\rightarrow \underline{N}$ a gerbe over stack". I could only come up with trivial examples as mentioned in the question.. Can you give a reference for "fibres of a gerbe are transitive (Lie) groupoids".. Thank you. @DavidRoberts $\endgroup$ – Praphulla Koushik Nov 28 '18 at 15:33
  • $\begingroup$ @PraphullaKoushik I added it as a parent tag, not to mean that you require algebraic geometry examples. The other two tags are too narrow; the parent tags I had in mind are ct.category-theory, at.algebraic-topology, ag.algebraic-geometry, and dg.differential-geometry, it's better with at least one of these (or another broad one) and don't hesitate changing. The tag not only yields an orientation, but makes the question visible. $\endgroup$ – YCor Nov 28 '18 at 15:50
  • $\begingroup$ @PraphullaKoushik it's essentially the definition. A gerbe is a stack of groupoids that has non-empty stalks ($X\to Y$ is an epimorphism), which are groupoids, and (more or less) the stalks are transitive ($X\to X\times_Y X$ is an epimorphism). Another way of thinking about it, is that a gerbe is like a bundle where the fibres are equivalent to a 1-object groupoid. $\endgroup$ – David Roberts Nov 28 '18 at 22:46
3
$\begingroup$

There are no other such gerbes. If $M$ and $N$ are manifolds, and $p\colon \underline{M}\to \underline{N}$ is a gerbe, then the corresponding map of manifolds is a diffeomorphism. The same holds if one merely has representable stacks, rather than something of the form $\underline{M}$ etc.

$\endgroup$
  • $\begingroup$ I am trying to say something like "If $\underline{M}\rightarrow \underline{N}$ is an epimorphism, then, $M\rightarrow N$ is a [---]" similarly, "If the diagonal map $\underline{M}\rightarrow \underline{M}\times_{\underline{N}}\underline{M}$ is an epimorphism, then, $M\rightarrow N$ is a [---]"... $\endgroup$ – Praphulla Koushik Nov 29 '18 at 5:13
  • $\begingroup$ I understand your comment " [---] has non empty stacks ($X\rightarrow Y$ is an epimorphism) [---] (more or less) stalks are transitive ($X\rightarrow X\times_YX$ is an epimorphism)".. I have said something like this in my answer here mathoverflow.net/questions/307123/….. So, more or less stalks are transitive Lie groupoids and if they are manifolds you are saying they have to singletons.. I am more or less ok with this.. $\endgroup$ – Praphulla Koushik Nov 29 '18 at 5:20
  • $\begingroup$ I am little confused... You said "The same holds if one merely if one merely has representable stacks, rather than something of the form $\underline{M}$".. As far as I know, a stack $\mathcal{D}$ is representable if there is an isomorphism $\mathcal{D}\rightarrow \underline{M}$ for a manifold $M$... Is there a more general definition that I am not aware of? $\endgroup$ – Praphulla Koushik Nov 29 '18 at 6:45
  • $\begingroup$ What I mean is that the notion is invariant under equivalence of stacks. If a stack is merely equivalent to one of the form $\underline{M}$, then it may have, for instance, fibres that are contractible groupoids. Also, a representable stack $X$ doesn't necessarily come with the data of a manifold $M$ and an equivalence $\underline{M}\simeq X$. (NB Isomorphisms are not enough here, since we are dealing with objects of a 2-category) $\endgroup$ – David Roberts Nov 29 '18 at 7:02
  • $\begingroup$ Ok. I think I got it. $\endgroup$ – Praphulla Koushik Nov 29 '18 at 7:08

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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