There is a notion of Gerbe over a Manifold and a notion of Gerbe over a stack. Given a manifold $M$, there is a way to associate a stack $\underline{M}$ with it and this gives an embedding of category of smooth manifolds into category of Categories fibered in groupoids which can be found in Orbifolds as Stacks.

This gives a hope that notion of gerbe on a manifold $M$ has to be some special case of notion of gerbe on stack $\underline{M}$. This question is about the difficulty I faced when realize that.

I have written definition of Gerbe over manifold and Gerbe over Stack in comments (hoping the question looks smaller with out the definitions).

The problem I am looking at is :

Is a gerbe over a manifold a special case of a gerbe over a stack? In particular, if $\mathfrak{X}=\underline{M}$, does a gerbe over $\mathfrak{X}=\underline{M}$ give a gerbe on manifold $M$? If so, How? How is the main question here.

Has any one looked at this before?

My previous question says the definition of gerbe over stack mean that

Given a manifold $U$, there exists an open cover $\{U_\alpha\rightarrow U\}$ such that the restriction functor $\mathcal{D}(U)\rightarrow \mathfrak{X}(U)$ has following properties

  • given an object $x\in \mathfrak{X}(U)$ there is an object $y_\alpha\in \mathcal{D}(U_\alpha)$ and an isomorphism $F(y_\alpha)\rightarrow x|_{U_\alpha}$ for each index $\alpha$.
  • given an arrow $\phi:a\rightarrow b$ in $\mathfrak{X}(U)$ there exists an arrow $\tau_\alpha:y_\alpha\rightarrow z_\alpha$ in $\mathcal{D}(U_\alpha)$ such that $F(\tau_\alpha)=\phi_\alpha$ for each index $\alpha$.

I start with a stack $\mathcal{D}\rightarrow \mathfrak{X}=\underline{M}$ and construct a gerbe over manifold $M$. I need to declare what is $\mathbb{G}(U)$ for each open set $U$ of $M$. I define that to be $\mathcal{D}(U)$ i.e., $\mathbb{G}(U)=\mathcal{D}(U)$.

motivation for declaring like that is first condition says that object set of $\mathcal{D}(U_\alpha)$ is non empty and second condition says that morphism set of $\mathcal{D}(U_\alpha)$ is non empty if I can produce some objects and morphisms in $\mathcal{C}(U_\alpha)=\underline{M}(U_\alpha)$.

I have to check that it is locally non empty and locally connected.

I was able to see that this makes $\mathcal{G}$ to be locally non empty stack. Proof I have written in comments.

Let $y,y'\in \mathbb{G}(U)=\mathfrak{R}(U)$. Then, we have $F(y),F(y')\in \mathfrak{X}(U)=\underline{M}(U)$. Assuming there is an arrow $F(y)\rightarrow F(y')$ in $\mathfrak{X}(U)$, second condition in my answer says that there is an arrow $y\rightarrow y'$ (locally atleast) in $\mathcal{D}(U)=\mathbb{G}(U)$ that maps to the arrow $F(y)\rightarrow F(y')$. This says $\mathbb{G}(U)$ is locally connected.

Thus, we have a Gerbe on manifold $M$ given a gerbe over stack $\underline{M}$ taking something for granted.

We have $F(x),F(y)\in \mathfrak{X}(U)=\underline{M}(U)$ i.e., $F(x):U\rightarrow M$ and $F(y):U\rightarrow M$ (smooth maps). An arrow $F(x)\rightarrow F(y)$ is a smooth map $\psi:U\rightarrow U$ such that $F(y)\circ \psi=F(x)$. One can quickly realize that this is not possible always (for example when $F(x)$ is constant map $p\in M$ and $F(y)$ is constant map $q\in M$ with $p\neq q$). Thus, this idea does not work.

We redefine $\mathbb{G}(U)$ as a category whose objects are objects of $\mathcal{D}(U)$ for which $F(y)$ is a diffeomorphism onto $U$. We have $$\mathbb{G}(U)=\{y\in \mathcal{D}(U): F(y)\text{ is a diffeomorphism onto } U\}.$$ Clearly, the $y$ we have choosen (locally) as preimage of inclusion map $i:U\rightarrow M$ is in $\mathbb{G}(U)$. Now, for $F(y),F(y)'\in \mathcal{C}(U)$, define $\Phi=F(y')^{-1}\circ F(y)$. Then, $F(y')\circ \Phi=F(y)$ which means there is an arrow $F(y)\rightarrow F(y')$. This solves the problem. Given $y,y'\in \mathcal{D}(U)$ there is an arrow $\phi:F(y)\rightarrow F(y')$ in $\mathcal{C}(U)$. As we are in case of gerbe, there exists arrow $\tau_{\alpha}:y|_{U_\alpha}\rightarrow y'|_{U_\alpha}$ such that $F(\tau_{\alpha})=\phi|_{U_\alpha}$, which in particular say that given $y,y'\in \mathbb{G}(U)$, they are locally connected by arrows $y|_{U_\alpha}\rightarrow y'|_{U_\alpha}$ which says $\mathbb{G}$ defines a gerbe on manifold $M$.

Is my construction correct? Can it be made more natural? For me, it was most reasonable thing to do. Any comments are welcome.

  • $\begingroup$ Given $a\in M$, there is an open set $U$ containing $a$. I claim that $\mathbb{G}(U)$ is non empty. First condition is that given an object $x\in \mathfrak{X}(U)$ there is an object $y\in \mathcal{D}(U)$ and an isomorphism $F(y)\rightarrow x$. As $\mathfrak{X}=\underline{M}$, $\mathfrak{X}(U)$ is just the collection of smooth maps $U\rightarrow M$. There is an obvious map we can go for, i.e., the inclusion map $x=i:U\rightarrow M$. For this, there is an object $y\in \mathcal{D}(U)$ and an isomorphism $F(y)\rightarrow x$. This in particular says that $\mathcal{D}(U)\neq \emptyset$. $\endgroup$ Commented Aug 3, 2018 at 16:58
  • $\begingroup$ A gerbe on a topological space $X$ is a stack $\mathbb{G}$ of groupoids on $X$ that is 1. Locally non empty i.e., i.e., given $x\in X$ there is an object/open set $U$ containing $x$ such that $\mathbb{G}(U)$ is non empty. 2. Locally connected i.e., given $a,b\in\mathbb{G}(U)$ and $x\in U$ there exists an open subset $V$ of $U$ containing $x$ such that $a|_V$ is isomorphic to $b|_V$. The definition is from arxiv.org/pdf/math/0212266.pdf $\endgroup$ Commented Aug 3, 2018 at 17:16
  • $\begingroup$ A stack $\mathcal{D}$ endowed with a morphism $\mathcal{D}\rightarrow \mathfrak{X}$ is called a gerbe over $\mathfrak{X}$ if 1. $\mathcal{D}\rightarrow \mathfrak{X}$ is an epimorphism and 2. $\mathcal{D}\rightarrow \mathcal{D}\times_{\mathfrak{X}}\mathcal{D}$ is an epimorphism. The definition is from arxiv.org/pdf/math/0605694.pdf $\endgroup$ Commented Aug 3, 2018 at 17:18
  • $\begingroup$ A slightly rash comment: Personally, I like to think of the latter condition as being: $D \simeq [X\times I,D]_X \to D\times_X D$ is an epi of stacks, where $I$ is the walking arrow, and $[-,-]_X$ is the hom-stack of maps over $X$, and the map to the fibred product is induced by the restriction to the source and target of the single non-trivial arrow in $I$. This makes it clear that condition 2. is about the fibres of $D\to X$ being (locally) connected. $\endgroup$
    – David Roberts
    Commented Aug 7, 2018 at 0:32

1 Answer 1


Yes. The category of manifolds embeds fully faithfully into the 2-category of stacks on Mfld, essentially by Yoneda, and the site structures are likewise compatible, so when restricted to the special case of your stack being a manifold, the definitions are equivalent.

  • $\begingroup$ I am trying to see how they are equivalent... $\endgroup$ Commented Aug 4, 2018 at 15:01
  • $\begingroup$ You didn't ask that. Also, it's past midnight here, so any further expansion of my answer will wait until a more appropriate time :-) $\endgroup$
    – David Roberts
    Commented Aug 4, 2018 at 15:03
  • $\begingroup$ He used the magic word "Yoneda", that's how they're equivalent... $\endgroup$ Commented Aug 4, 2018 at 15:36
  • $\begingroup$ @AndrejBauer It takes some time for me to understand the magic... For now, I want to write down in detail how one gets a gerbe on manifold $M$ from gerbe on stack $\underline{M}$ $\endgroup$ Commented Aug 4, 2018 at 16:03
  • 1
    $\begingroup$ @PraphullaKoushik have you looked at the Stacks Project much? It contains complete proofs of every statement and everything is linked. See eg stacks.math.columbia.edu/tag/06QB stacks.math.columbia.edu/tag/06NY May I also suggest that thinking in terms of Lie groupoids rather than general differentiable stacks makes things a lot easier, and what you seem to be interested in anyway. $\endgroup$
    – David Roberts
    Commented Aug 6, 2018 at 4:35

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.