I believe I know how to prove the following results. I also know to whom to cite fancy-shmancy results that have these as a very special case. My question is: what are the correct citations for these versions? Pre-question: Are the results correctly stated?

Version 1

Let $\mathrm{Man}_d$ denote the category of $d$-dimensional smooth manifolds and local diffeomorphisms — what are often called the "étale" maps of manifolds. Let us say that a presheaf $F: \mathrm{Man}_d^{\mathrm{op}} \to \mathrm{Set}$ is a topological sheaf if it is local for the usual (or "étale") topology on $\mathrm{Man}_d$ and also takes homotopic maps in $\mathrm{Man}_d$ to equal functions in $\mathrm{Set}$. Any topological sheaf $F$ then induces a $\mathbb Z/2 = \pi_0(\hom(\mathbb R^d,\mathbb R^d)$-action on $F(\mathbb R^d)$. The claim is that $F$ is determined by the set $F(\mathbb R^d)$ with its $\mathbb Z/2$-action.

Version 2

Same thing, but with embeddings: replace $\mathrm{Man}_d$ by the category $\mathrm{Emb}_d$ of $d$-dimensional manifolds and open embeddings; use the Weiss topology.

Version 1.$\infty$ and 2.$\infty$

Same thing but replace $\mathrm{Set}$ with $\mathrm{Spaces}$, use homotopy sheaves, and classify in terms of homotopy $\hom(\mathbb R^d,\mathbb R^d) \simeq O(d)$-actions.

  • $\begingroup$ I guess I am also interested in versions 1.1 and 2.1, where you study stacks of (1-)groupoids rather than fully-infinitized homotopy sheaves/stacks. $\endgroup$ Sep 30, 2015 at 16:13
  • $\begingroup$ Version 1 follows instantly from the existence of differentiable good open covers, which is definitely a folklore result. $\endgroup$ Oct 1, 2015 at 17:18
  • $\begingroup$ @DmitriPavlov Ah, good, and googling "differentiable good open covers" results in an nlab page with textbook references. Do you want to make your comment an answer? $\endgroup$ Oct 2, 2015 at 4:26

1 Answer 1


Version 1 is a simple corollary of the existence of differentiable good open covers, i.e., covers whose finite intersections are diffeomorphic to R^n.

This shows that sheaves on the site of n-manifolds and etale maps are equivalent to sheaves on the site with a single object R^n. The fact that the space of endomorphisms of R^n in this category is homotopy equivalent to O(n) completes the proof.

  • $\begingroup$ Dmitri: this is probably not helpful. Theo started his question by "I believe I know how to prove the following results". He was specifically asking for references. $\endgroup$ Oct 2, 2015 at 21:59
  • $\begingroup$ @AndréHenriques: There are references in the link I supplied. $\endgroup$ Oct 2, 2015 at 22:03
  • $\begingroup$ @AndréHenriques: Although I can see why one might be confused: a visited link is almost invisible in the new style. $\endgroup$ Oct 3, 2015 at 10:06
  • $\begingroup$ @AndréHenriques Given that I asked Dmitri to make his comment and answer, I think it reasonable to accept the answer, also since a few days is plenty for a question like this. In the best of worlds I would like original references, but Dmitri makes a convincing argument that Version 1, at least, is a folk theorem; the linked citations also answer my ulterior goal, which was to have something to cite that isn't obviously overkill. $\endgroup$ Oct 4, 2015 at 5:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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