Given a group $G$, one way to construct the classifying (topological) space $BG$ is to regard $G$ as a category with one object and morphisms $G$, take its nerve, and then apply the geometric realization functor.

Is an analogous statement true for topological groups? That is, given a topological group $G$, regarded as a topological category with one object, does the geometric realization of its homotopy coherent nerve model the classifying space $BG$?

Edit: I should have been clear about what I meant by the "homotopy coherent nerve" of a topological category; what I meant was the composition


where $\mathsf{Cat}_{\mathsf{Top}}$ and $\mathsf{Cat_{sSet}}$ are the category of small topologically/simplicially enriched categories, $\operatorname{Sing}$ is the base change along the singular complex functor, and $N$ is the homotopy coheret nerve functor.

  • $\begingroup$ Yes, that's all correct. See Graeme Segal's paper "Classifying spaces and spectral sequences", Section 3. $\endgroup$ Jul 1, 2022 at 7:24
  • $\begingroup$ related question: mathoverflow.net/questions/41616/… $\endgroup$ Jul 1, 2022 at 14:34
  • 1
    $\begingroup$ @Konrad Waldorf Thank you for your attention. As far as I can tell, the Segal's paper you cited does not answer the question I had in mind. I apologize for not being clear about what I meant by homotopy coheret nerves; I hope that the edited question addresses my problem clearly. $\endgroup$
    – Ken
    Jul 2, 2022 at 0:03
  • 2
    $\begingroup$ What exactly do you mean by BG? $\endgroup$ Jul 2, 2022 at 1:35
  • $\begingroup$ @TomGoodwillie By $BG$, I mean the base space of the universal principal $G$-bundle $p:EG\to BG$. And by the universal principal $G$-bundle, I mean that $p$ is a principal $G$-bundle with the property that, given a CW complex $B$, the pullback of the bundle $p$ induces a bijection between the set $\operatorname{Prin}_G(B)$ of isomorphism classes of principal $G$-bundles over $B$ and the set $[B,BG]$ of homotopy classes of maps $B\to BG$. $\endgroup$
    – Ken
    Jul 2, 2022 at 2:31

2 Answers 2


This is an answer to the edited question.

First, observe that the composition of functors $\def\N{{\rm N}}\def\Sing{{\rm Sing}}\N∘\Sing$ in the main post computes the homotopy colimit of the simplicial object $[n]↦(\Sing(G))^n$.

This can be seen as follows. Observe that the functors $\Sing$ and the homotopy coherent nerve are Quillen equivalences.

After the composition $\N∘\Sing$ we also implicitly apply the left Quillen functor given by the identity functor on the underlying categories from the Joyal model structure to the Kan–Quillen model structure on simplicial sets. (Indeed, we need to compare the value of $\N∘\Sing$ to a space (like $\B G$), not an (∞,1)-category.)

Altogether the whole composition is a homotopy cocontinuous functor. Such a functor can be uniquely specified (up to a contractible choice) by its values on generators and maps between them, which in our case correspond to categories of the form $\{0→1→2→⋯→n\}$ for all $n≥0$. As one can see from the definitions, in our case all these values are contractible spaces.

After we apply $\Sing$, we are free to apply a different Quillen equivalence, and then take the corresponding (unique up to a contractible choice) homotopy cocontinuous functor to simplicial sets. For example, we can apply the right Quillen equivalence given by the usual enriched nerve functor, which goes from simplicial categories to Segal categories, and then apply the left Quillen equivalence given by the inclusion of Segal categories into complete Segal spaces, i.e., simplicial objects in simplicial sets equipped with a certain model structure. (See Theorem 8.6 in Bergner's Three models for the homotopy theory of homotopy theories.) The enriched nerve functor yields the simplicial object $[n]↦(\Sing(G))^n$.

The generators $\{0→1→2→⋯→n\}$ correspond to simplicial objects in simplicial sets given by representable presheaves of simplices. The unique homotopy cocontinuous functor that sends these generators to contractible space is simply the homotopy colimit functor. Indeed, the homotopy colimit functor is homotopy cocontinuous and it sends all representable presheaves to contractible spaces. This proves the above claim.

Denote the above homotopy colimit by $\def\B{{\rm B}}\B(\Sing(G))$.

What is the easiest way to see that principal $G$-bundles over a topological group $G$ are classified by $\B(\Sing(G))$?

One way is provided by Theorem 8.5 in arXiv:2203.03120 (see also Theorem 1.1 in arXiv:1912.10544), which gives us an explicit formula for the classifying space of any homotopy coherent sheaf $F$ of spaces on the site of topological spaces equipped with numerable open covers. (Recall that any open cover of a CW-complex is numerable, so this is a strictly more general setting than in the original post.)

The classifying space is given by the homotopy colimit of simplicial sets $$\def\hocolim{\mathop{\rm hocolim}}\def\op{{\rm op}}\def\gs{{\bf Δ}}\hocolim_{n∈Δ^\op}F(\gs^n),$$ where $\gs^n$ denotes the topological space given by the $n$-dimensional simplex.

In our case, $F(X)$ is the nerve of the category of principal $G$-bundles over $X$ and their isomorphisms. Since any principal $G$-bundle over $\gs^n$ is trivial, we have $\def\C{{\rm C}}F(\gs^n)≃\B(\C(X,G))$, where $\C(X,G)$ denotes the group of continuous functions $X→G$.

Now $$\hocolim_{n∈Δ^\op}F(\gs^n)≃\hocolim_{n∈Δ^\op}\B(\C(\gs^n,G)).$$ Since the functor $\B$ is homotopy cocontinuous, we have $$\hocolim_{n∈Δ^\op}\B(\C(\gs^n,G))≃\B(\hocolim_{n∈Δ^\op}\C(\gs^n,G))≃\B(\Sing(G)),$$ as desired, proving that numerable principal $G$-bundles over an arbitrary topological space $X$ are classified by the simplicial set $\B(\Sing(G))$ (or the topological space $\B G$). (This computation appears as Example 8.6 in the cited paper.)

  • $\begingroup$ Thank you for your answer. Could you elaborate on your second paragraph? I am not experienced enough to understand the reasoning behind the claim that $N \operatorname{Sing}$ computes the homotopy colimit as stated. What do you mean when you say "$N$ is induced by homotopy cocontinuous functor from simplicial spaces to spaces that sends representable simplices to contractible spaces"? I thought that $N$ had nothing to do with simplicial spaces; rather, it is merely a part of the Quillen equivalence between $\mathsf{sSet}$ and $\mathsf{Cat}_{\mathsf{sSet}}$. $\endgroup$
    – Ken
    Jul 3, 2022 at 0:30
  • $\begingroup$ @Ken: I added more information about this statement. $\endgroup$ Jul 3, 2022 at 1:26
  • $\begingroup$ Thank you very much. Sorry for asking dumb questions, but can you explain why do the categories $\{0\to\cdots\to n\}$ generate $\mathsf{Cat}_{mathsf{Top}}$ (under homotopy colimit, I guess)? $\endgroup$
    – Ken
    Jul 3, 2022 at 2:05
  • $\begingroup$ @Ken: Sets of generators can be transported accross Quillen equivalences. Since Cat_Top is Quillen equivalent to sSet_Joyal, it suffices to see that {0→1→⋯→n}, i.e., Δ^n∈sSet_Joyal, generate sSet_Joyal under homotopy colimits. Indeed, any simplicial set X is the homotopy colimit of the canonical diagram Δ/X→sSet that sends (Δ^n→X)↦Δ^n, i.e., takes values in the generators. $\endgroup$ Jul 3, 2022 at 2:21
  • $\begingroup$ Aha! I greatly appreciate your patience and insight. $\endgroup$
    – Ken
    Jul 3, 2022 at 2:25

Konrad Waldorf answers this question in his comment, but I didn't want the question to linger in the "unanswered" queue, so here's a CW answer consisting of a page of Segal's 1968 paper that answers the question, in case the OP has trouble finding a copy of this paper.

enter image description here

  • 4
    $\begingroup$ Thank you very much. The paper you cited does not answer my question. I apologize for not being clear about what I meant by homotopy coherent nerves. Could you take a look at the edited question? $\endgroup$
    – Ken
    Jul 2, 2022 at 0:05

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.