Let $K(G,1)$ denote the Eilenberg-Maclane spaces with fundamental groups isomorphic to $G$.

Consider the category $\mathbf{K^1_{CW,*}}$ where objects are pointed $K(G,1)$ CW complex, and morphisms are equivalence classes of homotopy rel basepoint pointed continuous maps. Then the fundamental group functor $$ \Pi_1:\mathbf{K^1_{CW,*}}\to \mathbf{Grp} $$ is fully faithful. This is by Theorem 1B.8 in Hatcher. Moreover, this functor is essentially surjective, because for every group $G$, we can construct an element of $K(G,1)$ denoted $BG$ constructed in Hatcher 1B.7. Therefore $\Pi_1$ induces an equivalence of categories assuming axiom of choice.

My questions is:
1. Is the above correct?
2. Is the inverse functor given by the construction that assigns to $G$ the $K(G,1)$ $\Delta$-comple BG, and $\phi:G\to H$ the induced morphism $B\phi:BG\to BH$ given by Hatcher 1B.7?
3. Does it generalize to higher homotopy group construction $\Pi_n$ and higher Eilenberg-Maclane spaces $K(G,n)$?

  • $\begingroup$ How is choice used ? $\endgroup$ – Rene Schipperus Oct 1 '14 at 2:59
  • $\begingroup$ @ReneSchipperus I was told that fully faithful and essential surjective implies equivalence requires AC. $\endgroup$ – mez Oct 1 '14 at 3:00
  • $\begingroup$ Are t he objects spaces or homotopy equivalence classes of spaces ? $\endgroup$ – Rene Schipperus Oct 1 '14 at 3:01
  • $\begingroup$ @ReneSchipperus I'm thinking just spaces. $\endgroup$ – mez Oct 1 '14 at 3:02
  • 1
    $\begingroup$ My intuition tells me that in this case choice is not needed since you have a well defined method of constructing a $K(G,1)$ from $G$. $\endgroup$ – Rene Schipperus Oct 1 '14 at 3:15

I will work with Kan complexes (i.e. fibrant simplicial sets) instead of CW-complexes. For the purposes of homotopy theory there is no essential difference: the homotopy category of pointed Kan complexes is equivalent to the homotopy category of pointed CW-complexes.

Let $\mathbf{Kan}_*$ be the (ordinary) category of pointed Kan complexes. As is well known, we have a functor $$\pi_n : \mathbf{Kan}_* \to \mathbf{Ab}$$ for every integer $n \ge 2$. Slightly less well known is that we have a functor $$K (-, n) : \mathbf{Ab} \to \mathbf{Kan}_*$$ going in the other direction. The quickest way to define it is to say that $K (A, n)$ is (the underlying pointed Kan complex of) the simplicial abelian group corresponding (under Dold–Kan) to the chain complex $A [-n]$ that has $A$ in degree $n$ and $0$ otherwise. More explicitly, $$K (A, n)_m = \bigoplus_{[m] \twoheadrightarrow [n]} A$$ where the direct sum is indexed over the set of monotone surjections $\{ 0, \ldots, m \} \to \{ 0, \ldots, n \}$, with simplicial operators defined as follows: for any monotone map $\alpha : \{ 0, \ldots, l \} \to \{ 0, \ldots, m \}$, given a monotone surjection $\beta : \{ 0, \ldots, m \} \to \{ 0, \ldots, n \}$ and an element $a$ of $A$, if $\beta \circ \alpha$ is surjective, then $\alpha$ sends $(\beta, a)$ to $(\beta \circ \alpha, a)$; otherwise $\alpha$ sends $(\beta, a)$ to $0$.

The Dold–Kan correspondence says that, for any simplicial abelian group $G$, we have a natural isomorphism $\pi_n (G) \cong H_n (C)$, where $C$ is the chain complex corresponding to $G$. In particular, we have $$\pi_n (K (A, n)) \cong H_n (A [-n]) \cong A$$ naturally in $A$. This shows that $\pi_n : \mathbf{Kan}_* \to \mathbf{Ab}$ is essentially surjective on objects, while $K (-, n) : \mathbf{Ab} \to \mathbf{Kan}_*$ is faithful and conservative. With a bit more work it can be shown that $K (-, n) : \mathbf{Ab} \to \mathbf{Kan}_*$ is also full.

Of course, it should go without saying that $\pi_n : \mathbf{Kan}_* \to \mathbf{Ab}$ factors through the homotopy category to give a functor $\pi_n : \operatorname{Ho} \mathbf{Kan}_* \to \mathbf{Ab}$, which will be a left inverse for the functor $K (-, n) : \mathbf{Ab} \to \operatorname{Ho} \mathbf{Kan}_*$, which is full. Thus $\mathbf{Ab}$ is equivalent to the full subcategory of $\operatorname{Ho} \mathbf{Kan}_*$ spanned by the pointed Kan complexes that are homotopy equivalent to $K (A, n)$ for some abelian group $A$.

The case $n = 1$ requires some adjustments, but you seem to be familiar with that already.


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.