# What is the homotopy colimit of the Cech nerve as a bi-simplical set?

Let $f:Y_\bullet\to X_\bullet$ be an epimorphism of simplicial sets and define the bi-simplicial set $$F_{\bullet\bullet}=\ldots Y\times_X Y\times_X Y\underset{\to}{\underset{\to}{\to}}Y\times_X Y \underset{\to}{\to}Y$$ as usual (link). Now $F_{\bullet\bullet}$ can be viewed as a diagram in simplicial sets and one can take its homotopy colimit $\operatorname{hocolim}(F_{\bullet\bullet})$ which is a simplicial set. This is weakly equivalent to the diagonal (or the realization) of $F_{\bullet\bullet}$.

Is $\operatorname{hocolim}(F_{\bullet\bullet})$ weakly equivalent to $X_\bullet$? What is a reference for this?

I believe this to be true since there is a similar statement called the nerve theorem for spaces and simplicial spaces instead of simplicial sets and bi-simplicial sets. Perhaps it is not sufficient for $f$ only being an epimorphism for the statement to hold. In this case, my question would be what the exact conditions on $f$ are.

Comment on a comment to this question: If $X$ and $Y$ are discrete simplicial sets, the diagonal of $F_{\bullet \bullet}$ is the simplicial set $$F_{\bullet}=\ldots Y\times_X Y\times_X Y\underset{\to}{\underset{\to}{\to}}Y\times_X Y \underset{\to}{\to}Y$$ where each $X$ and $Y$ is viewed as a set (and not as a simplicial set). This simplicial set $F_\bullet$ is weakly equivalent to the colimit $C$ (in the category of sets) of $$Y\times_X Y \underset{\to}{\to}Y$$ and, if I understand it correctly, because an epimorphism $f$ is the same as an effective epimorphism of sets, $X$ is the colimit of this diagram and therefore $\operatorname{hocolim}(F_{\bullet\bullet})\cong diagonal(F_{\bullet\bullet})\cong F_{\bullet}\cong C \cong X$ in this discrete case.

On the other hand, in the example with $X$ being a point and $Y$ being two points, I think that $C$ is a space with two points which is mysterious. What am I doing wrong?

• The diagonal of a bisimplicial set is weakly equivalent to its homotopy colimit: see here. – Zhen Lin Apr 24 '13 at 0:12
• What happens if $X$ is a point? Furthermore what if $Y$ is two points? – Baby Dragon Apr 24 '13 at 1:40
• Dear @BabyDragon, I've made an edit to the question concerning your example. – Ronald Bernard Apr 24 '13 at 9:12
• I am not sure that you are doing anything wrong. Thank you for your reply. I thought that contemplating of the simplest possible nontrivial case would hopefully be enlightening. – Baby Dragon Apr 24 '13 at 14:02

I am not sure what the precise conditions under which the weak equivalence is true, because the simplicial space you've written down isn't homotopy invariant unless $Y \to X$ is a fibration, or unless you take homotopy fiber products at every stage. But if you take homotopy fiber products at every stage, then the geometric realization is always $X$ (if $Y \to X$ is an epimorphism on $\pi_0$): this is a sort of homotopical descent property. However, I'm not sure if this is what you want.

One way to see this is that the simplicial object above is augmented: it lives naturally in the category of spaces over $X$. The (homotopy) pullback functor from spaces over $X$ to spaces over $Y$ preserves homotopy colimits (e.g., at the level of model categories, it is a left Quillen functor), and it's conservative (reflects homotopy equivalences) since $\pi_0 Y \to \pi_0 X$ is surjective, so it's sufficient to check that after you pull the simplicial object over $X$ back to $Y$, its geometric realization is equivalent to $Y$. But when you pull it back, it's augmented over $Y$ and has an extra degeneracy, so its realization is weakly equivalent to $Y$.

• Dear @Akhil Mathew, thank you for the great answer. I am trying to grasp the things you said in the terms of model categories or $\infty$-topoi. Let us assume we know that $Sets^{\Delta^{op}}$ is such. Let $f:Y\to X$ be the fibration. The first property ($f^*$ preserves homotopy colimits) seems to be true since homotopy colimits are universal (Higher topos theory, definition 6.1.1.2) but I can't find the second property ($f^*$ is ''conservative''). Do you know how this property is called in Lurie's? Is it one of the $\infty$-topos properties? It seems to be more crucial than the first one. – Ronald Bernard Apr 25 '13 at 13:03
• @Ronald: The conservativity property follows from the long exact sequence of a fibration. One way to see it, though, is that you can think of fibrations over $X$ as being functors from $X$ (considered as an $\infty$-category) into spaces, and the pull-back of a fibration from $Y$ to $X$ corresponds to restricting the functor from $X$ to $Y$. Since every object in $X$ is isomorphic to something in the image of $Y$, the pull-back on functors is conservative. – Akhil Mathew Apr 25 '13 at 13:18
• Akhil gave excellent replies. As to where to find this in Lurie's book: around 7.2.1.14, 7.2.1.15. A digest is here: ncatlab.org/nlab/show/… – Urs Schreiber Apr 25 '13 at 14:15
• Dear AkhilMathew, thank you. Dear @Urs Schreiber, thank you for the references. – Ronald Bernard Apr 25 '13 at 14:42

This is true.

Start with the map $Y_\bullet \to X_\bullet$ of simplicial sets and form the bisimplicial set $F_{\bullet \bullet}$ as you've already done. Fiber products in simplicial sets are calculated levelwise, and so $F_{p,q}$ is the ${p+1}$-fold fiber product of $Y_q$ over $X_q$, and there is a natural augmentation $F_{0,q} \to X$.

To check that you get an equivalence, it suffices to check that $|F_{\bullet \bullet}| \to |X_\bullet|$ is an equivalence. The former can be realized several ways: realize with respect to $p$ first, with respect to $q$ first, or take the diagonal and then realize. All of these are homeomorphic.

Let's do realization in the $p$ component first first. Then for any $q$, we get a map of spaces $|F_{\bullet,q}| \to X_q$, where the former space is built out of the iterated fiber products $(Y_q)^{p+1}$ over $X_q$. It turns out that this is a homotopy equivalence whenever $Y_q \to X_q$ is surjective.

(The shortest proof I know is that, given a section $s: X_q \to Y_q$, there is a simplicial contraction down to $s(x)$; given a simplex $(y_0,\ldots,y_p)$ over $x \in X_q$, we have a $(p+1)$-simplex $(s(x),y_0,\ldots,y_p)$ that contracts it down to $s(x)$, and this respects boundary identifications.)

Therefore, the map $|F_{\bullet,q}| \to X_q$, viewed as a map of simplicial spaces, is a levelwise equivalence, and so it becomes an equivalence on geometric realization $|F_{\bullet \bullet}| \to |X_\bullet|$. (Geometric realization takes weak equivalences to weak equivalences for "good" simplicial spaces, and bisimplicial sets always give "good" simplicial spaces when you realize one direction.)