2
$\begingroup$

Suppose that we have a (combinatorial if necessary) model category $M$, and let $F:\Delta^{op}\rightarrow M$ a simplicial object in $M$, such that for any natural number $n$, $F([n])$ is a fibrant object in $M$. We define a new object $X= colim_{n} F([n]) $. Is it true that $X$ is a fibrant object ?

$\endgroup$
0

1 Answer 1

7
$\begingroup$

No, it is not.

If what you mean by $\rm colim_n$ is the actual colimit of $F$ as a diagram of shape $\Delta$, then this colimit is isomorphic to the coequalizer of the two maps $F([1]) \rightrightarrows F([0])$. Coequalizers rarely preserve fibrancy, and with a little thought we can think of a counterexample that extends to a simplicial object. Let $M$ be a model category in which coproducts of fibrant objects are fibrant (e.g. simplicial sets), and let $G:D\to M$ be any diagram of fibrant objects whose colimit is not fibrant (such as a cospan $X \leftarrow \Delta[0] \to Y$ in simplicial sets for almost any Kan complexes $X$ and $Y$). Let $F$ be the simplicial bar construction of $G$, with $F([n]) = \coprod_{d_0 \to \cdots \to d_n} G(d_0)$. Then each $F([n])$ is fibrant by our assumptions on $M$ and $G$, but the colimit of $F$ is the coequalizer of $\coprod_{d_0\to d_1} G(d_0) \rightrightarrows \coprod_d G(d)$, which is just the colimit of $G$ (it is the usual computation of colimits in terms of coproducts and coequalizers).

If instead what you mean by $\rm colim_n$ is actually the geometric realization, then there is an even easier counterexample. Let $M$ be simplicial sets, let $X$ be a simplicial set that is not a Kan complex, and let $F([n]) = X_n$ regarded as a discrete simplicial set. Discrete simplicial sets are Kan complexes, so each $F([n])$ is fibrant, but the geometric realization of $F$ is just $X$ itself, which is not fibrant.

$\endgroup$
2
  • $\begingroup$ I see, do you think it is still false if we impose that the two maps $F([1]) \rightrightarrows F([0])$ are fibrations ? $\endgroup$
    – Paris
    Feb 21, 2019 at 20:36
  • 4
    $\begingroup$ @Paris Yes, probably. Colimits rarely preserve fibrant objects and fibrations. $\endgroup$ Feb 22, 2019 at 0:00

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.