canonical model structure


Category theory

Model category theory

model category



Universal constructions


Producing new model structures

Presentation of (,1)(\infty,1)-categories

Model structures

for \infty-groupoids

for ∞-groupoids

for nn-groupoids

for \infty-groups

for \infty-algebras



for stable/spectrum objects

for (,1)(\infty,1)-categories

for stable (,1)(\infty,1)-categories

for (,1)(\infty,1)-operads

for (n,r)(n,r)-categories

for (,1)(\infty,1)-sheaves / \infty-stacks



In general, canonical model structures are model category structures on the categories of some flavor of n-categories for 1n1\le n\le \infty (note that n=n=\infty or ω\omega is allowed), which are intended to capture the correct “categorical” theory of these categories.

The archetypical examples are the canonical model structure on categories and the canonical model structure on groupoids.

Canonical model structures are sometimes called “folk” model structures, but the appropriateness of this term is very questionable, especially in the cases n>1n \gt 1. Other alternatives are ‘endogenous’, ‘standard’, ‘natural’, and ‘categorical’.

While ultimately the collection of all n-categories should form an (n+1)(n+1)-category, restricting that to just invertible higher morphisms will yield an (n+1,1)-category, and thus in particular an (∞,1)-category. It is this (∞,1)-category which is intended to be presented by a canonical model structure. In particular, the weak equivalences in a canonical model structure should be the category-theoretic equivalences.

This is to be contrasted with Thomason model structures in which the weak equivalences are the morphisms that induce a weak homotopy equivalence of nerves. This amounts to regarding each category, or rather its nerve, as a placeholder for its groupoidification (Kan fibrant replacement) and then considering the standard notion of equivalence.

In a canonical model structure for some flavor of nn-categories, usually

  • a fibration is a functor that lifts equivalences in all dimensions,
  • an acyclic fibration is a functor which is k-surjective for all 0kn0\le k\le n,
  • a weak equivalence is a functor which is essentially k-surjective for all 0kn0\le k\le n, and
  • a cofibration is a functor which is injective on objects and “relatively free” on kk-morphisms for 1k<n1\le k \lt n. These can also be described as the morphisms generated by the inclusions G kG k\partial G_k \hookrightarrow G_k of the boundary of the kk-globe into the kk-globe for 0k<0\le k \lt \infty.


A common problem is to transport the (a) model structure on plain ω\omega-categories, i.e. ω\omega-categories internal to SetsSets to another internal context, notably for the case that SetsSets is replaced with some kind of category of SpacesSpaces. This is relevant for the discussion of the homotopy theory of topological and smooth ω\omega-categories.

Usually, such internalization of model structures has the consequence that some properties invoked in the description of the original model structure, notably some of the lifting properties, will only continue to hold “locally”. One way to deal with this is to pass to a notion slightly weaker than that of a model category called a category of fibrant objects as used in homotopical cohomology theory.

But there are also full model structures for such situations. Notice that under a suitable nerve operation all n-categories usually embed into simplicial sets. The models for infinity-stack (infinity,1)-toposes given by the model structure on simplicial presheaves then serves to present the corresponding (,1)(\infty,1)-category of parameterized or internal nn-categories. See for instance also smooth infinity-stack.

Cofibrant resolutions


it is shown that cofibrant ω\omega-categories with respect to the canonical model structure are precisely the “free” ones, where “free” here means “generated from a polygraph” as described in

We had some blog discussion about this at Freely generated omega-categories.


The canonical model structure for 1-categories was known to experts for some time before being written down formally (this is the origin of the adjective “folk”).

The canonnical model structure for 1-groupoids is briefly described in section 5 of

  • D.W. Anderson, Fibrations and Geometric Relations, Bulletin of the AMS, Vol. 84: 765-788, 1978.

It was apparently first published (for categories internal to a Grothendieck topos) by Joyal and Tierney, Strong Stacks and Classifying Spaces, Category theory (Como, 1990) Springer LNM 1488, 213-236.

A more elementary writeup is in

A general internal version relative to a Grothendieck coverage can be found in

  • T. Everaert, R.W. Kieboom, T. Van der Linden, Model structures for homotopy of internal categories, Theory and Applications of Categories, Vol. 15, CT2004, No. 3, pp 66-94. (tac).

though it seems that the assumptions in this article apply to ambient categories that are semiabelian categories, but do not apply to ambient categories like Top.

A brief summary, together with a generalization when one assumes only weaker versions of the axiom of choice, can be found at folk model structure on Cat.

See also

The canonical model structures for 2-categories and bicategories are due to

  • Steve Lack,

    A Quillen Model Structure for 2-Categories, K-Theory 26: 171–205, 2002. (website)

    A Quillen Model Structure for Bicategories, K-Theory 33: 185-197, 2004. (website)

and for n=3n=3 (Gray-categories) in

For n=ωn = \omega (strict omega-categories)

For n=ωn = \omega and all morphisms invertible, there is the model structure on strict omega-groupoids:

  • Ronnie Brown and M. Golasinski, A model structure for the homotopy theory of crossed complexes, Cah. Top. Géom. Diff. Cat. 30 (1989) 61-82 (pdf)

  • Dimitri Ara, Francois Metayer, The Brown-Golasinski model structure on ∞-groupoids revisited (pdf) Homology, Homotopy Appl. 13 (2011), no. 1, 121–142.

Revised on April 13, 2015 11:36:40 by Urs Schreiber (