What is called geometric realization of categories is a functor that sends categories to topological spaces, namely the functor which first forms the simplicial set $N(\mathcal{C})$ that is the nerve of the category $\mathcal{C}$, and then forms the geometric realization ${\vert N(\mathcal{C})\vert}$ of this simplical set. Typically one is interested in this geometric realization up to weak homotopy equivalence.
By the homotopy hypothesis-theorem the geometric realization of simplicial sets constitutes a (Quillen)equivalence between the classical homotopy theory of simiplicial sets and the classical homotopy theory of topological spaces. This means that in as far as one is interested in geometric realization of categories up to weak homotopy equivalence, then the key part of the operation is in forming the simplicial nerve $N(\mathcal{C})$ of a category, with the latter regarded as a model for an ∞-groupoid. Indeed, equivalently one could consider the Kan fibrant replacement of the nerve $N(\mathcal{C})$ (which still has the same geometric realization, up to weak homotopy equivalence).
Therefore an equivalent perspective on geometric realization of categories is that it universally turns a category into an infinity-groupoid by freely turning all its morphisms into homotopy equivalences.
Geometric realization of categories has various good properties:
It sends equivalences of categories to weak homotopy equivalences (corollary 1 below). A more general sufficient criterion for the geometric realization of a functor is given by the seminal theorem known as Quillen’s theorem A (theorem 1 below.)
The existence of the Thomason model structure (below) implies that every homotopy type arises as the geometric realization of some category. In fact it arises alrady as the geometric realization of some poset ((0,1)-category).
Write
for the nerve functor from Cat to sSet. Write
for the geometric realization of simplicial sets from sSet to Top.
The geometric realization of categories is the composite of these two operstions:
There is a model category structure on Cat whose weak equivalences are those functors $F \colon \mathcal{C} \to \mathcal{D}$ which under geometric realization become weak equivalences in the classical model structure on topological spaces, hence which become weak homotopy equivalences. This is called the Thomason model structure.
The existence of the Thomas model structure implies that every homotopy type arises as the geometric realization of some category, in fact already as the realization of some poset/(0,1)-category:
For $C$ a category, let $\nabla C$ be the poset of simplices in the nerve $N C$, ordered by inclusion.
For every category $\mathcal{C}$ the poset $\nabla \mathcal{C}$ from def. 1 has weakly homotopy equivalent geometric realization
Let $\mathcal{C}, \mathcal{D}$ be two categories and let
be a functor between them.
(Quillen 72, theorem A)
If for all objects $d \in \mathcal{D}$ the geometric realization ${\vert N(F/d)\vert}$ of the comma category $F/d$ is contractible (meaning that $F$ is a “homotopy cofinal functor”, hence a cofinal (∞,1)-functor), then
(Quillen 72 theorem B)
If for all objects $d \in \mathcal{D}$ we have that ${\vert N(F/d)\vert}$ is weakly homotopy equivalent to a given topological space $X$ and all morphisms $f \colon d_1 \to d_2$ induce weak homotopy equivalences between these, then $X$ is the homotopy fiber of ${\vert N(F) \vert}$, hence we have a homotopy fiber sequence (in the classical model structure on topological spaces) of the form
As a consequence:
(McCord 66, theorem 6, Quillen 78, prop. 1.6)
Let $\mathcal{C}, \mathcal{D}$ be finite posets and consider $F \colon \mathcal{C} \to \mathcal{D}$ be a functor.
If for each element/object $y \in \mathcal{D}$ its preimage $f^{-1}( \{ y' \in Y \vert y' \leq y \})$ has contractible geometric realization, then ${\vert N(F)\vert}$ is a homotopy equivalence.
An alternative proof is given in (Barmak 10).
A natural transformation $\eta : F \Rightarrow G$ between two functors $F, G : \mathcal{C} \to \mathcal{D}$ induces under geometric realization a homotopy
The natural transformation is equivalently a functor of the form
out of the product category of $\mathcal{C}$ with the interval category.
Since geometric realization of simplicial sets preserves Cartesian products (see there) we have that
But this is a cylinder object in topological spaces, hence ${\vert N(\eta) \vert}$ is a left homotopy.
An equivalence of categories $\mathcal{C} \simeq \mathcal{D}$ induces a homotopy equivalence between their geometric realizations.
Notice that the converse is far from true: Very different categories can have geometric realizations that are (weakly) homotopy equivalent. This is because geometric realization implicitly involves Kan fibrant replacement: it freely turns morphisms into equivalences.
If a category $\mathcal{C}$ has an initial object or a terminal object, then its geometric realization is contractible.
Assume the case of a terminal object, the other case works formally dually. Write $*$ for the terminal category.
Then we have an equality of functors
where the first functor on the right picks the terminal object, and we have a natural transformation
whose components are the unique morphisms into the terminal object.
By prop. 3 it follows that we have a homotopy equivalence $\vert N(\mathcal{C}) \vert \to \vert N(\ast) \vert = \ast$.
For $F \colon \mathcal{D} \to Cat$ a functor, let
be its postcomposition with geometric realization of categories
Then we have a weak homotopy equivalence
exhibiting the homotopy colimit in Top over $\vert N(F (-)) \vert$ as the geometric realization of the Grothendieck construction $\int F$ of $F$.
This is due to (Thomason 79).
For general references see also nerve and geometric realization.
The original articles are
Michael C. McCord, Singular homology groups and homotopy groups of finite topological spaces, Duke Math. J. 33 (1966), 465-474
Daniel Quillen, Higher algebraic K-theory, I: Higher K-theories Lect. Notes in Math. 341 (1972), 85-1 (pdf)
Daniel Quillen, Homotopy properties of the poset of nontrivial p-subgroups of a group, Adv. Math. 28 (1978), 101-128.
The geometric realization of Grothendieck constructions has been analyzed in
Review is in
Further development includes
Clark Barwick, Daniel Kan, A Quillen theorem $B_n$ for homotopy pullbacks (arXiv:1101.4879)