Geometric Realization
Geometric realization is the functor that turns a simplicial set into a topological space. It replaces each abstract -simplex with a topological -simplex and glues them according to the face and degeneracy maps. Together with the singular simplicial set functor going the other way, geometric realization establishes a deep connection between combinatorial and topological homotopy theory.
The Topological Simplex
The topological standard -simplex is
For each order-preserving map , there is an induced continuous map given by
This defines a covariant functor .
In low dimensions:
- : a single point.
- : a line segment.
- : a filled triangle with vertices .
- : a solid tetrahedron.
The coface map embeds the -simplex as the -th face (the face opposite vertex ). The codegeneracy collapses the edge from vertex to vertex .
Definition of Geometric Realization
The geometric realization of a simplicial set is the topological space
where is given the discrete topology and the equivalence relation is generated by:
Equivalently, is the coend
which is the tensor product of the functor with .
The geometric realization is homeomorphic to (the topological -simplex). This follows from the Yoneda lemma and the coend formula:
The last isomorphism is the "co-Yoneda lemma" (or density theorem for coends).
The geometric realization is the boundary of the topological -simplex, i.e., the union of all -dimensional faces of . This is homeomorphic to the -sphere .
For : is the boundary of a triangle, homeomorphic to .
The simplicial set has one vertex and one non-degenerate edge (with both endpoints identified). Its geometric realization is obtained by taking the interval and identifying , giving the circle .
For a group , the geometric realization is the classifying space .
For : , the infinite lens space, which has and is a .
For (symmetric group on 3 elements): the classifying space has and all higher homotopy groups trivial.
Properties of Geometric Realization
If and are simplicial sets, then (where the right side uses the compactly generated product topology), provided at least one of or is locally finite (finitely many non-degenerate simplices in each dimension).
Without the locally finite condition, one uses compactly generated spaces to ensure the product formula holds.
The product is a simplicial set whose geometric realization is the square . This square is triangulated: the non-degenerate -simplices correspond to the two triangles in the standard triangulation of the square (upper-left and lower-right).
The Eilenberg--Zilber theorem provides a natural quasi-isomorphism between and at the chain level.
Geometric realization preserves all colimits (since it is a left adjoint). In particular:
and more generally, if , then .
This means is built by gluing together the geometric realizations of the standard simplices appearing in .
The Singular Simplicial Set
The singular simplicial set (or singular complex) of a topological space is the simplicial set defined by
Face maps and degeneracy maps are induced by precomposition with the coface maps and codegeneracy maps .
This defines a functor .
is a single element for each (the unique map ). So , the terminal simplicial set.
is an "enormous" simplicial set: its -simplices are all continuous paths , its -simplices are all continuous maps from a triangle into , etc. Despite being large, it is a Kan complex (every horn can be filled) and is weakly equivalent to a point (since is contractible).
is a Kan complex with and for . It has uncountably many -simplices (all loops and paths in ), but its homotopy type is completely described by .
The Adjunction
The geometric realization and singular simplicial set functors form an adjunction
with geometric realization as the left adjoint. The unit and counit are:
- Unit : sends an -simplex to the corresponding continuous map .
- Counit : sends a point (where and ) to .
For any topological space , the counit is a weak homotopy equivalence: it induces isomorphisms on all homotopy groups. This is a classical theorem in algebraic topology.
For CW complexes, this map is actually a homotopy equivalence (by Whitehead's theorem). For general spaces, is a CW approximation of .
This means that from the perspective of homotopy theory, every topological space is "modeled" by a simplicial set.
The unit is generally not an isomorphism (or even a weak equivalence for arbitrary ). However, when is a Kan complex, is a weak equivalence.
For example, if , then , which has uncountably many -simplices, while has only three (two degenerate, one non-degenerate).
CW Structure
The geometric realization of any simplicial set naturally has the structure of a CW complex. The cells are in bijection with the non-degenerate simplices of : each non-degenerate -simplex contributes an open -cell.
The -skeleton of the CW complex is precisely , the geometric realization of the -skeleton of .
The attaching maps are determined by the face operators. Specifically, a non-degenerate -simplex contributes a cell attached via the map induced by the faces .
Summary
The main ideas of geometric realization are:
-
Geometric realization turns a simplicial set into a topological space by replacing abstract simplices with topological simplices and gluing.
-
The singular functor goes the other way, producing a simplicial set from a topological space.
-
These form an adjunction that is the bridge between combinatorial and topological homotopy theory.
-
The counit is always a weak homotopy equivalence, showing that simplicial sets capture the homotopy theory of spaces.
-
Geometric realization is naturally a CW complex, with cells corresponding to non-degenerate simplices.