ConceptComplete

Geometric Realization

Geometric realization is the functor that turns a simplicial set into a topological space. It replaces each abstract nn-simplex with a topological nn-simplex βˆ£Ξ”n∣|\Delta^n| 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

Definition1.1Topological standard simplex

The topological standard nn-simplex is

βˆ£Ξ”n∣={(t0,t1,…,tn)∈Rn+1β€…β€Š|β€…β€Štiβ‰₯0,βˆ‘i=0nti=1}.|\Delta^n| = \left\{(t_0, t_1, \ldots, t_n) \in \mathbb{R}^{n+1} \;\middle|\; t_i \geq 0, \sum_{i=0}^n t_i = 1\right\}.

For each order-preserving map Ο†:[m]β†’[n]\varphi: [m] \to [n], there is an induced continuous map βˆ£Ο†βˆ£:βˆ£Ξ”mβˆ£β†’βˆ£Ξ”n∣|\varphi|: |\Delta^m| \to |\Delta^n| given by

βˆ£Ο†βˆ£(t0,…,tm)=(s0,…,sn),sj=βˆ‘iβˆˆΟ†βˆ’1(j)ti.|\varphi|(t_0, \ldots, t_m) = (s_0, \ldots, s_n), \quad s_j = \sum_{i \in \varphi^{-1}(j)} t_i.

This defines a covariant functor βˆ£β‹…βˆ£:Ξ”β†’Top|\cdot|: \Delta \to \mathbf{Top}.

ExampleLow-dimensional topological simplices

In low dimensions:

  • βˆ£Ξ”0∣={1}|\Delta^0| = \{1\}: a single point.
  • βˆ£Ξ”1∣={(t0,t1):t0+t1=1,tiβ‰₯0}β‰…[0,1]|\Delta^1| = \{(t_0, t_1) : t_0 + t_1 = 1, t_i \geq 0\} \cong [0, 1]: a line segment.
  • βˆ£Ξ”2∣|\Delta^2|: a filled triangle with vertices (1,0,0),(0,1,0),(0,0,1)(1,0,0), (0,1,0), (0,0,1).
  • βˆ£Ξ”3∣|\Delta^3|: a solid tetrahedron.

The coface map ∣δi∣:βˆ£Ξ”nβˆ’1βˆ£β†’βˆ£Ξ”n∣|\delta^i|: |\Delta^{n-1}| \to |\Delta^n| embeds the (nβˆ’1)(n-1)-simplex as the ii-th face (the face opposite vertex ii). The codegeneracy βˆ£Οƒj∣:βˆ£Ξ”n+1βˆ£β†’βˆ£Ξ”n∣|\sigma^j|: |\Delta^{n+1}| \to |\Delta^n| collapses the edge from vertex jj to vertex j+1j+1.


Definition of Geometric Realization

Definition1.2Geometric realization

The geometric realization of a simplicial set XX is the topological space

∣X∣=(∐nβ‰₯0XnΓ—βˆ£Ξ”n∣)/∼|X| = \left(\coprod_{n \geq 0} X_n \times |\Delta^n|\right) / {\sim}

where XnX_n is given the discrete topology and the equivalence relation ∼\sim is generated by:

(di(x),t)∼(x,∣δi∣(t))forΒ x∈Xn,β€…β€Štβˆˆβˆ£Ξ”nβˆ’1∣(d_i(x), t) \sim (x, |\delta^i|(t)) \quad \text{for } x \in X_n, \; t \in |\Delta^{n-1}| (sj(x),t)∼(x,βˆ£Οƒj∣(t))forΒ x∈Xn,β€…β€Štβˆˆβˆ£Ξ”n+1∣(s_j(x), t) \sim (x, |\sigma^j|(t)) \quad \text{for } x \in X_n, \; t \in |\Delta^{n+1}|

Equivalently, ∣X∣|X| is the coend

∣X∣=∫[n]βˆˆΞ”XnΓ—βˆ£Ξ”n∣=XβŠ—Ξ”βˆ£β‹…βˆ£|X| = \int^{[n] \in \Delta} X_n \times |\Delta^n| = X \otimes_\Delta |\cdot|

which is the tensor product of the functor X:Ξ”opβ†’SetX: \Delta^{\mathrm{op}} \to \mathbf{Set} with βˆ£β‹…βˆ£:Ξ”β†’Top|\cdot|: \Delta \to \mathbf{Top}.

ExampleRealization of standard simplices

The geometric realization βˆ£Ξ”[n]∣|\Delta[n]| is homeomorphic to βˆ£Ξ”n∣|\Delta^n| (the topological nn-simplex). This follows from the Yoneda lemma and the coend formula:

βˆ£Ξ”[n]∣=∫[m]Hom⁑Δ([m],[n])Γ—βˆ£Ξ”mβˆ£β‰…βˆ£Ξ”n∣.|\Delta[n]| = \int^{[m]} \operatorname{Hom}_\Delta([m], [n]) \times |\Delta^m| \cong |\Delta^n|.

The last isomorphism is the "co-Yoneda lemma" (or density theorem for coends).

ExampleRealization of the boundary

The geometric realization βˆ£βˆ‚Ξ”[n]∣|\partial\Delta[n]| is the boundary of the topological nn-simplex, i.e., the union of all (nβˆ’1)(n-1)-dimensional faces of βˆ£Ξ”n∣|\Delta^n|. This is homeomorphic to the (nβˆ’1)(n-1)-sphere Snβˆ’1S^{n-1}.

For n=2n = 2: βˆ£βˆ‚Ξ”[2]∣|\partial\Delta[2]| is the boundary of a triangle, homeomorphic to S1S^1.

ExampleRealization of the simplicial circle

The simplicial set S1=Ξ”[1]/βˆ‚Ξ”[1]S^1 = \Delta[1]/\partial\Delta[1] has one vertex and one non-degenerate edge (with both endpoints identified). Its geometric realization ∣S1∣|S^1| is obtained by taking the interval [0,1][0,1] and identifying 0∼10 \sim 1, giving the circle S1S^1.

ExampleClassifying spaces via geometric realization

For a group GG, the geometric realization ∣N(BG)∣|N(BG)| is the classifying space BGBG.

For G=Z/nZG = \mathbb{Z}/n\mathbb{Z}: ∣N(B(Z/nZ))βˆ£β‰ƒLn∞|N(B(\mathbb{Z}/n\mathbb{Z}))| \simeq L_n^\infty, the infinite lens space, which has Ο€1=Z/nZ\pi_1 = \mathbb{Z}/n\mathbb{Z} and is a K(Z/nZ,1)K(\mathbb{Z}/n\mathbb{Z}, 1).

For G=S3G = S_3 (symmetric group on 3 elements): the classifying space BS3BS_3 has Ο€1=S3\pi_1 = S_3 and all higher homotopy groups trivial.


Properties of Geometric Realization

Theorem1.3Geometric realization preserves finite products

If XX and YY are simplicial sets, then ∣XΓ—Yβˆ£β‰…βˆ£Xβˆ£Γ—βˆ£Y∣|X \times Y| \cong |X| \times |Y| (where the right side uses the compactly generated product topology), provided at least one of XX or YY 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.

ExampleProduct of simplicial sets

The product Ξ”[1]Γ—Ξ”[1]\Delta[1] \times \Delta[1] is a simplicial set whose geometric realization is the square [0,1]2[0,1]^2. This square is triangulated: the non-degenerate 22-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 Cβˆ—(XΓ—Y)C_*(X \times Y) and Cβˆ—(X)βŠ—Cβˆ—(Y)C_*(X) \otimes C_*(Y) at the chain level.

ExampleRealization preserves coproducts

Geometric realization preserves all colimits (since it is a left adjoint). In particular:

∣XβŠ”Y∣=∣Xβˆ£βŠ”βˆ£Y∣|X \sqcup Y| = |X| \sqcup |Y|

and more generally, if X=colim⁑iXiX = \operatorname{colim}_i X_i, then ∣X∣=colim⁑i∣Xi∣|X| = \operatorname{colim}_i |X_i|.

This means ∣X∣|X| is built by gluing together the geometric realizations of the standard simplices appearing in XX.


The Singular Simplicial Set

Definition1.4Singular simplicial set

The singular simplicial set (or singular complex) of a topological space YY is the simplicial set Sing⁑(Y)\operatorname{Sing}(Y) defined by

Sing⁑(Y)n=Hom⁑Top(βˆ£Ξ”n∣,Y)={continuousΒ mapsΒ βˆ£Ξ”nβˆ£β†’Y}.\operatorname{Sing}(Y)_n = \operatorname{Hom}_{\mathbf{Top}}(|\Delta^n|, Y) = \{\text{continuous maps } |\Delta^n| \to Y\}.

Face maps did_i and degeneracy maps sjs_j are induced by precomposition with the coface maps ∣δi∣|\delta^i| and codegeneracy maps βˆ£Οƒj∣|\sigma^j|.

This defines a functor Sing⁑:Topβ†’sSet\operatorname{Sing}: \mathbf{Top} \to \mathbf{sSet}.

ExampleSingular set of a point

Sing⁑(βˆ—)n\operatorname{Sing}(*)_n is a single element for each nn (the unique map βˆ£Ξ”nβˆ£β†’βˆ—|\Delta^n| \to *). So Sing⁑(βˆ—)=Ξ”[0]\operatorname{Sing}(*) = \Delta[0], the terminal simplicial set.

ExampleSingular set of the interval

Sing⁑([0,1])\operatorname{Sing}([0,1]) is an "enormous" simplicial set: its 11-simplices are all continuous paths [0,1]β†’[0,1][0,1] \to [0,1], its 22-simplices are all continuous maps from a triangle into [0,1][0,1], etc. Despite being large, it is a Kan complex (every horn can be filled) and is weakly equivalent to a point (since [0,1][0,1] is contractible).

ExampleSingular set of the circle

Sing⁑(S1)\operatorname{Sing}(S^1) is a Kan complex with Ο€1=Z\pi_1 = \mathbb{Z} and Ο€n=0\pi_n = 0 for nβ‰₯2n \geq 2. It has uncountably many 11-simplices (all loops and paths in S1S^1), but its homotopy type is completely described by Ο€1=Z\pi_1 = \mathbb{Z}.


The Adjunction

Theorem1.5Geometric realization and singular set adjunction

The geometric realization and singular simplicial set functors form an adjunction

βˆ£β‹…βˆ£βŠ£Sing⁑:sSet⇄Top|\cdot| \dashv \operatorname{Sing}: \mathbf{sSet} \rightleftarrows \mathbf{Top}

with geometric realization as the left adjoint. The unit and counit are:

  • Unit Ξ·X:Xβ†’Sing⁑(∣X∣)\eta_X: X \to \operatorname{Sing}(|X|): sends an nn-simplex x∈Xnx \in X_n to the corresponding continuous map βˆ£Ξ”nβˆ£β†’βˆ£X∣|\Delta^n| \to |X|.
  • Counit Ξ΅Y:∣Sing⁑(Y)βˆ£β†’Y\varepsilon_Y: |\operatorname{Sing}(Y)| \to Y: sends a point (f,t)(f, t) (where f:βˆ£Ξ”nβˆ£β†’Yf: |\Delta^n| \to Y and tβˆˆβˆ£Ξ”n∣t \in |\Delta^n|) to f(t)∈Yf(t) \in Y.
ExampleCounit is a weak equivalence

For any topological space YY, the counit Ξ΅Y:∣Sing⁑(Y)βˆ£β†’Y\varepsilon_Y: |\operatorname{Sing}(Y)| \to Y 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, ∣Sing⁑(Y)∣|\operatorname{Sing}(Y)| is a CW approximation of YY.

This means that from the perspective of homotopy theory, every topological space is "modeled" by a simplicial set.

ExampleThe unit is generally not an isomorphism

The unit Ξ·X:Xβ†’Sing⁑(∣X∣)\eta_X: X \to \operatorname{Sing}(|X|) is generally not an isomorphism (or even a weak equivalence for arbitrary XX). However, when XX is a Kan complex, Ξ·X\eta_X is a weak equivalence.

For example, if X=Ξ”[1]X = \Delta[1], then Sing⁑(βˆ£Ξ”[1]∣)=Sing⁑([0,1])\operatorname{Sing}(|\Delta[1]|) = \operatorname{Sing}([0,1]), which has uncountably many 11-simplices, while Ξ”[1]\Delta[1] has only three (two degenerate, one non-degenerate).


CW Structure

RemarkCW structure of geometric realization

The geometric realization ∣X∣|X| of any simplicial set XX naturally has the structure of a CW complex. The cells are in bijection with the non-degenerate simplices of XX: each non-degenerate nn-simplex contributes an open nn-cell.

The nn-skeleton of the CW complex ∣X∣|X| is precisely ∣skn(X)∣|\mathrm{sk}_n(X)|, the geometric realization of the nn-skeleton of XX.

The attaching maps are determined by the face operators. Specifically, a non-degenerate nn-simplex xx contributes a cell attached via the map βˆ‚βˆ£Ξ”nβˆ£β†’βˆ£sknβˆ’1(X)∣\partial|\Delta^n| \to |\mathrm{sk}_{n-1}(X)| induced by the faces d0(x),…,dn(x)d_0(x), \ldots, d_n(x).


Summary

RemarkKey points

The main ideas of geometric realization are:

  1. Geometric realization ∣X∣|X| turns a simplicial set into a topological space by replacing abstract simplices with topological simplices and gluing.

  2. The singular functor Sing⁑\operatorname{Sing} goes the other way, producing a simplicial set from a topological space.

  3. These form an adjunction βˆ£β‹…βˆ£βŠ£Sing⁑|\cdot| \dashv \operatorname{Sing} that is the bridge between combinatorial and topological homotopy theory.

  4. The counit ∣Sing⁑(Y)βˆ£β†’Y|\operatorname{Sing}(Y)| \to Y is always a weak homotopy equivalence, showing that simplicial sets capture the homotopy theory of spaces.

  5. Geometric realization ∣X∣|X| is naturally a CW complex, with cells corresponding to non-degenerate simplices.