ConceptComplete

\infty-Groupoid (Kan complex)

An \infty-groupoid is an \infty-category in which every morphism is an equivalence. In the quasi-categorical framework, \infty-groupoids are precisely Kan complexes. The homotopy hypothesis, attributed to Grothendieck, asserts that \infty-groupoids model homotopy types of topological spaces. This identification is one of the central pillars of higher category theory.


Definition and Characterization

Definition3.1Infinity-groupoid

An \infty-groupoid is a quasi-category C\mathcal{C} in which every morphism is an equivalence. Equivalently, it is a simplicial set satisfying the Kan condition: every horn ΛknC\Lambda^n_k \to \mathcal{C} (for all n1n \geq 1, 0kn0 \leq k \leq n) has a filler.

An \infty-groupoid is the same thing as a Kan complex. The \infty-category of all (small) \infty-groupoids is denoted S\mathcal{S} and is called the \infty-category of spaces.

ExampleOrdinary groupoids as infinity-groupoids

An ordinary groupoid G\mathcal{G} gives an \infty-groupoid N(G)N(\mathcal{G}) via the nerve. This is a Kan complex with unique horn fillers (it is 11-truncated: πn=0\pi_n = 0 for n2n \geq 2).

The fundamental groupoid Π1(X)\Pi_1(X) of a topological space XX is an ordinary groupoid whose nerve is the 11-truncation of Sing(X)\operatorname{Sing}(X).

ExampleClassifying spaces as infinity-groupoids

For a group GG, the nerve N(BG)N(BG) is a Kan complex with π0=\pi_0 = *, π1=G\pi_1 = G, and πn=0\pi_n = 0 for n2n \geq 2. This is the Eilenberg--MacLane space K(G,1)K(G, 1).

More generally, for a topological group GG, the classifying space BGBG is an \infty-groupoid with π1=π0(G)\pi_1 = \pi_0(G), π2=π1(G)\pi_2 = \pi_1(G), etc.

ExampleEilenberg--MacLane spaces

For an abelian group AA and n1n \geq 1, the Eilenberg--MacLane space K(A,n)K(A, n) is an \infty-groupoid with πk=0\pi_k = 0 for knk \neq n and πn=A\pi_n = A. It represents cohomology:

[X,K(A,n)]=Hn(X;A)[X, K(A, n)] = H^n(X; A)

These can be constructed as simplicial abelian groups via the Dold--Kan correspondence applied to the chain complex A[n]A[-n] concentrated in degree nn.


The Homotopy Hypothesis

Theorem3.2Homotopy hypothesis

The \infty-category of \infty-groupoids is equivalent to the \infty-category of homotopy types (spaces). More precisely, the Quillen equivalence Sing|\cdot| \dashv \operatorname{Sing} between sSetKQ\mathbf{sSet}_{\mathrm{KQ}} and Top\mathbf{Top} identifies Kan complexes with spaces up to weak homotopy equivalence.

At the level of homotopy categories: Ho(Kan)Ho(Top)\operatorname{Ho}(\operatorname{Kan}) \simeq \operatorname{Ho}(\mathbf{Top}).

ExampleEvidence for the homotopy hypothesis

The homotopy hypothesis holds in the following precise senses:

  1. Quillen equivalence: sSetKQQTop\mathbf{sSet}_{\mathrm{KQ}} \simeq_Q \mathbf{Top} via Sing|\cdot| \dashv \operatorname{Sing}.

  2. Homotopy groups agree: πn(X)πn(X)\pi_n(X) \cong \pi_n(|X|) for Kan complexes XX.

  3. Mapping spaces agree: MapKan(X,Y)MapTop(X,Y)\operatorname{Map}_{\mathrm{Kan}}(X, Y) \simeq \operatorname{Map}_{\mathrm{Top}}(|X|, |Y|).

  4. Classification: Homotopy types are classified by homotopy groups and Postnikov invariants, which are the same data as \infty-groupoids with their higher structure.


The Infinity-category of Spaces

ExampleThe infinity-category S

The \infty-category S\mathcal{S} of spaces is the central object of higher category theory. It can be constructed as:

  1. The coherent nerve NΔ(Kan)N_\Delta(\operatorname{Kan}) of the simplicial category of Kan complexes.
  2. The \infty-categorical localization N(sSet)[W1]N(\mathbf{sSet})[W^{-1}].
  3. The quasi-category of objects in the Kan--Quillen model structure.

S\mathcal{S} plays the role in \infty-category theory that Set\mathbf{Set} plays in ordinary category theory. It is the "base" for enrichment, and functors CS\mathcal{C} \to \mathcal{S} are the \infty-categorical analogues of presheaves.

ExampleS is the free presentable infinity-category

The \infty-category S\mathcal{S} is the free colimit-completion of the terminal \infty-category {}\{*\}:

SInd({})Fun({}op,S)\mathcal{S} \simeq \operatorname{Ind}(\{*\}) \simeq \operatorname{Fun}(\{*\}^{\mathrm{op}}, \mathcal{S})

This says that S\mathcal{S} is the "free presentable \infty-category on one generator." The generator is the point S* \in \mathcal{S}, and every space is a colimit of copies of *.

ExamplePresheaves valued in spaces

A presheaf on a quasi-category C\mathcal{C} valued in S\mathcal{S} is a functor F:CopSF: \mathcal{C}^{\mathrm{op}} \to \mathcal{S}. The \infty-category of presheaves is Fun(Cop,S)\operatorname{Fun}(\mathcal{C}^{\mathrm{op}}, \mathcal{S}), which is always a presentable \infty-category.

For C=N(C0)\mathcal{C} = N(\mathcal{C}_0) an ordinary category, presheaves valued in S\mathcal{S} generalize classical presheaves (valued in Set\mathbf{Set}): instead of sets of sections, we have spaces of sections.


Truncation and Connectivity

Definition3.3n-truncated and n-connected

A space (Kan complex) XX is nn-truncated if πk(X,x)=0\pi_k(X, x) = 0 for all k>nk > n and all basepoints xx. It is nn-connected if πk(X,x)=0\pi_k(X, x) = 0 for all knk \leq n.

  • (1)(-1)-truncated spaces are either empty or contractible (π01\pi_0 \leq 1, all higher trivial).
  • 00-truncated spaces are homotopy equivalent to discrete sets.
  • 11-truncated spaces are K(π1,1)K(\pi_1, 1)'s (plus π0\pi_0 components).
  • 00-connected spaces are simply connected (π0=\pi_0 = *).
ExampleExamples of truncation levels
  • S0={0,1}S^0 = \{0, 1\} is 00-truncated (discrete, πk=0\pi_k = 0 for k1k \geq 1).
  • K(Z,1)=S1K(\mathbb{Z}, 1) = S^1 is 11-truncated (π1=Z\pi_1 = \mathbb{Z}, πk=0\pi_k = 0 for k2k \geq 2).
  • S2S^2 is NOT finitely truncated: π3(S2)=Z\pi_3(S^2) = \mathbb{Z} (Hopf), etc.
  • K(Z,2)=CPK(\mathbb{Z}, 2) = \mathbb{C}P^\infty is 22-truncated.

The nn-truncation functor τn:SSn\tau_{\leq n}: \mathcal{S} \to \mathcal{S}_{\leq n} is left adjoint to the inclusion SnS\mathcal{S}_{\leq n} \hookrightarrow \mathcal{S}.

ExamplePostnikov towers

Every space XX has a Postnikov tower:

τ2Xτ1Xτ0X\cdots \to \tau_{\leq 2} X \to \tau_{\leq 1} X \to \tau_{\leq 0} X

with XlimnτnXX \simeq \lim_n \tau_{\leq n} X. Each stage τnXτn1X\tau_{\leq n} X \to \tau_{\leq n-1} X is a principal K(πn(X),n)K(\pi_n(X), n)-fibration, classified by the kk-invariant knHn+1(τn1X;πn(X))k_n \in H^{n+1}(\tau_{\leq n-1} X; \pi_n(X)).

The Postnikov tower completely determines the homotopy type of XX (given the kk-invariants). This is the \infty-groupoid analogue of describing a group by generators and relations.

ExampleLooping and delooping

For a pointed \infty-groupoid (X,x)(X, x), the loop space ΩxX=MapX(x,x)\Omega_x X = \operatorname{Map}_X(x, x) is an \infty-groupoid with πn(ΩX)=πn+1(X)\pi_n(\Omega X) = \pi_{n+1}(X).

Conversely, for an \infty-group (a group object in S\mathcal{S}, i.e., a connected Kan complex GG with group-like structure), the delooping BGBG satisfies ΩBGG\Omega BG \simeq G. This gives a bijection between \infty-groups and pointed connected spaces.

For nn-fold loop spaces: Ωn(K(A,n))A\Omega^n(K(A, n)) \simeq A (discrete), reflecting the identification K(A,n)=BnAK(A, n) = B^n A.


Summary

RemarkKey points

\infty-groupoids are the foundation of homotopy theory:

  1. \infty-groupoids = Kan complexes = spaces (the homotopy hypothesis).

  2. The \infty-category S\mathcal{S} of spaces is the \infty-categorical analogue of Set\mathbf{Set}.

  3. Presheaves valued in S\mathcal{S} generalize classical presheaves, replacing sets with spaces.

  4. Spaces are classified by homotopy groups and Postnikov invariants (kk-invariants).

  5. Truncation relates \infty-groupoids to ordinary groupoids (11-truncated) and sets (00-truncated).