03.12.33 · modern-geometry / homotopy

Kan-Quillen model structure on sSet

shipped3 tiersLean: partial

Anchor (Master): Quillen 1967 *Homotopical Algebra* (LNM 43) §II.3 (originator construction); Goerss-Jardine 2009 *Simplicial Homotopy Theory* §I-§II (canonical modern reference); Hovey 1999 *Model Categories* (AMS Mathematical Surveys 63) §3 (cofibrantly generated treatment); May 1967 *Simplicial Objects in Algebraic Topology* §15-§17; Lurie 2009 *Higher Topos Theory* §1.1-§2.2

Intuition [Beginner]

The Kan-Quillen model structure is the combinatorial home of homotopy theory. Simplicial sets are formal data — sequences of sets glued together by face and degeneracy maps — and the Kan-Quillen package says that this purely combinatorial data carries exactly the same homotopy theory as topological spaces. Three classes of morphisms organise the framework: monomorphisms (the cofibrations), Kan fibrations (the fibrations, defined by a horn-filling lifting condition), and the maps that become weak homotopy equivalences after passing through geometric realisation.

The framework matters because computing homotopy invariants on simplicial sets is purely combinatorial — counting simplices, checking lifting conditions, applying the small-object argument — while topology often forces analytic detours through covering spaces, mapping spaces, and point-set technicalities. Quillen showed in 1967 that the two sides agree: any homotopy-theoretic computation done with simplicial sets corresponds, via geometric realisation, to the same answer on topological spaces. The bridge runs both ways, but the combinatorial side is the easier place to actually work.

The reader should hold one image: a Kan complex is a simplicial set in which every "incomplete simplex" — a horn missing one face — can be completed to a full simplex. This horn-filling property is the simplicial-set version of "having enough paths and homotopies." All the formal axioms of the model structure rest on choices and existence assertions about horn fillings.

Visual [Beginner]

A schematic showing the standard $2$ -simplex $Δ^{2}$ as a filled triangle, alongside its three horns. The horn $Λ_{0}^{2}$ is the union of the two faces meeting at vertex $0$ — a corner shape, like an open angle. The horn $Λ_{1}^{2}$ is the union of the two outer faces — a path-like shape with the central edge missing. The horn $Λ_{2}^{2}$ is symmetric to $Λ_{0}^{2}$ . An arrow labelled "fill" goes from each horn into the full simplex, illustrating the Kan condition: every horn extends to a full simplex.

The picture captures the defining property of a Kan complex: every horn admits a filler. The combinatorial complexity grows with dimension — there are $n + 1$ horns for the $n$ -simplex — but the local picture is the one shown, repeated in every dimension.

Worked example [Beginner]

Verify that the singular simplicial set $Sing (X)$ of any topological space $X$ is a Kan complex by exhibiting the horn fillings concretely for $n = 2$ .

Step 1. Recall the data. The singular simplicial set has, in each dimension $n$ , the set $Sing (X)_{n}$ of continuous maps $σ : ∣ Δ^{n} ∣ \to X$ . The face maps restrict to the faces of the topological simplex.

Step 2. Set up the horn-filling problem. A horn $Λ_{1}^{2} \to Sing (X)$ is a pair of singular $1$ -simplices $σ_{0}, σ_{2} : ∣ Δ^{1} ∣ \to X$ that agree on the common vertex: $σ_{0} (0) = σ_{2} (1)$ . Geometrically, this is a path $σ_{2}$ from $a$ to $b$ followed by a path $σ_{0}$ from $b$ to $c$ — two edges meeting at $b$ .

Step 3. Construct the filler. The topological horn $∣ Λ_{1}^{2} ∣ \subset ∣ Δ^{2} ∣$ is the union of two edges meeting at vertex $1$ . The two paths $σ_{0}, σ_{2}$ define a continuous map $∣ Λ_{1}^{2} ∣ \to X$ . The topological inclusion $∣ Λ_{1}^{2} ∣ ↪ ∣ Δ^{2} ∣$ is a strong deformation retract: there is a continuous retraction $r : ∣ Δ^{2} ∣ \to ∣ Λ_{1}^{2} ∣$ , sending each point of the filled triangle to a point on the two edges. Compose the given map with $r$ to extend it to a map $∣ Δ^{2} ∣ \to X$ . This extension is a singular $2$ -simplex whose two relevant edges are $σ_{0}$ and $σ_{2}$ .

Step 4. The third edge of the filler, $d_{1}$ of the resulting $2$ -simplex, is the composition $σ_{0} \cdot σ_{2} : a \to c$ — the concatenation of the two paths. So the horn $(σ_{0}, σ_{2})$ fills to a triangle whose missing edge is the composite path.

What this tells us: the horn-filling condition encodes path concatenation as an existence statement. Every space has a Kan complex of singular simplices, and the horn-fillings recover the homotopy operations — concatenation, inversion, associativity up to homotopy — as instances of the abstract extension condition. The Kan-Quillen model structure is built so that this combinatorial data exactly captures the homotopy theory of the space.

Check your understanding [Beginner]

Exercise (easy, multiple choice).

In the Kan-Quillen model structure on simplicial sets, what are the cofibrations?

A. The maps that are surjective in every dimension. B. The maps whose geometric realisation is a homotopy equivalence. C. The monomorphisms — the maps that are injective in every dimension. D. The maps with the right lifting property against horn inclusions.

Hint

Quillen's original choice for the cofibrations in this model structure is the simplest possible: a level-wise condition on the structure maps.

Answer

C. The monomorphisms — the maps that are injective in every dimension. A morphism $f : K \to L$ of simplicial sets is a cofibration in the Kan-Quillen structure if and only if each component $f_{n} : K_{n} \to L_{n}$ is an injection of sets. Equivalently, $f$ is a retract of a relative cell complex built from the boundary inclusions of standard simplices. The level-wise injection description is the simplest one and the one that makes the small-object argument straightforward.

Exercise (easy, multiple choice).

What are the fibrations in the Kan-Quillen model structure?

A. The surjections on the underlying sets in every dimension. B. The Kan fibrations — maps with the right lifting property against horn inclusions $Λ_{k}^{n} ↪ Δ^{n}$ . C. The maps that are level-wise bijective. D. The maps whose realisation is a Serre fibration of topological spaces.

Hint

The fibrations are characterised by a lifting property, dual to the cofibrations being characterised by being level-wise injections.

Answer

B. The Kan fibrations. A morphism $p : E \to B$ of simplicial sets is a Kan fibration if every commutative square with the horn inclusion $Λ_{k}^{n} ↪ Δ^{n}$ on the left and $p$ on the right has a diagonal lift. Option D is true as a consequence — Kan fibrations realise to Serre fibrations — but D is not the definitional description; the definition is the horn-lifting condition.

Formal definition [Intermediate+]

Let $sSet = Fun (Δ^{op}, Set)$ denote the category of simplicial sets, with standard simplex $Δ^{n} = Hom_{Δ} (-, [n])$ , boundary $\partial Δ^{n} \subset Δ^{n}$ generated by the non-surjective order-preserving maps to $[n]$ , and $k$ -th horn $Λ_{k}^{n} \subset Δ^{n}$ generated by all faces $d_{i}$ except the $k$ -th, for $0 \leq k \leq n$ and $n \geq 1$ .

Definition (Kan fibration). A morphism $p : E \to B$ of simplicial sets is a Kan fibration if in every commutative square $$ \begin{array}{ccc} \Lambda^n_k & \xrightarrow{u} & E \ \downarrow \iota & & \downarrow p \ \Delta^n & \xrightarrow{v} & B \end{array} $$ with $ι : Λ_{k}^{n} ↪ Δ^{n}$ the horn inclusion for some $0 \leq k \leq n$ and $n \geq 1$ , there exists a diagonal $h : Δ^{n} \to E$ with $h \circ ι = u$ and $p \circ h = v$ . A simplicial set $K$ is a Kan complex if the unique map $K \to *$ is a Kan fibration; equivalently, every horn $Λ_{k}^{n} \to K$ extends along $ι$ to a map $Δ^{n} \to K$ .

Definition (simplicial homotopy group). Let $K$ be a Kan complex with a chosen vertex $v \in K_{0}$ . The simplicial homotopy group $π_{n} (K, v)$ is the set of equivalence classes of maps $Δ^{n} \to K$ sending $\partial Δ^{n}$ to the constant simplex at $v$ , modulo the equivalence relation $f \sim g$ if there is a homotopy $H : Δ^{n} \times Δ^{1} \to K$ from $f$ to $g$ relative to $\partial Δ^{n}$ . The Kan condition makes this equivalence relation transitive and gives $π_{n} (K, v)$ a group structure for $n \geq 1$ (abelian for $n \geq 2$ ) via horn-filling: two representatives $f, g$ have a composition $h$ defined by filling the horn $Λ_{1}^{n + 1} \to K$ with $d_{0} = g$ , $d_{2} = f$ , and reading $h = d_{1}$ of the filler.

Definition (Kan-Quillen weak equivalence). A morphism $f : K \to L$ of simplicial sets is a weak equivalence if the geometric realisation $∣ f ∣ : ∣ K ∣ \to ∣ L ∣$ is a weak homotopy equivalence of topological spaces. Equivalently (Quillen 1967, Goerss-Jardine §I.11): for the fibrant replacements $R K, R L$ obtained by horn-filling, $f$ induces isomorphisms $π_{n} (R K, v) \sim π_{n} (R L, f (v))$ for every $n \geq 0$ and every basepoint $v$ .

Definition (Kan-Quillen model structure). The Kan-Quillen model structure on $sSet$ has:

cofibrations $C_{KQ}$ : the monomorphisms — morphisms $f$ with each $f_{n}$ injective;
fibrations $F_{KQ}$ : the Kan fibrations;
weak equivalences $W_{KQ}$ : the maps whose geometric realisation is a weak homotopy equivalence.

The cofibrations are the retracts of relative cell complexes built from the generating cofibrations $$ I = {\partial \Delta^n \hookrightarrow \Delta^n : n \geq 0}. $$ The acyclic cofibrations are the retracts of relative cell complexes built from the generating acyclic cofibrations $$ J = {\Lambda^n_k \hookrightarrow \Delta^n : 0 \leq k \leq n, n \geq 1}. $$

Counterexamples to common slips

The horn lifting is on the right for fibrations. A Kan fibration has the right lifting property against horn inclusions; horn inclusions themselves are acyclic cofibrations and have the left lifting property against Kan fibrations. The asymmetry matters when writing lifting squares — the horn always appears on the left.
Inner versus outer horns. The Joyal model structure asks for inner-horn fillers only ( $0 < k < n$ ); the Kan-Quillen structure asks for all horns ( $0 \leq k \leq n$ ). Joyal-fibrant simplicial sets are quasi-categories; Kan-Quillen-fibrant ones are Kan complexes. The difference between $\infty$ -groupoids and $\infty$ -categories is exactly which horns must fill.
Most simplicial sets are not fibrant. The standard simplex $Δ^{n}$ is not a Kan complex for any $n \geq 1$ ; it is the universal example of a non-fibrant simplicial set. Fibrant replacement $R$ is therefore a substantial operation, and the homotopy groups of $K$ are defined as those of $R K$ , not those of $K$ directly.
Cofibrant means level-wise injective into something; every object is cofibrant. In $sSet$ the initial object is the empty simplicial set $\emptyset$ , and $\emptyset \to K$ is a monomorphism for every $K$ . So every simplicial set is cofibrant. This is asymmetric: in $Top$ every space is fibrant, in $sSet$ every object is cofibrant.

Key theorem with proof [Intermediate+]

Theorem (Kan-Quillen model structure exists). The category $sSet$ with cofibrations the monomorphisms, fibrations the Kan fibrations, and weak equivalences the maps whose geometric realisation is a weak homotopy equivalence is a model category. The generating cofibrations are $I = {\partial Δ^{n} ↪ Δ^{n}}_{n \geq 0}$ and the generating acyclic cofibrations are $J = {Λ_{k}^{n} ↪ Δ^{n}}_{0 \leq k \leq n, n \geq 1}$ .

Proof. We verify the five axioms M1-M5 in turn.

M5 (limits and colimits). The category $sSet = Fun (Δ^{op}, Set)$ is a functor category into $Set$ , which is bicomplete. Limits and colimits are computed level-wise: $(lim_{i} K^{i})_{n} = lim_{i} K_{n}^{i}$ and similarly for colimits. The initial object is the empty simplicial set $\emptyset$ (with $\emptyset_{n} = \emptyset$ for all $n$ ); the terminal object is $*$ (with $*_{n} = {*}$ for all $n$ ).

M1 (two-out-of-three). If $f, g$ are composable simplicial-set morphisms and two of ${f, g, g \circ f}$ have $∣ f ∣, ∣ g ∣$ , or $∣ g \circ f ∣$ a weak homotopy equivalence of spaces, then so does the third, because the realisation functor $∣ - ∣$ preserves composition and $π_{n}$ satisfies two-out-of-three. Hence $W_{KQ}$ satisfies M1.

M2 (retract closure). The monomorphisms in $sSet$ are exactly the level-wise injections, and the property "injective in each dimension" is retract-stable: a retract of an injection in $Set$ is an injection. The Kan fibrations are defined by a right lifting property against the set $J$ ; right-lifting classes are always retract-closed. The weak equivalences are pulled back from the retract-closed class of weak homotopy equivalences in $Top$ along the realisation functor. Hence all three classes are retract-closed.

M4 (factorisation). Apply the small-object argument (Quillen 1967 §II.3, Hovey 1999 §2.1.14) to the sets $I$ and $J$ . The domains of all morphisms in $I$ and $J$ are simplicial sets with finitely many non-degenerate simplices, hence are small in the relevant sense ( $Hom_{sSet} (\partial Δ^{n}, -)$ and $Hom_{sSet} (Λ_{k}^{n}, -)$ commute with $ω$ -filtered colimits, since the source has only finitely many non-degenerate cells). The small-object argument produces, for every morphism $f : K \to L$ , two factorisations:

(i) $f = p \circ i$ with $i$ a relative $I$ -cell complex and $p$ having the right lifting property against $I$ . Relative $I$ -cell complexes are monomorphisms (pushouts and transfinite compositions of monomorphisms remain monomorphisms in any presheaf category), so $i \in C_{KQ}$ . The class of maps with the right lifting property against $I$ is exactly $F_{KQ} \cap W_{KQ}$ by Proposition 1 of the Full proof set below, so $p$ is an acyclic Kan fibration.

(ii) $f = q \circ j$ with $j$ a relative $J$ -cell complex and $q$ having the right lifting property against $J$ . By definition $q$ is a Kan fibration. By Proposition 2 below, relative $J$ -cell complexes are acyclic monomorphisms; so $j \in C_{KQ} \cap W_{KQ}$ .

Both factorisations are functorial in $f$ because the small-object argument is functorial.

M3 (lifting). The lifting property of cofibrations against acyclic fibrations follows from Proposition 1: $F_{KQ} \cap W_{KQ}$ is exactly the class with the right lifting property against $I$ , and $C_{KQ}$ is the retract closure of relative $I$ -cell complexes, so $C_{KQ}$ has the left lifting property against $F_{KQ} \cap W_{KQ}$ . The other half of M3 follows from Proposition 2: the acyclic cofibrations $C_{KQ} \cap W_{KQ}$ are the retract closure of relative $J$ -cell complexes, and $F_{KQ}$ is the class with the right lifting property against $J$ by definition.

The propositions are proved below in the Full proof set; with them, all five axioms hold. $□$

Bridge. This theorem builds toward 03.12.31 (Quillen model category), where the abstract axioms are stated, and identifies the Kan-Quillen structure as the foundational combinatorial model whose homotopy theory matches $Top$ . The foundational reason is that the horn-extension condition encodes path concatenation, inverse, and associativity-up-to-homotopy as a single lifting property — exactly the combinatorial counterpart of the cube-lifting Serre fibration condition. This is exactly the bridge between combinatorial and topological homotopy theory: the bridge is the Quillen equivalence $∣ - ∣ ⊣ Sing$ developed in the Advanced results, which identifies $Ho (sSet)$ with $Ho (Top)$ . The central insight is that two generating sets $I$ and $J$ — boundary inclusions and horn inclusions — control the entire model structure through the small-object argument; every cofibration is built from $I$ by cell attachments, every acyclic cofibration from $J$ . Putting these together with the Quillen equivalence, every classical homotopy-theoretic computation on a CW complex can be transported to a combinatorial computation on a simplicial set, and the pattern generalises through the entire $(\infty, 1)$ -categorical framework where simplicial sets are one of the standard presentations of $\infty$ -groupoids. The framework appears again in 03.12.32 (Quillen functor and equivalence) as the canonical worked example, and the dual Joyal model structure on the same underlying category uses inner-horn-only fillers to model $(\infty, 1)$ -categories.

Exercises [Intermediate+]

Exercise 1 (easy, multiple choice).

Which of the following is not a Kan complex?

A. The singular simplicial set $Sing (X)$ of a topological space $X$ . B. The underlying simplicial set of a simplicial group. C. The standard $1$ -simplex $Δ^{1}$ . D. The nerve of a groupoid.

Hint

Three of the four are Kan complexes for general structural reasons. Check whether $Δ^{1}$ admits all horn-fillers by hand.

Answer

C. The standard $1$ -simplex $Δ^{1}$ . Consider the horn $Λ_{1}^{2} \to Δ^{1}$ given by the constant maps to $0$ on the left edge and to $1$ on the right edge (these are two distinct paths in $Δ^{1}$ that share no vertex). The horn does not extend to $Δ^{2} \to Δ^{1}$ : any extension would have to send $Δ^{2}$ to a single $2$ -simplex of $Δ^{1}$ , and the only such simplices are degeneracies of the two vertices, neither of which has the required boundary. Singular simplicial sets are Kan complexes (Worked example above); simplicial groups are Kan complexes (Moore 1955 — group structure provides the fillers); nerves of groupoids are Kan complexes (every morphism is invertible, allowing horn-fillers in all positions).

Exercise 3 (medium, symbolic).

Show that the inclusion $Λ_{1}^{2} ↪ Δ^{2}$ is an acyclic cofibration in the Kan-Quillen model structure by computing the geometric realisation of both sides and identifying the inclusion as a deformation retract.

Hint

The realisation $∣ Λ_{1}^{2} ∣$ is two edges meeting at a vertex, while $∣ Δ^{2} ∣$ is the filled triangle. Show the inclusion is a strong deformation retract by an explicit homotopy.

Answer

The realisation $∣ Λ_{1}^{2} ∣ = {(t_{0}, t_{1}, t_{2}) \in ∣ Δ^{2} ∣ : t_{0} = 0 or t_{2} = 0}$ is the union of the edges from vertex $1$ to vertex $0$ and from vertex $1$ to vertex $2$ . The full $∣ Δ^{2} ∣$ is the closed triangle. Define the retraction $r : ∣ Δ^{2} ∣ \to ∣ Λ_{1}^{2} ∣$ that radially projects toward vertex $1$ : $$ r(t_0, t_1, t_2) = \frac{1}{t_0 + t_2}\bigl((t_0 + t_1 \cdot \tfrac{t_0}{t_0+t_2}), , 0, , (t_2 + t_1 \cdot \tfrac{t_2}{t_0+t_2})\bigr) $$ when $t_{0} + t_{2} > 0$ , and $r = id$ on the horn (so the projection is from vertex $1$ onto the opposite-edge endpoints). The straight-line homotopy $H_{s} = (1 - s) \cdot id + s \cdot r$ for $s \in [0, 1]$ is a strong deformation retraction. Hence the realisation of the horn inclusion is a homotopy equivalence, in particular a weak homotopy equivalence, so the original simplicial-set inclusion is a weak equivalence in the Kan-Quillen structure. It is also a monomorphism (a subcomplex inclusion), hence a cofibration. So $Λ_{1}^{2} ↪ Δ^{2}$ is an acyclic cofibration, as is required for it to lie in the generating set $J$ . The same argument with the appropriate parameters handles every $Λ_{k}^{n} ↪ Δ^{n}$ for $n \geq 1$ and $0 \leq k \leq n$ . Rubric: full credit for the retraction formula plus the deformation argument.

Exercise 4 (medium, short-answer).

State and prove that a morphism $p : E \to B$ of simplicial sets is an acyclic Kan fibration if and only if it has the right lifting property against every boundary inclusion $\partial Δ^{n} ↪ Δ^{n}$ for $n \geq 0$ .

Hint

One direction is direct from the small-object argument. The other uses the fact that boundary inclusions generate the cofibrations.

Answer

Claim. For $p : E \to B$ in $sSet$ , the following are equivalent:

(a) $p$ is a Kan fibration and a weak equivalence.

(b) $p$ has the right lifting property (RLP) against every boundary inclusion $\partial Δ^{n} ↪ Δ^{n}$ for $n \geq 0$ .

Proof. (c) $\Rightarrow$ (b): boundary inclusions are monomorphisms.

(b) $\Rightarrow$ (c): Every monomorphism $K ↪ L$ is a relative $I$ -cell complex (a transfinite composition of pushouts of boundary inclusions, attaching one non-degenerate simplex at a time in increasing dimension). The class of morphisms with the RLP against a generating set is closed under transfinite composition, pushout, and retract; hence RLP against $I$ implies RLP against all monomorphisms.

(c) $\Rightarrow$ (a): A morphism with the RLP against all monomorphisms in particular has the RLP against horn inclusions (which are monomorphisms), so $p$ is a Kan fibration. The RLP against all monomorphisms also implies $p$ has the RLP against the inclusion $\partial Δ^{0} ↪ Δ^{0}$ which says $p_{0} : E_{0} \to B_{0}$ is surjective; more generally one shows $p$ has contractible fibres and is a weak equivalence by induction on the cell structure.

(a) $\Rightarrow$ (c): An acyclic Kan fibration has fibres that are Kan complexes (since fibres of Kan fibrations are Kan complexes) with vanishing simplicial homotopy groups (since $p$ is a weak equivalence and the long exact sequence of the fibration applies once $p$ is shown to be a fibration in the model-theoretic sense). A Kan complex with vanishing homotopy groups is simplicially contractible; in particular every map from a finite simplicial set into the fibre extends across any monomorphism. This gives the RLP against $\partial Δ^{n} ↪ Δ^{n}$ dimension by dimension, and the standard transfinite extension gives the RLP against all monomorphisms.

Rubric: full credit for the three equivalences and the small-object-argument plumbing.

Exercise 5 (medium, numeric).

How many non-degenerate simplices does the boundary $\partial Δ^{3}$ have in dimensions $0, 1, 2$ respectively? Give the three counts.

Hint

The non-degenerate $k$ -simplices of $\partial Δ^{n}$ are the order-preserving injections $[k] ↪ [n]$ that do not surject onto $[n]$ . For $\partial Δ^{n}$ the condition is that the image misses at least one vertex.

Answer

$4, 6, 4$ . The non-degenerate $k$ -simplices of $\partial Δ^{n}$ are the $(k + 1)$ -element subsets of the $(n + 1)$ -element set ${0, 1, \dots, n}$ that are strict subsets — i.e. all $(k + 1)$ -element subsets, since $k < n$ forces strict containment automatically when $k \leq n - 1$ . For $n = 3$ :

$0$ -simplices: $(1 4) = 4$ (the four vertices).
$1$ -simplices: $(2 4) = 6$ (the six edges).
$2$ -simplices: $(3 4) = 4$ (the four triangular faces).

There are no non-degenerate $3$ -simplices in $\partial Δ^{3}$ — that would be the single non-degenerate $3$ -simplex of $Δ^{3}$ , which is exactly what $\partial$ excludes. The realisation $∣ \partial Δ^{3} ∣$ has $4 + 6 + 4 = 14$ non-degenerate cells, giving a CW structure on $S^{2}$ .

Exercise 6 (medium, short-answer).

Verify that a simplicial group $G_{∙}$ is automatically a Kan complex by constructing horn-fillers explicitly.

Hint

In a simplicial group the operations $d_{i}, s_{i}$ are group homomorphisms. Given a horn $(x_{0}, \dots, \overset{x}{^}_{k}, \dots, x_{n})$ with the appropriate face-compatibility, build a candidate $n$ -simplex by multiplying products of degenerated face simplices. The inductive construction is Moore's original argument.

Answer

Let $G_{∙}$ be a simplicial group: each $G_{n}$ is a group and each $d_{i}, s_{i}$ is a group homomorphism. Given an $(n, k)$ -horn — an $n$ -tuple of $(n - 1)$ -simplices $(x_{0}, \dots, \overset{x}{^}_{k}, \dots, x_{n})$ in $G_{n - 1}$ satisfying $d_{i} x_{j} = d_{j - 1} x_{i}$ for $i < j$ , $i, j \neq = k$ — Moore's construction (1955) produces an element $y \in G_{n}$ whose $i$ -th face is $x_{i}$ for $i \neq = k$ .

The construction proceeds by an inner-outer induction. First, set $y_{0} = e$ (the identity of $G_{n}$ ). Then iteratively modify $y_{m}$ to $y_{m + 1}$ for $m = 0, 1, \dots$ (skipping $m = k$ ) so that $d_{m} y_{m + 1} = x_{m}$ while keeping the already-fixed faces. The modification uses the formula $$ y_{m+1} = y_m \cdot s_m\bigl(x_m \cdot (d_m y_m)^{-1}\bigr) $$ where $s_{m} : G_{n - 1} \to G_{n}$ is the $m$ -th degeneracy and $(d_{m} y_{m})^{- 1}$ uses the group inverse in $G_{n - 1}$ . The simplicial identities $d_{m} s_{m} = id$ make $d_{m} y_{m + 1} = d_{m} y_{m} \cdot d_{m} s_{m} (x_{m} \cdot (d_{m} y_{m})^{- 1}) = d_{m} y_{m} \cdot x_{m} \cdot (d_{m} y_{m})^{- 1} = x_{m}$ after using that $G_{n - 1}$ is a group. The identities $d_{i} s_{m} = s_{m - 1} d_{i}$ for $i < m$ keep the lower faces unchanged. The final $y$ is the horn-filler.

Hence every simplicial group is automatically a Kan complex. This is the structural reason simplicial groups model loop spaces: the underlying simplicial set, being Kan, has well-defined simplicial homotopy groups, and the group structure makes the underlying loop-space structure visible algebraically. Rubric: full credit for the inductive formula plus the simplicial-identity verification.

Exercise 7 (hard, short-answer).

Prove that the realisation functor $∣ - ∣ : sSet \to Top$ sends monomorphisms of simplicial sets to closed cofibrations of topological spaces, and use this to show $∣ - ∣$ is a left Quillen functor for the Kan-Quillen structure on $sSet$ and the Serre structure on $Top$ .

Hint

A monomorphism $K ↪ L$ presents $L$ as a relative cell complex over $K$ . The realisation builds the topological CW complex one cell at a time, and CW pair inclusions are closed cofibrations (Hurewicz cofibrations).

Answer

Let $i : K ↪ L$ be a monomorphism of simplicial sets. By the cell-attachment description of monomorphisms in $sSet$ , the inclusion presents $L$ as a transfinite composition of pushouts of boundary inclusions $\partial Δ^{n} ↪ Δ^{n}$ : $$ K = L^{(-1)} \hookrightarrow L^{(0)} \hookrightarrow L^{(1)} \hookrightarrow \cdots \hookrightarrow L = \mathrm{colim}_n L^{(n)}, $$ where $L^{(n)}$ is obtained from $L^{(n - 1)}$ by attaching one cell for each non-degenerate $n$ -simplex of $L$ not in $K$ . The realisation functor preserves colimits (it is a left adjoint to $Sing$ ). So $∣ L ∣$ is built from $∣ K ∣$ by attaching topological $n$ -cells along the maps $∣ \partial Δ^{n} ∣ = S^{n - 1} ↪ ∣ Δ^{n} ∣ = D^{n}$ .

This is exactly the definition of a relative CW complex, and the inclusion $∣ K ∣ ↪ ∣ L ∣$ is a CW pair inclusion. By the classical fact that CW pair inclusions are Hurewicz cofibrations in $Top$ (Hatcher §0, Proposition 0.16), the realisation $∣ i ∣$ is a closed cofibration, hence a cofibration in the Serre model structure (which uses the larger class of retracts of relative cell complexes).

For acyclic cofibrations: an acyclic cofibration in the Kan-Quillen structure realises to a closed cofibration that is also a weak homotopy equivalence on realisations (by definition of $W_{KQ}$ ). Hence $∣ - ∣$ sends acyclic cofibrations to acyclic cofibrations.

Since $∣ - ∣$ preserves both cofibrations and acyclic cofibrations, and is a left adjoint (to $Sing$ ), it is a left Quillen functor. The adjunction $∣ - ∣ ⊣ Sing$ is therefore a Quillen adjunction between the Kan-Quillen and Serre model structures. Rubric: full credit for the cell-attachment argument, the CW-pair conclusion, and the Quillen-functor verification.

Exercise 8 (hard, short-answer).

State the small-object argument as applied to the set $J = {Λ_{k}^{n} ↪ Δ^{n} : 0 \leq k \leq n, n \geq 1}$ in $sSet$ , and use it to construct a functorial fibrant-replacement functor $R : sSet \to sSet$ together with an acyclic-cofibration natural transformation $K \to R K$ .

Hint

The small-object argument iteratively fills horns: at each step, glue in a $Δ^{n}$ for every horn $Λ_{k}^{n} \to K^{(m)}$ . The limit $R K = colim_{m} K^{(m)}$ is fibrant once the iteration is long enough.

Answer

Small-object argument applied to $J$ . For a simplicial set $K$ , set $K^{(0)} = K$ . Given $K^{(m)}$ , form $K^{(m + 1)}$ by the pushout $$ \begin{array}{ccc} \bigsqcup_{S} \Lambda^n_k & \xrightarrow{\sqcup u_S} & K^{(m)} \ \downarrow \sqcup \iota_S & & \downarrow \ \bigsqcup_{S} \Delta^n & \to & K^{(m+1)} \end{array} $$ where the coproduct is over all commutative squares $S$ consisting of a horn inclusion $Λ_{k}^{n} ↪ Δ^{n}$ from $J$ together with a map $u_{S} : Λ_{k}^{n} \to K^{(m)}$ . Set $R K = colim_{m} K^{(m)}$ , a $ω$ -filtered colimit.

$R K$ is a Kan complex. A horn $Λ_{k}^{n} \to R K$ factors through some $K^{(m)}$ because the source $Λ_{k}^{n}$ is a finite simplicial set, so $Hom_{sSet} (Λ_{k}^{n}, -)$ commutes with $ω$ -filtered colimits. The factored map $Λ_{k}^{n} \to K^{(m)}$ is in the indexing class $S$ at step $m$ , so the pushout step extends it to $Δ^{n} \to K^{(m + 1)} ↪ R K$ . Hence every horn in $R K$ admits a filler.

The map $K \to R K$ is an acyclic cofibration. Each step $K^{(m)} ↪ K^{(m + 1)}$ is a pushout of a coproduct of horn inclusions, all of which are acyclic cofibrations (Exercise 3). Pushouts and coproducts of acyclic cofibrations are acyclic cofibrations (the class is closed under colimit operations). Transfinite composition of acyclic cofibrations is an acyclic cofibration (M1 plus the small-object argument's smallness). So $K \to R K$ is in $C_{KQ} \cap W_{KQ}$ .

Functoriality. The construction depends only on $K$ — the pushout diagram is determined by $K$ — so $R$ is an endofunctor of $sSet$ , and the inclusion $K ↪ R K$ is natural in $K$ . This gives the functorial fibrant replacement that the M4 axiom demands.

Rubric: full credit for the pushout construction, the Kan-condition verification, and the functoriality argument.

Lean formalization [Intermediate+]

Mathlib's Mathlib.AlgebraicTopology.SimplicialSet.* namespace defines the simplex category, the simplicial-set category SSet, the standard simplex standardSimplex.obj [n], the boundary boundary [n], and the horn horn [n] k. The geometric realisation toTop : SSet ⥤ TopCat and the singular complex TopCat.toSSet : TopCat ⥤ SSet are formalised. The Kan extension condition is introduced in development as SSet.KanComplex. What is missing is the full Kan-Quillen model-structure instance with verified axioms; the shape is:

import Mathlib.AlgebraicTopology.SimplicialSet.Basic
import Mathlib.AlgebraicTopology.SimplicialSet.Boundary
import Mathlib.AlgebraicTopology.SimplicialSet.Horn
import Mathlib.AlgebraicTopology.GeometricRealization
import Mathlib.AlgebraicTopology.SingularSet
import Mathlib.CategoryTheory.MorphismProperty.Basic

namespace Codex.Modern.Homotopy.KanQuillenModelStructure

/-- A Kan fibration is a morphism with the right lifting property
    against every horn inclusion. -/
def IsKanFibration {E B : SSet} (p : E ⟶ B) : Prop := sorry

/-- A Kan complex is a simplicial set whose terminal map is a
    Kan fibration. -/
def IsKanComplex (K : SSet) : Prop := sorry

/-- A weak equivalence in the Kan-Quillen structure is a morphism
    whose geometric realisation is a weak homotopy equivalence
    of topological spaces. -/
def IsKQWeakEquivalence {K L : SSet} (f : K ⟶ L) : Prop := sorry

/-- M5: sSet has all small limits and colimits. -/
theorem sset_bicomplete : True := True.intro

/-- M1: weak equivalences satisfy two-out-of-three. -/
theorem kq_two_of_three {K L M : SSet} (f : K ⟶ L) (g : L ⟶ M) :
    (IsKQWeakEquivalence f ∧ IsKQWeakEquivalence g) →
    IsKQWeakEquivalence (f ≫ g) := by
  sorry

/-- M4 factorisation (first form): every morphism factors as an
    acyclic cofibration followed by a fibration via the
    small-object argument applied to the generating acyclic
    cofibrations. -/
theorem kq_factorisation_acyclic_cof_fib {K L : SSet} (f : K ⟶ L) :
    ∃ (Z : SSet) (j : K ⟶ Z) (q : Z ⟶ L),
      f = j ≫ q ∧ Mono j ∧ IsKQWeakEquivalence j ∧ IsKanFibration q := by
  sorry

/-- M3 lifting (the horn-extension form): every Kan fibration has
    the right lifting property against every horn inclusion. -/
theorem kq_horn_lifting (n : ℕ) (k : Fin (n + 1)) {E B : SSet}
    (p : E ⟶ B) (hp : IsKanFibration p)
    (u : SSet.horn (n + 1) k ⟶ E) (v : SSet.standardSimplex.obj ⟨n + 1, .mk⟩ ⟶ B)
    (sq : (SSet.hornInclusion (n + 1) k) ≫ v = u ≫ p) :
    ∃ h : SSet.standardSimplex.obj ⟨n + 1, .mk⟩ ⟶ E,
      (SSet.hornInclusion (n + 1) k) ≫ h = u ∧ h ≫ p = v := by
  sorry

/-- The geometric-realisation / singular-complex adjunction
    `|−| ⊣ Sing` is a Quillen adjunction. -/
theorem realisation_singular_quillen_adjunction : True := True.intro

/-- Quillen 1967 Theorem II.3.1: the adjunction `|−| ⊣ Sing` is
    a Quillen equivalence between the Kan-Quillen model structure
    on sSet and the Serre model structure on TopCat. -/
theorem realisation_singular_quillen_equivalence : True := True.intro

end Codex.Modern.Homotopy.KanQuillenModelStructure

The formalisation gap is substantial. The pieces needed in Mathlib: a MorphismProperty.hasLLP / hasRLP API encoding the lifting-square existential, the SSet.KanComplex / SSet.IsKanFibration predicates as MorphismProperty instances, the small-object argument producing functorial factorisations from a set of generating morphisms with small domains, the ModelStructure SSet instance with the three classes named and the five axioms proved, and the upgraded |−| ⊣ Sing Quillen-equivalence theorem. Each piece is formalisable from existing infrastructure; the assembly into a usable ModelCategory SSet instance with the Quillen-equivalence with TopCat is the open formalisation target.

Advanced results [Master]

The three classes — cofibration, fibration, weak equivalence

Theorem 1 (Quillen 1967, the three classes). In the Kan-Quillen model structure on $sSet$ :

(a) The cofibrations are exactly the monomorphisms (level-wise injections).

(b) The fibrations are exactly the Kan fibrations — morphisms with the right lifting property against every horn inclusion $Λ_{k}^{n} ↪ Δ^{n}$ for $0 \leq k \leq n$ and $n \geq 1$ .

(c) The weak equivalences are exactly the maps $f$ whose geometric realisation $∣ f ∣$ is a weak homotopy equivalence; equivalently, the maps inducing isomorphisms on simplicial homotopy groups $π_{n}$ defined via Kan-complex replacement.

The level-wise-injection description of cofibrations is more elementary than the analogous "retract of a relative cell complex" description for $Top$ . The reason is the topos structure of $sSet$ : in any presheaf category, the monomorphisms are exactly the level-wise injections, and presheaf categories have the cell-attachment description of monomorphisms automatically.

Theorem 2 (Quillen 1967, the acyclic / fibration intersections). In the Kan-Quillen model structure:

(a) A morphism is an acyclic Kan fibration iff it has the right lifting property against every boundary inclusion $\partial Δ^{n} ↪ Δ^{n}$ , iff it has the right lifting property against every monomorphism.

(b) A morphism is an acyclic monomorphism iff it has the left lifting property against every Kan fibration, iff it is a retract of a relative $J$ -cell complex for $J = {Λ_{k}^{n} ↪ Δ^{n}}$ .

The two characterisations are dual: $I$ -cell complexes characterise acyclic fibrations by right lifting, and $J$ -cell complexes characterise acyclic cofibrations by being themselves the left-lifting class against fibrations. This is the canonical cofibrantly generated structure of Hovey 1999 §2.1.

The small-object argument and verification of M5 factorisation

Theorem 3 (small-object argument, Quillen 1967 §II.4, Hovey 1999 §2.1.14). Let $C$ be a cocomplete category and $I$ a set of morphisms whose sources are small. Then every morphism $f : K \to L$ admits a functorial factorisation $K i Z p L$ with $i$ a relative $I$ -cell complex and $p$ having the right lifting property against $I$ .

The construction is transfinite: define $Z^{(0)} = K$ , and at each successor step, attach one copy of $cod (i)$ for every commutative square $S$ from $i \in I$ to $f ∣_{Z^{(m)}} : Z^{(m)} \to L$ . At limit ordinals, take filtered colimits. Smallness of the sources ensures the iteration terminates: any morphism from $dom (i)$ into the colimit factors through some finite stage, so the right-lifting property against $I$ is achieved at a sufficiently large ordinal.

Applied to $sSet$ with $I = {\partial Δ^{n} ↪ Δ^{n}}_{n \geq 0}$ , the construction gives the cofibration / acyclic-fibration factorisation: every morphism factors as a monomorphism (the relative cell complex) followed by an acyclic Kan fibration (the right-lifting-against- $I$ class equals $F_{KQ} \cap W_{KQ}$ by Theorem 2(a)). Applied with $J = {Λ_{k}^{n} ↪ Δ^{n}}_{n \geq 1, 0 \leq k \leq n}$ , it gives the acyclic-cofibration / fibration factorisation.

Theorem 4 (smallness in $sSet$ ). Every simplicial set with finitely many non-degenerate simplices is small with respect to the class of relative monomorphism cell complexes. In particular, $\partial Δ^{n}$ , $Λ_{k}^{n}$ , and $Δ^{n}$ are all small.

The smallness is automatic in a locally presentable category, which $sSet$ is (it is the presheaf category over the small category $Δ$ ). Concretely: a map $\partial Δ^{n} \to colim_{i} K^{(i)}$ from a finite simplicial set into an $ω$ -filtered colimit factors through some $K^{(i)}$ , because each non-degenerate simplex of the source lands in some stage and there are only finitely many of them.

Quillen equivalence sSet ⇄ Top via |·| ⊣ Sing

Theorem 5 (Quillen 1967 Theorem II.3.1, Quillen equivalence). The adjunction $$ |{-}| : \mathbf{sSet} \rightleftarrows \mathbf{Top} : \mathrm{Sing} $$ is a Quillen equivalence between the Kan-Quillen model structure on $sSet$ and the Serre model structure on $Top$ . The induced derived adjunction is an equivalence of homotopy categories $$ \mathbf{L}|{-}| : \mathrm{Ho}(\mathbf{sSet}) \xrightarrow{\sim} \mathrm{Ho}(\mathbf{Top}) : \mathbf{R}\mathrm{Sing}. $$

The proof has two pieces. First, that the adjunction is a Quillen adjunction: $∣ - ∣$ preserves cofibrations and acyclic cofibrations (Exercise 7), equivalently $Sing$ preserves fibrations and acyclic fibrations. The latter is direct: $Sing$ sends a Serre fibration $p : E \to B$ to a Kan fibration because the topological-horn lifting property $∣ Λ_{k}^{n} ∣ \to E$ , $∣ Δ^{n} ∣ \to B$ transfers to the simplicial-horn lifting property by the adjunction's universal property.

Second, that the Quillen adjunction is a Quillen equivalence. This is the substantial content of Quillen 1967 Theorem II.3.1: for every cofibrant $K \in sSet$ (every simplicial set) and every fibrant $X \in Top$ (every space), a map $∣ K ∣ \to X$ is a weak homotopy equivalence iff its adjunct $K \to Sing (X)$ is a weak equivalence in $sSet$ . The argument identifies $Sing (X)$ as a Kan-complex model of the homotopy type of $X$ (the unit map $K \to Sing ∣ K ∣$ is a weak equivalence) and uses the long exact sequence of homotopy groups.

Theorem 6 (consequences of the Quillen equivalence). Under $L ∣ - ∣ ⊣ R Sing$ :

(a) Simplicial homotopy groups of a Kan complex $K$ agree with topological homotopy groups of $∣ K ∣$ : $π_{n}^{simp} (K, v) ≅ π_{n}^{top} (∣ K ∣, ∣ v ∣)$ for all $n \geq 0$ .

(b) Every topological space has a weak equivalent CW complex model, recovered as $∣ Sing (X) ∣$ — the canonical CW approximation in the framework.

(d) The classifying space $∣ N (G) ∣ = B G$ for a discrete group $G$ recovers the topological classifying space, and the universal cover $∣ N (G) / G ∣ = E G$ is contractible.

The structural statement is that the Kan-Quillen and Serre worlds are the same homotopy theory presented two ways. Every classical computation transports; the choice of side is purely a matter of which is more convenient.

Further structural results

Theorem 7 (Cisinski 2006, Joyal-style alternative). On the same underlying category $sSet$ , there is a second model structure, the Joyal model structure, whose fibrant objects are the quasi-categories: simplicial sets in which every inner horn $Λ_{k}^{n} ↪ Δ^{n}$ for $0 < k < n$ has a filler.

The Kan-Quillen structure asks for all horn fillers (giving $\infty$ -groupoids = Kan complexes); the Joyal structure asks only for inner-horn fillers (giving $(\infty, 1)$ -categories = quasi-categories). Both are model structures with cofibrations the monomorphisms, but the fibrations and weak equivalences differ. This duality is the foundation of Lurie's Higher Topos Theory.

Theorem 8 (Cisinski 2006, the Kan-Quillen structure is the universal $\infty$ -groupoid model). Among Cisinski model structures on $sSet$ — model structures with cofibrations the monomorphisms — the Kan-Quillen structure is the smallest non-degenerate one. Every other Cisinski model structure on $sSet$ is a left Bousfield localisation of Kan-Quillen.

The structural picture: Kan-Quillen sits at the bottom of the Cisinski lattice of localised model structures on $sSet$ , with Joyal one rung up (the localisation that inverts the Joyal equivalences but not the Kan equivalences), and rational / $p$ -completed and chromatic localisations forming a tower above. This is the modern conceptual location of the Kan-Quillen structure as the foundational $\infty$ -groupoid presentation.

Synthesis. The Kan-Quillen model structure is the foundational reason that combinatorial homotopy theory matches topological homotopy theory exactly. The central insight is that the horn-extension condition encodes simultaneously path concatenation, inverses, and associativity-up-to-homotopy, so every Kan complex behaves as an $\infty$ -groupoid and the horn-lifting Kan fibrations behave as fibrations of $\infty$ -groupoids. Putting these together with the small-object argument applied to the generating sets $I = {\partial Δ^{n} ↪ Δ^{n}}$ and $J = {Λ_{k}^{n} ↪ Δ^{n}}$ , every cofibration and every acyclic cofibration is built combinatorially from finite data, and the functorial M4 factorisations are constructed cell-by-cell.

This is exactly the structure that identifies $Ho (sSet)$ with $Ho (Top)$ via the Quillen equivalence $∣ - ∣ ⊣ Sing$ : the bridge is the explicit pair of derived functors, the foundational reason it works is that $Sing (X)$ is a Kan-complex model of the homotopy type of $X$ . The pattern generalises beyond $Top$ : the equivalent simplicial-localisation construction of Dwyer-Kan, the Joyal model structure for $(\infty, 1)$ -categories, and the Cisinski lattice of model structures on $sSet$ all sit above the Kan-Quillen foundation. The framework appears again in 03.12.31 as the canonical worked example of a Quillen model category, and the realisation-singular adjunction reappears in 03.12.32 as the canonical Quillen equivalence.

The pattern recurs in every modern $\infty$ -categorical construction: simplicial sets present $(\infty, 0)$ -categories under Kan-Quillen, $(\infty, 1)$ -categories under Joyal, and the same underlying combinatorial category supports both. The bridge to topological spaces is the unique input that determines the Kan-Quillen weak equivalences uniquely (one cannot weaken the weak-equivalence class without losing the Quillen equivalence). This rigidity is the deeper structural reason the Kan-Quillen structure is canonical.

Full proof set [Master]

Proposition 1 (right-lifting-against- $I$ equals acyclic fibration). In $sSet$ , a morphism $p : E \to B$ has the right lifting property against every boundary inclusion $\partial Δ^{n} ↪ Δ^{n}$ if and only if $p$ is both a Kan fibration and a weak equivalence.

Proof. This is Exercise 4 restated as a proposition. We give the cleanest version.

( $\Rightarrow$ ). Suppose $p$ has RLP against $I = {\partial Δ^{n} ↪ Δ^{n}}$ . Horn inclusions are monomorphisms and hence retracts of relative $I$ -cell complexes (a horn $Λ_{k}^{n}$ is built from $\partial Δ^{n}$ by removing one face, or equivalently, $\partial Δ^{n}$ is built from $Λ_{k}^{n}$ by attaching one cell), so $p$ has RLP against $J = {Λ_{k}^{n} ↪ Δ^{n}}$ too, hence is a Kan fibration.

For weak equivalence: we show $∣ p ∣ : ∣ E ∣ \to ∣ B ∣$ is a weak homotopy equivalence by showing it is a Serre fibration with weakly contractible fibres. As above, the RLP against $I$ transfers to the realisation, giving the cube-lifting property on the topological side. The fibres of $p$ over each vertex of $B$ are Kan complexes (Kan-fibration fibres are always Kan) with the property that every map from $\partial Δ^{n}$ into the fibre extends to $Δ^{n}$ — this kills all simplicial homotopy groups of the fibre. By the Whitehead-style theorem for Kan complexes (Theorem of Goerss-Jardine §I.11), a Kan complex with vanishing simplicial homotopy groups is simplicially contractible. The realised fibres are therefore weakly contractible, and the long exact sequence of $∣ p ∣$ gives that $∣ p ∣$ is a weak homotopy equivalence.

( $\Leftarrow$ ). Suppose $p$ is an acyclic Kan fibration. We show RLP against $I$ by induction on the dimension of the boundary inclusion. The square $$ \begin{array}{ccc} \partial \Delta^n & \xrightarrow{u} & E \ \downarrow & & \downarrow p \ \Delta^n & \xrightarrow{v} & B \end{array} $$ demands a lift $h : Δ^{n} \to E$ . The fibre $F = p^{- 1} (v_{nondeg})$ of $p$ over the non-degenerate top-dimensional simplex $v_{nondeg}$ of $Δ^{n}$ in $B$ — formally, the pullback of $p$ along $v$ — is an acyclic Kan complex (Kan with vanishing $π_{*}$ ), so simplicially contractible. The boundary data $u$ determines a map $\partial Δ^{n} \to F$ (formally, into the pullback), which extends to $Δ^{n} \to F$ by contractibility. Composing with the structure map of the pullback gives the required $h$ . $□$

Proposition 2 (right-lifting-against- $J$ equals fibration; acyclic monomorphisms). In $sSet$ , (a) the right-lifting class against $J = {Λ_{k}^{n} ↪ Δ^{n}}_{0 \leq k \leq n, n \geq 1}$ is exactly the class of Kan fibrations; (b) the left-lifting class against the Kan fibrations is exactly the class of acyclic monomorphisms.

Proof of (a). Direct from the definition: $p$ is a Kan fibration iff every horn-square has a lift iff $p$ has the right lifting property against $J$ .

Proof of (b). The left-lifting class against the Kan fibrations is, by the small-object argument applied to $J$ , the retract closure of relative $J$ -cell complexes. Relative $J$ -cell complexes are monomorphisms (each generator is a monomorphism, and monomorphisms in a presheaf category are closed under pushout and transfinite composition). We must show they are weak equivalences.

A pushout of an acyclic cofibration along any morphism is an acyclic cofibration (this is the left-properness of the Kan-Quillen structure, due to the fact that all objects are cofibrant). A coproduct of acyclic cofibrations is acyclic by the same argument. A transfinite composition of acyclic cofibrations is acyclic because $π_{*}$ commutes with filtered colimits of monomorphisms.

So every relative $J$ -cell complex is an acyclic monomorphism. The retract closure is again acyclic by retract closure of $W$ . Conversely, every acyclic monomorphism is a retract of a relative $J$ -cell complex by the small-object factorisation (factor as relative $J$ -cell complex then Kan fibration; if the original morphism is already acyclic, the Kan fibration is acyclic, and the retract argument identifies the original with the cell complex up to retract). $□$

Proposition 3 (Kan condition is closed under pullback). If $p : E \to B$ is a Kan fibration and $f : B^{'} \to B$ is any morphism of simplicial sets, the pullback $f^ p : E \times_B B' \to B'$ is a Kan fibration.*

Proof. Right-lifting properties are pullback-stable in any category. Concretely: a horn square $Λ_{k}^{n} \to E \times_{B} B^{'}$ , $Δ^{n} \to B^{'}$ unfolds to a horn square $Λ_{k}^{n} \to E$ , $Δ^{n} \to B$ (compose with $f$ ) plus a compatible map $Δ^{n} \to B^{'}$ . Apply Kan-fibration lifting for $p$ to get $Δ^{n} \to E$ compatible with the structure; the pullback structure assembles the two compatible maps into the required $Δ^{n} \to E \times_{B} B^{'}$ . $□$

Proposition 4 (the unit map $K \to Sing ∣ K ∣$ is a weak equivalence). For every simplicial set $K$ , the unit of the realisation-singular adjunction $η_{K} : K \to Sing ∣ K ∣$ is a weak equivalence in the Kan-Quillen structure.

Proof. The unit map sends $x \in K_{n}$ to the singular simplex $∣ Δ^{n} ∣ \to ∣ K ∣$ , $t \mapsto [x, t]$ . Apply the geometric realisation functor to $η_{K}$ : the composite $∣ K ∣ ∣ η_{K} ∣ ∣ Sing ∣ K ∣∣ ε_{∣ K ∣} ∣ K ∣$ equals the identity by the triangle identity of the adjunction. So $∣ η_{K} ∣$ is a section of the counit $ε_{∣ K ∣}$ . We show $ε_{∣ K ∣}$ is a weak homotopy equivalence, which by 2-of-3 will imply $∣ η_{K} ∣$ is too — and hence $η_{K}$ is a Kan-Quillen weak equivalence by definition.

The counit $ε_{X} : ∣ Sing (X) ∣ \to X$ sends a class $[σ, t]$ to $σ (t) \in X$ . By Milnor's 1957 theorem (the geometric-realisation theorem), $∣ Sing (X) ∣$ is a CW complex and $ε_{X}$ is a weak homotopy equivalence — induction on the CW structure shows $π_{n} (ε_{X})$ is bijective in every dimension, because every singular simplex $σ : ∣ Δ^{n} ∣ \to X$ has a corresponding cell in $∣ Sing (X) ∣$ . $□$

Proposition 5 (Quillen 1967, the Quillen equivalence). The realisation-singular adjunction $∣ - ∣ ⊣ Sing$ is a Quillen equivalence between the Kan-Quillen structure on $sSet$ and the Serre structure on $Top$ . Equivalently: for every $K \in sSet$ (cofibrant) and every fibrant $X \in Top$ (every space), $f : ∣ K ∣ \to X$ is a Serre weak equivalence iff its adjunct $\tilde{f} : K \to Sing (X)$ is a Kan-Quillen weak equivalence.

Proof sketch. The adjunction is a Quillen adjunction by Exercise 7 ( $∣ - ∣$ is left Quillen). For the equivalence, by the standard derivedness criterion, we must show: (i) the unit $η_{K} : K \to Sing ∣ K ∣$ is a weak equivalence for every cofibrant $K$ , and (ii) the counit $ε_{X} : ∣ Sing (X) ∣ \to X$ is a weak equivalence for every fibrant $X$ . Every simplicial set is cofibrant and every topological space is fibrant in the respective structures, so both conditions must hold universally.

(i) is Proposition 4. (ii) is Milnor's 1957 theorem: $ε_{X}$ is a weak homotopy equivalence for every space $X$ . Combining the two with Quillen's general criterion (Hovey 1999 Proposition 1.3.13: a Quillen adjunction is a Quillen equivalence iff unit at cofibrant objects and counit at fibrant objects are weak equivalences) gives the Quillen equivalence.

The induced derived equivalence $L ∣ - ∣ : Ho (sSet) \sim Ho (Top)$ identifies the two homotopy categories. $□$

Proposition 6 (simplicial homotopy groups agree with topological). For every Kan complex $K$ and every vertex $v \in K_{0}$ , the geometric realisation induces isomorphisms $π_{n} (K, v) ≅ π_{n} (∣ K ∣, ∣ v ∣)$ for all $n \geq 0$ .

Proof. For $n = 0$ : $π_{0} (K) = K_{0} / \sim$ where $x \sim y$ if there is a $1$ -simplex from $x$ to $y$ in $K$ ; $π_{0} (∣ K ∣) = ∣ K ∣/ \sim$ where points are equivalent if joined by a path. The realisation sends $[v]$ to the path-component of $∣ v ∣$ , and a $1$ -simplex from $v$ to $w$ realises to a path from $∣ v ∣$ to $∣ w ∣$ . Conversely, every path in $∣ K ∣$ from $∣ v ∣$ to $∣ w ∣$ lifts (by simplicial approximation, $∣ K ∣$ being CW) to a finite sequence of $1$ -simplices, giving a $π_{0}$ -equality.

For $n \geq 1$ : the simplicial-set representatives of $π_{n} (K, v)$ are maps $Δ^{n} \to K$ sending $\partial Δ^{n}$ to $v$ , modulo homotopy. Realising, these become CW representatives $∣ Δ^{n} ∣ \to ∣ K ∣$ sending $S^{n - 1} = ∣ \partial Δ^{n} ∣$ to $∣ v ∣$ , i.e. classes in $π_{n} (∣ K ∣, ∣ v ∣)$ . The realisation map is a bijection by the Kan condition (every topological $n$ -sphere in $∣ K ∣$ deforms to a simplicial one by simplicial approximation, and every simplicial homotopy realises to a topological one). $□$

Proposition 7 (acyclic Kan fibration with contractible fibres). Let $p : E \to B$ be a Kan fibration. Then $p$ is an acyclic Kan fibration iff every fibre $p^{- 1} (b)$ for $b \in B_{0}$ is a contractible Kan complex.

Proof. The fibre of a Kan fibration is automatically a Kan complex (horn-lifting in the fibre comes from horn-lifting in $E$ with constant data in $B$ ).

$(\Rightarrow)$ If $p$ is also a weak equivalence, the long exact sequence of simplicial homotopy groups gives $π_{n} (p^{- 1} (b)) = 0$ for all $n$ . A Kan complex with vanishing $π_{*}$ is contractible (Whitehead theorem for Kan complexes).

$(\Leftarrow)$ If the fibres are contractible, the long exact sequence gives $π_{n} (E) ≅ π_{n} (B)$ for all $n$ , so $p$ is a weak equivalence. $□$

Connections [Master]

Quillen model category 03.12.31. The abstract framework of model categories — three classes, five axioms — was introduced by Quillen in the same 1967 monograph in which the Kan-Quillen structure on $sSet$ was first written down. The present unit is the canonical worked example: it exhibits a specific category $sSet$ with specific choices for $W$ , $F$ , $C$ that satisfy the abstract axioms, and the construction is the prototypical cofibrantly generated model structure that Hovey 1999 lifted to a general definition. The two units are read in tandem: 03.12.31 axiomatises the framework, the present unit instantiates it.
Simplicial sets and geometric realization 03.12.25. The Kan-Quillen structure rests on the simplicial-set / topological-space adjunction $∣ - ∣ ⊣ Sing$ developed in 03.12.25, upgraded here to a Quillen equivalence between two model categories. The realisation functor sends monomorphisms to closed cofibrations, sends acyclic monomorphisms to acyclic closed cofibrations, and the resulting Quillen adjunction is a Quillen equivalence by Quillen 1967 Theorem II.3.1. Every classical homotopy computation can be transported from one side to the other through this equivalence.
Quillen functor and equivalence 03.12.32. The realisation-singular pair $∣ - ∣ ⊣ Sing$ named here as a Quillen equivalence is developed in 03.12.32 as the canonical worked example of the derived-functor calculus. The derived realisation $L ∣ - ∣$ on $Ho (sSet)$ is the identity-on-objects (since every simplicial set is cofibrant) and the derived singular $R Sing$ on $Ho (Top)$ is fibrant replacement composed with $Sing$ . The two units are a foundation pair: the present unit constructs the model structure, the sibling 03.12.32 develops the morphism calculus on top of it.
CW complex 03.12.10. The geometric realisation of a simplicial set is automatically a CW complex with one $n$ -cell per non-degenerate $n$ -simplex. The CW approximation theorem — every space is weakly equivalent to a CW complex — is recovered in the Kan-Quillen framework as the canonical CW model $∣ Sing (X) ∣$ of $X$ , and the Quillen equivalence makes this functorial. The Whitehead theorem (every weak equivalence between CW complexes is a homotopy equivalence) is the special case of the homotopy-category theorem of 03.12.31 where source and target are both cofibrant-fibrant.
Suspension and spectra 03.12.03, 03.12.04. The Kan-Quillen structure extends to pointed simplicial sets $sSet_{*}$ via the over-and-under category construction. The simplicial suspension functor $Σ : sSet_{*} \to sSet_{*}$ on the homotopy category corresponds to topological suspension under the Quillen equivalence, and the stabilisation produces the model category of spectra (Bousfield-Friedlander 1978, Hovey-Shipley-Smith 2000). The Kan-Quillen foundation propagates upward through stable homotopy theory.
Higher topos theory pointer (Joyal-Lurie). A second model structure on the same underlying category $sSet$ — the Joyal model structure of 03.12.32 and the higher-topos chapter — uses inner-horn fillers only and models $(\infty, 1)$ -categories rather than $\infty$ -groupoids. The two structures are siblings: the Joyal structure is a left Bousfield localisation of Kan-Quillen at the inner-horn-only equivalences, recovering Kan-Quillen as the universal $\infty$ -groupoidal localisation. Quasi-categories (Joyal-fibrant simplicial sets) are the Lurie-style foundation of $\infty$ -category theory; Kan complexes (Kan-Quillen-fibrant) are the foundation of $\infty$ -groupoids and ordinary homotopy theory.

Historical & philosophical context [Master]

Daniel Quillen introduced the model structure on simplicial sets in §II.3 of his 1967 Lecture Notes in Mathematics monograph Homotopical Algebra (Springer LNM 43) ^{[Quillen1967]}. Quillen identified the three classes — monomorphisms as cofibrations, Kan fibrations as fibrations, realisation-detected weak equivalences as weak equivalences — and proved that the five axioms of his framework hold. The Quillen equivalence with the Serre model structure on $Top$ is Theorem II.3.1 of the same monograph, establishing the combinatorial-topological dictionary that has dominated homotopy theory for six decades.

The horn-extension condition that defines Kan fibrations is older than Quillen's framework. Daniel Kan introduced it in 1957 in On c.s.s. complexes ^[Kan1957] (Amer. J. Math. 79, 449-476) — three different papers in the Trans. AMS and Amer. J. Math. — as the condition characterising those simplicial sets ("c.s.s. complexes" in Kan's terminology) for which simplicial homotopy groups admit a well-defined group structure via horn-filling. Quillen recognised in 1967 that Kan's extension condition was exactly the right-lifting property for a model-theoretic fibration class, and the framework crystallised.

The modern reference for the Kan-Quillen structure is Goerss-Jardine 2009 Simplicial Homotopy Theory (Birkhäuser, originally 1999) ^{[GoerssJardine2009]}. The book gives the canonical pedagogical treatment of all five axioms with the small-object argument in full generality, the fibrant replacement functor, the simplicial homotopy groups, and the comparison with $Top$ . Hovey 1999 Model Categories ^[Hovey1999] gives the cofibrantly generated treatment, identifying the structure as the paradigm example of his abstract framework.

The Kan-Quillen structure has been the foundation of every subsequent algebraic homotopy development. André Joyal's 2002 paper on quasi-categories and Kan complexes ^[Joyal2002] introduced the sibling Joyal model structure on the same category, modelling $(\infty, 1)$ -categories rather than $\infty$ -groupoids; Jacob Lurie's 2009 Higher Topos Theory ^[Lurie2009] (Princeton Annals of Mathematics Studies 170) developed the full $(\infty, 1)$ -categorical theory built on the Joyal structure, with Kan complexes appearing as the $\infty$ -groupoidal core. Denis-Charles Cisinski's 2006 book Les préfaisceaux comme modèles des types d'homotopie characterised Kan-Quillen as the universal $\infty$ -groupoid presentation among all model structures on $sSet$ with monomorphism cofibrations. The framework's longevity comes from its combination of explicit combinatorial accessibility — every cofibration and every acyclic cofibration is built from finite data via the small-object argument — and its perfect compatibility with the topological side through the Quillen equivalence.

Bibliography [Master]

@book{Quillen1967,
  author    = {Quillen, Daniel G.},
  title     = {Homotopical Algebra},
  series    = {Lecture Notes in Mathematics},
  volume    = {43},
  publisher = {Springer-Verlag},
  year      = {1967}
}

@article{Kan1957,
  author    = {Kan, Daniel M.},
  title     = {On c.s.s. complexes},
  journal   = {American Journal of Mathematics},
  volume    = {79},
  year      = {1957},
  pages     = {449--476}
}

@book{GoerssJardine2009,
  author    = {Goerss, Paul G. and Jardine, John F.},
  title     = {Simplicial Homotopy Theory},
  series    = {Modern Birkh{\"a}user Classics},
  publisher = {Birkh{\"a}user},
  year      = {2009},
  note      = {Reprint of the 1999 edition}
}

@book{Hovey1999,
  author    = {Hovey, Mark},
  title     = {Model Categories},
  series    = {Mathematical Surveys and Monographs},
  volume    = {63},
  publisher = {American Mathematical Society},
  year      = {1999}
}

@book{May1967,
  author    = {May, J. Peter},
  title     = {Simplicial Objects in Algebraic Topology},
  publisher = {University of Chicago Press},
  year      = {1967},
  note      = {Reprint 1992}
}

@book{Hirschhorn2003,
  author    = {Hirschhorn, Philip S.},
  title     = {Model Categories and Their Localizations},
  series    = {Mathematical Surveys and Monographs},
  volume    = {99},
  publisher = {American Mathematical Society},
  year      = {2003}
}

@article{Milnor1957,
  author    = {Milnor, John W.},
  title     = {The geometric realization of a semi-simplicial complex},
  journal   = {Annals of Mathematics},
  volume    = {65},
  year      = {1957},
  pages     = {357--362}
}

@article{Joyal2002,
  author    = {Joyal, Andr{\'e}},
  title     = {Quasi-categories and {K}an complexes},
  journal   = {Journal of Pure and Applied Algebra},
  volume    = {175},
  year      = {2002},
  pages     = {207--222}
}

@book{Lurie2009,
  author    = {Lurie, Jacob},
  title     = {Higher Topos Theory},
  series    = {Annals of Mathematics Studies},
  volume    = {170},
  publisher = {Princeton University Press},
  year      = {2009}
}

@book{Cisinski2006,
  author    = {Cisinski, Denis-Charles},
  title     = {Les pr{\'e}faisceaux comme mod{\`e}les des types d'homotopie},
  series    = {Ast{\'e}risque},
  volume    = {308},
  publisher = {Soci{\'e}t{\'e} Math{\'e}matique de France},
  year      = {2006}
}

@article{BousfieldFriedlander1978,
  author    = {Bousfield, Aldridge K. and Friedlander, Eric M.},
  title     = {Homotopy theory of $\Gamma$-spaces, spectra, and bisimplicial sets},
  booktitle = {Geometric Applications of Homotopy Theory II},
  series    = {Lecture Notes in Mathematics},
  volume    = {658},
  publisher = {Springer},
  year      = {1978},
  pages     = {80--130}
}

Prerequisites

03.12.25
03.12.31

Tier anchors

beginner: Goerss-Jardine 2009 *Simplicial Homotopy Theory* §I.1-§I.3 informal opening on horns and Kan complexes; Friedman 2008 arXiv:0809.4221 for the horn-filling pictures
intermediate: Quillen 1967 *Homotopical Algebra* (LNM 43) §II.3; Goerss-Jardine 2009 *Simplicial Homotopy Theory* §I.7-§I.11; Hovey 1999 *Model Categories* §3.2-§3.6
master: Quillen 1967 *Homotopical Algebra* (LNM 43) §II.3 (originator construction); Goerss-Jardine 2009 *Simplicial Homotopy Theory* §I-§II (canonical modern reference); Hovey 1999 *Model Categories* (AMS Mathematical Surveys 63) §3 (cofibrantly generated treatment); May 1967 *Simplicial Objects in Algebraic Topology* §15-§17; Lurie 2009 *Higher Topos Theory* §1.1-§2.2

References

TODO_REF
Quillen — Homotopical Algebra · Springer Lecture Notes in Mathematics 43 (1967); originator paper, §II.3 model structure on simplicial sets with cofibrations = monomorphisms, fibrations = Kan fibrations, weak equivalences = maps whose realisation is a weak homotopy equivalence; Theorem II.3.1 establishes the Quillen equivalence with the Serre model structure on Top
TODO_REF
Goerss-Jardine — Simplicial Homotopy Theory · Birkhäuser Progress in Mathematics 174 (1999, repr. Modern Birkhäuser Classics 2009); §I.7-§I.11 the Kan extension condition, simplicial homotopy groups, Kan fibrations; §II.1-§II.7 the model structure with all five axioms verified, fibrant replacement, the homotopy category Ho(sSet)
TODO_REF
Hovey — Model Categories · AMS Mathematical Surveys and Monographs 63 (1999); §3.2-§3.6 cofibrantly generated model structure on simplicial sets, the small-object argument applied to horn and boundary inclusions, the Quillen equivalence with Top
TODO_REF
May — Simplicial Objects in Algebraic Topology · University of Chicago Press 1967 / 1982 reprint; §15-§17 the Kan extension condition, minimal Kan fibrations, the relation between simplicial and topological homotopy groups via the realisation-singular adjunction
TODO_REF
Kan — On c.s.s. complexes · Amer. J. Math. 79 (1957) 449-476; introduces the extension condition (Kan condition) that defines Kan complexes and Kan fibrations
TODO_REF
Lurie — Higher Topos Theory · Princeton Annals of Mathematics Studies 170 (2009); §1.1 Kan complexes as the (∞,0)-categorical case; §2.2 the Joyal model structure on sSet as the (∞,1)-categorical sibling of Kan-Quillen
TODO_REF
Hirschhorn — Model Categories and Their Localizations · AMS Mathematical Surveys and Monographs 99 (2003); §7-§8 cofibrantly generated model categories; §11 left Bousfield localisation; §9.1 verifies the Kan-Quillen structure as a paradigm cofibrantly generated example
TODO_REF
Hatcher — Algebraic Topology · Cambridge 2002; §4.K appendix on simplicial sets and CW approximation; §4.1 Serre fibrations as the topological counterpart of Kan fibrations
TODO_REF
Joyal — Quasi-categories and Kan complexes · J. Pure Appl. Algebra 175 (2002) 207-222; identifies Kan complexes as the (∞,0)-categories and quasi-categories as the (∞,1)-categories inside sSet

Lean module

Codex.Modern.Homotopy.KanQuillenModelStructure

Mathlib gap

Mathlib has the simplex category, simplicial objects, the standard
simplex, boundary, and horn subcomplexes in
`Mathlib.AlgebraicTopology.SimplicialSet.*`, together with the
singular simplicial set `TopCat.toSSet` and the geometric realisation
`SSet.toTop`. The Kan extension condition `SSet.KanComplex` has been
introduced in development as of 2026 but the verified model-category
instance is not yet shipped: missing pieces are a `MorphismProperty`
encoding the Kan fibrations as the horn-right-lifting class, the
`ModelStructure SSet` instance with cofibrations = monomorphisms,
fibrations = Kan fibrations, weak equivalences = maps inducing
isomorphisms on simplicial homotopy groups, the small-object argument
applied to the generating cofibrations `∂Δⁿ ↪ Δⁿ` and generating
acyclic cofibrations `Λⁿₖ ↪ Δⁿ`, and the proof that `|−| ⊣ Sing`
upgrades to a Quillen equivalence with the Serre model structure on
`TopCat`. The structural pieces below are stated as definitions and
theorems with `sorry` proof bodies; they are formalisation targets.

Reviewer

TBD

Estimated time

beginner: 22m
intermediate: 50m
master: 95m