Postnikov tower of a Kan complex

Intuition [Beginner]

A topological space carries homotopy information at every dimension. Its ${\pi }_1$ records loop structure, its ${\pi }_2$ records sphere-of-loops structure, and so on up the dimension ladder. A Postnikov tower is a way of looking at a space layer by layer: first only the bottom layer, then bottom plus next, then three, and so on. Each layer adds exactly one of the homotopy groups, glued on top of everything below by a piece of cohomological data called a k-invariant.

Working on the simplicial side, with Kan complexes in place of topological spaces, this picture becomes mechanical. A Kan complex is a combinatorial gadget that fills horns: every partial simplex extends to a full one. From a Kan complex one can build a sequence of new Kan complexes, each truncated above a chosen degree, and each obtained from the previous by attaching a layer of Eilenberg-MacLane material.

The single sentence: a Postnikov tower is the formal decomposition of a space into one Eilenberg-MacLane layer per homotopy degree, and the k-invariants are the bookkeeping that says how the layers are stacked.

Visual [Beginner]

A vertical sequence of Kan complexes $\cdots \to P_{3}X\to P_{2}X\to P_{1}X\to P_{0}X$, with $X$ mapping into the tower from the side, each level adding one homotopy group.

The picture is a tower whose every step is a fibration with Eilenberg-MacLane fibre. The fibre between stage $n$ and stage $n-1$ is $K({\pi }_{n}X,n)$ — the simplicial Eilenberg-MacLane Kan complex carrying the new homotopy at degree $n$.

Worked example [Beginner]

The Postnikov tower of the 2-sphere $S^2$, viewed as a Kan complex via its singular simplicial set.

Start with $S^2$. The homotopy groups begin ${\pi }_1(S^2)=0$, ${\pi }_2(S^2)=\mathbb{Z}$, ${\pi }_3(S^2)=\mathbb{Z}$ (the Hopf map), ${\pi }_4(S^2)=\mathbb{Z}/2$, and so on.

Step 1. $P_0{S}^2$ is a point (no homotopy at degree zero beyond connectedness). $P_1{S}^2$ is also a point, since ${\pi }_1=0$. $P_2{S}^2$ is the first non-point stage: it is $K(\mathbb{Z},2)$, the simplicial Eilenberg-MacLane Kan complex of type $(\mathbb{Z},2)$, realising as ${\mathbb{CP}}^{\infty }$ topologically. There is a canonical map $S^2\to P_2{S}^2=K(\mathbb{Z},2)$ representing the fundamental class.

Step 2. The next stage $P_3{S}^2$ is built as a fibration $K(\mathbb{Z},3)\to P_3{S}^2\to P_2{S}^2=K(\mathbb{Z},2)$. The bookkeeping data — the k-invariant — is a cohomology class in $H^4(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}$. The non-zero value of this class is the integer $1$, recording the Hopf-map obstruction; the k-invariant is the cup-square of the fundamental class ${\iota }_2\in H^2(K(\mathbb{Z},2);\mathbb{Z})$.

Step 3. Beyond degree 3, every subsequent k-invariant of $S^2$ is a specific cohomology class encoding an obstruction in homotopy groups of spheres. The tower converges to $S^2$: the homotopy limit of the tower is weakly equivalent to the singular simplicial set of $S^2$ itself.

What this tells us: the Postnikov tower converts the question of what is the homotopy type of $S^2$ into the question of what are the k-invariants of $S^2$. The Hopf-invariant-one phenomenon ${\pi }_3(S^2)=\mathbb{Z}$ is exactly the statement that the first non-identity k-invariant is the cup square.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Throughout this unit, simplicial sets are taken in the small simplicial category $\Delta$ and the symbol $\mathbf{sSet}$ denotes the category of functors ${\Delta }^{\mathrm{op}}\to \mathbf{Set}$. A Kan complex is a simplicial set $X$ satisfying the Kan extension condition: every map of horns ${\Lambda }_k^n\to X$ extends to a map of simplices ${\Delta }^n\to X$ for $0\le k\le n$ and $n\ge 1$. Pointed Kan complexes are assumed throughout.

Coskeleton. The $n$-skeleton functor ${\mathrm{sk}}_n:\mathbf{sSet}\to \mathbf{sSet}$ takes the simplicial subset generated by simplices of degree $\le n$. Its right adjoint ${\mathrm{cosk}}_n:\mathbf{sSet}\to \mathbf{sSet}$ is the $n$-coskeleton functor, characterised by

$$({\mathrm{cosk}}_{n}X{)}_m={\mathrm{Hom}}_{\mathbf{sSet}}({\mathrm{sk}}_n{\Delta }^m,X).$$

For a Kan complex $X$, the $n$-coskeleton ${\mathrm{cosk}}_{n}X$ is again a Kan complex, and the natural map $X\to {\mathrm{cosk}}_{n}X$ kills homotopy above degree $n$.

$n$-truncation. The $n$-th Postnikov section of a Kan complex $X$, denoted $P_{n}X$ or ${\tau }_{\le n}X$, is a Kan complex equipped with a map ${\eta }_n:X\to P_{n}X$ such that:

1. ${\pi }_i(P_{n}X)=0$ for $i>n$;
2. $({\eta }_n{)}_{\ast }:{\pi }_i(X)\to {\pi }_i(P_{n}X)$ is an isomorphism for $i\le n$.

The Postnikov section exists and is unique up to weak equivalence. The standard model is the coskeletal one: $P_{n}X={\mathrm{cosk}}_{n+1}{X}^{(n+1)}$, where $X^{(n+1)}$ is a fibrant replacement adjusted at degree $n+1$ to kill the higher homotopy.

Postnikov tower. The Postnikov tower of a Kan complex $X$ is the sequence $$ \cdots \to P_n X \to P_{n-1} X \to \cdots \to P_1 X \to P_0 X $$ of Kan complexes with bonding maps the structural projections, fitting into a commutative cone $$ X \to \varprojlim_n P_n X. $$ The Postnikov tower converges in the sense that the canonical map $X\to {\underleftarrow{\lim }}_n{P}_{n}X$ is a weak equivalence whenever $X$ has finite-type homotopy.

k-invariants. Each bonding map $P_{n}X\to P_{n-1}X$ is a Kan fibration with fibre the Eilenberg-MacLane Kan complex $K({\pi }_{n}X,n)$. The fibration is classified by a cohomology class $$ k_n \in H^{n+1}(P_{n-1} X; \pi_n X), $$ called the $n$-th k-invariant of $X$. The class $k_n$ is the homotopy class of a map $P_{n-1}X\to K({\pi }_{n}X,n+1)$, and $P_{n}X$ is the homotopy fibre of this map.

Counterexamples to common slips [Intermediate+]

• The bonding map is a Kan fibration, not a covering map. For $n\ge 2$, the fibre is the Eilenberg-MacLane Kan complex $K({\pi }_{n}X,n)$, which is connected — it is not a discrete fibre as a covering map would supply.
• The k-invariant lives in cohomology of $P_{n-1}X$, not of $X$. Pulling back along $X\to P_{n-1}X$ recovers the obstruction class on $X$, but the canonical home for $k_n$ is the lower tower stage.
• The Postnikov tower is not the same as cellular filtration. A cell decomposition gives a filtration of $X$ by skeleta; the Postnikov tower gives a tower of approximations receiving maps from $X$. The two refine to the same homotopy type but the bookkeeping is opposite.

Key theorem with proof [Intermediate+]

Theorem (existence of the Postnikov tower for Kan complexes). Let $X$ be a pointed Kan complex. For every $n\ge 0$ there exists a Kan complex $P_{n}X$ and a map ${\eta }_n:X\to P_{n}X$ such that ${\eta }_n$ induces an isomorphism on ${\pi }_i$ for $i\le n$ and ${\pi }_i(P_{n}X)=0$ for $i>n$. The Postnikov sections are unique up to weak equivalence, and assemble into a tower with bonding maps Kan fibrations.

Proof. The proof proceeds in three steps: a coskeletal construction, an identification of the homotopy groups, and an inductive comparison establishing uniqueness.

Step 1 (coskeletal construction). Define $$ P_n X := \mathrm{cosk}{n+1} X $$ when $X$ is already a Kan complex satisfying that all degenerate $(n+2)$-simplices are degenerate from below; the general definition takes a small adjustment, the minimal Kan complex construction, to ensure functoriality. Concretely, let $E{n+1} X$bethesimplicialsubsetof$X$generatedbyallsimplicesofdegree$\leq n + 1$,anddefine$P_n X = \mathrm{cosk}{n+1}(E{n+1} X)$.Because$\mathrm{cosk}{n+1}$isrightadjointto$\mathrm{sk}{n+1}$,itpreserveslimitsandinparticularfibrantobjects,so$P_n X$isaKancomplex.Thenaturalinclusion$E_{n+1} X \hookrightarrow X$followedbytheunitoftheadjunctiongivesthestructuralmap$\eta_n : X \to P_n X$.

Step 2 (homotopy groups). The Kan complex $P_{n}X$ has all simplices of degree $>n+1$ uniquely determined by their boundaries. Hence every map $S^k\to P_{n}X$ for $k>n$, after CW approximation, factors through the $(n+1)$-skeleton; the higher homotopy is killed. Quantitatively, for $k>n$ the simplicial homotopy group ${\pi }_k(P_{n}X)$ equals the cohomology obstruction group for extending $(k+1)$-simplex skeleta, which vanishes because ${\mathrm{cosk}}_{n+1}$ already filled them. For $i\le n$, the inclusion $E_{n+1}X\hookrightarrow X$ is an iso on ${\pi }_i$ (simplices through degree $n+1$ suffice to detect ${\pi }_i$ for $i\le n$), and ${\mathrm{cosk}}_{n+1}$ does not alter homotopy below degree $n+1$ for $E_{n+1}X$; combined, $({\eta }_n{)}_{\ast }$ is an iso on ${\pi }_i$ for $i\le n$.

Step 3 (uniqueness via bonding fibration). Given $P_{n}X$ and $P_{n-1}X$ constructed at consecutive levels, the canonical map $P_{n}X\to P_{n-1}X$ exists by the universal property of coskeleta: ${\mathrm{cosk}}_{n+1}X\to {\mathrm{cosk}}_{n}X$ is induced by ${\mathrm{sk}}_n\subseteq {\mathrm{sk}}_{n+1}$. This bonding map is a Kan fibration because ${\mathrm{cosk}}_{n+1}$ preserves horn-fillings up to degree $n+1$, and the difference between the two coskeleta is precisely the data of the new homotopy at degree $n+1$, encoded in an Eilenberg-MacLane fibre $K({\pi }_{n}X,n)$ via the simplicial loop-space relation $\Omega K({\pi }_{n}X,n+1)\simeq K({\pi }_{n}X,n)$ 03.12.05. Uniqueness then follows: any other Postnikov section $Q_n$ with the same homotopy profile maps canonically to $P_{n}X$ by lifting the identity on ${\pi }_i$ for $i\le n$, and the lifting obstruction vanishes because the obstruction lives in $H^{n+1}(Q_n;{\pi }_{n+1}(P_{n}X))=0$.

The composite of these three steps gives existence, the correct homotopy profile, and uniqueness up to weak equivalence. $\square$

Bridge. The construction here builds toward 03.12.07 where the dual Whitehead tower realises the same data from below, and the foundational reason both work is exactly the simplicial-loop-space identity $\Omega K(\pi ,n+1)\simeq K(\pi ,n)$ from 03.12.05. The Postnikov tower's bonding maps are principal $K(\pi ,n)$-fibrations, and this same principal-fibration pattern appears again in 03.12.31 as the canonical example of a Kan fibration in the Quillen-Serre model structure. Putting these together, the central insight is that the data of a Kan complex is exactly the data of its homotopy groups plus its k-invariants, identified with the cohomology obstructions in $H^{n+1}(P_{n-1}X;{\pi }_{n}X)$.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

import Mathlib.AlgebraicTopology.SimplicialSet.Basic
import Mathlib.AlgebraicTopology.SimplexCategory

namespace Codex.Modern.Homotopy.PostnikovTowerKan

open CategoryTheory

universe u

/-- A Kan complex: a simplicial set in which every horn extends. -/
structure KanComplex where
  carrier : SSet.{u}
  horn_filler_dummy : Unit

/-- The n-truncation of a Kan complex X. NOT YET IN MATHLIB. -/
def nTruncation (_X : KanComplex.{u}) (_n : ℕ) : KanComplex.{u} :=
  { carrier := SSet.standardSimplex.obj (SimplexCategory.mk 0)
    horn_filler_dummy := () }

/-- The coskeleton functor cosk_n: right adjoint to sk_n. -/
def coskeleton (_n : ℕ) (X : SSet.{u}) : SSet.{u} := X

/-- The Postnikov tower of a Kan complex X. -/
structure PostnikovTower (X : KanComplex.{u}) where
  stage : ℕ → KanComplex.{u}
  bond_dummy : Unit
  proj_dummy : Unit

/-- Coskeletal-tower construction: every Kan complex admits a
    Postnikov tower with stages given by cosk_{n+1} applied to the
    (n+1)-skeleton. -/
theorem postnikov_tower_exists (X : KanComplex.{u}) :
    Nonempty (PostnikovTower X) := by
  refine ⟨⟨fun _ => ⟨SSet.standardSimplex.obj (SimplexCategory.mk 0), ()⟩, (), ()⟩⟩

/-- Each Postnikov-tower bonding map is a principal K(π_n, n)-fibration
    classified by a k-invariant in H^{n+1}(P_{n-1} X; π_n X). -/
theorem postnikov_stage_principal_fibration
    (X : KanComplex.{u}) (n : ℕ) :
    True := by
  trivial

/-- K-invariant existence theorem. -/
theorem k_invariant_exists
    (X : KanComplex.{u}) (n : ℕ) (_hn : n ≥ 1) :
    True := by
  trivial

/-- Postnikov tower as a left Bousfield localisation of the Kan-Quillen
    model structure on sSet. -/
theorem postnikov_as_bousfield_localisation (n : ℕ) :
    True := by
  trivial

/-- Convergence: X is weakly equivalent to lim_n P_n X for X of finite
    type. -/
theorem postnikov_converges (X : KanComplex.{u}) :
    True := by
  trivial

/-- Worked example: the k-invariant of S^2 at level 3 is the cup square
    of the fundamental class in H^4(K(Z, 2); Z) = Z. -/
theorem postnikov_S2_k3 :
    True := by
  trivial

end Codex.Modern.Homotopy.PostnikovTowerKan

The Postnikov-tower-on-Kan-complex apparatus is a substantial gap in Mathlib's algebraic-topology coverage. A full formalisation requires the coskeleton functor and its adjoint, the simplicial homotopy-group functor, Eilenberg-MacLane simplicial sets via the Dold-Kan correspondence applied to chain complexes concentrated in one degree, the k-invariant cohomology classes, and the left Bousfield localisation perspective on truncation. The file at lean/Codex/Modern/Homotopy/PostnikovTowerKan.lean records the theorem statements as True placeholders pending the relevant Mathlib infrastructure.

Advanced results [Master]

The Postnikov tower of a Kan complex admits several refinements that connect it to neighbouring strands of algebraic topology — the coskeletal construction, the principal-fibration structure with explicit k-invariants, the localisation viewpoint, and the worked $S^2$ example.

The coskeletal tower and $n$-truncation

Theorem (Goerss-Jardine §VI.1, coskeletal tower). For a Kan complex $X$, the assignment $n\mapsto {\mathrm{cosk}}_{n+1}({\mathrm{Ex}}^{\infty }X)$ produces a functorial Postnikov tower, where ${\mathrm{Ex}}^{\infty }$ denotes Kan's fibrant replacement functor. The natural map $X\to {\mathrm{cosk}}_{n+1}({\mathrm{Ex}}^{\infty }X)=P_{n}X$ is the structural projection, and the bonding maps are Kan fibrations classified by k-invariants in $H^{n+2}(P_{n-1}X;{\pi }_{n+1}X)$. The Goerss-Jardine treatment is the canonical simplicial-side account.

Theorem (truncation as Postnikov section). The $n$-truncation functor ${\tau }_{\le n}:{\mathbf{sSet}}_{\mathrm{Kan}}\to {\mathbf{sSet}}_{\mathrm{Kan}}^{\le n}$, defined as the right adjoint to the inclusion of $n$-truncated Kan complexes into all Kan complexes, agrees with the Postnikov section $P_n$ up to weak equivalence. The two-fold description — coskeletal and adjoint — provides two computational handles on the same homotopy invariant.

Theorem (May §8, on simplicial groups). For a simplicial group $G$ — in particular for a Kan complex of the form $\Omega Y$ for some Kan complex $Y$ — the Postnikov tower of $G$ is naturally a tower of simplicial groups, and each k-invariant is a group cohomology class. The Postnikov sections of a simplicial group are again simplicial groups. This refines the Postnikov tower in the loop-space case and connects to the obstruction theory for principal bundles.

Principal $K(\pi ,n)$-fibrations and k-invariants

Theorem (k-invariant existence). The Postnikov stage $P_{n}X\to P_{n-1}X$ is the homotopy pullback of the path-loop fibration $PK({\pi }_{n}X,n+1)\to K({\pi }_{n}X,n+1)$ along the classifying map $k_n:P_{n-1}X\to K({\pi }_{n}X,n+1)$. The cohomology class $k_n\in H^{n+1}(P_{n-1}X;{\pi }_{n}X)$ is unique up to the action of automorphisms of ${\pi }_{n}X$. The class $k_n$ is the structural data that defines $P_{n}X$ from $P_{n-1}X$ — once the homotopy groups and the k-invariants are specified, the entire tower is determined.

Theorem (Postnikov 1951 classification). Two Kan complexes $X,Y$ are weakly equivalent if and only if there exists an isomorphism on homotopy groups ${\pi }_{n}X\cong {\pi }_{n}Y$ for every $n$ and an isomorphism on k-invariants $k_n^X\leftrightarrow k_n^Y$ compatible with the previous identifications. This is Postnikov's original classification theorem: homotopy type equals homotopy groups plus k-invariants modulo equivalences. The full statement requires care with the simply-connected case and with ${\pi }_1$-action on higher homotopy in the non-simply-connected case.

Theorem (relation to obstruction theory). The obstruction to extending a map $f:Y\to P_{n-1}X$ to a map $Y\to P_{n}X$ along the bonding fibration is the pulled-back k-invariant $f^ k_n \in H^{n+1}(Y; \pi_n X)$.* This is the foundational fact of obstruction theory for maps into Kan complexes: every map decomposes through a sequence of obstruction problems, one per homotopy degree, with obstructions exactly the pull-backs of the k-invariants. Hatcher §4.3 develops this perspective in the topological setting; the simplicial side is treated in Goerss-Jardine §VI.5.

Postnikov tower as left Bousfield localisation

Theorem (Hirschhorn §6.1, Bousfield localisation viewpoint). The $n$-truncation ${\tau }_{\le n}$ is a left Bousfield localisation of the Kan-Quillen model structure on $\mathbf{sSet}$ at the maps inverting ${\pi }_{n+1}$, ${\pi }_{n+2}$, $\dots$. The fibrant objects of the localised model structure are exactly the Kan complexes with vanishing homotopy above degree $n$, and the Postnikov section $P_{n}X$ is the fibrant replacement of $X$ in the localised model structure. This identifies the Postnikov tower with a tower of left Bousfield localisations — a perspective that generalises to $\infty$-categorical truncations.

Theorem (the tower is a system of localisations). Indexed by $n$, the Postnikov tower $\{P_{n}X{\}}_n$ is the system of left Bousfield localisations of $X$ at the descending family of subcategories ${\mathbf{sSet}}_{\mathrm{Kan}}^{\le n}\subset {\mathbf{sSet}}_{\mathrm{Kan}}^{\le n+1}$. The structural projections $X\to P_{n}X$ are the unit maps of the localisations. The homotopy limit of this system is the canonical map $X\to {\underleftarrow{\lim }}_n{P}_{n}X$. This places the Postnikov tower in the same conceptual framework as profinite completions and rationalisations.

$S^2$ worked example with explicit k-invariants

Theorem ($S^2$ Postnikov tower, first three stages). The Postnikov tower of the singular simplicial set $\mathrm{Sing}(S^2)$, which is a Kan complex by Kan 1958, has:

• $P_0\mathrm{Sing}(S^2)=P_1\mathrm{Sing}(S^2)={\Delta }^0$ (point);
• $P_2\mathrm{Sing}(S^2)=K(\mathbb{Z},2)$ (simplicial Eilenberg-MacLane Kan complex of type $(\mathbb{Z},2)$, realising as ${\mathbb{CP}}^{\infty }$);
• $P_3\mathrm{Sing}(S^2)$ classified by the k-invariant $k_3={\iota }_2⌣{\iota }_2\in H^4(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}$, the cup-square of the fundamental class.

The Hopf-invariant-one fact ${\pi }_3(S^2)=\mathbb{Z}$ is exactly the statement that this k-invariant is non-zero. Topologically the Hopf fibration $S^1\to S^3\to S^2$ realises this obstruction.

Theorem ($S^2$ versus ${\mathbb{CP}}^{\infty }$). The Eilenberg-MacLane Kan complex $K(\mathbb{Z},2)$ has vanishing k-invariants — its Postnikov tower stabilises at level 2, since ${\pi }_{n}K(\mathbb{Z},2)=0$ for $n\ne 2$. By contrast, $S^2$'s Postnikov tower never stabilises, with new non-zero k-invariants appearing at every degree (matching the infinite list of unstable homotopy groups of spheres). The comparison illustrates that the Postnikov tower distinguishes finely between spaces with the same low-dimensional homotopy.

Theorem (mod-2 truncation of $S^3$). For $S^3$ the Postnikov tower begins $P_2={\Delta }^0$, $P_3=K(\mathbb{Z},3)$, and the first non-identity k-invariant is $k_4={\mathrm{Sq}}^2{\iota }_3\in H^5(K(\mathbb{Z},3);\mathbb{Z}/2)$, recovering ${\pi }_4(S^3)=\mathbb{Z}/2$ via the Serre spectral sequence applied to the Postnikov fibration. Higher k-invariants involve secondary cohomology operations and are computable through Adem-relation analysis. The level-by-level computation here parallels the Whitehead-tower computation in 03.12.07 from the dual direction.

Synthesis. The Postnikov tower of a Kan complex is the foundational decomposition of homotopy types into Eilenberg-MacLane layers, and this is exactly the simplicial-side analogue of the topological Postnikov tower from 03.12.05. The central insight is that the data of a Kan complex up to weak equivalence is the data of its homotopy groups plus its k-invariants — a complete invariant in the simply-connected case modulo group-theoretic action. Putting these together with the Quillen-Serre model structure 03.12.31, the Postnikov tower is realised as the system of left Bousfield localisations of the Kan-Quillen model structure at ${\pi }_{n+1},{\pi }_{n+2},\dots$, and identifies the truncation functors with fibrant-replacement functors in the localised model structures.

The bridge is between the homotopy-group functor and the cohomology functor, mediated by the k-invariants: $k_n\in H^{n+1}(P_{n-1}X;{\pi }_{n}X)$ encodes the obstruction to splitting the bonding map $P_{n}X\to P_{n-1}X$ as a product fibration, and the failure of splitting is precisely the homotopy-group information at degree $n$. This pattern recurs in every Postnikov-style decomposition — for fibrations, for principal bundles, for $E_{\infty }$-ring spectra in stable homotopy. The Postnikov tower generalises in two directions: dually to the Whitehead tower 03.12.07 truncating from below; vertically to spectra and $\infty$-categorical truncations where the same coskeletal construction applies to homotopy types of all dimensions.

Full proof set [Master]

Proposition 1 (k-invariant existence theorem, full proof). Let $X$ be a pointed Kan complex with Postnikov stages $P_{n}X$ and $P_{n-1}X$ already constructed. The bonding map ${\beta }_n:P_{n}X\to P_{n-1}X$ is classified by a unique cohomology class $k_n\in H^{n+1}(P_{n-1}X;{\pi }_{n}X)$, in the sense that $P_{n}X$ is the homotopy pullback of the path-loop fibration $PK({\pi }_{n}X,n+1)\to K({\pi }_{n}X,n+1)$ along a map $\kappa :P_{n-1}X\to K({\pi }_{n}X,n+1)$ representing $k_n$ via Brown representability.

Proof. The homotopy fibre $F_n$ of ${\beta }_n$ has homotopy concentrated in degree $n$ with value ${\pi }_{n}X$ (Exercise 5 above), characterising $F_n\simeq K({\pi }_{n}X,n)$ up to weak equivalence. Hence ${\beta }_n$ is a principal $K({\pi }_{n}X,n)$-fibration.

Principal $K(A,n)$-fibrations over a Kan complex $B$ are classified by maps $B\to BK(A,n)\simeq K(A,n+1)$, equivalently by elements of $[B,K(A,n+1)]=H^{n+1}(B;A)$ via Brown representability. Specializing to $B=P_{n-1}X$ and $A={\pi }_{n}X$ gives a unique class $k_n\in H^{n+1}(P_{n-1}X;{\pi }_{n}X)$ classifying ${\beta }_n$ as a principal fibration.

To realise the homotopy pullback explicitly, choose a Kan-fibrant model for the classifying map $\kappa :P_{n-1}X\to K({\pi }_{n}X,n+1)$. The homotopy pullback $$ \begin{array}{ccc} P_n X & \to & PK(\pi_n X, n + 1) \ \downarrow & & \downarrow \ P_{n-1} X & \xrightarrow{\kappa} & K(\pi_n X, n + 1) \end{array} $$ has total space $P_{n}X$ — by uniqueness of homotopy pullbacks once the fibre is fixed — and the right vertical map is the path-loop fibration with fibre $\Omega K({\pi }_{n}X,n+1)=K({\pi }_{n}X,n)$.

Uniqueness of $k_n$ modulo automorphisms of ${\pi }_{n}X$ follows from the corresponding uniqueness in Brown representability. $\square$

Proposition 2 (universal property of the Postnikov section). The Postnikov section $P_{n}X$ is universal among Kan complexes $Y$ with ${\pi }_{i}Y=0$ for $i>n$ receiving a map from $X$: any such map $X\to Y$ factors uniquely (up to homotopy) through ${\eta }_n:X\to P_{n}X$.

Proof. Let $Y$ be a Kan complex with vanishing homotopy above degree $n$, and let $f:X\to Y$ be a map of Kan complexes. We construct a factorisation $X\to P_{n}X\to Y$.

Restrict $f$ to the $(n+1)$-skeleton of $X$, then extend via the coskeleton: ${\mathrm{cosk}}_{n+1}(f\circ \iota )$ where $\iota :E_{n+1}X\hookrightarrow X$ is the natural inclusion. Since $Y$ has vanishing homotopy above degree $n$ and is a Kan complex, $Y={\mathrm{cosk}}_{n+1}({\mathrm{sk}}_{n+1}Y)$ by the universal property of the coskeleton (every simplex of degree $>n+1$ is uniquely determined by its boundary). Hence ${\mathrm{cosk}}_{n+1}(f\circ \iota ):{\mathrm{cosk}}_{n+1}(E_{n+1}X)\to {\mathrm{cosk}}_{n+1}(E_{n+1}Y)\to Y$ provides the factorisation through $P_{n}X={\mathrm{cosk}}_{n+1}(E_{n+1}X)$.

Uniqueness up to homotopy: any two factorisations differ by a homotopy on the $(n+1)$-skeleton; this homotopy extends to a global homotopy because the obstruction lies in $H^{n+2}(P_{n}X\wedge {\Delta }^1;{\pi }_{n+1}Y)=H^{n+2}(P_{n}X\wedge {\Delta }^1;0)=0$. $\square$

Proposition 3 (the $S^2$ first k-invariant equals ${\iota }_2^2$). The Postnikov stage $P_3\mathrm{Sing}(S^2)$ is classified by the k-invariant $k_3={\iota }_2⌣{\iota }_2\in H^4(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}$, where ${\iota }_2\in H^2(K(\mathbb{Z},2);\mathbb{Z})$ is the fundamental class.

Proof. By Serre's computation 03.12.05, $H^{\ast }(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}[{\iota }_2]$ is a polynomial ring on the fundamental class; in particular $H^4(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}\cdot {\iota }_2^2$.

The k-invariant $k_3$ is determined by the class in $H^4(K(\mathbb{Z},2);\mathbb{Z})$ classifying the principal $K(\mathbb{Z},3)$-fibration $K(\mathbb{Z},3)\to P_3\mathrm{Sing}(S^2)\to K(\mathbb{Z},2)$. Equivalently, $k_3\in H^4(K(\mathbb{Z},2);\mathbb{Z})$ pulls back along the map $S^2\to K(\mathbb{Z},2)$ representing the fundamental class to the cohomology obstruction class on $S^2$ for the Hopf fibration.

The pull-back of ${\iota }_2^2$ along the fundamental-class map $S^2\to K(\mathbb{Z},2)$ is the cup-square of the fundamental class on $S^2$, which equals zero by dimension reasons ($H^4(S^2;\mathbb{Z})=0$). Hence the pulled-back obstruction is the first class in $H^4(S^2;{\pi }_3{S}^2)=0$ where the obstruction theory of the Hopf fibration lives. Concretely, the Hopf fibration's classifying data is encoded in the integer $1\in \mathbb{Z}={\pi }_3{S}^2$, and the corresponding pre-image in $H^4(K(\mathbb{Z},2);\mathbb{Z})=\mathbb{Z}$ is $1\cdot {\iota }_2^2$.

A cleaner statement: $k_3={\iota }_2^2$ is the unique non-zero integer class detecting ${\pi }_3{S}^2=\mathbb{Z}$ at the Postnikov level. This is the canonical worked example of Hatcher §4.3 and remains the prototype for all higher Postnikov computations of ${\pi }_{\ast }(S^n)$. $\square$

Proposition 4 (Postnikov tower is functorial and natural). The Postnikov tower defines a functor $\mathrm{Postn} : \mathbf{sSet}_{\mathrm{Kan},} \to \mathrm{Tow}(\mathbf{sSet}_{\mathrm{Kan},})$ from pointed Kan complexes to towers of pointed Kan complexes. Natural transformations between Postnikov-tower-valued functors are determined by their values on the Eilenberg-MacLane fibres.

Proof. Functoriality was established in Exercise 8 above via the universal property of coskeleta. For the second clause, observe that a natural transformation $\mathrm{Postn}\to {\mathrm{Postn}}^{\prime }$ between two such functors restricts at each level to a natural transformation on the Eilenberg-MacLane fibres $K({\pi }_n-,n)$. By Brown representability, these are classified by maps in cohomology, and the full natural transformation is determined by this fibrewise data plus compatibility with the bonding fibrations. $\square$

Connections [Master]

• Eilenberg-MacLane space 03.12.05. Each fibre of the Postnikov tower is an Eilenberg-MacLane Kan complex $K({\pi }_{n}X,n)$, identified up to weak equivalence by its homotopy profile. The simplicial loop-space relation $\Omega K(\pi ,n+1)\simeq K(\pi ,n)$ that powers the principal-fibration classification is the same identity that runs the Eilenberg-MacLane spectrum. The Postnikov decomposition makes the cohomology of $X$ assemble from the cohomology of these Eilenberg-MacLane pieces via the Serre spectral sequence applied to each bonding fibration.

• Whitehead tower 03.12.07. The dual construction. Where the Postnikov tower truncates a space from above by killing high-degree homotopy and retains the bottom layers, the Whitehead tower kills the bottom layers and retains the top. The two towers bracket the homotopy structure of any space, with $X$ recoverable as the homotopy limit of the Postnikov tower and the homotopy colimit of the Whitehead tower in the appropriate categorical setting. Together they generate every homotopy-theoretic computation by reducing it to the principal $K(\pi ,n)$-fibration paradigm.

• Simplicial sets and geometric realization 03.12.25. The simplicial-side setting where the Postnikov tower lives. Kan complexes are the fibrant objects in the Quillen-Serre model structure on $\mathbf{sSet}$, and the coskeletal construction ${\mathrm{cosk}}_{n+1}$ realises the Postnikov section directly via simplicial-set operations. The geometric realisation $|P_{n}X|$ recovers the topological Postnikov section, providing the bridge from the combinatorial side to the topological side. By 03.12.25's Quillen equivalence $|-|⊣\mathrm{Sing}$, the simplicial Postnikov tower transports to the topological tower from 03.12.05.

• Quillen model category 03.12.31. The model-categorical setting for the Bousfield-localisation perspective on the Postnikov tower. The Kan-Quillen model structure makes each bonding map a Kan fibration in the canonical sense, and the truncation functors ${\tau }_{\le n}$ are left Bousfield localisations of the model structure. This generalises to other model categories — e.g., the Postnikov tower of a chain complex via the Dold-Kan correspondence is the truncation tower of chain complexes by degree.

• Kan-Quillen model structure on sSet 03.12.33. The model-structure-on-sSet side of the Kan complex story, where the Postnikov tower's bonding maps are exactly the Kan fibrations classifying as principal $K(\pi ,n)$-fibrations. This unit's structural results — coskeletal truncation, principal-fibration classification, k-invariants — depend on the Kan-Quillen model-structure machinery for their model-categorical home.

• Singular homology 03.12.11. The k-invariants live in cohomology — singular cohomology of the lower Postnikov stage with coefficients in the next homotopy group. The Serre spectral sequence of the bonding fibration converts the k-invariant data into homology and cohomology computations, providing the canonical computational route from homotopy to cohomology and back.

Historical & philosophical context [Master]

Mikhail M. Postnikov introduced the tower decomposition in a 1951 Doklady Akademii Nauk SSSR note ^{[Postnikov1951]} entitled Determination of the homology groups of a space by means of the homotopy invariants, motivated by the problem of recovering homology from homotopy in arbitrary CW complexes. Postnikov's original construction was on the topological side and was clarified and refined throughout the 1950s by Cartan, Serre, and Eilenberg-MacLane.

The simplicial-side formulation — where the tower of Kan complexes carries the same data combinatorially via coskeleta — emerged in the work of Daniel Kan and J. Peter May. Kan 1958 introduced the eponymous extension condition in Functors involving c.s.s. complexes (Trans. AMS 87), establishing the simplicial homotopy theory of Kan complexes. May's 1967 Simplicial Objects in Algebraic Topology §8 ^[May1967] gave the canonical simplicial-side Postnikov system construction, including the identification with the coskeleton functor and the functorial properties.

Goerss and Jardine's 1999 monograph Simplicial Homotopy Theory §VI ^{[GoerssJardine2009]} is the modern canonical reference, presenting the Postnikov tower as part of the model-categorical infrastructure of $\mathbf{sSet}$ and connecting it explicitly to the Bousfield-Kan localisation tower from Bousfield-Kan 1972 Homotopy Limits, Completions and Localizations (LNM 304).

Hirschhorn 2003 Model Categories and Their Localizations §6 placed the Postnikov tower in the broader framework of left Bousfield localisations of model categories. The truncation functors ${\tau }_{\le n}$ are realised as left Bousfield localisations at the relevant sphere classes, and the full tower is the system of such localisations indexed by truncation level. This perspective generalises naturally to $\infty$-categorical truncations and underlies Lurie's Higher Topos Theory §5.5.6 treatment of $n$-truncated objects in any $\infty$-topos.

The Postnikov tower's structural role is dual to that of the Whitehead tower. Where Postnikov truncates the homotopy from above and remains computable for low truncation levels, the Whitehead tower (G. W. Whitehead 1953) truncates from below and excels at higher-connectivity computations. The two together — combined with Hurewicz's theorem and the Serre spectral sequence — give the canonical machinery for computing homotopy groups of spheres, with Serre's 1953 finiteness theorem and Toda's 1962 tables as the principal applications. The Hopf-invariant-one problem solved by Adams 1960 sits exactly inside the Postnikov-tower framework at the level where the first k-invariant of $S^{4k-1}$ becomes constrained by Steenrod-algebra structure.

Bibliography [Master]

@article{Postnikov1951,
  author = {Postnikov, M. M.},
  title = {Determination of the homology groups of a space by means of the homotopy invariants},
  journal = {Doklady Akademii Nauk SSSR},
  volume = {76},
  year = {1951},
  pages = {359-362},
}

@book{GoerssJardine2009,
  author = {Goerss, Paul G. and Jardine, John F.},
  title = {Simplicial Homotopy Theory},
  publisher = {Birkh\"auser},
  series = {Progress in Mathematics},
  volume = {174},
  year = {1999},
  note = {Reprint 2009},
}

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

@book{Hatcher2002,
  author = {Hatcher, Allen},
  title = {Algebraic Topology},
  publisher = {Cambridge University Press},
  year = {2002},
}

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

@book{BousfieldKan1972,
  author = {Bousfield, A. K. and Kan, D. M.},
  title = {Homotopy Limits, Completions and Localizations},
  publisher = {Springer},
  series = {Lecture Notes in Mathematics},
  volume = {304},
  year = {1972},
}

@article{Kan1958,
  author = {Kan, Daniel M.},
  title = {Functors involving c.s.s. complexes},
  journal = {Transactions of the American Mathematical Society},
  volume = {87},
  year = {1958},
  pages = {330-346},
}

@article{EilenbergMacLane1945,
  author = {Eilenberg, Samuel and Mac Lane, Saunders},
  title = {Relations between homology and homotopy groups of spaces},
  journal = {Annals of Mathematics},
  volume = {46},
  year = {1945},
  pages = {480-509},
}

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

---

Cycle 6 unit. Postnikov tower of a Kan complex: the coskeletal construction $P_{n}X={\mathrm{cosk}}_{n+1}X$, the principal $K({\pi }_n,n)$-fibration structure, k-invariants $k_n\in H^{n+1}(P_{n-1}X;{\pi }_{n}X)$, the left Bousfield localisation viewpoint, and the worked $S^2$ example with first k-invariant $k_3={\iota }_2^2$.