03.12.61 · modern-geometry / homotopy

Nilpotent groups and nilpotent spaces

shipped3 tiersLean: none

Anchor (Master): Hilton-Mislin-Roitberg 1975 *Localization of Nilpotent Groups and Spaces* (North-Holland Math Studies 15) Ch I-II (canonical monograph); May-Ponto 2012 *More Concise Algebraic Topology* (Chicago) Ch 3-4; Bousfield-Kan 1972 *Homotopy Limits, Completions and Localizations* (LNM 304) §I.5 (nilpotent fibrations); Dror 1971 *Ann. of Math.* 94 (generalized Whitehead theorem for nilpotent spaces)

Intuition Beginner

Some spaces are easy to take apart layer by layer, and some are not. The well-behaved ones have a name: nilpotent spaces. The name comes from group theory, where a nilpotent group is one you can build up from the centre outward in finitely many steps. A space is nilpotent when its fundamental group is nilpotent in that sense, and when the way loops act on the higher layers can also be reduced to nothing in finitely many steps.

The reason to care is practical. The two big tools of modern homotopy theory — localisation (keeping only the information at a chosen set of primes) and completion (filling in the rest) — work cleanly only on nilpotent spaces. On wilder spaces they misbehave. Nilpotent spaces are exactly the spaces where the layer-by-layer decomposition has enough room to run.

The single sentence: a nilpotent space is one whose fundamental group is nilpotent and whose loops act mildly on every higher layer, and that mildness is what makes the whole computational machinery of the subject apply.

Visual Beginner

A tower of layers stacked above a base, where each step from one layer to the next adds a single sheet of homotopy and the gluing between sheets is recorded by one piece of data.

The picture is the Postnikov tower with one extra feature: for a nilpotent space the tower can be sliced even more finely, so that every single step adds exactly one Eilenberg-MacLane sheet, attached by exactly one class. Spaces that are not nilpotent cannot be sliced this finely — their layers tangle.

Worked example Beginner

A circle, written $S^{1}$ , and a figure-eight, written as the wedge of two circles.

The circle has fundamental group the integers, $Z$ . The integers form a commutative group, and every commutative group is nilpotent in the simplest way: the centre is the whole group, so the build-up from the centre finishes in one step. The circle has no higher layers to worry about. So $S^{1}$ is a nilpotent space.

Now the figure-eight. Its fundamental group is the free group on two generators. A free group on two or more generators is not nilpotent: the build-up from the centre never terminates, because the commutator of the two generators, then its commutator with a generator, and so on, keep producing new elements forever. So the figure-eight is not a nilpotent space.

What this tells us: nilpotency is sensitive to the fundamental group. A commutative fundamental group is always fine. A free group on several generators is the standard example that fails. The cut-off between the two cases is exactly whether the lower central series — the repeated commutator build-up — reaches the identity in finitely many steps.

Check your understanding Beginner

Exercise (easy, multiple choice).

A space whose fundamental group is the free group on two generators is:

A. Nilpotent, because free groups are commutative. B. Not nilpotent, because the lower central series never reaches the identity. C. Nilpotent, because every space with a fundamental group is nilpotent. D. Undecidable.

Hint

Track whether repeated commutators of the two generators ever stop producing new elements.

Answer

B. Feedback-correct: yes; in the free group on two generators the repeated commutators keep producing new elements, so the lower central series does not terminate, and the space fails to be nilpotent. Feedback-wrong: A is false because free groups on two or more generators are not commutative; C is false because nilpotency is a genuine restriction; D is false because the criterion is a definite calculation.

Formal definition Intermediate+

Let $G$ be a group. The lower central series of $G$ is the descending chain of subgroups $$ G = \Gamma_1 G \supseteq \Gamma_2 G \supseteq \Gamma_3 G \supseteq \cdots, \qquad \Gamma_{i+1} G = [G, \Gamma_i G], $$ where $[G, Γ_{i} G]$ is the subgroup generated by all commutators $g h g^{- 1} h^{- 1}$ with $g \in G$ and $h \in Γ_{i} G$ . Each $Γ_{i} G$ is normal in $G$ , and the successive quotients $Γ_{i} G / Γ_{i + 1} G$ are central in $G / Γ_{i + 1} G$ . The group $G$ is nilpotent of class $\leq c$ when $Γ_{c + 1} G = {e}$ ; the least such $c$ is the nilpotency class. Nilpotency of class $\leq 1$ is precisely commutativity 01.02.05.

Now let $G$ act on an abelian group $A$ by automorphisms; equivalently let $A$ be a module over the group ring $Z [G]$ . The action is nilpotent when the descending chain $$ A = \Gamma_G^0 A \supseteq \Gamma_G^1 A \supseteq \cdots, \qquad \Gamma_G^{j+1} A = \langle, g a - a : g \in G,\ a \in \Gamma_G^j A ,\rangle, $$ of the augmentation-ideal filtration reaches $0$ in finitely many steps, $Γ_{G}^{k} A = 0$ for some $k$ . Writing $I \subseteq Z [G]$ for the augmentation ideal, $Γ_{G}^{j} A = I^{j} A$ , so a nilpotent action is one for which $I^{k} A = 0$ : the augmentation ideal acts nilpotently on $A$ .

Definition (nilpotent space). A connected space $X$ with a chosen basepoint is nilpotent when

$π_{1} (X, x_{0})$ is a nilpotent group, and
for every $n \geq 2$ the action of $π_{1} (X, x_{0})$ on $π_{n} (X, x_{0})$ — the standard action of $π_{1}$ on higher homotopy 03.12.52 — is a nilpotent action in the sense above.

A connected space with $π_{1}$ acting as the identity on every $π_{n}$ (including by conjugation on $π_{1}$ itself, so $π_{1}$ abelian) is called a simple space; a simple space is nilpotent with all the actions of class $\leq 1$ . The convention throughout is that "nilpotent space" means pointed and connected; the basepoint choice does not affect the property for a connected space.

Counterexamples to common slips Intermediate+

Nilpotent $π_{1}$ alone is not enough. One also needs each action on $π_{n}$ to be nilpotent. A space can have abelian $π_{1}$ yet a non-nilpotent action on some $π_{n}$ — for instance certain non-simple lens-space-like quotients where the deck action on $π_{n}$ has no finite augmentation filtration.
Nilpotent is weaker than simple. A simple space requires the action to be the identity; a nilpotent space only requires it to be filtered to zero. The class of nilpotent spaces sits strictly between simple spaces and all spaces.
The augmentation filtration, not the lower central series of $π_{n}$ , governs condition 2. The group $π_{n}$ is already abelian for $n \geq 2$ ; what must be nilpotent is the $π_{1}$ -action on it, measured by powers of the augmentation ideal $I^{k} π_{n} = 0$ .

Key theorem with proof Intermediate+

Theorem (principal refinement of the Postnikov tower). A connected pointed space $X$ of the homotopy type of a CW complex is nilpotent if and only if its Postnikov tower admits a principal refinement: a tower $$ \cdots \to X_{s} \to X_{s-1} \to \cdots \to X_1 \to X_0 = * $$ converging to $X$ in which every map $X_{s} \to X_{s - 1}$ is a principal fibration with fibre an Eilenberg-MacLane space $K (A_{s}, n_{s})$ , classified by a single k-invariant $k_{s} \in H^{n_{s} + 1} (X_{s - 1}; A_{s})$ with untwisted (identity-action) coefficient system.

Proof. ( $\Rightarrow$ ) Suppose $X$ is nilpotent. Build the ordinary Postnikov tower ${P_{n} X}$ 03.12.40; the stage $P_{n} X \to P_{n - 1} X$ has fibre $K (π_{n} X, n)$ but is in general classified by a k-invariant with the $π_{1}$ -twisted coefficient system $π_{n} X$ . Refine each stage. For $n = 1$ , nilpotency of $π_{1}$ gives a finite central series $π_{1} = Γ_{1} \supseteq \dots \supseteq Γ_{c + 1} = e$ with each $Γ_{i} / Γ_{i + 1}$ central abelian; interpolate $P_{1} X = K (π_{1}, 1)$ by the tower of principal fibrations with fibres $K (Γ_{i} / Γ_{i + 1}, 1)$ , each classified by a single class because the quotient is central, hence carries the identity action. For $n \geq 2$ , nilpotency of the action gives the finite augmentation filtration $π_{n} X = I^{0} π_{n} \supseteq I^{1} π_{n} \supseteq \dots \supseteq I^{k} π_{n} = 0$ , with each $I^{j} / I^{j + 1}$ a $π_{1}$ -module on which $π_{1}$ acts as the identity (the augmentation ideal kills the successive quotient by construction). Interpolate the single Postnikov stage at degree $n$ by the finite tower of principal fibrations with fibres $K (I^{j} / I^{j + 1}, n)$ ; each is classified by one class in untwisted cohomology because the coefficient action is the identity. Splicing these refinements across all $n$ yields the principal refinement.

( $\Leftarrow$ ) Conversely, a principal refinement with untwisted single-class k-invariants exhibits, at each homotopy degree, a finite filtration of $π_{n} X$ by sub- $π_{1}$ -modules with identity-action quotients — read off from the fibres $K (A_{s}, n)$ at that degree — which is exactly a finite augmentation filtration, so the action is nilpotent; the $n = 1$ refinement displays a finite central series, so $π_{1}$ is nilpotent. Both conditions of the definition hold, and $X$ is nilpotent. $□$

Bridge. This theorem builds toward the localisation unit 03.12.60, where the principal refinement is the object localisation is applied to one Eilenberg-MacLane fibre at a time, and the foundational reason the construction converges is exactly that each fibre is a single $K (A, n)$ with abelian $A$ on which localisation is the elementary algebraic operation $A \mapsto A \otimes Z_{T}$ . The principal refinement appears again in 03.12.45, where the arithmetic-square fracture is assembled stage by stage along the same tower; this is dual to the way completion is built fibre by fibre, and putting these together, the central insight is that nilpotency is precisely the hypothesis under which the homotopy-theoretic operations reduce to abelian-group operations on the tower's fibres. The bridge is the augmentation filtration: it converts the non-abelian datum of a $π_{1}$ -action into a finite stack of identity-action abelian layers, which is what makes the layers individually computable.

Exercises Intermediate+

Exercise 7 (hard, proof). Nilpotent fibrations preserve nilpotency.

Let $F \to E \to B$ be a fibration of connected spaces with $B$ nilpotent and $F$ nilpotent, and suppose $π_{1} (E)$ acts nilpotently on each $π_{n} (F)$ (a nilpotent fibration). Prove $E$ is nilpotent.

Hint

Use the long exact sequence of the fibration together with the fact that an extension of nilpotent modules by nilpotent modules, under a single group, is nilpotent.

Answer

The long exact homotopy sequence gives, for each $n \geq 2$ , a short exact sequence of $π_{1} (E)$ -modules $$ \mathrm{coker} \to \pi_n(E) \to \ker \to 0 $$ assembled from $π_{n} (F) \to π_{n} (E) \to π_{n} (B)$ , where $π_{n} (B)$ carries the $π_{1} (B)$ -action pulled back along $π_{1} (E) \to π_{1} (B)$ and $π_{n} (F)$ carries the nilpotent $π_{1} (E)$ -action by hypothesis. An augmentation filtration of $π_{n} (E)$ is obtained by splicing a finite filtration of the sub from $π_{n} (F)$ with a finite filtration of the quotient from $π_{n} (B)$ : since $I^{a}$ kills the $π_{n} (F)$ -part and $I^{b}$ kills the $π_{n} (B)$ -part, $I^{a + b}$ kills $π_{n} (E)$ . For $n = 1$ , $π_{1} (E)$ is an extension of the nilpotent $π_{1} (B)$ by a quotient of the nilpotent $π_{1} (F)$ with nilpotent action, hence nilpotent. So $E$ satisfies both conditions. $□$

This is the Bousfield-Kan nilpotent-fibration lemma, the structural engine behind fibrewise localisation.

Exercise 8 (hard, proof). The generalized Whitehead theorem.

State and prove the homology form of the Whitehead theorem on the class of nilpotent spaces: a map $f : X \to Y$ of nilpotent spaces inducing an isomorphism on integral homology is a weak equivalence.

Hint

Reduce to the principal refinement and induct up the tower, using that on each $K (A, n)$ fibre homology detects the homotopy by the Hurewicz theorem.

Answer

Both $X$ and $Y$ admit principal refinements with identity-action $K (A_{s}, n_{s})$ fibres (Key theorem). The map $f$ induces a map of refinements stage by stage. Argue by induction up the tower: on the bottom stage $f$ is an iso on $π_{1}$ because an integral-homology iso of nilpotent $K (π_{1}, 1)$ 's is an iso of the abelianisation at each central layer, and nilpotency lets the central-series layers be recovered from homology. At an inductive stage the fibre comparison is a map of $K (A, n)$ 's; an integral-homology isomorphism of Eilenberg-MacLane spaces is an isomorphism on the single homotopy group by the Hurewicz theorem applied to the $(n - 1)$ -connected fibre. The five-lemma on the homotopy long exact sequences of the two principal fibrations then propagates the isomorphism up one stage. Passing to the limit, $f$ is an isomorphism on all homotopy groups, hence a weak equivalence. $□$

This is Dror's theorem; it fails for non-nilpotent spaces, the acyclic non-contractible spaces being the standard obstruction.

Advanced results Master

The structural content of nilpotency divides into the group-theoretic input, the homotopy-theoretic structure theorem, and the detection criteria that make the class recognisable in practice.

Nilpotent groups and modules

Theorem (lower central series characterisation). For a group $G$ the following are equivalent: the lower central series $Γ_{i} G$ reaches the identity in finitely many steps; the upper central series $Z_{i} G$ reaches $G$ in finitely many steps; $G$ admits some finite central series $G = G_{0} \supseteq G_{1} \supseteq \dots \supseteq G_{k} = e$ with each $G_{i} / G_{i + 1}$ central in $G / G_{i + 1}$ . The least length is the nilpotency class and is the same number computed from either canonical series 01.02.05. The lower central series is the fastest-descending and the upper the fastest-ascending such series, sandwiching every other.

Theorem (augmentation-ideal criterion for modules). A $Z [G]$ -module $A$ carries a nilpotent $G$ -action if and only if $I^{k} A = 0$ for some $k$ , where $I$ is the augmentation ideal; equivalently the $I$ -adic filtration is finite, equivalently $A$ admits a finite filtration by sub- $G$ -modules with identity-action quotients. The three formulations are the module-theoretic shadow of the three group series, and the third is the one that reads off directly from a principal Postnikov refinement.

Theorem (closure properties). Nilpotent groups are closed under subgroups, quotients, and finite products, but not under extensions in general; finitely generated nilpotent groups are polycyclic, hence Noetherian, and their torsion elements form a finite characteristic subgroup. The Hirsch-length and torsion structure of finitely generated nilpotent groups (P. Hall) is what makes $T$ -localisation of such groups — inverting primes outside $T$ on the torsion-free quotients — a finite well-controlled operation.

The structure theorem and its consequences

Theorem (principal Postnikov refinement, full form). A connected CW space $X$ is nilpotent if and only if it admits a Postnikov-type tower converging to $X$ in which each stage is a principal fibration induced from the path-loop fibration over a single Eilenberg-MacLane space $K (A, n)$ by one k-invariant in untwisted cohomology. The refinement is not unique, but any two are related by a finite zig-zag of stage subdivisions; the multiset of fibres $K (A_{s}, n_{s})$ refines the homotopy groups via the augmentation filtrations and is an invariant of the homotopy type once the filtration is fixed.

Theorem (Bousfield-Kan, the $R$ -good criterion). A nilpotent space of finite type is $R$ -good for every solid ring $R$ : the $R$ -completion map $X \to R_{\infty} X$ induces an isomorphism on $R$ -homology and the Bousfield-Kan completion tower converges. The class of nilpotent spaces is precisely the natural domain on which the completion functor is homotopy-invariant and idempotent. This is the converse-direction justification for centring the entire localisation-and-completion theory on nilpotent spaces.

Theorem (Dror, generalized Whitehead). On the class of nilpotent spaces, an integral-homology isomorphism is a weak equivalence; equivalently, homology detects the homotopy type. The class of nilpotent spaces is the largest class containing the simply-connected spaces on which the homology-Whitehead theorem holds without further hypotheses. The failure outside the class is exactly the existence of acyclic non-contractible spaces, whose fundamental groups are perfect and hence as far from nilpotent as possible.

Detection and recognition

Theorem (detection criteria). Each of the following implies $X$ nilpotent: $X$ is simply connected; $X$ is simple (abelian $π_{1}$ acting identically on every $π_{n}$ ); $X = B G$ for $G$ a nilpotent group; $X$ is an $H$ -space; $π_{1} (X)$ is finite of prime-power order acting on each $π_{n}$ through a finite $p$ -group. The $H$ -space case follows because the multiplication forces the $π_{1}$ -action to be the identity, so an $H$ -space is simple, hence nilpotent; this places Lie groups, loop spaces, and infinite loop spaces inside the class.

Synthesis. Nilpotency is the foundational reason the two central operations of the subject — localisation 03.12.60 and completion 03.12.45 — are well-defined and idempotent on spaces, and this is exactly the hypothesis under which the augmentation filtration converts the non-abelian $π_{1}$ -action into a finite stack of identity-action abelian layers. The central insight is that a nilpotent space's principal Postnikov refinement presents it as built from single Eilenberg-MacLane fibres $K (A, n)$ glued by single untwisted k-invariants, so every functor that is understood on $K (A, n)$ and respects principal fibrations is thereby understood on the whole class; localisation $A \mapsto A \otimes Z_{T}$ and $p$ -completion $A \mapsto A_{p}^{\land}$ are exactly such functors, which is the bridge from abelian-group algebra to space-level homotopy. Putting these together with the Dror theorem, homology detects homotopy on this class, so the class is simultaneously the domain of the arithmetic operations and the domain on which homological computation suffices — the two facts that make nilpotent spaces the computable class on which the rest of the theory is built, and which is dual to the way the non-nilpotent acyclic spaces are precisely the obstruction to both. This generalises the simply-connected case that classical rational homotopy theory had treated, widening it to the full class on which the Sullivan and Quillen models extend.

Full proof set Master

Proposition 1 (the two central series have equal length). For a nilpotent group $G$ , the lower central series and the upper central series terminate after the same number of steps, namely the nilpotency class $c$ .

Proof. Write $Γ_{i}$ for the lower and $Z_{j}$ for the upper central series, so $Γ_{1} = G$ , $Γ_{i + 1} = [G, Γ_{i}]$ , and $Z_{0} = e$ , $Z_{j + 1} / Z_{j} = Z (G / Z_{j})$ . We show $Γ_{i + 1} \subseteq Z_{c - i}$ for all $i$ by descending induction once a finite central series of length $c$ exists. If $Γ_{c + 1} = e$ , then $Γ_{c}$ is central, so $Γ_{c} \subseteq Z_{1}$ ; inductively $Γ_{c - j} \subseteq Z_{j + 1}$ because $[G, Γ_{c - j}] = Γ_{c - j + 1} \subseteq Z_{j}$ forces the image of $Γ_{c - j}$ in $G / Z_{j}$ to be central. Hence $G = Γ_{1} \subseteq Z_{c}$ , so the upper series reaches $G$ in at most $c$ steps. Symmetrically, any finite central series of length $ℓ$ caps the lower series at length $ℓ$ from above (each $Γ_{i}$ is contained in the $i$ -th term from the top of any central series), so the lower series has length at most that of the upper. The two bounds give equality. $□$

Proposition 2 (principal refinement of a nilpotent $K (π, 1)$ ). For a nilpotent group $π$ of class $c$ , the space $K (π, 1)$ admits a tower of principal fibrations of length $c$ with fibres $K (Γ_{i} π / Γ_{i + 1} π, 1)$ , each abelian, classified by single untwisted k-invariants.

Proof. The lower central series quotients $Q_{i} = Γ_{i} π / Γ_{i + 1} π$ are abelian and central in $π / Γ_{i + 1} π$ . The quotient maps $π / Γ_{i + 1} π \to π / Γ_{i} π$ are central extensions by $Q_{i}$ . A central extension $1 \to Q_{i} \to π / Γ_{i + 1} \to π / Γ_{i} \to 1$ is classified by a single class in $H^{2} (π / Γ_{i}; Q_{i})$ with $Q_{i}$ an identity-action module, by the standard classification of central extensions. Applying the classifying-space functor $K (-, 1)$ turns each central extension into a principal fibration $$ K(Q_i, 1) \to K(\pi/\Gamma_{i+1}, 1) \to K(\pi/\Gamma_i, 1), $$ with k-invariant the image of the extension class in $H^{2} (K (π / Γ_{i}, 1); Q_{i})$ , an untwisted cohomology group because $Q_{i}$ is central. Stacking from $i = c$ down to $i = 1$ gives the asserted tower of length $c$ , converging to $K (π, 1)$ . $□$

Proposition 3 ( $H$ -spaces are simple, hence nilpotent). A connected $H$ -space $X$ has $π_{1} (X)$ abelian and acting identically on every $π_{n} (X)$ ; in particular $X$ is nilpotent.

Proof. Let $μ : X \times X \to X$ be the multiplication with two-sided homotopy unit. The standard Eckmann-Hilton argument shows $π_{1} (X)$ is abelian: the two compositions of loops, concatenation and pointwise $μ$ -product, share a unit and satisfy the interchange law, forcing both to coincide and to be commutative. For the action on $π_{n}$ , the unit component of $μ$ provides a homotopy from the $π_{1}$ -action map to the identity: translation by a loop $γ \in π_{1}$ is homotopic, through the $H$ -multiplication by the unit, to the identity self-map of $X$ , so the induced automorphism of $π_{n} (X)$ is the identity. Thus the action is the identity on each $π_{n}$ , $X$ is simple, and by Exercise 2 it is nilpotent. $□$

Proposition 4 (nilpotency is a homotopy-type invariant). If $X ≃ Y$ are weakly equivalent connected pointed spaces and $X$ is nilpotent, then $Y$ is nilpotent.

Proof. A weak equivalence $f : X \to Y$ induces isomorphisms $f_{*} : π_{n} (X) \to π_{n} (Y)$ for all $n$ , compatible with the $π_{1}$ -actions: $f_{*} (γ \cdot a) = f_{*} (γ) \cdot f_{*} (a)$ for $γ \in π_{1} (X)$ , $a \in π_{n} (X)$ . Hence $f_{*}$ carries the lower central series of $π_{1} (X)$ isomorphically to that of $π_{1} (Y)$ , so $π_{1} (Y)$ is nilpotent of the same class. It carries the augmentation filtration $I_{π_{1} (X)}^{j} π_{n} (X)$ onto $I_{π_{1} (Y)}^{j} π_{n} (Y)$ , because $f_{*}$ intertwines the two augmentation ideals; finiteness of the filtration transfers. Both conditions hold for $Y$ . $□$

Connections Master

Localisation of nilpotent spaces at a set of primes 03.12.60. The unit that this one is built for. Localisation $X \to X_{T}$ is applied one Eilenberg-MacLane fibre at a time along the principal Postnikov refinement established here; nilpotency is the exact hypothesis that the refinement exists, that each fibre $K (A, n)$ localises by the elementary operation $A \mapsto A \otimes Z_{T}$ , and that the localised fibres re-assemble into a localised space. Without nilpotency the localisation of a space is not controlled by the localisation of its homotopy groups.
Postnikov tower of a Kan complex 03.12.40. The ambient construction this unit refines. Every connected space has a Postnikov tower, but its k-invariants live in twisted cohomology with the $π_{1}$ -action as coefficient system. Nilpotency is precisely the condition that the tower can be subdivided into principal stages with untwisted single-class k-invariants, turning the general tower into the computable refined tower.
Eilenberg-MacLane spaces and k-invariants 03.12.05. The atoms of the refinement. Each stage of the principal refinement is a principal $K (A, n)$ -fibration classified by one cohomology class, so the entire homotopy type of a nilpotent space is encoded by a finite-per-degree list of abelian groups $A$ and untwisted k-invariants. The cohomology of $K (A, n)$ is the input to every k-invariant computation on the class.
Arithmetic square and integral fracture 03.12.45. Where the nilpotent-space definition was first introduced in one line and where completion is built. The fracture square reassembles a nilpotent space from its rationalisation and $p$ -completions; the present unit supplies the structural reason the square is a homotopy pullback, namely that on each principal fibre the square is the abelian-group fracture of $A$ and these assemble up the refinement.
Solvable and nilpotent groups, Jordan-Hölder 01.02.05. The group-theoretic foundation. The lower central series, nilpotency class, and central-series characterisations are imported wholesale from group theory; the homotopy-theoretic novelty is the augmentation-ideal version for the $π_{1}$ -action on the abelian higher homotopy groups, which specialises to ordinary nilpotency when applied to $π_{1}$ acting on itself by conjugation.
Relative homotopy group and the $π_{1}$ -action 03.12.52. The action whose nilpotency is condition 2 of the definition. The standard action of $π_{1}$ on $π_{n}$ is the datum the augmentation filtration is applied to; the relative-homotopy long exact sequence and the action it carries are what make "nilpotent action on $π_{n}$ " a well-posed condition.

Historical & philosophical context Master

The group-theoretic notion of nilpotency originates with Philip Hall's 1934 study of prime-power groups ^{[Hall 1934]}, where the lower central series, the basis theorem, and the collection process were introduced; the term "nilpotent" reflects that the associated graded Lie ring has nilpotent bracket. The transposition into homotopy theory came through the recognition that the $π_{1}$ -action on higher homotopy groups carries exactly the same algebraic shape, with the augmentation ideal of the group ring playing the role of the commutator.

Bousfield and Kan's 1972 Lecture Notes volume ^{[Bousfield-Kan 1972]} isolated nilpotent spaces as the natural domain of their completion and localisation functors, proving the $R$ -completion convergence and idempotence results on this class and identifying nilpotent fibrations as the structural building block. Independently, Hilton, Mislin, and Roitberg's 1975 monograph ^{[Hilton-Mislin-Roitberg 1975]} gave the systematic account of localisation of nilpotent groups and spaces, establishing the principal Postnikov refinement as the technical heart and treating the pullback/fracture theory in detail.

Emmanuel Dror's 1971 Annals paper ^{[Dror 1971]} proved the generalized Whitehead theorem on the nilpotent class, showing that integral homology detects weak equivalence there, and pinpointing the acyclic spaces with perfect fundamental group as the obstruction outside the class. May and Ponto's 2012 More Concise Algebraic Topology ^{[May-Ponto 2012]} organised the modern textbook treatment around nilpotent spaces as the central computable class, deriving localisation, completion, and the fracture squares uniformly from the principal-refinement structure.

Bibliography Master

@article{Hall1934,
  author = {Hall, Philip},
  title = {A contribution to the theory of groups of prime-power order},
  journal = {Proceedings of the London Mathematical Society},
  series = {2},
  volume = {36},
  year = {1934},
  pages = {29-95},
}

@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},
}

@book{HiltonMislinRoitberg1975,
  author = {Hilton, Peter and Mislin, Guido and Roitberg, Joseph},
  title = {Localization of Nilpotent Groups and Spaces},
  publisher = {North-Holland},
  series = {Mathematics Studies},
  volume = {15},
  year = {1975},
}

@article{Dror1971,
  author = {Dror, Emmanuel},
  title = {A generalization of the Whitehead theorem},
  journal = {Annals of Mathematics},
  volume = {94},
  year = {1971},
  pages = {530-538},
}

@book{MayPonto2012,
  author = {May, J. Peter and Ponto, Kate},
  title = {More Concise Algebraic Topology: Localization, Completion, and Model Categories},
  publisher = {University of Chicago Press},
  year = {2012},
}

Nilpotent groups and nilpotent spaces: the lower central series and nilpotency class; the nilpotent $π_{1}$ -action on $π_{n}$ via the augmentation filtration $I^{k} π_{n} = 0$ ; a nilpotent space as one with nilpotent $π_{1}$ acting nilpotently on each $π_{n}$ ; the structure theorem giving a principal Postnikov refinement with single $K (A, n)$ fibres and untwisted single-class k-invariants; the $R$ -good and Dror generalized-Whitehead consequences; detection criteria (simply connected, simple, $B G$ for $G$ nilpotent, $H$ -spaces). The foundation 03.12.60 localisation builds on.

Prerequisites

03.12.40
03.12.05
01.02.05
03.12.52

Tier anchors

beginner: Hatcher *Algebraic Topology* §4.2-§4.3 (the action of π₁ on πₙ, simple spaces) at informal level; May-Ponto *More Concise Algebraic Topology* Ch 3 opening (the lower central series picture)
intermediate: May-Ponto *More Concise Algebraic Topology* Ch 3-4; Hilton-Mislin-Roitberg *Localization of Nilpotent Groups and Spaces* Ch I-II; Bousfield-Kan *Homotopy Limits, Completions and Localizations* (LNM 304) §I.4-§I.5
master: Hilton-Mislin-Roitberg 1975 *Localization of Nilpotent Groups and Spaces* (North-Holland Math Studies 15) Ch I-II (canonical monograph); May-Ponto 2012 *More Concise Algebraic Topology* (Chicago) Ch 3-4; Bousfield-Kan 1972 *Homotopy Limits, Completions and Localizations* (LNM 304) §I.5 (nilpotent fibrations); Dror 1971 *Ann. of Math.* 94 (generalized Whitehead theorem for nilpotent spaces)

References

Hilton-Mislin-Roitberg — Localization of Nilpotent Groups and Spaces · North-Holland Mathematics Studies 15 (1975); Ch I nilpotent groups and the lower central series, Ch II nilpotent spaces and the principal Postnikov refinement; the canonical monograph treatment
May-Ponto — More Concise Algebraic Topology: Localization, Completion, and Model Categories · University of Chicago Press 2012; Ch 3 nilpotent spaces and Postnikov towers, Ch 4 detection theorems; nilpotent spaces presented as the central computable class on which localization and completion behave
Bousfield-Kan — Homotopy Limits, Completions and Localizations · Springer Lecture Notes in Mathematics 304 (1972); §I.4 nilpotent groups and modules, §I.5 nilpotent fibrations and nilpotent spaces, with the R-completion convergence criterion
Dror — A generalization of the Whitehead theorem · Annals of Mathematics 94 (1971), 530-538; the homology-Whitehead theorem valid on the class of nilpotent spaces, a structural justification for the class
Hall — A contribution to the theory of groups of prime-power order · Proc. London Math. Soc. (2) 36 (1934), 29-95; the lower central series, the basis theorem, and the group-theoretic foundation of nilpotency

Estimated time

beginner: 20m
intermediate: 50m
master: 85m