the Whitehead theorem
/ WYT-hed /
Two spaces that have identical homotopy groups in every dimension feel like they ought to be the same up to deformation. Whitehead's theorem says: yes, but only if you are honest about hypotheses. The catch is that matching homotopy groups abstractly is not enough — you need a single map that realizes the matching, and you need the spaces to be CW complexes. With those provisos, the theorem is the long-awaited converse that lets you certify a homotopy equivalence by checking homotopy groups.
Call a map f: X -> Y a weak homotopy equivalence if it induces isomorphisms on all homotopy groups: f_*: pi_n(X, x_0) -> pi_n(Y, f(x_0)) is an isomorphism for every n >= 0 and every basepoint (including a bijection on pi_0). Whitehead's theorem states: if X and Y are CW complexes and f is a weak homotopy equivalence, then f is an actual homotopy equivalence — there is a homotopy inverse g with g ∘ f ≃ id and f ∘ g ≃ id. The proof leans on cellular approximation and the homotopy extension property of CW pairs, building the inverse and the homotopies cell by cell up the skeleta. The result fails without the CW hypothesis and fails if you only know the groups abstractly match without a map inducing the isomorphisms.
This theorem is why CW complexes are the natural category for homotopy theory and why 'compute all the homotopy groups' is a meaningful strategy: a map of CW complexes inducing isomorphisms on every pi_n is as good as an equivalence. It pairs with the long exact sequence of a fibration (to verify f_* is iso) as a standard toolkit. Two honest cautions that trip people up. First, weak equivalence is about a MAP — two CW complexes with abstractly isomorphic homotopy groups but no map realizing the isomorphisms need NOT be homotopy equivalent (there are classic examples). Second, homotopy groups, not homology, are what Whitehead needs: a map of simply connected CW complexes inducing homology isomorphisms is an equivalence too (a homology Whitehead theorem), but for non-simply-connected spaces homology isomorphism is genuinely weaker and the theorem in that form is false.
To show a CW complex X with pi_n(X) = 0 for all n is contractible, build a map from a point into X (picking a basepoint); it induces isomorphisms on all pi_n (both sides are trivial), so by Whitehead it is a homotopy equivalence and X is contractible. This is the clean reason 'all homotopy groups vanish + CW' implies contractible.
Whitehead converts 'all pi_n vanish' into 'contractible' — but only for CW complexes.
The theorem needs a MAP inducing the isomorphisms, not merely abstractly isomorphic groups: there exist CW complexes (e.g. certain products versus wedges) with isomorphic homotopy groups in every degree that are NOT homotopy equivalent, because no single map realizes the isomorphisms.