03.06.03 · modern-geometry / characteristic-classes

Stiefel-Whitney classes

shipped3 tiersLean: none

Anchor (Master): Milnor-Stasheff §4–§9; Bott-Tu §23; Husemoller §16

Intuition [Beginner]

Stiefel-Whitney classes are the mod-2 characteristic classes of real vector bundles. They live in the integer-coefficient cohomology, but with a twist: they take values in $Z /2$ , the integers mod 2. So each class is a finite-degree obstruction whose value is either zero or one.

The first Stiefel-Whitney class $w_{1}$ measures orientability: it is zero if and only if the bundle is orientable. The second class $w_{2}$ measures the obstruction to a spin structure. Higher classes measure deeper obstructions to additional structure (string structures, fivebrane structures, etc.).

Stiefel-Whitney classes are the prototype of characteristic-class theory: a small list of cohomology classes capturing global invariants of bundles by purely topological means. They are computable, axiomatically characterised, and underlie much of the topology of manifolds.

Visual [Beginner]

A real vector bundle over a base manifold; Stiefel-Whitney classes are mod-2 cohomology elements detecting structural obstructions — orientability, spin lift, and beyond.

The classes are natural: pulling back the bundle pulls back the classes; the product bundle has all classes zero.

Worked example [Beginner]

The tangent bundle of the sphere $S^{n}$ has Stiefel-Whitney classes that vanish in low degrees:

$w_{1} (T S^{n}) = 0$ (the sphere is orientable for all $n \geq 1$ ).
$w_{2} (T S^{n}) = 0$ (the sphere is spin for all $n$ , since it is simply connected for $n \geq 2$ and orientable).
All higher $w_{k} (T S^{n}) = 0$ except possibly the top class (which is the mod-2 reduction of the Euler class, vanishing for odd $n$ and nonzero for even $n$ ).

The tangent bundle of the real projective plane $RP^{2}$ has $w_{1} (T RP^{2}) \neq = 0$ — projective space is not orientable in even dimension. So $w_{1}$ correctly detects the non-orientability.

The tangent bundle of $RP^{4}$ has $w_{2} (T RP^{4}) \neq = 0$ — projective 4-space is orientable but not spin. So $w_{2}$ correctly detects the spin obstruction.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $E \to M$ be a real vector bundle of rank $n$ over a topological space $M$ 03.05.02, 02.01.01. The Stiefel-Whitney classes of $E$ are cohomology classes

w_{k} (E) \in H^{k} (M; Z /2) for k = 0, 1, 2, \dots

with $w_{0} (E) = 1$ and $w_{k} (E) = 0$ for $k > n$ . They are characterised axiomatically by the Stiefel-Whitney axioms:

Naturality: for a continuous map $f : M^{'} \to M$ and the pullback bundle $f^{*} E$ ,

w_{k} (f^{*} E) = f^{*} w_{k} (E) .

Whitney product formula: for two real vector bundles $E_{1}, E_{2}$ ,

w_{k} (E_{1} \oplus E_{2}) = i + j = k \sum w_{i} (E_{1}) \cup w_{j} (E_{2}),

equivalently $w (E_{1} \oplus E_{2}) = w (E_{1}) \cup w (E_{2})$ for the total Stiefel-Whitney class $w := \sum_{k} w_{k}$ .

Normalisation: for the tautological line bundle $γ_{1} \to RP^{\infty}$ ,

w_{1} (γ_{1}) \in H^{1} (RP^{\infty}; Z /2) ≅ Z /2

is the nonzero generator.

Dimension/vanishing: $w_{k} (E) = 0$ for $k > rank (E)$ .

These four axioms determine the Stiefel-Whitney classes uniquely.

Equivalent formulations.

Via Steenrod squares on the Thom class: $w_{k} (E)$ is the unique cohomology class such that $Sq^{k} (u) = (π^{*} w_{k} (E)) \cup u$ where $u \in H^{n} (Th (E); Z /2)$ is the Thom class and $π : Th (E) \to M$ is the bundle projection.
Via the cohomology of the classifying space: $H^{*} (B O (n); Z /2) = Z /2 [w_{1}, w_{2}, \dots, w_{n}]$ , a polynomial ring in the universal Stiefel-Whitney classes; pullback along the classifying map of $E$ gives $w_{k} (E)$ .

Standard interpretations:

$w_{1} (E) = 0$ iff $E$ is orientable.
For an orientable bundle, $w_{2} (E) = 0$ iff $E$ admits a spin structure.
For a spin bundle, $\frac{1}{2} p_{1} (E) = 0$ iff $E$ admits a string structure (a higher cohomological lift); see 03.09.04.

Key theorem with proof [Intermediate+]

Theorem (existence and uniqueness of Stiefel-Whitney classes). There is a unique assignment of cohomology classes $w_{k} (E) \in H^{k} (M; Z /2)$ to real vector bundles satisfying the four axioms (naturality, Whitney product, normalisation, dimension).

Proof sketch. Uniqueness. Any Stiefel-Whitney theory is determined by its values on the universal bundle $γ_{n} \to B O (n)$ , by naturality. The cohomology ring $H^{*} (B O (n); Z /2) = Z /2 [w_{1} (γ_{n}), \dots, w_{n} (γ_{n})]$ is a polynomial ring in $n$ generators in degrees $1, 2, \dots, n$ . The normalisation axiom fixes $w_{1} (γ_{1})$ for the universal line bundle on $B O (1) = RP^{\infty}$ . The Whitney product formula extends to all $w_{k}$ via the identification of $B O (n)$ with the $n$ -fold product of universal line bundle factors (after splitting principle), forcing the $w_{k} (γ_{n})$ to be the elementary symmetric polynomials in line-bundle Chern roots (mod 2).

So the universal classes are uniquely determined, and naturality forces all classes for all bundles.

Existence. Construct $w_{k}$ as follows: for the universal bundle on $B O (n)$ , take $w_{k}$ to be the elementary symmetric polynomials in the Stiefel-Whitney classes of the universal line bundle factors (after the natural map $B O (1)^{n} \to B O (n)$ — the splitting). Verify the axioms hold (the constraints are compatible with the cohomology ring structure of $B O (n)$ ). Naturality propagates this to general bundles via the classifying-map construction. $□$

The Steenrod square approach gives an alternative direct construction: $Sq^{k}$ on the Thom class $u$ gives a class $Sq^{k} (u) \in H^{n + k} (Th (E); Z /2)$ , which under the Thom isomorphism corresponds to a class in $H^{k} (M; Z /2)$ — defined to be $w_{k} (E)$ . The Steenrod-square approach is the most computational.

Bridge. The construction here builds toward 03.06.04 (pontryagin and chern classes), where the same data is upgraded, and the symmetry side is taken up in 03.09.04 (spin structure on an oriented riemannian manifold). The defining pattern appears again in those units in a sharpened form, where the local data is glued or quotiented. Putting these together, the foundational insight is that the data of this unit gives the structural signature that the rest of the strand reads off.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

lean_status: none — Mathlib lacks bundled characteristic classes.

[object Promise]

Advanced results [Master]

Wu's formula. Stiefel-Whitney classes are computable from Steenrod squares on the manifold's cohomology. Wu's formula expresses the Wu classes $v_{k}$ (defined by $Sq^{i} x = v_{i} \cup x$ on top-degree classes) in terms of Stiefel-Whitney classes of the tangent bundle: $v = Sq^{- 1} (w (T M))$ in suitable graded sense. Wu's formula reduces computation of Stiefel-Whitney classes to computation of Steenrod squares plus knowledge of the cohomology ring.

Cobordism invariants. The total Stiefel-Whitney number $w_{i_{1}} w_{i_{2}} \dots w_{i_{k}} [M]$ (evaluated on the fundamental class) is a cobordism invariant: two closed manifolds are unoriented-cobordant iff all their Stiefel-Whitney numbers agree. This is the bridge between Stiefel-Whitney classes and Thom's cobordism theory: $Ω_{n}^{O} ≅ π_{n} (M O)$ , the homotopy of the universal Thom spectrum.

Higher obstructions and Whitehead tower. The vanishing of successive Stiefel-Whitney / Pontryagin classes corresponds to lifts up the Whitehead tower:

$w_{1} = 0$ : orientation, lifting $O (n) \to SO (n)$ .
$w_{2} = 0$ : spin, lifting $SO (n) \to Spin (n)$ .
$\frac{1}{2} p_{1} = 0$ : string, lifting $Spin (n) \to String (n)$ .
Higher: fivebrane, etc.

Each lift is gated by a cohomological obstruction in $H^{k} (M; Z /2)$ or $H^{k} (M; Z)$ .

Classifying space cohomology. $H^{*} (B O; Z /2) ≅ Z /2 [w_{1}, w_{2}, w_{3}, \dots]$ , the polynomial ring in countably many generators. This is the "universal Stiefel-Whitney ring" — every real vector bundle's classes pull back from here. Compare with $H^{*} (B U; Z) ≅ Z [c_{1}, c_{2}, \dots]$ for complex bundles (Chern classes).

Connection to Stiefel manifolds. The original construction by Stiefel and Whitney was via Stiefel manifolds $V_{k} (R^{n})$ of orthonormal $k$ -frames. The Stiefel-Whitney class $w_{n - k + 1} (E)$ is the obstruction to extending an $(n - k + 1)$ -frame on a $k$ -skeleton to a $k$ -frame on $(k + 1)$ -skeleton. This was the historical origin of the theory and remains the most geometric viewpoint.

Synthesis. This construction generalises the pattern fixed in 03.05.02 (vector bundle), with the symmetric data replaced by its skew or twisted analogue. Read in the opposite direction, the construction is dual to the metric story: complements and orthogonality are taken with respect to the bilinear datum of this unit, not a metric, and the resulting category of subobjects is the one the rest of the strand classifies. The central insight is that this datum identifies algebra with geometry: functions become vector fields, subspaces become quotients, and invariants become cohomology classes — and that identification is the engine driving every theorem downstream.

Full proof set [Master]

Existence and uniqueness via universal bundle. Sketched in §"Key theorem". Full uniqueness uses the polynomial-ring structure of $H^{*} (B O (n); Z /2)$ . Full existence uses the Steenrod-square / Thom-class construction.

Whitney product formula. Proved on the universal bundle (where $B O (m) \times B O (n) \to B O (m + n)$ classifies the direct sum), then propagated by naturality.

$w_{1}$ orientability. A bundle is orientable iff its $SO (n)$ -reduction exists iff the classifying map $M \to B O (n)$ lifts through $B S O (n) \to B O (n)$ , with fibre $K (Z /2, 0) = Z /2$ . Obstruction theory: lift exists iff a single class in $H^{1} (M; Z /2)$ vanishes — namely $w_{1} (E)$ .

$w_{2}$ spin obstruction. Proved in Exercise 6 — obstruction theory applied to $B S p in (n) \to B S O (n)$ with fibre $K (Z /2, 1)$ .

Pullback formula. Naturality is direct from the classifying-map formulation: $w_{k} (f^{*} E) = w_{k}$ pulled back along $M^{'} \to M \to B O (n)$ .

Connections [Master]

Vector bundle 03.05.02 — the input.
De Rham cohomology 03.04.06 — Stiefel-Whitney classes live in mod-2 cohomology, related to integral / rational cohomology by reduction.
Spin structure 03.09.04 — $w_{2}$ is the obstruction.
Pontryagin and Chern classes 03.06.04 — the integral / complex companions.
Smooth manifold 03.02.01 — the underlying setting.

Historical & philosophical context [Master]

Eduard Stiefel (1936) introduced the classes that bear his name in his Zürich thesis, in the context of the obstruction to extending sections of frame bundles. Hassler Whitney (1937) independently introduced the same classes from the cohomological side and proved the product formula. Steenrod (1947) gave the modern Steenrod-square interpretation that makes the classes mod-2 cohomology operations.

Milnor and Stasheff's monograph (1974) systematised the theory: axioms, classifying spaces, cobordism applications, computational examples. The relation to the Whitehead tower and higher gauge theory (string structures, fivebrane structures) is a more recent development (Stolz-Teichner, Sati-Schreiber-Stasheff), tying Stiefel-Whitney classes to elliptic cohomology and the chromatic structure of stable homotopy theory.

In modern physics, vanishing of Stiefel-Whitney classes signals the existence of fermionic / spinorial physical structures: $w_{1} = 0$ for matter to have a definite chirality (orientation), $w_{2} = 0$ for a spin structure to define Dirac fermions, $\frac{1}{2} p_{1} = 0$ for heterotic string anomaly cancellation.

Bibliography [Master]

Milnor, J. & Stasheff, J., Characteristic Classes, Princeton University Press, 1974. §4–§9.
Bott, R. & Tu, L. W., Differential Forms in Algebraic Topology, Springer GTM 82, 1982. §23.
Husemoller, D., Fibre Bundles, Springer, 3rd ed., 1994. §16.
Stiefel, E., "Richtungsfelder und Fernparallelismus in n-dimensionalen Mannigfaltigkeiten", Comm. Math. Helv. 8 (1936), 305–353.
Whitney, H., "On the topology of differentiable manifolds", Lectures in Topology (1941), 101–141.

Wave 4 Strand B unit #5. Stiefel-Whitney classes — the mod-2 characteristic classes of real vector bundles; foundation for orientability, spin structures, and the Whitehead-tower obstruction theory.