Topological K-theory

Intuition [Beginner]

Topological K-theory is a way to count vector bundles when ordinary counting is too rigid. Two bundles can be added by direct sum, just as two piles can be combined. But subtraction is not always visible as an actual bundle.

K-theory fixes this by allowing formal differences. A class can look like $E-F$, meaning "bundle $E$ minus bundle $F$" in a controlled bookkeeping system.

This bookkeeping is powerful because vector bundles carry geometric information about the base space. K-theory packages that information into an abelian group.

Visual [Beginner]

Bundles add by stacking fibers. K-theory then permits stable differences between stacks.

The result is not a single bundle in general. It is a stable class that remembers how bundles compare after extra summands are allowed.

Worked example [Beginner]

Over a point, a complex vector bundle is just a complex vector space. Its only invariant is dimension.

The bundle ${\mathbb{C}}^3$ represents $3$, and ${\mathbb{C}}^1$ represents $1$. In K-theory, the formal difference ${\mathbb{C}}^3-{\mathbb{C}}^1$ represents $2$.

So $K(\mathrm{point})$ behaves like the integers. Ordinary bundles give nonnegative dimensions, and K-theory adds the missing negative and difference classes.

What this tells us: K-theory turns vector-bundle addition into group-valued topology.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a compact Hausdorff space. Let ${\mathrm{Vect}}_{\mathbb{C}}(X)$ be the commutative monoid of isomorphism classes of complex vector bundles over $X$, with addition induced by direct sum 03.05.02. The complex topological K-group $K^0(X)$ is the Grothendieck group completion of this monoid ^{[Atiyah §1]}.

Elements of $K^0(X)$ may be represented by formal differences $[E]-[F]$. The relation is stable: $[E]-[F]=[E^{\prime }]-[F^{\prime }]$ when there exists a bundle $H$ such that

$$E\oplus F^{\prime }\oplus H\cong E^{\prime }\oplus F\oplus H.$$

A continuous map $f:Y\to X$ induces a pullback homomorphism $f^{\ast }:K^0(X)\to K^0(Y)$.

Key theorem with proof [Intermediate+]

Theorem (Universal property of group completion). Let $M$ be a commutative monoid. Its Grothendieck group $G(M)$ has a monoid map $\eta :M\to G(M)$ such that every monoid map from $M$ to an abelian group $A$ factors uniquely through a group homomorphism $G(M)\to A$.

Proof. Construct $G(M)$ from pairs $(m,n)\in M\times M$ with the relation

$$(m,n)\sim (m^{\prime },n^{\prime })⟺m+n^{\prime }+r=m^{\prime }+n+r$$

for some $r\in M$. Addition is defined by $(m,n)+(p,q)=(m+p,n+q)$, and the inverse of $(m,n)$ is $(n,m)$. The map $\eta (m)$ is the class of $(m,0)$.

Let $u:M\to A$ be a monoid map into an abelian group. Define $\overline{u}([(m,n)])=u(m)-u(n)$. The defining relation makes this independent of representatives, and the formula is additive. It satisfies $\overline{u}\circ \eta =u$.

If $v:G(M)\to A$ also satisfies $v\circ \eta =u$, then $v([(m,n)])=v(\eta (m)-\eta (n))=u(m)-u(n)$. Thus $v=\overline{u}$. $\square$

Bridge. The construction here builds toward 03.08.07 (bott periodicity), where the same data is developed in the next layer of the strand. 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 is recorded because the geometric vector-bundle construction of $K^0(X)$, its functoriality, and its connection to Bott periodicity are not yet packaged in Mathlib.

Advanced results [Master]

For compact Hausdorff $X$, $K^0(X)$ is the Grothendieck group of complex vector bundles, and reduced K-theory is the kernel

$${\tilde{K}}^0(X)=\ker (K^0(X)\to K^0(\ast ))$$

for a based connected compact space ^{[Atiyah §1]}. Higher groups are defined by suspension, for example ${\tilde{K}}^{-n}(X)={\tilde{K}}^0({\Sigma }^{n}X)$ in one common convention.

The representing-space viewpoint identifies stable vector-bundle data with maps into $BU\times \mathbb{Z}$. Bott periodicity then supplies the two-periodic structure of complex K-theory 03.08.07.

Synthesis. Topological K-theory converts the geometric data of vector bundles into an algebraic invariant through the Grothendieck group completion: direct sum becomes addition, stable equivalence becomes equality, and the contravariant pullback makes $K^0$ a functor from compact spaces to abelian groups. The reduced theory measures the failure of a bundle to be stably isomorphic to a product, Bott periodicity collapses the infinite suspension tower into a two-periodic pattern, and the Chern character relates K-theory to ordinary cohomology after rationalization. The representing space $BU\times \mathbb{Z}$ turns K-theoretic computations into homotopy classes of maps, while the Atiyah-Singer index theorem connects the K-theoretic topological index to the analytical index of elliptic operators — bridging topology, analysis, and geometry through the single invariant of virtual bundle rank.

Full proof set [Master]

The construction of $K^0(X)$ is the Grothendieck group completion applied to ${\mathrm{Vect}}_{\mathbb{C}}(X)$. The theorem above proves its universal property. Pullback functoriality follows because $f^{\ast }$ preserves direct sums and isomorphism classes, so it defines a monoid homomorphism before group completion and a group homomorphism afterward.

For the point, ${\mathrm{Vect}}_{\mathbb{C}}(\ast )\cong \mathbb{N}$ by dimension, and the Grothendieck group of $\mathbb{N}$ is $\mathbb{Z}$. Therefore $K^0(\ast )\cong \mathbb{Z}$.

Connections [Master]

• Topological K-theory depends on vector bundles 03.05.02 and classifying spaces 03.08.04. It is the input for Bott periodicity 03.08.07, which turns the suspension-defined groups into a periodic theory. It also supplies the symbol class target in the Atiyah-Singer index theorem 03.09.10.

• Characteristic classes 03.06.04 give natural transformations out of K-theory after rationalization through the Chern character. This ties the vector-bundle bookkeeping of K-theory to the differential-form representatives built by Chern-Weil theory 03.06.06.

• Chern character ring homomorphism 03.06.18. The Chern character $\mathrm{ch}:K(X)\to H^{\mathrm{even}}(X;\mathbb{Q})$ is the explicit natural ring homomorphism that becomes an isomorphism after rationalising: $K(X)\otimes \mathbb{Q}\cong H^{\mathrm{even}}(X;\mathbb{Q})$ for finite CW complexes. The foundational reason topological K-theory is a rational refinement of even cohomology is exactly that the Chern character of 03.06.18 gives the comparison, with integral discrepancies controlled by torsion phenomena tracked through the Atiyah-Hirzebruch spectral sequence. Anchor phrase: the rational $K\to H$ iso.

Historical & philosophical context [Master]

Atiyah's K-Theory develops the subject from vector bundles, Grothendieck groups, exact sequences, and Bott periodicity ^{[Atiyah §1]}. The notation and functorial viewpoint used here follow that source.

Husemoller presents K-theory in the language of fibre bundles, emphasizing the direct-sum monoid and its group completion ^{[Husemoller §10]}.

Bibliography [Master]

• Michael Atiyah, K-Theory, §1-§2. ^{[Atiyah §1]}
• Dale Husemoller, Fibre Bundles, §10. ^{[Husemoller §10]}
• Alexander Grothendieck, work on the group completion idea in algebraic geometry and K-theory. ^{[Grothendieck K-theory]}