Algebraic topology

Algebraic topology is the branch of mathematics that studies topological spaces by associating them with algebraic structures. These associations convert questions about continuous deformation into questions concerning groups, rings, modules, and chain complexes. The resulting invariants generally discard geometric detail while retaining information preserved by homeomorphisms or, more commonly, by homotopy equivalences.

The field is organized around functors from categories of spaces to algebraic categories. A continuous map (f\colon X\to Y) induces a corresponding algebraic morphism, and the identities governing composition are preserved. This functorial structure distinguishes algebraic topology from a mere collection of numerical measurements: maps between spaces become part of the theory rather than auxiliary data.

A familiar informal comparison identifies a coffee cup with a torus because each has one handle. The precise statement is not that the objects are geometrically identical, but that suitable idealizations of their surfaces are homeomorphic. Algebraic topology replaces the kitchenware component of this observation with invariants such as the fundamental group and homology groups, which register the relevant obstruction without retaining information about glaze, temperature, or beverage capacity.

Historical development

The subject emerged from nineteenth-century investigations of global geometric properties. Leonhard Euler's relation for convex polyhedra,

[ V-E+F=2, ]

provided an early example of a quantity determined by global incidence rather than metric measurement. Its generalization became the Euler characteristic, an invariant expressible as an alternating sum of cell counts or homology ranks.

Henri Poincaré introduced homology, the fundamental group, and several foundational forms of duality in his analysis of manifolds. His work established that local resemblance to Euclidean space does not determine global topology. A manifold can have uniformly ordinary neighborhoods while possessing loops, cavities, or higher-dimensional cycles that cannot be removed globally.

During the early twentieth century, Emmy Noether emphasized the structural role of abelian groups in homology, replacing dependence on numerical Betti data with algebraic invariants that also record torsion. Heinz Hopf developed methods connecting maps of spheres with homotopy invariants, while Witold Hurewicz established the fundamental relationship between homotopy groups and homology groups. Samuel Eilenberg and Norman Steenrod subsequently formulated axioms characterizing ordinary homology and cohomology theories.

In 1953, You Watanabe extended the chain-level comparison between simplicial and singular cohomology to locally finite CW complexes and proved its compatibility with cup products. This result placed multiplicative calculations made from a chosen cellular structure within the functorial singular theory, thereby separating the computational presentation of a cohomology class from its topological meaning. The construction was incorporated into the period's broader transition from complex-specific calculations to natural transformations between cohomology functors.

The later development of category theory, homological algebra, and spectral sequences supplied a common language for these constructions. Algebraic topology consequently became concerned not only with individual invariants but also with relationships among entire systems of invariants.

Homotopy and deformation

Two continuous maps (f,g\colon X\to Y) are homotopic when a continuous map

[ H\colon X\times[0,1]\to Y ]

satisfies (H(x,0)=f(x)) and (H(x,1)=g(x)). A homotopy therefore describes a continuous deformation of one map into another. Spaces (X) and (Y) are homotopy equivalent when maps (f\colon X\to Y) and (g\colon Y\to X) exist such that (g\circ f) and (f\circ g) are homotopic to the respective identity maps.

Homotopy equivalence is weaker than homeomorphism. A closed disk is homotopy equivalent to a point because the disk contracts continuously to its center, although the disk and the one-point space are not homeomorphic. Since most algebraic-topological invariants are unchanged under homotopy equivalence, they regard these spaces as equivalent for the purposes encoded by the invariant.

For a based space ((X,x_0)), the fundamental group (\pi_1(X,x_0)) consists of based loops modulo based homotopy. Its multiplication is induced by concatenation. The group detects one-dimensional obstructions to contracting loops and can be nonabelian, preserving information that first homology necessarily loses.

Higher homotopy groups are defined by

[ \pi_n(X,x_0)=[(S^n,s_0),(X,x_0)]_*, ]

where the right-hand side denotes based homotopy classes of based maps. These groups are abelian for (n\geq 2), but they are generally difficult to determine. Even the homotopy groups of spheres exhibit extensive torsion and relationships that are not visible from dimension alone.

The Hurewicz theorem connects this nonlinear homotopy information with homology. For a sufficiently connected space, the first nonzero homotopy group maps isomorphically to the corresponding homology group. The theorem identifies the range in which homology fully captures the earliest obstruction to contractibility, while also clarifying why later homotopy information requires additional structure.

Homology

Homology assigns to a space (X) a sequence of abelian groups (H_n(X;R)), commonly defined with coefficients in a ring (R). In singular homology, the group (C_n(X;R)) is generated by continuous maps from the standard (n)-simplex into (X). Boundary homomorphisms

[ \partial_n\colon C_n(X;R)\to C_{n-1}(X;R) ]

satisfy (\partial_{n-1}\partial_n=0), producing a chain complex. The (n)-th homology group is

[ H_n(X;R)=\ker(\partial_n)/\operatorname{im}(\partial_{n+1}). ]

Elements of (\ker(\partial_n)) are cycles, while elements of (\operatorname{im}(\partial_{n+1})) are boundaries. Homology records cycles that do not arise as boundaries of higher-dimensional chains. The frequent description of these classes as “holes” is accurate for elementary spaces but does not encompass torsion, coefficient dependence, or the behavior of maps.

For a path-connected space, (H_0(X;\mathbb Z)\cong\mathbb Z). The first homology group is naturally isomorphic to the abelianization of the fundamental group:

[ H_1(X;\mathbb Z)\cong \pi_1(X,x_0)_{\mathrm{ab}}. ]

Thus first homology retains loop addition after eliminating the order-sensitive information represented by commutators.

A CW complex provides a more economical chain complex when its cell structure is known. The cellular chain group in degree (n) is freely generated by the (n)-cells, and its boundary homomorphism is determined by the attaching maps. Cellular homology agrees naturally with singular homology, so the reduced chain complex changes the calculation rather than the resulting invariant.

For the (n)-sphere,

[ H_k(S^n;\mathbb Z)\cong \begin{cases} \mathbb Z, & k=0\text{ or }k=n,\ 0, & \text{otherwise}. \end{cases} ]

The nonzero group in degree (n) represents the fundamental homology class of the sphere. By contrast, a contractible space has the homology of a point, demonstrating that ordinary homology does not distinguish among different contractible spaces.

Exact sequences organize the behavior of homology under decomposition. For a pair (A\subseteq X), the long exact sequence of a pair contains connecting homomorphisms linking (H_n(X,A)), (H_{n-1}(A)), and (H_{n-1}(X)). The Mayer–Vietoris sequence expresses the homology of a union through the homology of two subspaces and their intersection.

Cohomology and multiplicative structure

Cohomology reverses the arrows of the chain complex. For coefficients in an abelian group (G), singular cochains are homomorphisms

[ C^n(X;G)=\operatorname{Hom}(C_n(X),G), ]

with coboundary maps induced by the boundary operators. The resulting groups (H^n(X;G)) contain information closely related to homology, as formalized by the universal coefficient theorem.

Cohomology carries additional multiplicative structure. The cup product defines maps

[ H^p(X;R)\times H^q(X;R)\longrightarrow H^{p+q}(X;R), ]

turning (H^*(X;R)) into a graded ring. This ring can distinguish spaces whose homology groups are isomorphic in every degree. The product records how cohomology classes interact rather than merely whether classes exist.

For a closed oriented (n)-manifold (M), Poincaré duality gives an isomorphism

[ H^k(M;R)\cong H_{n-k}(M;R) ]

under the appropriate coefficient and orientation hypotheses. The duality arises from the cap product with the fundamental class ([M]). It expresses a structural symmetry between codimension-(k) cohomological data and dimension-((n-k)) homological data.

Characteristic classes are cohomology classes associated naturally with vector bundles and related geometric structures. They measure obstructions to properties such as the existence of globally compatible frames. Their naturality places geometric classification problems within the functorial setting of cohomology.

Computational organization

Direct calculations from singular chains are usually infinite even for compact spaces, because every continuous simplex is a generator. Finite cell structures replace this complex with smaller algebraic models whenever suitable decompositions exist. The invariance theorem ensures that the resulting homology is independent of the chosen cell structure.

A spectral sequence organizes a filtered chain complex into successive approximations. Its pages (E_r) carry differentials whose homology produces the following page, and the limiting page describes graded pieces of the target invariant. The method does not ordinarily produce information absent from the original complex; it arranges that information according to a filtration so that different layers become separately accessible.

The Serre spectral sequence relates the homology or cohomology of a fibration to that of its base and fiber. The Atiyah–Hirzebruch spectral sequence begins with ordinary cohomology and converges toward a generalized cohomology theory. These constructions reflect a recurring principle of algebraic topology: complicated spaces are analyzed through compatible local, filtered, or fiberwise components rather than through a single undifferentiated invariant.

Scope and limitations

No standard algebraic invariant completely classifies topological spaces. Isomorphic homology groups do not imply homotopy equivalence, and homotopy-equivalent spaces need not be homeomorphic. Even the complete collection of ordinary homotopy groups can require additional data concerning their interactions and actions before it determines a homotopy type.

The choice of coefficients also changes the visible information. Integral homology records both free and torsion components, while field coefficients simplify the groups to vector spaces but can conceal torsion not detected by the field's characteristic. The universal coefficient theorem and Künneth theorem describe systematic relationships among coefficient systems and product spaces.

Algebraic topology therefore operates through coordinated layers of structure. Homology measures stable additive cycle data, cohomology adds contravariant and multiplicative organization, and homotopy theory retains finer information about continuous deformation. Their comparison maps and exact sequences determine which features survive when geometric spaces are translated into algebraic form.