Cellular Homology

Homology from a handful of cells: the cellular chain complex, degrees of attaching maps, and the computations that finally make homology practical.

The idea

Cellular homology is a method for computing the homology groups $H_{n}(X)$ of a CW complex, and it is the method that makes those groups computable in practice. Simplicial homology needs a triangulation, which is laborious and not always available. Singular homology is defined for every space, but its chain group in each degree is free on all continuous maps of a simplex into $X$ — an uncountable basis, from which nothing can be computed by hand.

A CW structure is already a finite presentation: a few cells and their attaching maps. Cellular homology converts it directly into a chain complex whose $n$-th group $C_{n}$ is free abelian on the $n$-cells, joined by boundary maps $d$. Where a surface carried uncountably many singular simplices, it now carries one or two cells per dimension.

Theorem (Cellular homology).

Let $X$ be a CW complex, let $C_{n}(X)$ be the free abelian group on its $n$-cells, and let $d_{n} \colon C_{n}(X) \to C_{n-1}(X)$ be the cellular boundary maps. Then for every $n$, $\ker d_{n} \,/\, \operatorname{im} d_{n+1} \;\cong\; H_{n}(X),$ the singular homology of $X$.

In particular the answer does not depend on which cell structure we chose. The computation then rests on the boundary maps, and each entry of one is an integer: the degree of a map of spheres, recording how many times, and with which signs, the boundary of one cell sweeps across another.

Ways to work on it

Not sure where to start? Take the ten-question placement test.