A CW-complex is a topological space constructed by successively attaching closed disks in increasing dimension, with the topology specified below. It generalizes the notion of a simplicial complex (Hatcher 2002, pp. 5-6).
Start with a discrete space ,
whose points are called 0-cells. For each integer
, attach a collection
of
-dimensional closed
disks
to
by continuous
maps
from their boundaries, where
is the
-sphere. The resulting n-skeleton
is the quotient
space of the topological disjoint union of
and the closed
disks, identifying each boundary point
with
.
The images of the interiors
of the closed disks are the
-cells, each homeomorphic
to an
-dimensional
open disk. There may be infinitely many
-cells (Hatcher 2002, p. 5).
Set . A subset
is declared closed iff
is closed in
for every nonnegative
integer
. This condition specifies the topology
on the union, including when the cells have unbounded dimension (Hatcher 2002, pp. 5 and 519).
For example, a circle has a CW-complex structure with one 0-cell and one 1-cell. Attach both endpoints of
the closed interval to the same point; its interior
is the 1-cell. Thus an attaching map need not be injective
(Hatcher 2002, p. 6).
By the CW-approximation theorem, every topological space admits a weak homotopy
equivalence
from a CW-complex
. This map
also induces isomorphisms on singular
homology and singular cohomology groups
(Hatcher 2002, pp. 352-353 and 356). Whitehead's
theorem states that a weak homotopy equivalence
between CW-complexes is a homotopy equivalence
(Hatcher 2002, p. 352).