Why is a disk with a $0$-cell for every boundary point not a CW-complex by the inductive definition?

82 Views Asked by At

In Hatcher's Algebraic Topology, he defines a CW complex as follows:

Definition of CW complex

He proves in Proposition A.1. that any space constructed this was will satisfy the closure finiteness property, and then gives an example of a cell complex which doesn't:

CW complex counterexample

My question is, how is this a counterexample? It seems to me that this can be constructed in the way he presents in his CW complex definition. Can someone explain to me what's going on?

1

There are 1 best solutions below

0
On BEST ANSWER

In part (2) of the definition, "map" means continuous map, so the attaching maps $\varphi_\alpha: S^{n-1}\to X^{n-1}$ are required to be continuous. This is not true in the counterexample: the $0$-skeleton is $S^1$ with the discrete topology (by part (1) of the definition), and so the identity map $S^1\to X^0=S^1$ which would attach the $2$-cell is not continuous (the domain has the usual topology of $S^1$ but the codomain has the discrete topology).