# CW complex structure on a disk with 2 smaller disks removed

For some reason, I'm having trouble visualizing how to put a CW structure on a disk with 2 smaller disks removed. What I'd like to do is have three 0-cells, five 1-cells, and a single 2-cell. Three of the one cells would be glued to the three vertices to give three disjoint circles, and the other two 1-cells would be used to connect the three circles together. Then I can't see how to glue the 2-cell in, because when I visualize it, it always ends up covering one of the holes in the disk.

I realize my ability to visualize CW-complexes is not so good. For another example, it seems to me that if you take a wedge of two circles A and B, and glue a 2-cell on by $AB$, you should get a wedge of two disks, which would be contractible. But using cellular homology I see this has a nontrivial first homology group. I'd appreciate help sorting these two examples out to aid my intuition.

-
A disk with 2 smaller disks removed has nontrivial first homology group $H_1(S)=\Bbb{Z}\oplus\Bbb{Z}$, being homotopy equivalent to the wedge of two circles, no? – Managu Aug 3 '12 at 1:25
Sure. The problem really is to put a CW structure on this, not to calculate anything. – John Aug 3 '12 at 1:27

Your thoughts on how to arrange the 0- and 1-cells are spot on. If you can draw the structure with just the 0- and 1-cells, you should see a good place to put the 2-cell. If you get stuck, I've drawn a picture:

http://imgur.com/P4Bc6

For the second example, recall that you're not actually filling in either of $A$ or $B$. Attaching a 2-cell via $AB$ means creating something like what you get when you pinch one of these:

http://imgur.com/QcZfi

by holding a point on the top and bottom rings in your fist. Alternately, you can picture a piece of macaroni noodle with the smaller arc connecting the ends shrunk to a point. In fact, such a space is homotopy equivalent to a cylinder- simply unscrunch the join point of the circles into an interval. If you're curious about this sort of thing, some treatment is given in chapter 0 of Hatcher.

-
Thank you, this was exactly what I had thought, but for some reason didn't think it worked. – John Aug 3 '12 at 1:46

Please allow me to present my terrible GIMP sketch:

So we have points p1, p2, and p3, lines L1, L2, L3, L4, and L5, and I didn't label the obvious 2-cell. If I were to describe the attaching map of the 2-cell, I might start at p1, attach along L4, attach along L1, attach along -L4, attach along L5, attach along L2, attach along -L5, and attach along -L3 (ending at p1 again). I use +- to indicate direction, using the arrows on my rough sketch as the guide.

And as we can see - the fundamental group is very clearly nontrivial.

-