Last time:
Lemma 24: Fix basepoints as usual. There is a lift
of
such that
if and only if
.
Furthermore, if the lift exists, it is unique.
Lemma 25: Given a choice of , there exists an elevation
of
at
. Furthermore,
is unique in the sense that if
is another elevation of
and
then there is a homeomorphism
and the diagram

commutes.
Proof:

Let be defined by
. By Lemma 24, the composition
lifts at , call this lift
. Suppose

Then . But by Lemma 24,
. This implies
.
Another, more categorical construction uses the fibre product.

The fibre product is defined by
.
There are obvious maps
given by forgetting factors.
Exercise: If is a covering map then
is a covering map.
Lemma 26: Fix . Let
and let
for
. Let
, and let
be the connected component containing
. Then
is an elevation of
at
, and every elevation of
arises in this way.
Proof: To prove that is an elevation we just observe that
. Now suppose

is an elevation. Then , with
.
The covering map factors trhough
, and so
is a covering map. Because
is an elevation,
is a homeomorphism onto its image, a connected component of
.
What has this got to do with graphs of spaces/groups?
Let be a vector space, and let
be an edge map. Define
to be the mapping cylinder
comes with a map
such that
and
. This is an inclusion
, and
. Let
be a covering map.

Let be the fibre product. There’s a map
;
. Clearly,
.
Therefore, is an injection. It’s easy to see that
induces a bijection at the level
.
Lemma 27: Any covering space arises as the fibre product of a covering map
.
Proof: Let be a covering map

let be the fibre product of
and
.
Define by
. As before,
is a covering map.

2 comments
Comments feed for this article
30 November 2011 at 10.24 am
Ronnie Brown
You might like to look at the exposition of covering spaces and in particular fibre products, or pullbacks, in Chapter 10 of
R. Brown, _Topology and groupoids_, Booksurge, 2006 (available from amazon)
a revised version of editions with different titles published in 1968, 1988. This uses the notion of _covering morphism of groupoids_ and proves an equivalence of categories by the fundamental groupoid functor $\pi_1$
(covering spaces of $X$) $ \to $ (covering groupoids of $\pi_1 X$)
for spaces satisfying the usual local conditions. This also leads to a family of Mayer-Vietoris type exact sequences for a pullback of coverings, which contains the most common uses, including double cosets.
12 December 2011 at 5.13 am
Henry Wilton
Thanks for pointing this out, Ronnie. I think there are several different treatments of this material – there’s also a related paper by Bass. This arose out of reading the work of Wise.