This lemma is one crucial result in general point-set topology, essential to help us establish conditions for topological spaces to be metrizable. I’m most certainly not the one who came up with this idea, but this is the lemma in its full glory.
Theorem 1 (Urysohn’s Lemma). Let be a topological space. Then
is normal if and only if for disjoint closed subsets
, there exists a continuous function
such that
and
.
Proof. For the easy direction , the disjoint closed subsets
are separated by the neighbourhoods
and
.
For the difficult direction , fix disjoint closed subsets
. Define
and
. We note that
. By normality, there exists a neighbourhood
of
such that
In fact, for any two rational numbers , there exists a neighbourhood
of
such that
Define the dyadic rationals as rational numbers of the form
, where
are positive integers and
. Then inductively, for each
, a
exists as per our construction above.
The rough idea is to define to be the “effectively smallest” possible
such that
. If
for any
, then
, which means that
. Define
whenever
, so that
. Otherwise,
for some
. Define the we;;-defined mapping
by
We leave it as an exercise in real analysis to verify the following implications:
- If
, then
.
- If
, then
.
In particular, if , then
, which means
. All that remains is to prove that
is continuous on
.
Fix and
. We need to prove that
is open. Since
, find dyadic rationals
such that
By the construction of ,
. Furthermore, for any
,
. This implies
Hence, , so that
is open, as required.
With Urysohn’s lemma, we can state the big result that we aim to prove…eventually.
Theorem 2 (Urysohn Metrization Theorem). If is regular and second-countable, then it is metrizable.
The goal is to find a continuous injection that is a homemorphism with
, where
is metrizable. Since subspaces of metrizable spaces are metrizable,
is metrizable, and thus
will be metrizable too. The challenge is constructing an
that is metrizable, and then constructing the required injection
. As such, we need to take a detour into metrizable spaces in order to establish this result. This we will do, next time.
—Joel Kindiak, 25 Apr 25, 1531H
Leave a comment