The separation axioms are meant to encode a certain concept of distance on your space. Full metric spaces satisfy all of the separation axioms; as you satisfy fewer and fewer of these axioms, your concept of distance weakens accordingly.
Pseudometric Spaces
When points aren't closed (ie. not T1) we can have generic points whose closures are massive.
Zariski Topology on Spec(R)
T0 but not T1.
Scott Topology
T0 but not T1
Alexandrov Topology on a Poset
T0 but not T1.
A topological space is T1 if every singleton is a closed set.
Zariski Topology
T1 but not T2 unless the space is finite Every non-empty set is dense
Has unique limits for nets and filters Compact sets are closed.
K-Topology
Let K = {1/n ∣ n ∈ ℕ > 0}. The K-topology on ℝ contains all the open intervals plus all sets of the form (a,b) \ K
It's T2 but not T3.
A space is T3 if it is T1 and every closed set C and point x ∉ C can be separated by disjoint open sets.
The space is locally uniform You get to say that closed sets have sharp edges, and points outside of them have a certain amount of distance.
Moore-Niemytzki Plane
Consider ℍ ≥ 0 = {(x,y) ∈ ℝ2 ∣ y ≥ 0}
For points above the x-axis, open sets are the ordinary open sets in ℝ2. For points on the x-axis, basic open neighbourhoods are open disks that hit the x-axis at exactly that point
T3 but not T4
Sorgenfrey Product
ℝℓ × ℝℓ is T3 but not T4.
Helps you to assign things like energy levels, coordinate charts, measures, etc.
Sorgenfrey Line
The real line with lower limit topology is T4 but not metrizable.