Closures and completions
Closures and completions
There is a general pattern in mathematics wherein:
Certain structures are considered somehow “complete” or “closed”
These complete structures can be generated from simpler, non-complete structures
Completing a complete structure produces the same thing (taking closure is idempotent)
Structure | Generator | Generation operation | Idempotent? |
---|---|---|---|
Groups | Group generator sets | inducing the group | Yes? |
Linear spaces | Linear bases | inducing the space | Yes? |
Topological spaces | Topological bases or subbases | inducing the topology | Yes |
Topologically closed sets | Sets | Yes | |
Complete metric spaces | Metric spaces | completion | Yes |
Transitively-closed directed graphs | Directed graphs | transitive closure | Yes |
Details: precisely,
Let me express myself a fourth time. Let’s say we’re working with the group . How I think about is not as a set satisfying the group axioms. Rather, it is the completion of the generating set . Further, under Cayley's theorem, the single element of this generating set should really be thought of as . In other words, what really is, to me, is the structure constructed by starting with the mod- successor function and then “completed” by the allowance of composition and inverses.
I feel like I’m not quite expressing myself well, so hopefully this makes some sense.
It’s worth noting that the situation isn’t always quite so simple. Given a linear basis or group generating set, I can only generate one (smallest) linear space or group. However, given a set and partial operator , there may be more than one magmas I can form. As such, by asking for a magma, I let the theorem invoker choose which closure to use.