Name for a subset of real space being nowhere a manifold with boundary

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
5 replies · 2K views
Science Advisor
Messages
1,864
Reaction score
34
I was wondering if anyone knew of a name for such a set, namely a subset [itex]S \subseteq \mathbb{R}^n[/itex] which at every point [itex]x \in S[/itex] there exists no open subset [itex]U[/itex] of [itex]\mathbb{R}^n[/itex] containing [itex]x[/itex] such that [itex]S \cap U[/itex] is homeomorphic to either [itex]\mathbb{R}^m[/itex] or the half-space [tex]\mathbb{H}^m = \{(y_1,...,y_m) \in \mathbb{R}^m : y_m \geq 0\}[/tex] for any integer [itex]m \geq 0[/itex]. Of course, any set for which such open sets [itex]U[/itex] exists for every [itex]x[/itex] is called an embedded manifold with boundary. I'm looking for the opposite notion, in a sense.
 
Last edited:
Physics news on Phys.org
I've corrected my question a bit, it now more accurately reflects the title. The complement of a dense set for the original question (where [itex]m[/itex] had the fixed value [itex]n[/itex]) could very well be the correct answer. I can't prove it right now, but I'll look into it. For general [itex]m[/itex] however, it's not the right notion. For example, I'd like the graph of a nowhere continuous function to fit the bill, but not the graph of a continuous function.

EDIT: Upon some further reflection it dawns on me that the half-space condition is unecessary. If a space is isomorphic to the half-space locally around any point, then it is necessarily isomorphic to euclidean space at a nearby point. Hence the half-space condition can be dropped entirely without changing the question.
 
Last edited:
Office_Shredder said:
I missed the ##m \neq n## idea, sorry.

It was my fault, I didn't include it in the original formulation :smile:
 
It still needs to be the complement of a dense set I think, but not every complement works (because {0} is the complement of a dense set for example).