Question #116445

Prove that the boundary of a subset A of a metric space X is always a closed set

Expert's answer

Let's fix some notation first.

∂A\partial A is the set denotes the boundary of AA .

M:=X\∂A=∂AcM:=X\backslash \partial A= \partial A^{c} is the complement of the boundary of AA .

We have given that A⊆XA\subseteq X ,where XX is a metric space.


Claim1: No points in MM can be the limit points of ∂A\partial A

Proof: Suppose on the contrary there exist at least one point q∈Mq \in M is limit point of ∂A\partial A .

Thus, for every ϵ>0\epsilon >0 there exist an open ball

B(q,ϵ):={y∈X∣ d(q,y)<ϵ}B(q,\epsilon):=\{y \in X |\: d(q,y)<\epsilon\}

such that

∂A∩B(q,ϵ)\{q}≠φ\partial A\cap B(q,\epsilon)\backslash \{q\}\neq \varphi

but if we choose ϵ\epsilon small enough ,such ϵ\epsilon always exist ,we get

∂A∩B(q,ϵ)\{q}=φ\partial A\cap B(q,\epsilon)\backslash \{q\}= \varphi

Hence contradicting the hypothesis that qq is limit point of AA . This can be easily seen in the below rough figure


Clearly, the ball KK inside AA and FF outside AA does not intersect with boundary ∂A\partial A . Hence, proved.

Thus claim1 guaranteed that if ∂A\partial A has limit points then it must be on ∂A\partial A.



Claim2: Every point on the boundary ∂A\partial A is limit point of ∂A\partial A

Proof:let for any arbitrary point p∈∂Ap\in \partial A ,thus for every ϵ>0\epsilon>0 there exist an open ball B(p,ϵ)B(p,\epsilon) such that

∂A∩B(p,ϵ)\{p}≠φ\partial A \cap B(p,\epsilon)\backslash \{p\}\neq \varphi

which is very clear from the above rough sketch, hence pp is the limit point of ∂A\partial A.

As,pp is arbitrary,thus every point of ∂A\partial A is limit point of ∂A  ⟹  \partial A \implies every limit points of ∂A\partial A are contained in ∂A\partial A . Hence proved.


Therefore immediately from Claim 2, ∂A\partial A is close set.

Hence, we are done.


LATEST TUTORIALS
APPROVED BY CLIENTS