0

I would like general feedback on my solution to this exercise, as well as any ideas for a more direct approach.

Exercise

Let $(X,d)$ be a metric space. Denote the closure of any set $A\subseteq X$ by $\bar{A}$.

Prove that $\bar{\bar{A}}=\bar{A}$.

Solution

Clearly $\bar{A}\subseteq\bar{\bar{A}}$, so it suffices to prove that any limit point of $\bar{A}$ is also a limit point of $A$.

Let $x\in X$ be a given limit point of $\bar{A}$, and let $\varepsilon >0$ be given.

If $x\in A$ then clearly $x\in\bar{A}$ so for the remainder of the proof suppose $x\notin A$.

So $\exists y\in\bar{A}$ such that $x\neq y$ and $d(x,y)<\frac{\varepsilon}{2}$.

There are two possibilities:

Case 1 ($y\in A$)

In this case, $y\in A$ is such that $x\neq y$ and $d(x,y)<\frac{\varepsilon}{2}<\varepsilon$.

So $x$ is a limit point of $A$, as required.

Case 2 ($y$ is a limit point of A)

In this case, $\exists z\in A$ such that $z\neq y$ and $d(z,y)<\frac{\varepsilon}{2}$.

So $z\in A$ is such that $z\neq x$ (since $x\notin A$) and $d(x,z)\leq d(x,y)+d(y,z)<\frac{\varepsilon}{2}+\frac{\varepsilon}{2}=\varepsilon$.

So $x$ is a limit point of $A$, as required.

  • 1
    the author wants feedback, not (just) a solution. And there is certainly feedback to give; I feel this is yet another premature closure – FShrike Sep 02 '23 at 10:21
  • 1
    Your proof is correct. You can shorten it twice: 1) No need to treat separately $x\in A,x\notin A.$ There always exists $y\in\overline A$ s.t. $d(x,y)<\varepsilon/2.$ 2) No need to treat separately $y\in A,y\notin A.$ There always exists $z\in A$ s.t. $d(y,z)<\varepsilon/2.$ – Anne Bauval Sep 02 '23 at 10:28
  • It’s wrong to conclude $x$ is a limit point at the end of each of your cases. You only have enough data to conclude once both cases have been covered (imagine if case $1$ was true but case $2$ was not, what if the only points ‘really close’ to $x$ lived in $\overline{A}\setminus A$?) – FShrike Sep 02 '23 at 12:31
  • 1
    Note that $A$ is not really involved here, and the proof is complicated by its presence. What is really being proved is that the closure of a closed set $C$ is itself. – copper.hat Sep 02 '23 at 14:34
  • Closure of a set A is itself closed. – MathRookie2204 Sep 02 '23 at 14:36
  • 2
    I feel that it is easier to show that ${\rm Cl} A$ is closed for any $A$, and for any closed set $C$, we have that $C = {\rm Cl} C$. Therefore, ${\rm Cl} A = {\rm Cl} {\rm Cl} A$. – K. Jiang Sep 02 '23 at 16:55

0 Answers0