26

I'm trying to prove something here which isn't necessarily hard, but I believe it to be somewhat tricky. I've looked online for the proofs, but some of them don't seem 'strong' enough for me or that convincing. For example, they use the argument that since $A\subset \overline{B} $, then $ \overline{A} \subset \overline{B} $. That, or they use slightly altered definitions. These are the definitions that I'm using:

Definition #1: The closure of $A$ is defined as the intersection of all closed sets containing A.

Definition #2: We say that a point x is a limit point of $A$ if every neighborhood of $x$ intersects $A$ in some point other than $x$ itself.

Theorem 1: $ \overline{A} = A \cup A' $, where $A'$ = the set of all limit points of $A$.

Theorem 2: A point $x \in \overline{A} $ iff every neighborhood of $x$ intersects $A$.

Prove: If $ A \subset B,$ then $ \overline{A} \subset \overline{B} $

Proof: Let $ \overline{B} = \bigcap F $ where each $F$ is a closed set containing $B$. By hypothesis, $ A \subset B $; hence, it follows that for each $F \in \overline{B} $, $ A \subset F \subset \overline{B} $. Now that we have proven that $ A \subset \overline{B} $, we show $A'$ is also contained in $\overline{B} $.

Let $ x \in A' $. By definition, every neighborhood of x intersects A at some point other than $x$ itself. Since $ A \subset B $, every neighborhood of $x$ also intersects $B$ at some other point other than $x$ itself. Then, $ x \in B \subset \overline{B} $.

Hence, $ A \cup A' \subset \overline{B}$. But, $ A \cup A' = \overline{A}$. Hence, $ \overline{A} \subset \overline{B}.$

Is this proof correct?

Be brutally honest, please. Critique as much as possible.

Kenta S
  • 16,151
  • 15
  • 26
  • 53
Daavid M.
  • 261
  • It looks fine to me. Your proof is correct. : ) – Rudy the Reindeer Mar 17 '12 at 08:35
  • Thank you! I learned from the information below that I could've written a simpler proof, but I just wanted to see if this was correct. – Daavid M. Mar 17 '12 at 08:58
  • I thought so : ) – Rudy the Reindeer Mar 17 '12 at 09:03
  • I think there's a small error in the first paragraph of the proof. Instead of $F\in \overline{B}$, seems like you mean $F\subseteq \overline{B}$. – Patrick Mar 19 '12 at 07:54
  • (Re previous comment): Perhaps I'm missing something, but if $F$ is one of the closed sets in the intersection $\cap F$, then $F\subset \overline{B}$ doesn't follow. And if $F$ is just an arbitrary subset of $\overline{B}$ then $A \subset F$ doesn't follow. I don't see how to fix this. – Patrick Mar 19 '12 at 08:02

7 Answers7

41

I think it's much simpler than that. By definition #1, the closure of A is a subset of any closed set containing A; and the closure of B is certainly a closed set containing A (because it contains B, which contains A). QED.

14

Using Definition #1 makes it quite easy. For each $A \subseteq X$, let $\mathcal{C}_A = \{ F \subseteq X : F\text{ is closed and }A \subseteq F \}$. Then by Definition #1 it follows that $\overline{A} = \bigcap \mathcal{C}_A$.

Note that if $A \subseteq B$, then $\mathcal{C}_B \subseteq \mathcal{C}_A$, and therefore $\bigcap \mathcal{C}_A \subseteq \bigcap \mathcal{C}_B$.

user642796
  • 52,188
  • 1
    Ah, okay. When I read the proofs with one or two lines previously, I didn't know why/how they justified it so swiftly. It just seemed very awkward to me. I guess this is because I learned about the 2nd definition immediately after the first. – Daavid M. Mar 17 '12 at 08:55
  • @DaavidM.: Yeah, this is a somewhat unnatural way of looking at the closure. The notion of limit points make the closure something that can be visualised, whereas the formal definition seems a bit odd. David Wallace's proof above is very nice, and the same idea is used quite often: If $A \subseteq F$ and $F$ is closed, then $\overline{A} \subseteq F$. – user642796 Mar 17 '12 at 11:35
6

I think it's simplest to see from the first definition.

Let $\mathcal{A}$ be the collection of closed sets containing $A$ and $\mathcal{B}$ the collection of closed sets containing $B$. Since $A \subset B$, we know $\mathcal{B} \subset \mathcal{A}$, and so $\bigcap \mathcal{A} \subset \bigcap \mathcal{B}$ (i.e. $\overline{A} \subset \overline{B}$).

Loosely speaking, adding more sets to an intersection can only make it smaller.

Austin Mohr
  • 25,662
2

You say that some of the proofs you have looked use the argument "that since $A$ is contained in $\overline{B}$, then $\overline{A}\subseteq\overline{B}$" and that they don't seem strong enough for you but this follows directly from definition #1. Any closed subset containing $B$ contains $A$ and consequently $A\subseteq \overline{B}$. Since $\overline{B}$ is closed, $\overline{A}\subseteq\overline{B}.$

Kiku
  • 178
2

Using the first theorem from the question which is given as a definition in Rudin.

Theorem 1: $ \overline{A} = A \cup A' $, where $A'$ = the set of all limit points of $A$ (Definition 2.26 in Rudin).

Since $A \subset B$, we have that $A \subset B \cup B'$ or $A \subset \overline{B}$

Next consider a point $x \in A'$ or $x$ is a limit point of $A$. This gives that every neighbourhood of $x$ contains a point $y \neq x$ such that $y \in A$ (Definition 2.18. (b) in Rudin). But this $y \in B$ since $A \subset B$ and hence we have that $x$ is a limit point of $B$ or $x \in B'$ or $x \in B\cup B'$ or $x \in \overline{B}$.
We have shown that if $x \in A'$ then $x \in \overline{B}$ or $A' \subset \overline{B}$.

Since $A \subset \overline{B}$ and $A' \subset \overline{B}$ we have $A \cup A'=\overline{A} \subset \overline{B}$

QED.

texmex
  • 800
  • But when $A \subset S$ and $B \subset S$, could be that $A \cup B = S$ (not necessarily $A \cup B \subset S$). – Ari Royce Hidayat Feb 06 '22 at 17:19
  • @AriRoyce If A = B, then A is a subset of B. Since A is a subset of itself. We are not necessarily looking for proper subsets. And please do not up or down vote (especially down vote) unless you are completely sure. – texmex Mar 23 '22 at 05:40
  • May be you could change it to:

    $A \subset B$ and $A' \subset B'$;

    $(A \cup A') \subset (B \cup B')$;

    $\overline{A} \subset \overline{B}$.

    Then I could undo the downvote.

    – Ari Royce Hidayat Mar 24 '22 at 10:16
0

I think there is one more important portion that is missing - we need to state (if not rigorously prove) that the closure $\overline{A} = A\cup L$ is in fact a closed set. Otherwise, as Stephen Abbott pointed out in his book Understanding Analysis, the union of sets could potentially create new limit points. That is, we need to prove that if $x$ is a limit point of $\overline{A}$, then $x$ is infact a limit point of $A$. To see this, we consider $\epsilon$-neighbourhood of $x$ and show that it contains some point in $A$ that is not equal to $x$.

-4

Consider a point $x \in A'$. This gives that every neighborhood of $x$ contains a point $y \neq x$ such that $y \in A$. But this $y \in B$ since $A \subset B$ and hence we also have that $x \in B'$ (because $x$ has neighborhood in $B$ that is not $x$).

Now we have that:

$A \subset B$ and $A' \subset B'$

$(A \cup A') \subset (B \cup B')$

$\overline{A} \subset \overline{B}$