Show that if $\{x_n\}_{n=1}^\infty$ converges to $x\in \mathbb{R}$ then $\overline{\{x_n:n\in\mathbb{N}\}}=\{x_n:n\in\mathbb{N}\}\cup\{x\}$

My definition of the closure is that $x\in \overline{A}$ if and only if for any open set $U$ containing $x$, $U\cap A\neq \emptyset$

So showing $\{x_n:n\in\mathbb{N}\}\cup\{x\}\subseteq\overline{\{x_n:n\in\mathbb{N}\}}$

Let $A=\{x_n:n\in\mathbb{N}\}$

I have $A\subseteq \overline{A}$

Need to show $\{x\}\subseteq \overline{A}$

let $U$ be an open set containing $x$. Since $U$ is open there exists an epsilon ball $B_\epsilon(x)\subseteq U$. And since $x_n$ converges to $x$, there exists an $N$ such that $x_n\in B_\epsilon(x)\subseteq U$ for any $n>N$. thus $x_n\in U$ for any $n>N$ and $U\cap A\neq \emptyset$. then $\{x\}\subseteq \overline{A}$

Showing $A\cup \{x\}\supseteq \overline{A}$

Let $y\in \overline{A}$ and $y\not\in A$ then for any open set $U$ containing $y$ $U\cap A\neq \emptyset$.

Then for any $\epsilon>0$, since $B_\epsilon(y)$ is open, $B_\epsilon(y)\cap A\neq \emptyset$, thus there is an $x_m\in B_\epsilon(y)$

So I believe I just need to show that if $n>m$ then $x_n\in B_\epsilon(y)$ so that $A$ converges to $y$ which means $y=x$ but I'm not sure how to show that.

  • $\begingroup$ You can use the fact that all the subsequences of $\{ x_n \}_{n \in \mathbb{N}}$ converge to the same limit $x$, thus, the closure of the set should just contain $x$ in addition to $\{x_n\}$. $\endgroup$ – sudeep5221 Dec 21 '19 at 15:29
  • $\begingroup$ Yes, if a sequence converges, it only has one limit point: the limit of the sequence. $\endgroup$ – bjorn93 Dec 21 '19 at 15:48
  • $\begingroup$ @bjorn93 But how do I show that if $x_m\in B_\epsilon(y)$ that any $n>m$, $x_n\in B_\epsilon(y)$? $\endgroup$ – AColoredReptile Dec 21 '19 at 16:01
  • $\begingroup$ @AColoredReptile If $y\in\overline{A}$ and $y\not\in A$, then since every open set $U$ containing $y$ must contain element of $A$, it follows that $y$ is a limit point. The only such point is $x$ by convergence. Perhaps you don't want to use limit points? $\endgroup$ – bjorn93 Dec 21 '19 at 16:13
  • $\begingroup$ $x$ is the only limit point of the set. $\endgroup$ – monoidaltransform Dec 26 '19 at 18:13

The problem in your argument is that, having one term $x_m$ in the open ball $B_{\epsilon}(y)$ does not imply all successive terms $x_n \in B_{\epsilon}(y)$, where $n > m$. It needn't even be the case that the next term $x_{m+1}$ is in the open ball. I would rethink your approach, but this time utilize convergence $x_n \to x$. Said convergence is going to guarantee you that, eventually, successive terms $x_N,x_{N+1},...$ are all going to be close to $x$. For any point $y$ other than $x$, we know that a tail of successive terms will be pulled away from $y$ and towards $x$. That leaves you with only finitely many terms $x_1,...,x_N$ to deal with...

Given any positive integer $N$, let $$S_N \; = \; \Big\{x_1,...,x_N\Big\}$$ and $$T_N \; = \; \Big\{x_n \; : \; n > N \Big\},$$ so that clearly $$A = S_N \cup T_N.$$

Now suppose $y \in \bar{A} \backslash A$, but $y \neq x$. We may choose radii $r,s > 0$ such that $$B_{r}(x) \cap B_s(y) = \varnothing.$$ Due to convergence $x_n \to x,$ there exists a positive integer $N$ such that $$T_N \subseteq B_r(x).$$ In particular, $$T_N \cap B_s(y) = \varnothing.$$ Moreover, $\mathbb{R} \backslash S_N$ is an open set of the real line (being the compliment of a finite set) that contains $y$ (as otherwise $y \in S_N \subseteq A$). It follows that there exists a radius $t > 0$ for which $B_t(y) \subseteq \mathbb{R} \backslash S_N$, or rather, $$S_N \cap B_t(y) = \varnothing.$$

All that is left to do is take $\varepsilon = \min\{s,t\}$. With this radius we have, $$T_N \cap B_{\varepsilon}(y) = \varnothing$$ and $$S_N \cap B_{\varepsilon}(y) = \varnothing.$$ Combining both results tells us that $$A \cap B_{\varepsilon}(y) = \varnothing.$$ Here is where the contradiction arises. We supposed $y$ was in the closure of $A$. It could not be the case that an open ball around $y$ is completely disjoint from $A$, but under the assumption that $y \neq x$ we found such an open ball. Our assumption that $x,y$ were distinct must have been incorrect.

| cite | improve this answer | |
  • $\begingroup$ I don't understand the part about $\mathbb{R}\setminus S_N$. Doesn't that contain $T_N$? $\endgroup$ – AColoredReptile Dec 21 '19 at 17:08
  • $\begingroup$ Up till that part, we managed to separate $T_N$ from $y$. Now we want to separate $S_N$ from $y$. Also, note that $S_N$ and $T_N$ aren't necessarily disjoint. It is true that members of $T_N$ are close to $x$. But who knows, so may be some members of $S_N$. $\endgroup$ – joeb Dec 21 '19 at 17:29
  • $\begingroup$ I think maybe it's a typo then, you said $\mathbb{R}\setminus S_N$ is an open set that does not contain $y$. But doesn't it have to contain $y$? $\endgroup$ – AColoredReptile Dec 21 '19 at 17:32
  • $\begingroup$ Ah yes it was a typo. I corrected it $\endgroup$ – joeb Dec 21 '19 at 20:01

It's clear that all points $\{x_n: n \in \Bbb N\} \cup \{x\}$ are points of $\overline{\{x_n: n \in \Bbb N\}}$, all $x_n$ because always (by the definition) $A \subseteq \overline{A}$ (if $x \in A$, then any open point containing $x$ intersects $A$ (as $x \in A \cap U$)), and $x$ because each neighbourhood of $x$ contains infinitely many $x_n$ (all $n$ with $n \ge N$ for some $N$) by definition of convergence (note that these need not be distinct points, but there is at least one, say $x_N$).

Suppose for the converse that $y \in \overline{\{x_n: n \in \Bbb N\}}$. Then there is a sequence $(y_n)$ from $\{x_n: n \in \Bbb N\}$ such that $y_n \to y$. So each $y_n = x_{m_n}$ for some $m_n \in \Bbb N$. If the $m_n$ are cofinal in $\Bbb N$ it's easy to see we get a subsequence of the sequence $(x_n)$ so $y=x$ by unicity of limits. Of not, some $m_n$ occurs infinitely often and we have an eventually constant sequence, so converging to some term of $\{x_n: n \in \Bbb N\}$. In either case $y \in \{x_n: n \in \Bbb N\} \cup \{x\}$ and we're done.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.