1
$\begingroup$

Are the two subspaces $X$ and $\operatorname{cl}(X)$ of Euclidean space $\mathbb R^2$ locally compact?

$$X = \{(x,\sin 1/x) \mid 0 < x \le 4\}\cup\{(x,\sin 1/x) \mid -4 \le x \lt 0\} \cup \{(0,0)\}$$

Clearly since $\operatorname{cl}(X)$ is closed and bounded so $\operatorname{cl}(X)$ is compact and thus locally compact.

However, I'm having trouble proving that $X$ is not locally compact. My conjecture is that I need to show there is no open set $U$ and compact set $C$ such that $(0,0) \in U \subset C$. But I can't come up with a rigorous proof for this part. How can I construct a written proof of this?

$\endgroup$
  • $\begingroup$ If $X$ were locally compact, then for every sufficiently small $r > 0$, the set $\overline{B_r(0)}\cap X$ would be compact, in particular, a closed subset of $\mathbb{R}^2$. $\endgroup$ – Daniel Fischer Nov 28 '14 at 22:38
  • $\begingroup$ Note: \bigcup is used for things like $\displaystyle\bigcup_{n=1}^\infty A_n$, and \cup for things like $A\cup B$ and $A_1\cup\cdots\cup A_n$. Also note that "a\sin b" yields $a\sin b$ and "a sin b" yields $a sin b$ without proper spacing and with "$\sin$" improperly italicized. And in $\{(x,\sin1/x)\mid 0<x\le 4\}$, the $\{\text{curly braces}\}$ are INSIDE the math environment. I did these and a variety of other edits to the question. ${}\qquad{}$ $\endgroup$ – Michael Hardy Nov 28 '14 at 22:44
  • $\begingroup$ @DanielFischer I see the existence of such closed compact balls if X were locally compact but how does the fact that such closed subsets exist lead to a contradiction? $\endgroup$ – nomadicmathematician Nov 28 '14 at 22:46
  • $\begingroup$ The point is that such sets are not closed in $\mathbb{R}^2$. $\endgroup$ – Daniel Fischer Nov 28 '14 at 22:51
2
$\begingroup$

Hint. In a compact set in the plane every sequence has a convergent subsequence. For convenience, instead of an open ball, consider a rectangular neighborhood of $(0,0)$, $U=[-\epsilon,\epsilon]^2$ where $\epsilon>0$. Take $0<\delta<\epsilon$ and try to come up with a sequence $x_n\to 0$ such that $\delta=\sin1/{x_n}$ for all $n$. What does the sequence of points in the plane $(x_n,\sin1/{x_n})$ converge to? Is the limit of this sequence going to belong to $X$?

$\endgroup$
3
$\begingroup$

Every open nbhd of the origin contains a set of the form

$$U_\epsilon=X\cap\Big((-\epsilon,\epsilon)\times(-\epsilon,\epsilon)\Big)$$

for some $\epsilon>0$. Show that the closure of this set in $X$ is

$$\operatorname{cl}_XU_\epsilon=X\cap\Big([-\epsilon,\epsilon]\times[-\epsilon,\epsilon]\Big)\;.$$

Then let $a=\dfrac{\epsilon}2$ and find a sequence in

$$\big(\Bbb R\times\{a\}\big)\cap\operatorname{cl}_XU_\epsilon$$

that has no convergent subsequence in $X$ (because in $\Bbb R^2$ it converges to a point not in $X$).

$\endgroup$

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.