1
$\begingroup$

Ok, I have been informed that the below lemma is incorrect. I needed it to prove the following statement. Could someone else provide a proof?

Statement: If m(E) is finite, there exists a compact set K with $K\subset E$ and $m(E-K)\le \epsilon$.

m(E) is the Lebesgue measure of set E.

Thanks!

Lemma: Let F be a closed subset of $\mathbb{R^d}$. Let $B_n(0)$ be a ball of radius n centered at the origin. Let $K_n=F\cap B_n$. Show $K_n$ is compact.

$\endgroup$
3
  • 2
    $\begingroup$ This is not true at all. $\endgroup$
    – user223391
    Aug 28, 2015 at 16:27
  • $\begingroup$ Interesting. I'm almost certain this is the argument that Stein and Shakarchi use to show that "If m(E) is finite, there exists a compact set K with $K\subset E$ and $m(E-K)\le \epsilon$", where m(E) is the Lebesgue measure of set E. For reference, Thm 3.4(iii). $\endgroup$
    – manofbear
    Aug 28, 2015 at 16:30
  • $\begingroup$ interesting, I see your point. $\endgroup$
    – user178826
    Aug 28, 2015 at 16:32

3 Answers 3

1
$\begingroup$

This is definitely false, for $F = \mathbb{R}^n$ you get $F \cap B_n = B_n$ which is not compact.

However, the closed ball $\overline{B}_n(0)$ is compact and the intersection between a compact and a closed set is always compact. Therefore $\overline{B}_n(0) \cap F$ is compact.

Another option would be to consider $\overline{F \cap B_n(0)}$. This set is closed and bounded since the closure of a bounded set is bounded aswell.

$\endgroup$
0
$\begingroup$

Take $K_n:=\overline{F\cap B_n}$, then $$\lim_{n\to\infty}\lambda(K_n)=\lambda(F),$$ where $\lambda$ denotes the Lebesgue measure. It should be easy to conclude from here.

$\endgroup$
0
$\begingroup$

(The OP drastically changed the question after this answer was posted.)

I'm not sure if you intended the $n$ in the superscript of $\mathbb{R}^n$ to be the same as the $n$ in $B_{n}(0)$.

Nevertheless: Your question asks about the intersection of a closed and open set, so I assume that by "ball" you mean an open ball.

The statement is false.

Let $n = 1$ so that we are dealing with $\mathbb{R}$.

Then $B_{1}(0)$ is the open interval $(-1, 1) \subset \mathbb{R}$.

Let $F$ be the closed subset $[-1, 1] \subset \mathbb{R}$.

Then $K_{1} = [-1, 1] \cap (-1,1) = (-1, 1)$ is the same open set (i.e., open interval) as $B_{1}(0)$.

In particular: This provides a counterexample to the statement in question.

$\endgroup$
3
  • $\begingroup$ Whoops, I did not intend that. Thanks for pointing it out. $\endgroup$
    – manofbear
    Aug 28, 2015 at 16:42
  • 1
    $\begingroup$ Isn't $K_1=\{0\}$, not $(-1,1)?$ Since $(-1,1)\backslash \{0\}$ is not in $\{0\}$. $\endgroup$
    – manofbear
    Aug 28, 2015 at 18:01
  • $\begingroup$ @manofbear Good point! Sorry for the brain freeze. I've changed $F$ to be a closed subset of $\mathbb{R}$ that contains $B_{n}(0)$; in such a case, their intersection will always be $B_{n}(0)$, which is open by assumption hence provides a counterexample. (In my answer, I use the open ball of radius $1$.) $\endgroup$ Aug 28, 2015 at 19:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .