0
$\begingroup$

I want to show that if $E_1 \supset E_2 \supset....$ is a monotonic sequence of Jordan measurable sets, so $$Z = \bigcap_{k=1}^\infty E_k$$ is of Lebesgue measure null, then Z is also Jordan measurable and it's volume is 0.

Clearly, if $Z$ is indeed Jordan measurable, then it has no volume. So it's enough to show that $Z$ is Jordan measurable.
I started by showing that $\overline{Z} \subset \bigcap_{k=1}^\infty \overline{E_k}$ . I did it by taking some $x\in\overline{Z}$, since the closure $Z$ is the set of all limits of sequences in $Z$, we get that there is a sequence $\{x_i \} \subset Z$ so $x_i \rightarrow x$ as $i \rightarrow \infty$.
Since for all natural $k$ : $\{ x_i \} \subset E_k$ and by the assumption $x_i \rightarrow x$ we get that for all natural $k$: $x \in \overline{E_k}$. Hence $\overline{Z} \subset \bigcap_{k=1}^\infty \overline{E_k}$.
From the given information that $Z$ is of Lebesgue measure null, we infer that $Z$ has no interior, so $\partial Z = \overline{Z} \smallsetminus int(Z) = \overline{Z}$
By the previews claim - $\partial Z \subset \bigcap_{k=1}^\infty \overline{E_k}$.
Now it's enough to show that $\partial \bigcap_{k=1}^\infty \overline{E_k}$ is of Lebesgue measure null.

I don't know how to proceed.

$\endgroup$
3
$\begingroup$

First of all, I think you're mixing the terms "Jordan measurable" and "measure zero (=null set)". The first usually means that the indicator function is Riemann integrable, and the second means that the set can be covered by a countable union of cubes that with total volume less that $\varepsilon$ for every $\varepsilon > 0$.

The confusion might arise from the fact that a set is Jordan measurable iff it's boundary is of measure zero. That means that $\operatorname{Vol}(E)=\operatorname{Vol}(\operatorname{int}(E))$, not immediately $0$ (but $\operatorname{Vol}(\partial E) = 0$).

EDIT: perhaps it is immediate that the volume is $0$ (once we know it's measurable) since it's a null set... So you were right there :)

EDIT 2: OOPS. I didn't actually proved anything here, since I mistakenly claimed that the discontinuity set of $1_Z$ is $Z$, instead of $\partial Z$, so that proof was absolutely wrong! I'm truly sorry for misleading.

Here's an actual proof:

Defining $A_k = E_1 \setminus E_{k+1}$ for $k\geq 1$ gives us an increasing sequence, so $B_k = A_{k+1} \setminus A_k$ is a sequence of disjoint sets, and $\bigcup _{k=1} ^{\infty} B_k = \bigcup _{k=1} ^{\infty}A_k = E_1 \setminus Z$.

Note that Jordan measure is finitely additive, so $\operatorname{Vol}(A_k) = \operatorname{Vol}(E_1) - \operatorname{Vol}(E_k+1)$, thus:

$\operatorname{Vol}(B_k) = \operatorname{Vol}(A_k+1) - \operatorname{Vol}(A_k) = \operatorname{Vol}(E_k) - \operatorname{Vol}(E_k+1)$

Now: $\operatorname{Vol}(E_1 \setminus Z) = \sum _{k=1} ^{\infty} \operatorname{Vol}(B_k) = \sum _{k=1} ^{\infty} [\operatorname{Vol}(E_k) - \operatorname{Vol}(E_{k+1})] = \operatorname{Vol}(E_1) - \lim_{k \to \infty} \operatorname{Vol}(E_k)$

So far everything was legal because a union of Jordan measurable sets is a Jordan measurable set + Jordan volume is finitely additive (you can also exchange every $\operatorname{Vol}(X)$ by $\int_{X} 1$).

Exchanging sides, we get that $V(Z) = \lim_{k\to\infty} \operatorname{Vol}(E_k)$ if it exists. The limit exists since the sequence of volumes is bounded and monotonicly decreasing: $\operatorname{Vol}(E_k) \geq \operatorname{Vol}(E_{k+1})$, so $\operatorname{Vol}(Z)$ also exists, meaning $Z$ is Jordan measurable.

If this looks like cheating, you can take any other measure (for example, Lebesgue measure, and do the same \ use sigma-additivity) and follow the same steps (replacing $\operatorname{Vol}$ by $\mu$) - a set is Jordan measurable if it's inner measure equals it's outer measure, and here the inner measure of $Z$ is 0, so it's enough to show that the outer measure approches 0, but I believe that this is beyond the material of the calculus 3 course we're taking.

$\endgroup$
  • 1
    $\begingroup$ BTW, I'm taking Hedva 3 as well and believe that this solution is enough (if you did in fact took the question from an old exam), since they said you don't have to be 100% formal (but it never hurts). Good luck! $\endgroup$ – Trouble Jan 16 '16 at 22:56
  • $\begingroup$ You say that $Z$ is the discontinuity set of $1_Z : E_1 \rightarrow \mathbb R$. Isn't it $\,\partial Z$ ? $\endgroup$ – Tony Piccolo Jan 17 '16 at 17:44
  • $\begingroup$ It is! That means that my proof wasn't actually a proof but just a circular argument and a lie. I changed it into a different proof which is (hopefully) better. Thank's for noticing! $\endgroup$ – Trouble Jan 18 '16 at 19:13
  • $\begingroup$ $\operatorname{Vol}(E_1 \setminus Z) = \sum _{k=1} ^{\infty} \operatorname{Vol}(B_k)$: aren't you using countable additivity ? $\endgroup$ – Tony Piccolo Jan 18 '16 at 20:02
  • $\begingroup$ I'm using the Monotone convergence theorem (interchanging integral and sum)... anyhow, sigma-additivity can be used if we switch to Lebesgue outer measure. $\endgroup$ – Trouble Jan 18 '16 at 21:30
0
$\begingroup$

I know this is an old question, but I have an interest in it as well and I think I have a better solution, or at least a simpler more readable solution.

I will show that $\partial Z \subset Z \cup(\bigcup_{k=1}^{\infty}\partial E_k)$. It will then follow that $\partial Z$ is zero measure since it is a countable union of zero measure sets.

Let $x \in \partial Z$. If $x \in Z$ then we are done, so assume that $x\notin Z$ to handle the non trivial case.

We will show that $x \in \bigcup_{k=1}^{\infty}\partial E_k$. Suppose on the contrary that $x \notin \bigcup_{k=1}^{\infty}\partial E_k$.

This means that for every $k$, $x$ is not on the boundary of $E_k$. So there's either a punctured open ball $C_k = B(x,\epsilon_k)\setminus\{x\}$ that is completely inside $E_k$ or completely outside $E_k$, for all $k$.

This leaves us with three options:

(1) - There's a punctured open ball $C_k$ completely inside $E_k$ for all $k$. Because they are all nested, we have $\lim_{k \to \infty}C_k \subset \bigcap_{i=1}^{\infty}E_i = Z$, so $C_k$ is a punctured open ball completely inside $Z$, so $x \notin \partial Z$ which is a contradiction.

(2) There's a punctured open ball inside some sets, and outside other sets. Because they are nested, this means $C_k$ is completely inside $E_1,E_2,\dots,E_k$ and completely outside $E_{k+1},E_{k+2}, \dots$, then $C_k$ is completely outside $Z$ because it's completely outside $E_{k+1}$ so it's outside of the intersection. Then again $x \notin \partial Z$ as $C_k$ is a punctured open ball centered at $x$ completely outside $Z$.

(3) There's a punctured open ball centered around $x$ that's completely outside all $E_k$. Again, very easy to see that $x \notin \partial Z$.

To sum up (assuming $x\notin Z$) - if $x \notin \bigcup_{k=1}^{\infty}E_k$ it follows that $x \notin \partial Z$. Thus if $x \in \partial Z$ it follows that $x \in \bigcup_{k=1}^{\infty}E_k$.

This concludes the proof.

$\endgroup$

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.