Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $A$ be a set, let $B$ be a Borel set such that $B \subseteq A$.

Because $B$ is a Borel set, can I automatically say that I can represent it as a countable union of closed sets. Thus I can demonstrate that $m(A\setminus B)=0$.

Alternatively, let $C$ is a set and $D$ is a Borel set such that $C \subseteq D$. Then, since $D$ is Borel we can write it as a countable intersection of open sets. So $m(D \setminus C)=0$.

I'm trying to find a relationship between Borel sets and the Inner/Outer approximation theorems.

share|cite|improve this question
up vote 4 down vote accepted

Sadly, we can do nothing so simple. Countable unions (intersections) of closed (open) sets are called $F_\sigma$ $(G_\delta)$ sets. Countable unions (intersections) of $G_\delta$ $(F_\sigma)$ sets are called $G_{\delta\sigma}$ $(F_{\sigma\delta})$ sets. Similarly, we have $G_{\delta\sigma\delta}$ and $F_{\sigma\delta\sigma}$ sets, $G_{\delta\sigma\delta\sigma}$ and $F_{\sigma\delta\sigma\delta}$ sets, and so on. All of these are Borel sets (along with basic open and closed sets)--and in fact comprise the entirety of the collection of Borel sets. For a more explicit description of this transfinitely recursive construction of the Borel heirarchy--in $\omega_1$ steps, not countably many (Thanks, Trevor, for pointing that out!)--see here.

Taking $A$ to be the overlying set, $B'$ to be any Borel subset of $A$ of positive measure, and $B=A\smallsetminus B'$, we have that $B$ is a Borel set and $A\smallsetminus B=B'$, and furthermore $m(A\smallsetminus B)=m(B')>0$. On the other hand, take $D$ to be any Borel set of positive measure, and let $C=\emptyset$, so that $D\smallsetminus C=D$, and so $m(D\smallsetminus C)=m(D)>0$.

It is worth noting that we can approximate any measurable set from without (within) by some $G_\delta$ $(F_\sigma)$ set, in exactly the way you described. We just can't do it with every $G_\delta$ $(F_\sigma)$ superset (subset). Also, we can't necessarily do this for arbitrary sets, which may not be measurable.

share|cite|improve this answer
It is perhaps worth mentioning that the "and so on" in the first paragraph means "for $\omega_1$ many steps" and not "for countably many steps". – Trevor Wilson Oct 17 '12 at 3:52
Interesting, Trevor! I'd never heard or suspected that. Can you link me to a source? – Cameron Buie Oct 17 '12 at 4:05
Wait...I fumbled my positioning of $B$ and $A$ in the set differences. – emka Oct 17 '12 at 4:39
Ah! In that case, you run into a similar problem, taking $A$ to be the overlying set, and $B$ to be a set not of full measure. I'll correct my answer to reflect that. – Cameron Buie Oct 17 '12 at 5:06
@CameronBuie See The reason is that each of the finite stages you mentioned properly contains the previous ones, so we can pick for each $n$ a set that doesn't appear until stage $n$. Then we can take a single set coding these countably many sets, and this set will not appear at any finite stage. – Trevor Wilson Oct 17 '12 at 15:55

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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