5
$\begingroup$

This question is based on the following from Elements of Integration and Lebesgue Measure by Bartle. We are given the following example and exercise:

Let X be the set ${\mathbf R}$ of real numbers. The Borel algebra is the $\sigma$-algebra generated by all open intervals $(a,b)$ in ${\mathbf R}$. Observe that the Borel algebra ${\mathbf B}$ is also the $\sigma$-algebra generated by all closed intervals $[a,b]$ in ${\mathbf R}$

Exercise 2.B. Show that the Borel algebra ${\mathbf B}$ is also generated by the collection of all half-open intervals $(a,b]=\{x\in {\mathbf R}: a<x\leqslant b\}$

The definition of sigma algebras we are given is that they are unions and complements of a family of sets on a universal set $X$.

My thought on the way to solve this was to use something like $$ (a,b)=\cup_{n=1}^\infty[a+1/n,b-1/n] $$ and similarly $$ (a,b)=\cup_{n=1}^\infty(a,b-1/n] $$ to show that we can construct $(a,b)$ as a union of either the closed or the half-open intervals.

What I'm wondering about is, first I would like to check that the above is a correct answer of course, and also I wonder whether I've missed some other trick which doesn't involve these constructions.

$\endgroup$
4
$\begingroup$

Yes, you have the right idea. But the solution you gave is not complete.

Let $\mathfrak{B}$ be the Borel set generated by intervals of the form $[a,b]$ and $\mathfrak{B}^*$ be the Borel set generated by elements of the form $(a,b)$

By the relation you gave,

$$(a,b)=\bigcup_{n=1}^\infty [a+1/n,b-1/n]$$

You showed that any open interval $(a,b)$ can be written as a countable union of closed sets.

Hence, $(a,b)\in \mathfrak{B}$

Now since each element of $\mathfrak{B}^*$ is a countable union, intersection and complement of open intervals. Thus you have

$$\mathfrak{B}^*\subseteq\mathfrak{B}$$

Now, you also need to show $\mathfrak{B}\subseteq\mathfrak{B}^*$.

I'm sure, since you have the right idea, that you'll be able to show that too.

$\endgroup$
  • $\begingroup$ Thanks for your confirmation. I was hoping there was some kind of clever trick without all these constructions. $\endgroup$ – Suzu Hirose Nov 29 '14 at 2:05
  • $\begingroup$ "Now since each element of is a countable union, intersection and complement of open intervals. " Why would any Borel set in $B^*$ be of that form? $\endgroup$ – Aman Nov 29 '14 at 5:38
  • $\begingroup$ @Aman: Let $x\in\mathfrak{B}^*$ then $x$ is a countable union, intersection and complement. Wherever, there is an open interval in the expansion of $x$ you can replace it by the above expression and thus you have. $x\in\mathfrak{B}$ $\endgroup$ – Swapnil Tripathi Nov 29 '14 at 8:57
  • $\begingroup$ The point I wanted to make was that not every borel set is a countable union, intersection, complement of open sets, even though borel sigma algebra is generated by open sets. It is just the smallest sigma algebra containing open sets. $\endgroup$ – Aman Nov 29 '14 at 12:59
  • $\begingroup$ Ok. So just showing the containment of an open interval suffices? $\endgroup$ – Swapnil Tripathi Nov 29 '14 at 13:57
3
$\begingroup$

You showed that the $\sigma$-algebra generated by the closed (resp. half-open) intervals is a superset of the Borel algebra. The other inclusion is still missing: Obtain $[a,b]$ using nothing but countable unions and complements of open intervals. A straight-forward possiblity is $$[a,b]=\mathbb R\setminus\left(\bigcup_{n=1}^\infty\left(\mathbb R\setminus(a-1/n,b+1/n)\right)\right)\text.$$

Reusing this representation, we obtain $$(a,b]=(a,(a+b)/2)\cup[(a+b)/2,b]$$ for half-open intervals.

$\endgroup$
  • 5
    $\begingroup$ $$[a,b]=\bigcap_{n=1}^\infty (a-1/n,b+1/n)\text{ would be a cleaner version}$$ :) $\endgroup$ – Swapnil Tripathi Nov 29 '14 at 2:07
  • $\begingroup$ Thank you; I edited the answer. $\endgroup$ – user193271 Nov 29 '14 at 2:18

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.