I am reading these notes on the measure which say (pg4) that the Lebesgue measure on $[0,1]$ can be got by considering the algebra: $$A=\{\text{all finite unions of subintervals}\}$$ firstly does this include both open and closed intervals? And secondly this family generates the Borel sigma-algebra on $[0,1]$ is there any easy way to show this?

  • $\begingroup$ Yes, that's probably the intention of the author: intervals are just subsets that contain everything between any two members. Some people take $A$ = finite unions of "half open" intervals.The issue is $A$ being closed under complementation. $\endgroup$ – Tim kinsella Jul 28 '17 at 18:02
  • $\begingroup$ The smallest $\sigma$-algebra containing all subintervals of $[0,1]$ is the Borel $\sigma$-algebra, where subinterval can be taken to mean open, closed, or half-open interval. To show this, use the fact that every open subset of $\mathbb{R}$ is a countable disjoint union of open intervals, and that every open interval is a countable union of closed intervals. $\endgroup$ – Reveillark Jul 28 '17 at 18:02
  • $\begingroup$ To your second question, the Borel algebra is generated by the open intervals since they form a basis for the topology on the interval. $\endgroup$ – Tim kinsella Jul 28 '17 at 18:03
  • $\begingroup$ The author claims that $A$ is an algebra, so probably he means to include both open and closed subintervals. If he only used open intervals, for example, then $A$ would not be an algebra: the complement of an open interval is not open. Alternatively, he could use only intervals of the form $[a,b)$, $(-\infty, b)$, or $[a, \infty)$. $\endgroup$ – Bungo Jul 28 '17 at 18:03
  • $\begingroup$ @Reveillark just to clarify do you mean that every open subset of $\Bbb{R}$ is a countable union of disjoint open intervals, rather then a countable disjoint union? $\endgroup$ – Quantum spaghettification Jul 28 '17 at 18:53

Yes, it contains all finite and infinite open, closed, half-open, half-closed intervals of $\mathbb{R}$, is the way I read it.

Note that all intervals are Borel (open intervals are open, closed intervals are closed, and $[a,b) = [a,b] \cap [0, b)$ is in the $\sigma$-algebra generated by the open sets. So $\mathcal{A} \subset Bor(\mathbb{R})$, and so $\sigma(\mathcal{A})$ (its generated $\sigma$-algebra) is a subset of $Bor(\mathbb{R})$.
So $\sigma(\mathcal{A}) \subseteq Bor(\mathbb{R})$.

But also all open sets $O$ are in $\sigma(\mathcal{A})$, as open sets of $[0,1]$ are unions of at most countably many open intervals ,all of which lie in $\mathcal{A}$. So $Bor(\mathbb{R}) \subseteq \sigma(\mathcal{A})$ so we have equality.

  • $\begingroup$ Please note that in the question, we are working inside $[0,1]$, so there are no infinite intervals. $\endgroup$ – Nate Eldredge Jul 28 '17 at 23:30
  • $\begingroup$ @NateEldredge thx, I changed it. $\endgroup$ – Henno Brandsma Jul 29 '17 at 6:30

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.