I was wondering what is the relation between the first and second Borel–Cantelli lemmas and Kolmogorov's zero-one law?

The former is about limsup of a sequence of events, while the latter is about tail event of a sequence of independent sub sigma algebras or of independent random variables. Both have results for limsup/tail event to have either probability 0 or 1. I guess there are relations between but cannot identify them.

Can the former be viewed as a special case of the latter? How about in reverse direction?

Thanks and regards!


The first Borel-Cantelli is simply the fact that the probability of a union is at most the sum of the probabilities; it has nothing in common with Kolmogorov zero-one law.

What Kolmogorov zero-one law tells you in the setting of the second Borel-Cantelli lemma is that the probability of the limsup is $0$ or $1$, because (1) the limsup is always in the tail $\sigma$-algebra, (2) you are considering independent events, hence their tail $\sigma$-algebra is trivial. Then the second Borel-Cantelli lemma itself tells you that this probability is in fact $1$ under the non summability condition which you know.

  • 1
    $\begingroup$ Thanks! Could you elaborate (1) why "considering independent events, hence their tail σ-algebra is trivial"? What is their tail σ-algebra? (2) why "this probability is in fact 1 under the (non) summability condition"? Doesn't the second Borel-Cantelli lemma only consider nonsummarbility condition which implies probability 1? Does it consider nonsummability condition? $\endgroup$ – Tim Feb 25 '11 at 15:10
  • $\begingroup$ @Tim: (1) The definition of a tail event is on the page you link to. (2) Doesn't the second Borel-Cantelli lemma only consider nonsummability condition which implies probability $1$? Yes, this is the statement of second BC lemma. Does it consider nonsummability condition? Same question, same answer. Sorry but I fail to see what you want me to add to what I wrote in my post. $\endgroup$ – Did Feb 25 '11 at 16:47
  • $\begingroup$ Thanks! For (2), I made a typo, in my comment, but after your update, I don't have doubts any more. For (1), what is the tail σ-algebra here (not asked for definition)? I want to see how trivial it is. Thanks! $\endgroup$ – Tim Feb 25 '11 at 16:57
  • $\begingroup$ @Tim: As the WP page you linked to explains, the tail $\sigma$-algebra of a sequence of $\sigma$-algebras $(\mathcal{F}_n)_n$ is the $\sigma$-algebra $\bigcap_{n=1}^\infty G_n$, where each $G_n$ is the $\sigma$-algebra generated by $\bigcup_{k=n}^\infty \mathcal{F}_k$. $\endgroup$ – Did Feb 25 '11 at 17:04

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.