Mathematics
The Borel-Cantelli Lemma and Almost Sure Convergence
Quick fact
The Borel-Cantelli lemma states that if the sum of probabilities of a sequence of events is finite, then the probability that infinitely many of them occur is zero. This means that in an infinite sequence of independent trials, any event with positive constant probability must occur infinitely often—a principle that explains the 'infinite monkey theorem'.
Why this is interesting
Flip a fair coin repeatedly. It seems inevitable that heads will appear infinitely often, but can probability theory actually prove it? The Borel-Cantelli lemma gives a surprisingly simple answer—and it justifies the entire field of almost sure convergence.
Read the full explanation
Understanding The Borel-Cantelli Lemma and Almost Sure Convergence
Imagine a sequence of events E1, E2, E3, ...—each might be 'a coin lands heads on the nth flip' or 'the sample average deviates from the mean by more than ε'. The Borel-Cantelli lemma (first part) says: if you add up all their probabilities and that sum is finite (for instance, probabilities like 1/2^n sum to 1), then almost surely only finitely many of these events can happen. In other words, beyond some point, none of them occur. This is intuitive if you think of probability as 'mass': a finite total mass can't be spread out to cover infinitely many distinct positive-probability events. The second part goes the other way: if the events are independent and the sum of their probabilities is infinite, then almost surely infinitely many of them occur. This is the brilliant insight that links the sum of probabilities to the behavior over infinitely many trials. It forms the backbone of proving that averages converge almost surely—the strong law of large numbers.
A deeper explanation
Why does the Borel-Cantelli lemma work? The first part relies on the countable subadditivity of probability: the probability that infinitely many events occur (the event 'limsup En') is at most the sum of probabilities from any point onward. If the total sum is finite, the tail sums shrink to zero, forcing the probability of infinitely many occurrences to zero. This is a direct consequence of the axioms of probability—no independence needed. The second part requires independence. It uses the fact that the probability that none of a set of independent events occurs is the product of their complement probabilities, and for independent events with divergent sum, the product of (1 - pn) tends to zero, implying the probability that infinitely many occur is 1. This lemma is the engine behind proving almost sure convergence: to show that a sequence of random variables Xn converges to X almost surely, you often show that for every ε 0, the events '|Xn - X| ε' occur only finitely often, which follows from the Borel-Cantelli lemma if the sum of their probabilities is finite. This is exactly how the strong law of large numbers is proven: the sum of probabilities of large deviations is finite, so deviations happen only finitely often, and the sample average converges to the expected value almost surely. Thus, the Borel-Cantelli lemma provides a rigorous, quantitative link between finite-probability conditions and infinite-limit behavior, making it an indispensable tool in probability theory and statistics.