13

Note:

Borel-Cantelli Lemma says that

$$\sum_{n=1}^\infty P(A_n) \lt \infty \Rightarrow P(\lim\sup A_n)=0$$

$$\sum_{n=1}^\infty P(A_n) =\infty \textrm{ and } A_n\textrm{'s are independent} \Rightarrow P(\lim\sup A_n)=1$$

Then,

if $$\sum_{n=1}^\infty P(A_nA_{n+1}^c )\lt \infty$$

by using Borel-Cantelli Lemma

I want to show that

firstly,

$\lim_{n\to \infty}P(A_n)$ exists

and secondly,

$\lim_{n\to \infty}P(A_n) =P(\lim\sup A_n)$

Please help me showing these two parts. Thank you.

Zen
  • 21,786
  • 3
  • 72
  • 114
1190
  • 848
  • 3
  • 8
  • 20
  • 5
    No, the Borel-Cantelli lemma doesn't say (all of) that, at least, not without further assumptions. – cardinal Mar 29 '15 at 20:27
  • @cardinal well, how can i show these two statements? please can you explain it to me? i dont have any enough idea. i'll be glad if you'll show a solutin way:) thank you – 1190 Mar 29 '15 at 21:33
  • 2
    Added one "further assumption". – Zen Mar 30 '15 at 02:13
  • Minor note: as mentioned [here](https://en.wikipedia.org/wiki/Borel%E2%80%93Cantelli_lemma#Converse_result), for instance, we can get by with only pairwise independence of the $A_n$ in the second part of the lemma – jld Nov 15 '16 at 19:56

1 Answers1

2

None of the assertions are true.

Let $A_n$ be the chance of heads in a coin flip, with probability $1/n^2$ when $n$ is odd and $1-\frac{1}{n^2}$ when $n$ is even. Then:

$$\sum_{n=1}^\infty P(A_n,A_{n+1}^c)=\sum_{odd \ n}^\infty \frac{1}{n^2}\left(1-\frac{1}{(n+1)^2}\right)+\sum_{even \ n}\frac{1}{n^2}\left(1-\frac{1}{(n+1)^2}\right)<\sum_{n=1}^\infty \frac{1}{n^2}<\infty.$$

However, $\lim_nP(A_n)$ clearly does not exist. The best you can conclude is $\lim_n P(A_n,A_{n+1}^c)\rightarrow 0$.

Alex R.
  • 13,097
  • 2
  • 25
  • 49