4
$\begingroup$

Let $X_i$ be independent Poisson with mean $\lambda_i$. Show that

i)$\sum_1^\infty\lambda_i<\infty$ implies $\sum_1^\infty X_i$ converges almost surely to a finite limit.

ii)$\sum_1^\infty\lambda_i=\infty$ implies $\sum_1^\infty X_i=\infty$ almost surely

I could prove (i) by using Kolmogorov one series lemma just setting $Y_i=X_i-\lambda_i$, but I could not prove the second part. Any help/hint/solution for (ii) is appreciated. Thanks in advance

$\endgroup$
2
  • $\begingroup$ Have you tried Borel-Cantelli? $\endgroup$
    – Xi'an
    Feb 12, 2015 at 12:02
  • $\begingroup$ we could solve it by Kolmogorov 2-series theorem, i ll write answer soon $\endgroup$
    – hohoho
    Feb 12, 2015 at 15:16

1 Answer 1

1
$\begingroup$

Assume that $\sum_{i=1}^{+\infty}\lambda_i$ is infinite. Then we can construct inductively an increasing sequence of integers $\left(n_k\right)_{k\geqslant 1}$ such that for each $k\geqslant 1$, $$\tag{*}\sum_{i=n_k+1}^{n_{k+1}}\lambda_i\geqslant k.$$ This can be done by exploiting the fact that for each $k$, $\lim_{N\to +\infty}\sum_{i=n_k+1}^N\lambda_i=+\infty$.

Now define $Y_k:=\sum_{i=n_k+1}^{n_{k+1}}X_i$. In order to show that $\sum_{i=1}^{+\infty}X_i$ diverges almost surely, it suffices to prove that the sequence $\left(Y_k\right)_{k\geqslant 1}$ does not converge to $0$ almost surely. Define $A_k:=\left\{Y_k\geqslant 1\right\}$. Then the sequence $\left(A_k\right)_{k\geqslant 1}$ is independent and $$\sum_{k=1}^{+\infty}\mathbb P\left(A_k\right)=\sum_{k=1}^{+\infty}\left(1-\exp\left(-\sum_{i=n_k+1}^{n_{k+1}}\lambda_i\right)\right) \overset{\mbox{ by }(*)}{\geqslant}\sum_{k=1}^{+\infty}\left(1-\exp\left(-k\right)\right) =+\infty.$$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.