There is a sequence of independent random Poisson variables $Y_1, Y_2,\ldots$ with divergent parameter $Y_i \sim \text{Poisson}(\frac{1}{n})$. Does this sequence converge almost surely? In $L^2$ space? In probability?

I am really confused as I did some research on my own and almost every question I come up to has to do with the sum of a sequence; but I do not believe this has to do anything with the sum of $Y_n$ right? to prove it converges almost surely, I have to prove that $Y_n$ approaches the limit of $e^{-1}*\sum_{i=1}^{\infty} 1/i$? Any hints would be appreciated. Thanks.

  • 4
    $\begingroup$ Why is the parameter $\frac1n$ divergent? $\endgroup$ – zoli Apr 1 '18 at 9:41

Since $Y_n\geqslant 0$ a.s. we have $$ \|Y_n\|_2 = \mathbb E[Y_n^2] = \mathsf{Var}(Y_n) + \mathbb E[Y_n^2] = \frac1n + \frac1n^2\stackrel{n\to\infty}\longrightarrow0, $$ so that $Y_n\to 0$ in $L^2$. This implies that $Y_n\to 0$ in probability. Moreover, $$ \sum_{n=1}^\infty \mathbb P(Y_n\geqslant 1) = \sum_{n=1}^\infty \left(1-e^{-\frac1n}\right) = +\infty, $$ so from the second Borel-Cantelli lemma we conclude that $$ \mathbb P\left(\limsup_{n\to\infty} \{Y_n\geqslant 1\}\right) = 1, $$ and hence $Y_n$ does not converge to zero almost surely.


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.