The problem:

Let $\xi$ and $\eta$ be independent random variables distributed by Poisson law with parameters $\lambda_1$ and $\lambda_2$ correspondingly. Show that random variable $\xi + \eta$ also distributed by Poisson (with parameters $\lambda{1} + \lambda{2}$)

My attempt:

By Poisson law $P(\xi + \eta ; \lambda_1 + \lambda_2) = \frac{(\lambda_1 + \lambda_2)^{\xi + \eta} * e^{-(\lambda_1 + \lambda_2)}}{(\xi + \eta)!}$ . And since we know that they are independent we can say that $P(\xi + \eta ; \lambda_1 + \lambda_2)$ is also equal to $P(\xi ; \lambda_1) + P(\eta ; \lambda_2) - P(\xi ; \lambda_1) * P(\eta ; \lambda_2)$.

I tried simplifying the long equation that I get but I still cannot get them equal to each other. Should I try simplifying them from both sides?

  • How about calculating the fourrier transform? – clark Oct 17 '17 at 20:42
  • I am not familiar with that. – Nick202 Oct 17 '17 at 20:45
  • 1
    This might help. – drhab Oct 17 '17 at 20:48
  • Or the moment generating function – clark Oct 17 '17 at 20:49

For greater clarity, let's use Latin capital letters for random variables, lower case for their values, and Greek letters for the parameters so the question becomes

Let $X$ and $Y$ be independent random variables distributed by Poisson law with parameters $\lambda_1$ and $\lambda_2$ correspondingly. Show that random variable $Z=X+Y$ is also distributed by Poisson (with parameters $\lambda_{1} + \lambda_{2}$)

Your "$P(\xi ; \lambda_1) + P(\eta ; \lambda_2) - P(\xi ; \lambda_1) * P(\eta ; \lambda_2)$" is related to independent events rather than to random variables.

A better approach here would be to look at something like $$\mathbb{P}(Z=z)=\sum_x \mathbb{P}(X=x)\, \mathbb{P}(Y=z-x)$$ which since $x$ and $z-x$ must be non-negative integers gives $$\mathbb{P}(Z=z)=\sum_{x=0}^{z} \lambda_1^x \frac{e^{-\lambda_1}}{x!}\, \lambda_2^{z-x} \frac{e^{-\lambda_2}}{(z-x)!}$$

You want this to be equal to $(\lambda_1+\lambda_2)^z \, \dfrac{e^{-(\lambda_1+\lambda_2)}}{z!}$. I will leave that to you (hint: think binomial expansion).

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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