Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Given a Poisson process $N(t),t\geq 0$ with rate $\lambda$ and another r.v. $T$ independent of $N(t)$ with mean $\mu$ and variance $\sigma^2$, I would like to compute the following quantities:

$$ \mathbb{Cov}(T,N(T)) \ \ \mbox{ and } \ \ \mathbb{Var}(N(T))$$

My guess is respectively: $\lambda \mu + \lambda \sigma^2$ and $\sigma^2\lambda$. But I am not sure it is correct nor how to justify some steps.

Anyone knows? Thank you very much!

share|cite|improve this question
For the variance, use the conditional variance formula: $$\text{var}(N(T)) = E[\text{var}(N(T)\mid T] + \text{var}(E[N(T)\mid T).$$ This will show why your guessed answer is not correct (it is missing one term from the displayed expression above). – Dilip Sarwate Feb 28 '13 at 21:48

Since $$ \mathbb{Cov}\left(T, N(T)\right) = \mathbb{E}\left(T N(T)\right) - \mathbb{E}(T) \mathbb{E}\left(N(T)\right) = \mathbb{E}\left(T \, \mathbb{E}\left(N(T)\mid T\right)\right) - \mathbb{E}(T) \, \mathbb{E}\left(\mathbb{E}\left(N(T)\mid T\right)\right) $$ But $\mathbb{E}\left(N(T) \mid T\right) = \lambda T$, thus $$ \mathbb{Cov}\left(T, N(T)\right) = \mathbb{E}\left(\lambda T^2\right) - \lambda \mathrm{E}(T)^2 = \lambda \mathbb{Var}(T) $$ Similarly: $$ \begin{eqnarray} \mathbb{Var}\left(N(T)\right) &=& \mathbb{E}\left( \mathbb{E}\left(N(T)^2|T\right) \right) - \mathbb{E}\left(\mathbb{E}\left( N(T) \mid T\right)\right)^2 \\ &=& \mathbb{E}\left( \lambda^2 T^2 + \lambda T\right) - \mathbb{E}\left(\lambda T\right)^2 \\ &=& \lambda^2 \mathbb{Var}(T) + \lambda \mathbb{E}(T) \end{eqnarray}$$

share|cite|improve this answer
Thanks a lot! Yes I see it, this is exactly what I did but something seemed strange to me.... Since we compose $N(T)$, it could be that we know the random variable $T$ but not $N$, what I mean is that $N(T) = N(T(\omega))(\tilde{\omega})$ where $\omega \in \Omega$, $\tilde{\omega} \in \tilde{\Omega}$ are the sample spaces of each variable. So when averaging out one, we should obtain a random variable? shouldn't we? – Dan Feb 28 '13 at 21:53
Yes, I think you are correct. The conditioning by $T$ indicates that we average over $\tilde\omega$ while keeping $\omega$ fixed, and then average over $\omega$. – Sasha Feb 28 '13 at 22:03

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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