1
$\begingroup$

Consider two Poisson processes, independent $N_1 (t)$ and $N_2(t)$ of parameters $\mu_1$ and $\mu_2$ and $G(t) = N_1 (t) + N_2(t)$.

I have to find the conditional distribution of $N_1 (t)$ given $G(t)$ = k, with $k = 0, 1 , \ldots$

$P(N_1(t) = n_1 | G(t) = k) = P(N_2(t) = k - n_1) = e^{-\mu_2 \cdot t} \cdot \frac{(\mu_2t)^{k-n_1}}{(k-n_1)!}$

and

$F(a) = \sum_{i=0}^{a} P(N_1(t) = i | G(t) = k)$

it's correct ?

$\endgroup$
1
$\begingroup$

Guide:

Use Bayes rule, \begin{align} P(N_1(t)=n_1 |G(t)=k) &= \frac{P(N_1(t)=n_1 )}{P(G(t)=k)} \cdot P(G(t)=k|N_1(t)=n_1) \\ &= \frac{P(N_1(t)=n_1 )}{P(G(t)=k)} \cdot P(N_2(t)=k-n_1) \end{align}

Hopefully, you can evaluate the expression above.

Remark: Correction to your mistake.

$$P(N_1(t) = n_1 | G(t) = k) = P(N_2(t) = k - n_1 \color{red}{|G(t)=k})$$

$\endgroup$
  • $\begingroup$ Ok perfect, but I can not understand why can we write that $P(G(t)=k|N_1(t)=n_1) = P(N_2(t)=k-n_1)$ and not $P(N_1(t) = n_1 | G(t) = k) = P(N_2(t) = k - n_1)$ $\endgroup$ – Gabriele Picco Dec 17 '17 at 17:57
  • $\begingroup$ $P(G(t)=k|N_1 = n_1) = P(N_2(t)=k-n, N_1(t)=k|N_1(t)=k)=P(N_2(t)=k-n,|N_1(t)=k) = P(N_2(t)=k-n)$ because $N_2$ and $N_1$ are independent. For the other case $P(N_1(t)=n_1|G(t)=k) = P(N_2(t)=k-n_1|G(t)=k)$ and we can't drop the condition because $G(t)$ and $N_2(t)$ are not independent. $\endgroup$ – Siong Thye Goh Dec 17 '17 at 18:02
  • $\begingroup$ Perfect, all clear thank you ! $\endgroup$ – Gabriele Picco Dec 17 '17 at 18:06

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.