0
$\begingroup$

Here's what I have:

Let X be a simple birth-death process where individuals have independent $\text{Exp}(\mu)$ lifetimes and, during their lifetime give birth at rate $λ$ independently of other individuals.

Let $T = \inf\{t ≥ 0: X_{t} = 0\}$ be the extinction time for the population.

I have to find the density of T.

Earlier in the question I had to show that if $G(s, t) = E(s^{X_{t}})$, then $G$ satisfies :

$$ \frac{∂}{∂t}G(s, t) = (λs − \mu)(s − 1) \frac{∂}{∂s}G(s, t)$$

which I did.

The hint is that the solution to the above is given by:

$$\frac{\mu(s − 1) − (\lambda s − \mu)e^{-(\lambda-\mu)t}}{\lambda(s − 1) − (\lambda s − \mu)e^{-(\lambda-\mu)t}} $$ or

$$\frac{\lambda t(s-1)-s}{\lambda t(s-1)-1}$$

depending on whether $\mu \neq \lambda$ or $\mu = \lambda$.

Another observation is that $G(0, t) = P(X_t = 0)$ and that $$\inf\{t ≥ 0: X_{t} = 0\} \leq k \Leftrightarrow X_k=0$$

I just can add all those things up to a solution.

Any help is appreciated

$\endgroup$
1
$\begingroup$
  • Note that $G(s,t) = E[s^{X_t}]$ is the probability generating function of $X_t$. Therefore, $$P(X_t = k) = \frac{d^kG(s,t)}{ds^k}(0,t)$$ As a special case, $G(0,t)=P(X_t=0)$.

  • Next, the hint tells you that $P(T \leq t) = P(X_t=0)$. Can yo you use the previous item to determine $P(T \leq t)$?

  • Finally, use $$f_{T}(t)=\frac{dP(T \leq t)}{dt}$$

$\endgroup$
  • $\begingroup$ $P(T \leq k) = G(0,k)$ but then if I've not made a mistake, the density comes out quite nasty $\endgroup$ – asdf Nov 13 '17 at 15:00
  • $\begingroup$ I don't understand your comment. $\endgroup$ – madprob Nov 13 '17 at 15:02
  • $\begingroup$ We have that $P(T \leq k)=G(0,k)$ which is given by the above solutions. But differentiating this doesn't give anything sensible. Or I've made a mistake somewhere $\endgroup$ – asdf Nov 13 '17 at 15:03
  • $\begingroup$ You didn't indicate anywhere in your question that you had tried this path already. How can I know? $\endgroup$ – madprob Nov 13 '17 at 15:04
  • $\begingroup$ No way you could, but I was just wondering whether we are expected to get something familiar or is this badly looking in general $\endgroup$ – asdf Nov 13 '17 at 15:05

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.