1
$\begingroup$

I'm trying to work my way through a problem which defines $N_t$ as a Poisson process of rate $\lambda$ and $ X_n = N_n − n,\quad\text{for }\; n = 0, 1, 2, \ldots $

I've explained why $X_n$ is a Markov chain and I've found the transition probabilities to be

$$ P(X_{n+1} = k+i \mid X_n = k) = e^{-\lambda}\frac{\lambda^{i+1}}{(i+1)!} $$ for $i \geq0$, $$ P(X_{n+1} = k-i \mid X_n = k) = e^{-\lambda} $$ for $i = 1$, and $0$ otherwise.

Now I'm supposed to use the strong law of large numbers to show that the train is transient if $\lambda \neq 1$. The strong law relies on each random variable in the sum being i.i.d, and I really have no idea how to approach this.

Any help you could offer would be much appreciated.

$\endgroup$
2
$\begingroup$

Hint:

  1. Clearly, $$X_n = N_n-n = \sum_{j=1}^n \underbrace{(N_j-N_{j-1}-1)}_{=:\xi_j}$$ Since the Poisson process $(N_t)_{t \geq 0}$ has independent and stationary increments, the random variables $\xi_j$, $j \geq 1$, are independent and identically distributed. Show that $$\mathbb{E}(\xi_1) = \lambda-1.$$
  2. By step 1 and the strong law of large numbers, we get $$\lim_{n \to \infty} \frac{X_n}{n} = \mathbb{E}(\xi_1)=\lambda-1 \quad \text{a.s.}$$
  3. Deduce that $\lim_{n \to \infty} X_n = \infty$ if $\lambda>1$, and $\lim_{n \to \infty} X_n = -\infty$ almost surely if $\lambda<1$.
$\endgroup$
  • $\begingroup$ That's really helpful - thanks so much! $\endgroup$ – oxonian1 Sep 29 '18 at 12:55
  • $\begingroup$ @oxonian1 You are welcome. $\endgroup$ – saz Sep 29 '18 at 13:29

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.