2
$\begingroup$

I am attempting exercise 1.8.1 from Brémaud's Markov Chains: Gibbs Fields, Monte Carlo Simulation, and Queues.

Let $S_1, S_2, \ldots$ be a sequence of i.i.d. random variables, with $\mathbb{P}(0 < S_1 < \infty) = 1$ and $\mathbb{E}\left[S_1\right] < \infty$. For $t \geq 0$, let $N_t = \sum_{n=1}^\infty {1_{\left(0, t\right]}\left(T_n\right)}$ where $T_n = S_1 + \cdots + S_n$. Prove that, almost surely, $$\lim_{t \to \infty} {\frac{N_t}{t}} = \frac{1}{\mathbb{E}\left[S_1\right]}$$

I am not sure how to get started. Apparently I should be applying the Strong Law of Large Numbers but I don't see immediately how it applies.

Hints please; thank you!

$\endgroup$
2
  • 3
    $\begingroup$ Hint: $N_{S_n}/S_n = n/S_n \to 1/\mathbb{E}[S_1]$ a.s. $\endgroup$
    – zhoraster
    Oct 6 '16 at 5:49
  • $\begingroup$ @zhoraster Did you mean $N_{T_n} / T_n = n / T_n$? $\endgroup$ Oct 6 '16 at 20:49
2
$\begingroup$

Since $S_{N_t}\leqslant t\leqslant S_{N_t+1}$, we have $$\frac{S_{N_t}}{N_t}\leqslant \frac t{N_t} \leqslant \frac{S_{N_t+1}}{N_t}. $$ Now, $N_t\stackrel{t\to\infty}\longrightarrow\infty$ almost surely, so by the strong law of large numbers, $$\mathbb P\left(\lim_{t\to\infty} \frac{S_{N_t}}{N_t}= \mu \right) = 1. $$ Similarly, $$\frac{S_{N_t+1}}{N_t} = \frac{S_{N_t+1}}{N_t+1}\frac{N_t+1}{N_t}\stackrel{t\to\infty}\longrightarrow \mu $$ with probability $1$, so we conclude that $$\mathbb P\left(\lim_{t\to\infty} \frac t{N_t} =\mu\right)=1, $$ or equivalently $$\mathbb P\left(\lim_{t\to\infty} \frac{N_t}t =\frac1\mu\right)=1. $$

$\endgroup$
4
  • $\begingroup$ Sorry, I don't entirely follow. Did you mean $T_{N_t} \leq t \leq T_{N_t + 1}$? $\endgroup$ Oct 6 '16 at 20:48
  • $\begingroup$ @Arman Your notation is a bit nonstandard; generally $S_n$ denotes the time of the $n^{\mathrm{th}}$ renewal (where here you've used $T_n$). It is a bit more intuitive, as $S_n$ denotes a Sum whereas a capital $T$ is often used to denote a fixed Time. $\endgroup$
    – Math1000
    Oct 6 '16 at 21:09
  • $\begingroup$ I thought as much, but this was the notation chosen by Brémaud in the book. $\endgroup$ Oct 6 '16 at 21:11
  • $\begingroup$ Fair enough. It's worth mentioning that the notation I used in my post is that from the course in which I studied renewal theory three years ago, as well as the (entirely different) text that I was reviewing the other day. $\endgroup$
    – Math1000
    Oct 6 '16 at 21:13

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.