Basic Limit Theorem for Markov Chain (Knowing the odds)

922 Views Asked by At

In the book "Knowing the Odds", Basic Limit Theorem for Markov Chain is stated as follows.

Theorem 7.41 (Basic Limit Theorem). Suppose $j$ is a recurrent aperiodic state in an irreducible Markov chain. Then for each i, $ \lim_{n\to\infty} P_{ij}^{(n)}= \dfrac{1}{E^{j}\left\{T_{j}\right\}} $.

In the second line of the proof of this theorem, the author mentions that "To simplify the proof, suppose that $P_{jj}>0$, so that $f_{1}>0$. " But what if $P_{jj}=0$? Is it related to the condition "aperiodic" in this theorem? I had checked the rest of the proof, but I don't think aperiodic in the case $P_{jj}>0$ is needed.

Thanks!