Basic Limit Theorem of Markov Chains

549 Views Asked by At

Consider the following theorem from the text "An Introduction to Stochastic Modeling" by Mark A. Pinsky and Samuel Karlin

Theorem 4.3 (a.) Consider a recurrent irreducible aperiodic Markov chain. Let $P_{ii}^{(n)}$ be the probability of enetering state i at the nth transition, $n=0,1,2, \cdots,$ given that $X_{0} =i$ (the initial state is $i$). By convention $P_{ii}^{(0)}=1$. Let $f_{ii}^{(n)}$ be the probability of first returning to state $i$ at the nth transition, $n=0,1,2 \cdots $, where $f_{ii}^{(0)} = 0$. Then, $lim_{n\rightarrow \infty}P_{ii}^{(n)} = \frac{1}{\sum_{n=0}^\infty nf_{ii}^{(n)}}=\frac{1}{m_i}$. (b.)Under the same conditions as in (a.), $lim_{n\rightarrow \infty}P_{ji}^{(n)} = lim_{n \rightarrow \infty}P_{ii}^{(n)}$ for all states $j$.

I am having a difficult time setting the proof for this up. I am tempted to do an $\epsilon$ proof for sequences here. My first attempt was doing this proof direct: $\lim_{n \rightarrow \infty} P_{ii}^{(n)} = \lim_{n \rightarrow \infty}\sum_{k=0}^n f_{ii}^{(n)}P_{ii}^{(n-k)} = \sum_{k=0}^\infty f_{ii}^{(n)}P_{ii}^{(n-k)}$. Then here I do not know where to go. Similar to the second part of this theorem. Please provide a detailed proof and references if necessary to this theorem.