I have the following Markov Chain with infinite state space $I=\{0,1,2,3,4,...\}$ and transition matrix $$P = \begin{bmatrix}q_0 & p_0 & 0 & 0 & 0 & 0 & ...\\q_1 & 0 & p_1 & 0 & 0 & 0 & ...\\ q_2 & 0 & 0 & p_2 & 0 & 0 & ... \\ q_3 & 0 & 0 & 0 & p_3 & 0 & ... \\ ... & ... & ... & ... & ... &... & ...\end{bmatrix}$$ where $p_j\in (0,1) \forall j\ge0$
I have to find whether the chain is positive recurrent when $p_j=e^{-\frac{1}{(j+1)^2}}, p_j=e^{-\frac{1}{(j+1)}}$ or $p_j=1-\frac{1}{\sqrt{j+2}}$ where $j \ge 0$
My idea is the following,
I know that if I find a stationary distribution $\pi = \pi P$ the expected return time $m_i = E_i(T_i) = \frac{1}{\pi_i}$ has to be less than infinity.
Should I try to find the stationary distribution, or is there an easier way to solve find whether the chain is positive recurrent?
Thanks in advance!