stuck with system of equations coming from markov chain's stationary distribution

68 Views Asked by At

In each of the following cases determine whether the stochastic matrix P, which you may assume is irreducible, is reversible:

$ \begin{pmatrix} 0 & p & 1-p\\ 1-p & 0 & p\\ p & 1-p & 0 \end{pmatrix} $

So I am stuck with the following system of equations.

$ \pi_1p = \pi_2(1-p) \\ \pi_1(1-p) = \pi_3p \\ \pi_2p = \pi_3(1-p) \\ \pi_1 + \pi_2 + \pi_3 = 1 $

I think I used all possibilities but never my answer comes to the one provided:

$ \pi_1 = \frac{1}{1+p/(1-p) + (1-p) + p^2/(1-p)} \\ \pi_2 = \frac{p/(1-p)}{1+p/(1-p) + (1-p) + p^2/(1-p)} \\ \pi_3 = \frac{1-p+p^2/(1-p)}{1+p/(1-p) + (1-p) + p^2/(1-p)} $

Important factor is that p could be 0 (I think...). Could someone point me to the right path so that I get the provided answer? Also when I solve it myself my solutions somehow fail to fulfill the condition that all $\pi$ should sum up to one even though I use this property to solve the equation. What I am doing is first I substitute one of variables say $\pi_1$ with $\pi_1 = 1 - \pi_2 - \pi_3$ in both equations where it appears and then I try to substitute one of $\pi_2$ or $\pi_3$ with other equation. Basically I am stuck with high school math, any direction to unstuck me would be highly appreciated!

One of my attempts:

$ \pi_1p = \pi_2(1-p) \\ \pi_1(1-p) = \pi_3p \\ \pi_2p = \pi_3(1-p) \\ \pi_1 + \pi_2 + \pi_3 = 1 \\ $

$ (1-\pi_2-\pi_3)p = (1-p)\pi_2 \\ (1-\pi_2-\pi_3)(1-p) = \pi_3p \\ \pi_2p = \pi_3(1-p) \\ $

$ p-\pi_2p-\pi_3p = \pi_2-\pi_2p \\ 1-\pi_2-\pi_3-p+\pi_2p+\pi_3p = \pi_3p \\ \pi_2p = \pi_3(1-p) \\ $

$ p-\pi_3p = \pi_2 \\ 1-\pi_2-\pi_3-p+\pi_2p = 0 \\ \pi_2p = \pi_3(1-p) \\ $

$ p-\pi_3p = \pi_2 \\ 1-p+\pi_2p-\pi_2-\pi_3 = 0 \\ \pi_2p = \pi_3(1-p) \\ $

$ p-\pi_3p = \pi_2 \\ 1-p+\pi_2(p-1)-\pi_3 = 0 \\ \pi_2p = \pi_3(1-p) \\ $

$ p-\pi_3p = \pi_2 \\ \pi_2(p-1) = p-1+\pi_3 \\ \pi_2p = \pi_3(1-p) \\ $

$ (p-\pi_3p)(p-1) = p-1+\pi_3 \\ (p-\pi_3p)p = \pi_3(1-p) \\ $

$ \pi_3(1-p)-(p-\pi_3p) = p-1+\pi_3 \\ $

$ p = \frac{1}{2} \\ $

But answer remains a mystery