Random walk with different probability at the origin

396 Views Asked by At

I'm trying to solve the following problem. Consider the random walk on the integers. At zero the probability to go right is $1/2$ and the probability to go left is $1/2$. Everywhere else going to the right has probability $p$ and going to the left has probability $1-p$. I want to find the probability of ever returning back to zero if the random walk starts at zero.

My solution:

We know random walk is markov chain, so call it markov chain $(X_n)_{n\geq 0}$. Let's define the probability of ever returning to $i$ if the random walk starts in $i$ as $f_{ii}$ and $f_{ii}^n$ as returning to $i$ with exactly $n$ steps while $X_k\neq i$ if $0<k<n$. In this random walk there is only returning back with even number of steps, so $f_{ii}^{2n+1}=0$. Clearly:

\begin{align} f_{00} = \sum\limits_{k=1}^{\infty} f_{00}^{2k} \end{align}

The first term: \begin{align} f_{00}^{2}=p_{0,1} \cdot p_{1,0} + p_{0,-1}\cdot p_{-1,0} = \frac{1}{2} (1-p) + \frac{1}{2} p = \frac{1}{2} \end{align}

For the next terms we have something like (I dont write it down in formal way, because I want to know if the idea is right): \begin{align} \sum\limits_{k=2}^{\infty} f_{00}^{2k} = p_{0,1} \cdot \mathbb{P}(\text{Ever returning to 1 without going less than 1} | X_1 = 1)\cdot p_{1,0}+ p_{0,-1} \cdot \mathbb{P}(\text{Ever returning to -1 without going greater than -1} | X_1 = -1) \cdot p_{-1,0} \end{align} And that is equal to: \begin{align} \frac{1-p}{2} \mathbb{P}(\text{Ever returning to 1 without going less than 1} | X_1 = 1) + \frac{p}{2} \mathbb{P}(\text{Ever returning to -1 without going greater than -1} | X_1 = -1) \end{align} If the random walk is at $1$ and has positive number of steps left before returning to $0$, then the random walk can do whatever it "wants" as long as it stays to the right side of 1. And when it eventually returns to 1 then it must take one step to the left to reach 0. The same holds for $-1$. I know $\mathbb{P}(\text{Ever returning to 1 without going less than 1} | X_1 = 1)$ is like the probability of ever returning to 1 in one-sided random walk and that is equal to: $\frac{1}{2}(1-|2p-1|)$. The same holds for $\mathbb{P}(\text{Ever returning to -1 without going greater than -1} | X_1 = -1)$ because of symmetry.

So I end up with: \begin{align} f_{00}= \frac{1}{2}+\left(\frac{1-p}{2}+\frac{p}{2}\right)\cdot \frac{1}{2}(1-|2p-1|) = \frac{3}{4}-\frac{1}{4} |2p-1| \end{align}

Question: I know this is wrong, because that must be 1 when $p=\frac{1}{2}$ (recurrent random walk). However that is not the case with my solution. Can you guys tell me where I went wrong? Thanks!