Bond percolation games on the $2$-dimensional square lattice, and ergodicity of associated probabilistic cellular automata

We consider bond percolation games on the $2$-dimensional square lattice in which each edge (that is either between the sites $(x,y)$ and $(x+1,y)$ or between the sites $(x,y)$ and $(x,y+1)$, for all $(x,y) \in \mathbb{Z}^{2}$) has been assigned, independently, a label that reads "trap" with probability $p$, "target" with probability $q$, and "open" with probability $1-p-q$. Once a realization of this labeling is generated, it is revealed in its entirety to the players before the game starts. The game involves a single token, initially placed at the origin, and two players who take turns to make moves. A move involves relocating the token from where it is currently located, say the site $(x,y)$, to any one of $(x+1,y)$ and $(x,y+1)$. A player wins if she is able to move the token along an edge labeled a target, or if she is able to force her opponent to move the token along an edge labeled a trap. The game is said to result in a draw if it continues indefinitely (i.e.\ with the token always being moved along open edges). We ask the question: for what values of $p$ and $q$ is the probability of draw equal to $0$? By establishing a close connection between the event of draw and the ergodicity of a suitably defined probabilistic cellualar automaton, we are able to show that the probability of draw is $0$ when $p > 0.157175$ and $q=0$, and when $p=q \geqslant 0.10883$.
Comment: Main body of the paper: 21 pages. Appendix: from page 21 till page 43. No. of figures: 4