A discrete stochastic integral inequality and balanced random walk in a random environment (Q1060777)

Let \(\Lambda \subset {\mathbb{Z}}^ d\) be a fixed finite set, \(S^{\pi}(n)\) a Markov chain on \({\mathbb{Z}}^ d\) with transition probability function \[ \pi (x,y)=P(S^{\pi}(n+1)=y| S^{\pi}(n)=x), \] x,y\(\in {\mathbb{Z}}^ d\), \(n\in {\mathbb{Z}}^+\), \(\pi (x,y)=0\) if y-x\(\not\in \Lambda\). We say that \(\pi\) (x,y) is balanced if \(\pi (x,x+e)=\pi (x,x-e)\). For \(\alpha >0\), let \(C^{\alpha}\) denote the set of all balanced \(\pi\) such that \(\sum_{e\in \Lambda}| ea| \pi (x,x+e)\geq \alpha | a|\). The main result: If \(\mu\) is a stationary ergodic probability measure on \(C^{\alpha}\) and if there exists \(y\in {\mathbb{Z}}^ d\), \(y\neq 0\), such that, for \(\mu\)- almost every \(\pi\), \((P=y\) for some \(n)>0\) then, for \(\mu\)-almost every \(\pi\), \(X_ n(t)=S^{\pi}([nt])/\sqrt{n}\) converges in distribution to a diffusion with zero drift and constant covariance.