Einstein relation and steady states for the random conductance model (Q2406569)

Let \(\{\omega:\omega(x,x+e)\in (\kappa^{-1},\kappa); \kappa>1; x\in \mathbb{Z}^d, d \geq2; |e|=1\}\) be i.i.d. random elliptic conductances which form the environment. The nearest-neighbor random walk \(\{X_n\}_{n\geq 1}\) in the environment \(\omega\) is a Markov chain with transition probabilities \[ P_\omega^x(X_{n+1}=y+e|X_{n+1}=y)=\frac{\omega(y,y+e)} {\sum_{|e'|=1}\omega(y,y+e')} \] and \(P_\omega^x(X_{0}=x)=1\). For \(\lambda>0\) and \(\ell \in \mathbb{R}^d\), \(|\ell|=1\), a perturbed environment is defined by \(\omega^\lambda(x,y)=\omega(x,y)\exp(\lambda \ell \cdot (x+y)\) where ``\(\cdot\)'' is a scalar product. Let \(v(\lambda)=\lim_{n \to \infty} X_n/n\) which is known to exist \(\mathbb{P}_\lambda\)-a.s, where \(\mathbb{P}_\lambda\) is annealed law. In this article, the Einstein relation \[ \lim_{\lambda \to +0} \frac{v(\lambda)}{\lambda} =\Sigma \ell \] is established, where \(\Sigma\) is a limiting (as \(n \to \infty\)) diagonal covariance matrix of \(\{X_n\}_{n\geq 1}\) calculated under diffusive scaling at \(\lambda=0\). When \(\theta_{z}(\omega)=\omega(x+z,y+z)\) denotes the environment shifted by \(z\), the Markov chain \(\{\theta_{X_n}(\omega)\}\) is called the environment seen from the particle. For a bounded function \(f:\Omega \to R\), an invariant measure \(Q_\lambda\) is defined as \(\mathbb{P}_\lambda\)-a.s. limit \[ Q_\lambda(f)=\lim_{n \to\infty}n^{-1}\sum_{k=1}^n f(\theta_{X_k}(\omega)). \] The invariant measure \(Q_\lambda\) is called steady states. The Einstein relation follows from the first-order expansion of \(Q_\lambda\) at \(\lambda=0\).