Lipschitz regularity for elliptic equations with random coefficients (Q5962862)

The authors prove that under certain mixing conditions on the coefficients a quasi-linear elliptic equation in divergence form \[ \operatorname{div}(A(\nabla u(x), x)) = 0 \] with random coefficients satisfies due to statistical effects (in contrast to a general equation of the above form) the large scale estimate \[ \frac{1}{|B_r|} \int_{B_r} |\nabla u(x)|^2 \, dx \leq C \left( 1 + \frac{1}{|B_R|} \int_{B_R} |\nabla u(x)|^2 \, dx \right) \] for all \(r \in [X, \frac{1}{2} R]\). Here \(X\) is a random variable giving the minimal radius for the estimate to hold. Furthermore, optimal stochastic integrability estimates on \(X\) are established for a broad range of mixing conditions on the coefficients.