Coarse ellipticity and De Giorgi-Nash-Moser theory in the optimal range

Abstract

We extend the theory of De Giorgi-Nash-Moser to elliptic equations with symmetric coefficients $\mathbf{a}(x)$ which are possibly degenerate and unbounded but satisfy a coarse ellipticity condition. We prove local upper bounds for weak subsolutions, a weak Harnack inequality for nonnegative supersolutions, and a Harnack inequality for nonnegative solutions. The coarse ellipticity hypothesis requires spatial moments of coarse-grained matrices, suitably discounted and summed across scales, to be finite. In particular, it holds if, for some $α,β\geq0$ and $1 0$. For $α=β=0$, it corresponds to the results of Bella and Schäffner [8].