Volume estimates on the critical sets of solutions to elliptic PDEs

Abstract

In this paper we study solutions to elliptic linear equations $L(u)=\partial_i(a^{ij}(x)\partial_j u) + b^i(x) \partial_i u + c(x) u=0$, either on $R^n$ or a Riemannian manifold, under the assumption of Lipschitz control on the coefficients $a^{ij}$. We focus our attention on the critical set $Cr(u)\equiv\{x:|\nabla u|=0\}$ and the singular set $S(u)\equiv\{x:u=|\nabla u|=0\}$, and more importantly on effective versions of these. Currently, under the coefficient control we have assumed, the strongest results in the literature say that the singular set is n-2-dimensional, however at this point it has not even been shown that $H^{n-2}(S)<\infty$ unless the coefficients are smooth. Fundamentally, this is due to the need of an $\epsilon$-regularity theorem which requires higher coefficient control as the frequency increases. We introduce new techniques for estimating the critical and singular set, which avoids the need of any such $\epsilon$-regularity. Consequently, we prove that if the frequency of u is bounded by $\Lambda$ then we have the estimates $H^{n-2}(C(u))\leq C^{\Lambda^2}$, $H^{n-2}(S(u))\leq C^{\Lambda^2}$, depending on whether the equation is critical or not. More importantly, we prove corresponding estimates for the {\it effective} critical and singular sets. Even under the assumption of analytic coefficients these results are much sharper than those currently in the literature. We also give applications of the technique to the nodal set of solutions, and to give estimates on the corresponding eigenvalue problem.