Eigenvalue Inequalities for Fully Nonlinear Elliptic Equations via the Alexandroff-Bakelman-Pucci Method
Dimitrios Gazoulis
TL;DR
This work develops eigenvalue inequalities for elliptic operators by exploiting the Alexandroff-Bakelman-Pucci (ABP) method and extends these ideas to fully nonlinear equations. Starting from the ABP estimate for subsolutions, the authors derive a bound on the normal derivative at the boundary and then establish lower bounds for the $L^n$ (and $L^p$ to infinity) norms of the Laplacian in terms of boundary data, with sharp cases linking equality to symmetry. These ABP-driven bounds yield eigenvalue inequalities for the Laplacian under Dirichlet and Robin conditions, and they are extended to the Monge-Ampère operator, providing analogous eigenvalue lower bounds. The paper then broadens the framework to fully nonlinear elliptic equations via Pucci operators, obtaining an $L^{\infty}$ gradient bound and a $C^3$ regularity estimate, and establishing corresponding Dirichlet and Robin eigenvalue inequalities in the viscosity-solution setting. Collectively, the results demonstrate that ABP-type estimates are powerful tools for controlling nonlinear elliptic operators and yield sharp, symmetry-attained bounds in several classical eigenvalue problems, with clear implications for fully nonlinear theory and regularity.
Abstract
In this work we establish eigenvalue inequalities for elliptic differential operators either for Dirichlet or for Robin eigenvalue problems, by using the technique introduced by Alexandroff, Bakelman and Pucci. These inequalities can be extended for fully nonlinear elliptic equations, such as for the Monge-Ampère equation and for Pucci's equations. As an application we establish a lower bound for the $ L^p -$norm of the Laplacian and this bound is sharp, in the sense that, when equality is achieved then a symmetry property is obtained. In addition, we obtain an $ L^{\infty} $ bound for the gradient of solutions to fully nonlinear elliptic equations and as a result, a $ C^3 $ estimate.
