Interior estimates for the \(n\)-dimensional Abreu's equation (Q2445366): Difference between revisions

The authors derive interior estimates of solutions of the Abreu equation in polytopes under uniform \(K\)-stability assumptions. More precisely, the main result is a new upper bound of determinants of the Hessian of solutions. Previous results in dimension \(2\) were obtained in [\textit{S. K. Donaldson}, Collect. Math. 56, No. 2, 103--142 (2005; Zbl 1085.53063)]. The Abreu equation is a non-linear fourth-order partial differential equation, whose interest arises from the study of metrics with prescribed scalar curvature and extremal metrics on toric varieties. Its solvability is closely related to stability conditions. Let \(\Delta\) be a bounded open polytope in \(\mathbb{R}^n\), and \(A\) be a smooth function on \(\bar{\Delta}\). The Abreu equation is \[ \sum_{i,j=1}^{n} \frac{\partial^2 u^{ij}}{\partial \xi_i \partial \xi_j} = - A , \] where \(\left( u^{ij} \right)\) denotes the inverse of the Hessian \(\left( u_{ij} \right)\). It is the Euler-Lagrangian equation for the ``Mabuchi functional'' \[ \mathcal{F}_A(u) = -\int_{\Delta} \log\det\left(u_{ij}\right)d\mu+\mathcal{L}_A(u) \] where \[ \mathcal{L}_A(u) = \int_{\partial\Delta} u d\sigma - \int_{\Delta} Au d\mu \] is related to the Futaki invariant. Following Donaldson, \((\Delta,A)\) is called uniformly \(K\)-stable if \(\mathcal{L}_A\) vanishes on affine-linear functions and there exists a constant \(\lambda>0\) such that, for any \(u \in C\left(\bar\Delta\right)\) convex on \(\bar\Delta\) and smooth on \(\Delta\) such that \(u\geq u(p_0)=0\), (for a fixed \(p_0\in\Delta\),) the inequality \(\mathcal{L}_A(u) \geq \lambda \int_{\partial\Delta} u d\sigma\) holds. Uniform \(K\)-stability is a necessary condition for the existence of solutions of the Abreu equation satisfying the Guillemin boundary conditions, see [\textit{B. Chen} et al., J. Differ. Equations 257, No. 5, 1487--1500 (2014; Zbl 1300.53067)]. In dimension \(2\), it is also sufficient, as follows from results by Donaldson, and by Chen, Li, and Sheng. The main result of the paper is Theorem 1.2, equivalently Theorem 4.2. Assume that \((\Delta, A)\) is uniformly \(K\)-stable. Let \(u\) be a solution of the Abreu equation in the space \(\mathcal{S}_{p_0}\) of functions \(u \in C\left(\bar\Delta\right)\) such that \(u\) is convex on \(\bar\Delta\) and \(u-v\) is smooth on \(\bar\Delta\) and \(u\geq u(p_0)=0\), (for a fixed \(p_0\in\Delta\),) where \(v=\sum_k\delta_k\log\delta_k\) with \(\delta_k\) defining the polytope. Then, for any \(\Omega\Subset\Delta\), for any \(k\in\mathbb{N}\), for any \(\alpha\in (0,1)\), there exists a constant \(C>0\) depending only on \(n\), \(k\), \(\alpha\), \(\Omega\), \(\lambda\) such that \[ \| u \|_{C^{k+3,\alpha}(\Omega)} \leq C\cdot \| A \|_{C^k\left(\bar{\Delta}\right)} . \] In fact, the authors establish interior estimates for the Legendre transform of the solution.