Monge-Ampère equation with Guillemin boundary condition in high dimension

Genggeng Huang, Weiming Shen

Published 2024-06-08Version 1

The Guillemin boundary condition appears naturally in the study of K\"ahler geometry of toric manifolds. In the present paper, the following Guillemin boundary value problem is investigated \begin{align} \label{eq1} &\det D^2 u=\frac{h(x)}{\prod_{i=1}^N l_i(x)},\quad\text{in}\quad\quad P\subset\mathbb R^n, \quad\quad \quad \quad\quad \quad \quad \quad\quad (1)\\ \label{bdy1} &u(x)-\sum_{i=1}^N l_i(x)\ln l_i(x)\in C^\infty(\overline{P}). \quad\quad\quad\quad \quad \quad\quad \quad \quad \quad\quad\quad (2) \end{align} Here \begin{equation*} 0<h(x)\in C^\infty(\overline{P}),\quad P=\cap_{i=1}^N \{l_i(x)>0\} \end{equation*} is a simple convex polytope in $\mathbb R^n$. The solvability of (1)-(2) is given under the necessary and sufficient condition. The key issue in the proof is to obtain the boundary regularity of $u(x)-\displaystyle \sum_{i=1}^N l_i(x)\ln l_i(x)$. Due to the difficulty caused by the structure of the equation itself and the singularity of $\partial P$, we need to pay special attention to the influence of the difference of singularity types at different positions on $\partial P$ on the behavior of $u$ in its vicinity.