Convexity estimates for the Green's function (Q2349162): Difference between revisions

The author investigates the elliptic boundary value problem \[ \begin{cases}\Delta u=-\delta (x-x_0)\quad &\text{in }\Omega,\\ u=0\quad &\text{on }\partial\Omega,\end{cases}\leqno(1) \] where \(\Omega\) is a smooth, bounded convex domain in \(\mathbb{R}^n\), \(x_0\in\Omega\) and \(\delta(x-x_0)\) denotes the Dirac measure at the point \(x_0\). If \(u>0\) is the solution of problem \((1)\), and \(v=u^{\frac{1}{2-n}}\) for \(n\geq 3\), or \(v=e^{-\alpha u}\), \(\alpha>2\pi\) for \(n=2\) is a strictly convex function, then he shows that \(\psi=v^{2-n^2}\text{det} \,D^2v\) satisfies the elliptic differential inequality \(\Delta \psi\leq 0\mod (\nabla \psi)\), in \(\Omega\setminus \{x_0\}\), and \(\psi\) attains its minimum on the boundary \(\partial \Omega\). In the case where \(\Omega\) is strictly convex (that is, at every point of \(\partial\Omega\) all principal curvatures are strictly positive), then \(v=u^{\frac{1}{2-n}}\) \((n\geq 3)\) is strictly convex in \(\Omega\setminus\{x_0\}\).