The singularities of the distance function near convex boundary points (Q1021401)

Let \(\Omega \subset \mathbb{R}^{n}\) be an open set and consider the viscosity solution of the eikonal equation \[ \begin{cases} \left\langle A(x)\nabla d(x),\nabla d(x)\right\rangle =1\text{ in }\Omega \\ d(x)=0\text{ for }x\in \partial \Omega , \end{cases} \] where \(\overline{\Omega }\ni x\mapsto A(x)\) is a matrix valued map of class \(C^{1,1}\) taking values in the set of the positive definite matrices. The above Dirichlet problem admits a unique solution which is the distance function from \(\partial \Omega \) with respect to the Riemann metric \( g_{x}(\xi ,\xi )=\left\langle A(x)\xi ,\xi \right\rangle .\) The author is interested in the structure of the singular set of the distance function: \[ \Sigma (d):=\left\{ x\in \overline{\Omega }:d\text{ is not differentiable at }x\right\} . \] He proves that if \(x_{0}\in \partial \Omega \) is such that \(\Omega \cap B_{\rho }(x_{0})\) is convex for a suitable \(\rho >0,\) then \(x_{0}\in \Sigma (d)\) if and only if \(\partial \Omega \) is not differentiable at \(x_{0}.\)