Busemann functions and barrier functions (Q681319)

On a Riemannian manifold \(M\), a \textit{ray} is a minimizing geodesic \(\gamma:[0,\infty)\to M\), while a \textit{line} is a minimizing geodesic \(\gamma:\mathbb R\to M\), consisting of the rays \(\gamma_+(t)=\gamma(t)\) and \(\gamma_-(t)=\gamma(-t)\). The Busemann function \(b_{\gamma}:M \to\mathbb R\) of a ray \(\gamma\) is \[ b_{\gamma}(x) = \lim_{t \to \infty} (d(x,\gamma(t))-t). \] The \textit{barrier} function \(B_{\gamma}:M \to\mathbb R\) of a line \(\gamma\) is \[ B_{\gamma} = b_{\gamma_+}+b_{\gamma_-}. \] The authors reprove a known convexity result for Busemann functions, and prove that Busemann functions are viscosity solutions of the eikonal equation. They prove that a Busemann function \(b_{\gamma}\) fails to be differentiable at a point \(x_0 \in M\) precisely when there are two or more geodesic rays emerging from \(x_0\) which, in a suitable sense, approximate \(\gamma\). They prove that a barrier function has some convexity properties, and some control on its growth. They prove that the zero set of a barrier function \(B_{\gamma}\) is the disjoint union of lines, each approximating \(\gamma\) in a suitable sense.