How to show that some rays are maximal transport rays in Monge Problem (Q2504942)

If the ambient space is the closure \(\Omega\) of an open, bounded, and convex subset \(\Omega_ 0\subset \mathbb R^ 2\), and \(f^ +\) and \(f^ -\) are two probability measures, then a transport map from \(f^ +\) to \(f^ -\) is a Borel map \(t:\Omega\to \Omega\) such that \(t_{\#}f^ +=f^ -\), where \(t_{\#}: {\mathcal M}^ +(\Omega)\to{\mathcal M}^ +(\Omega)\) is defined as \(t_{\#}f^ +(B)=f^ +(t^{-1}(B))\) for a Borel set \(B\subset \Omega\). If \(f^ +\) and \(f^ -\) are thought to measure respectively the distribution of some mass and the depth of a hole, then any transport map is a strategy to move all the mass inside the hole to fill it. For every transport map there is a cost function \(C(t)=\int_\Omega|t(x)-x|\,df^+(x)\). The mass transportation problem is to find an optimal transport map, that is, a transport map minimizing the cost \(C\). In [Caffarelli, Luis A. (ed.) et al., Optimal transportation and applications. Lectures given at the C.I.M.E. summer school, Martina Franca, Italy, September 2-8, 2001. Berlin: Springer. Lect. Notes Math. 1813, 123--160 (2003; Zbl 1065.49026)], \textit{Ambrosio} and the author proved that a transport plan is optimal if and only if all its maximal transport rays are horizontal segments. In this paper, the author generalizes this theorem by constructing a balanced ray configuration (i.e. a set of segments with suitable properties), and then checking that this ray configuration satisfies a certain condition. The theorem states that the given ray configuration together with its open subsegments is just all the admissible transport rays for the given problem.