Unidimensional semi-discrete partial optimal transport
We study the semi-discrete formulation of one-dimensional partial optimal transport with quadratic cost, where a probability density is partially transported to a finite sum of Dirac masses of smaller total mass. This problem arises naturally in applications such as risk management, the modeling of crowd motion, and sliced partial transport algorithms for point cloud registration. Unlike higher-dimensional settings, the dual functional in the unidimensional case exhibits reduced regularity. To overcome this difficulty, we introduce a regularization procedure based on thickening the density along an auxiliary dimension. We prove that the maximizers of the regularized dual problem converge to those of the original dual problem, with quadratic rate in the introduced thickness. We further provide a numerical scheme that leverages the regularized functional, and we validate our analysis with simulations that confirm the quadratic convergence rate. Finally, we compare the semi-discrete and fully discrete settings, demonstrating that our approach offers both improved stability and computational efficiency for unidimensional partial transport problems.