I have a delaunay triangulation. Each point has an absolute coordinate in R2. Each edge has a known x, y distance between each point. There is error, and so the solution of absolute coordinates cannot satisfy the system, its over-determined. I would like to find the solution of absolute coordinates that minimizes the inf norm of this system. I have been using openCV HEAVILY for this project and so I intended to use cv::solveLP(). However, the answer cv::solveLP produced was way off. After looking at the documentation, I saw that solvLP() imposes the constraint x>=0. It seems this is not an option you can change. I realize that many real world LP problems resolve around numbers of objects, and often x>=0 makes sense. But why would they impose this constraint without giving you the option? and is there a way to still use this function for a solution where x can take on negative values?