This is a course project for Computational Geometry & GPGPU in NUS SoC Summer Workshop 2018.
Linear Programming (LP) is a technique for the optimization of a linear objective function, subject to linear inequality constraints. Its feasible region is a convex polytope, which is a set defined as the intersection of finitely many half-spaces. Each linear programming algorithm finds a point in the polyhedron where this function has the smallest (or largest) value if such a point exists.
We used CUDA\THRUST toolkit to solve the 2D Linear Programming problem on the GPU. This repo includes a parallel implementation of Migiddo's Algorithm and an advanced version with some tricks called V-Shape that runs faster than the original one.
Thanks to my friends Xintong and Xinyi.