high performance computing on graphics processing units: hgpu.org

Simulating frictional contact is a challenging computational task and there exist a variety of techniques to do so. One such technique, the staggered projections algorithm, requires the solution of two convex quadratic program (QP) subproblems at each iteration. We introduce a method, SCHURPA, which employs a primal-dual active-set strategy to efficiently solve these QPs based on a Schur-complement method. A single factorization of the initial saddle point system and a smaller dense Schur-complement is maintained to solve subsequent saddle point systems. Exploiting the parallelizability and warm-starting capabilities of the active-set method as well as the problem structure of the QPs yields a novel approach to the problem of frictional contact. Numerical results of a parallel GPU implementation using NVIDIA’s CUDA applied to a physical simulator of highly deformable bodies are presented.