MAS-PNCG: Efficient Preconditioner for Contact Simulation

A new technique known as MAS-PNCG has been introduced by researchers, merging multilevel additive Schwarz preconditioning with nonlinear conjugate gradient optimization specifically for incremental potential contact (IPC) simulations. While IPC ensures that simulations remain free of intersections, it is resource-intensive due to the need for Hessian assembly and linear solves in Newton's method. Although PNCG eliminates Hessian assembly, it has faced challenges with convergence in stiff, contact-heavy situations, as basic Jacobi preconditioners do not effectively manage global coupling, and more complex hierarchy-based preconditioners like MAS are costly to update at each nonlinear iteration. The breakthrough lies in a Sparse-Input Woodbury update algorithm that adjusts fine-level MAS components in response to changing contact sets without requiring complete recomputation. This research was published on arXiv (ID: 2604.19892) on April 24, 2025, marking a significant advancement in hierarchical preconditioning for nonlinear optimization in IPC, which could lead to quicker and more reliable simulations in computer graphics and engineering fields related to contact mechanics.