Robust and Efficient Penetration-Free Elastodynamics without Barriers

The paper proposes a barrier-free optimization framework for non-penetration elastodynamic simulation, aiming to keep the robustness of Incremental Potential Contact while removing two common pain points: ill-conditioned systems from logarithmic barriers and TOI locking that slows active-set progress. Instead of increasing penalty stiffness, it uses a custom augmented Lagrangian approach that updates multipliers and pulls in contact pairs from CCD immediately.

For developers, the practical takeaway is performance and solver stability in contact-rich scenes. The authors report fewer Newton iterations, a compact active set via filtering/decay, and a GPU-optimized simulator that reaches up to 103x speedup over GIPC on hard benchmarks. They also claim finite-step termination and first-order time integration accuracy under a cumulative TOI-based stopping rule, which is the kind of detail that matters if you care about predictable convergence rather than just pretty demos.