Reproduce “Geometric Clifford Algebra Networks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
David Ruhe, Jayesh K. Gupta, Steven de Keninck, Max Welling, Johannes Brandstetter
Geometric Clifford Algebra Networks (GCANs) encode objects in geometric algebra and use learnable sandwich transformations and weighted combinations to model dynamical changes. The authors evaluate motion prediction and solver-generated fluid fields. On Tetris motion with 1,024 training trajectories, GCA-GNN reaches test MSE 0.0012 versus 0.0432 for a parameter-matched GNN; fluid gains vary by benchmark and architecture.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.