Reproduce “Low-Rank Clebsch--Gordan Coupling for Equivariant Spherical Forecasting” 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

Yaojun Li, Zhixuan Zhao, Dongzhe Zheng, Christine Allen-Blanchette

LCG-SFNO forecasts global atmospheric fields while targeting consistency under rotations. It combines full-spectrum degree-wise processing with learned Clebsch–Gordan interactions in a restricted low-degree band, and projects a quadratic nonlinearity without aliasing. On one-day WeatherBench, its relative equivariance error was 6.6 × 10⁻⁷ versus 9.5 × 10⁻² for SFNO. This is near numerical precision in the reported finite diagnostic, not a guarantee for all inputs.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.