NVIDIA/torch-harmonics

Differentiable signal processing on the sphere for PyTorch

67
/ 100
Established

Implements spherical harmonic transforms (SHT) and vector SHT using quadrature rules paired with FFTs for efficient projection onto associated Legendre polynomials and harmonic bases. Built entirely on PyTorch primitives for full differentiability, with support for distributed quadrature across multiple ranks and custom CUDA kernels for spherical convolutions. Enables applications like Spherical Fourier Neural Operators and differentiable PDE solvers on spherical domains.

650 stars. Actively maintained with 21 commits in the last 30 days.

No Package No Dependents
Maintenance 23 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 18 / 25

How are scores calculated?

Stars

650

Forks

65

Language

Jupyter Notebook

License

Last pushed

Mar 12, 2026

Commits (30d)

21

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/NVIDIA/torch-harmonics"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.