healpix-resample#
Reproject unstructured lon/lat data onto a HEALPix grid — fast, sparse, GPU-ready.
healpix-resample takes a cloud of geographic measurements (each with a longitude, a latitude, and a value) and maps them onto a uniform spherical grid called HEALPix. It is designed for large datasets and runs efficiently on both CPU and GPU via PyTorch sparse operators.
lon/lat points (N) → sparse operator M → HEALPix cells (K)
HEALPix is a standard equal-area pixelization of the sphere used in astrophysics, climatology and oceanography. The resolution is controlled by a level parameter: nside = 2**level, so level=10 gives ~12 million cells.
Where to go next#
Hands-on notebooks to get started quickly.
In-depth explanation of each resampler and key parameters.
Complete documentation for every public class and function.
How to install the package in your environment.