healpix-resample 2026.5.1

Sparse GPU-enabled regridding from lon/lat to HEALPix

healpix-resample

healpix-resample is a lightweight Python package designed to regrid data defined on longitude--latitude coordinates onto a HEALPix grid.

The package provides GPU-accelerated operators (via PyTorch) to construct sparse linear mappings between input geodetic coordinates and a target HEALPix tessellation at a chosen resolution level.

This package manages the HEALPix authalic definition and the Earth ellipsoid using the WGS84 reference system.

Objectives

The main goals of the package are:

• Provide a generic regridding framework from (lon, lat) to HEALPix.
• Support different interpolation strategies:
  • Nearest-neighbor mapping
  • PSF / multi-point weighted interpolation
• Enable efficient handling of:
  • Large numbers of input points
  • Batched data (B, N)
  • CUDA acceleration
• Offer a reusable linear operator that can be:
  • Applied forward (data → HEALPix)
  • Used inside inverse problems or iterative solvers

Design Principles

• Modular architecture:
  • knn module: generic operator construction
  • nearest: nearest-neighbor specialization
  • psf: weighted multi-point interpolation
• Sparse matrix representation for scalability
• Torch-based implementation for CPU/GPU flexibility
• Resolution controlled via HEALPix level parameter

Installation (Private Repository)

This package is distributed as a private repository and must be installed from source.

---

Clone the repository

git clone https://github.com/EOPF-DGGS/healpix-resample.git
cd healpix-resample

If you use SSH access:

git clone [e-mail]:EOPF-DGGS/healpix-resample.git
cd healpix-resample

---

Install in editable mode (recommended for development)

pip install -e .

Editable mode allows you to modify the source code without reinstalling the package after each change.

---

Standard installation (optional)

If you do not need editable mode:

pip install .

---

Requirements

Make sure you are using:

• Python ≥ 3.8
• A working PyTorch installation (CPU or CUDA)
• numpy
• healpix-geo

Verifying the installation

After installation:

import healpix_resample
print(healpix_resample.__file__)

If no error occurs, the installation is successful.

Typical Use Case

from healpix_resample.nearest import NearestResampler

op = NearestResampler(lon_deg=lon, lat_deg=lat, level=level, device="cuda")
healpix_values = op.resample(values)

Target Applications

• Earth observation data remapping
• Oceanographic or atmospheric gridding
• Astronomical sky projections
• Large-scale geospatial data harmonization