geomfum 1.0.0

First version of Geomfum!

This is the first stable version of geomfum library!
Differences with previous versions:

• removed the dependencies with geostats
• implemented automatic creation of docs
• improved docstrings

geomfum 0.1.1

GeomFuM v0.1.1 — First Release

We are happy to announce the first official release of GeomFuM, a Python library for geometry processing and machine learning with functional maps. This release lays the foundation for modular, reproducible, and extensible pipelines in spectral geometry, shape matching, and geometric deep learning.

New features

• Core geometry operators: Normals, Tangent Frames, Laplace–Beltrami operator, Gradient.
• Spectral descriptors: Heat Kernel Signature (HKS), Wave Kernel Signature (WKS).
• Matching framework: classical functional map optimization and refinement (ICP, ZoomOut, Fast Sinkhorn Filters, Adjoint Bijective ZoomOut, Neural ZoomOut).
• Learning pipelines: integration of DiffusionNet, PointNet, point-based transformers, FMNet, and RobustFMNet with differentiable PyTorch support.
• Support for NumPy and PyTorch backends.
• Example notebooks and documentation to get started quickly.

Dependencies and compatibility

• Requires Python ≥ 3.9.
• Compatible with both NumPy and PyTorch ecosystems.

Licensed under MIT.

Notes
This release sets the stage for further development. Future versions will expand support for graph-based data, statistical analysis tools, and additional geometric deep learning pipelines.
As always, thanks to all contributors and collaborators who made this first release possible!