A Python Geometry Processing Toolbox

https://gpytoolbox.org

Authors: Silvia Sellán, Columbia University and Oded Stein, University of Southern California

This is a very young library of general geometry processing Python research utility functions that evolves from our personal student codebases. It is compatible with Python 3.10, 3.11, 3.12 and 3.13.

Installation

Latest stable release (recommended)

You should be able install the latest release of Gpytoolbox with pip:

python -m pip install gpytoolbox

From Git

If you want to build Gpytoolbox from a specific git commit; for example, because you want to develop for Gpytoolbox or because you want some functionality that is in the main branch but hasn't been pushed to any release yet, you should be able to do so by cloning Gpytoolbox's github repo and running

python -m pip install numpy
python -m pip install .

Documentation

You can find documentation for all our functions in our website. You can also view the documentation for a specific function by running help(function_name) or function_name.__doc__; for example,

>>> from gpytoolbox import grad
>>> help(grad)
Finite element gradient matrix

Given a triangle mesh or a polyline, computes the finite element gradient matrix assuming piecewise linear hat function basis.

Parameters
----------
V : numpy double array
    Matrix of vertex coordinates
F : numpy int array, optional (default None)
    Matrix of triangle indices

Returns
-------
G : scipy sparse.csr_matrix
    Sparse FEM gradient matrix

See Also
--------
cotangent_laplacian.

Notes
-----

Contribute

We hope you find our current version of our library useful. At the same time, we encourage you to ask not what Gpytoolbox can do for you, but what you can do for Gpytoolbox.

Since Gpytoolbox is a very young library, we want to make it as easy as possible for others to contribute to it and help it grow. You can contribute by adding a new function in a new file inside src/gpytoolbox/, or by adding to existing functions, and submitting a Pull Request.

We also want to make the contribution process as unintimidating as possible. We will gladly review and edit your code to make sure it acommodates to our standards and we have set up many tests that will let us know if your contribution accidentally breaks anything. If there's any functionality that is not already in this library, is remotely related to geometry processing, and you have used or used in any of your past projects, we encourage you to submit it as-is in a Pull Request. We will gladly credit you in the individual function as well as on this home page.

Note that the code that you contribute will be licensed under the MIT license. Everybody will be able to use this code as long as they credit gpytoolbox (and not you individually).

License

Gpytoolbox's is released under an MIT license (see details), except for files in the gpytoolbox.copyleft module, which are under a GPL one (see details). Functions in the copyleft module must be imported explicitly; this way, if you import only the main Gpytoolbox module

import gpytoolbox

or individual functions from it,

from gpytoolbox import regular_square_mesh, regular_cube_mesh

you are only bound by the terms of the permissive MIT license. However, if you import any functionality from gpytoolbox.copyleft; e.g.,

from gpytoolbox.copyleft import mesh_boolean

you will be bound by the more restrictive GPL license.

Attribution

If you use our library in your research paper, please cite us! You can use the bibtex block below:

@misc{gpytoolbox,
  title = {{gptyoolbox}: A Python Geometry Processing Toolbox},
  author = {Silvia Sell\'{a}n and Oded Stein and others},
  note = {https://gpytoolbox.org/},
  year = {2023}
}

Acknowledgements

Several people have, knowingly or unknowingly, greatly contributed to this library. We are thankful to them:

• Alec Jacobson is the author of the original Matlab gptoolbox on which we inspired ourselves to create this library. Several of our functions are line-by-line translations of his Matlab ones. Thanks, Alec!

• Nicholas Sharp, the author of the game-changing geometry visualization library polyscope, was extremely helpful in guiding us through setting up and distributing a Python package. Thanks, Nick!

Contributors

• We would like to thank Michael Jäger for being Gpytoolbox's first external contributor (PR #45, PR #69).
• Towaki Takikawa (PR #49)
• Otman Benchekroun (PR #59)
• Abhishek Madan (PR #103)
• Lukas Hermann (PR #135)
• Haoyang Wu (PR #142)
• Dylan Rowe (PR #144)
• Alice Wei (PR #157)
• Xiana Carrera (PR #184)
• Carmen Rubio García (PR #188)
• Jeongseok Lee (PR #185)

Basic Optimistic Roadmap

Here are some things we think would be nice to incorporate to future versions of gpytoolbox. If there's one you are missing, feel free to submit a PR adding your item to this bullet list. If you want to contribute to gpytoolbox, a great way to start is by picking any of the items below that does not have an associated PR yet

To-do

• Iterative closest point for mesh alignment
• Basic FEM (cotangent matrix, mass matrix, linear elasticity) for tetrahedral meshes
• ARAP for deformation and parametrization
• Exact geodesic distances
• Heat (approximate) geodesic distance
• Blue noise in random mesh sampling
• Intrinsic triangulation routines
• Fracture mode computation
• Pure-python version of in_element_aabb
• Make all grid sizes, resolutions, etc. into tuples not necessarily numpy arrays
• Add notes on every docstring mentioning libigl implementations
• Tetrahedral mesh implementation of subdivide.py
• Dihedral angle computation
• regular_square_mesh and regular_cube_mesh should support different resolutions in x and y direction (sensible default when n_y is None, to n_y=n_x)