pytetwild is a Python library for mesh tetrahedralization. It is a
Python wrapper around the efficient C++ library for tetrahedral meshing provided by
fTetWild.
We have pre-built wheels for Python 3.10 - Python 3.14 for Windows, Linux, and macOS.
The recommended way to install pytetwild is via PyPI:
pip install pytetwild[all]This installs pyvista by default, which you can use to tetrahedralize
sufrace meshes from PyVista. Alternatively you can just install with pip
install pytetwild for a lighter install.
You can also clone the repository and install it from source, but since there's
C++ involved, the build is a bit more complicated. See CONTRIBUTING.md for
more details.
To tetrahedralize a surface mesh from PyVista,
you'll need to first install pyvista as it's not a dependency and then run:
import pyvista as pv
import pytetwild
# Load or create a PyVista PolyData surface mesh
# Here, we'll create a simple sphere mesh as an example
surface_mesh = pv.Icosphere(nsub=2)
# Convert the surface mesh to a tetrahedral mesh. For this example let's
# use a coarse mesh
tetrahedral_mesh = pytetwild.tetrahedralize_pv(surface_mesh, edge_length_fac=1)
# Visualize the tetrahedral mesh in an "exploded" view
tetrahedral_mesh.explode(0.5).plot(
show_edges=True, zoom=1.6, ssao=True, anti_aliasing="ssaa"
)
You can also work with raw arrays. Here's a simple cube that we turn into tetrahedra.
import numpy as np
import pytetwild
# Define vertices of the cube
vertices = np.array([
[0, 0, 0], # Vertex 0
[1, 0, 0], # Vertex 1
[1, 1, 0], # Vertex 2
[0, 1, 0], # Vertex 3
[0, 0, 1], # Vertex 4
[1, 0, 1], # Vertex 5
[1, 1, 1], # Vertex 6
[0, 1, 1] # Vertex 7
])
# Define faces using vertex indices
# Each face is a rectangle (also accepts triangles)
faces = np.array([
[0, 1, 2, 3], # Front face
[1, 5, 6, 2], # Right face
[5, 4, 7, 6], # Back face
[4, 0, 3, 7], # Left face
[4, 5, 1, 0], # Bottom face
[3, 2, 6, 7] # Top face
])
v_out, tetra = pytetwild.tetrahedralize(vertices, faces, optimize=False)We've surfaced a several parameters to each of our interfaces
tetrahedralize and tetrahedralize_pv:
Parameters
----------
edge_length_fac : float, default: 0.05
Tetrahedral edge length as a function of bounding box diagonal. The
default ideal edge length is ``bb/20`` (bounding box divided by
20). Ignored when ``edge_length_abs`` is input.
edge_length_abs : float, optional
Absolute ideal edge length. When input ``edge_length_fac`` is ignored.
optimize : bool, default: True
Improve the minimum scaled Jacobean for each cell. This leads to higher
cell quality at the expense of computation time. Optimization level is
dependent on ``stop_energy`` and ``num_opt_iter``.
simplify : bool, default: True
Simplfiy the input mesh surface before tetrahedralization.
epsilon : float, default 1e-3
Envelop size, specifying the maximum distance of the output surface
from the input surface, relative to the bounding box size.
stop_energy : float, default: 10.0
The mesh optimization stops when the conformal AMIPS energy reaches
``stop_energy``.
coarsen : bool, default: False
Coarsen the output as much as possible, while maintaining the mesh
quality.
num_threads : int, default: 0
Set number of threads used. 0 (default) uses all available cores.
num_opt_iter : int, default: 80
Maximum number of optimization iterations if ``optimize=True``.
loglevel : int, default: 6
Set log level (0 = most verbose, 6 = minimal output).
quiet : bool, default: False
Disable all output. Overrides ``loglevel``.
This project relies on fTetWild and credit goes to the original authors for
their efficient C++ library for tetrahedral meshing. That work is licensed
under the Mozilla Public License v2.0.
The work in this repository is also licensed under the Mozilla Public License v2.0.
If you are having issues, please feel free to raise an Issue.