A Python package to build, manipulate and analyze polygonal meshes.
Project description
PolyMesh - A Python Library for Compound Meshes with Jagged Topologies
Warning PolyMesh is in the early stages of it's lifetime, and some concepts may change in the future. If you want long-term stability, wait until version 1.0, which is planned to be released if the core concepts all seem to sit and the documentation covers all major concepts.
The PolyMesh library aims to provide the tools to build and analyse meshes with complex topologies. Meshes can be built like a dictionary, using arbitarily nested layouts and then be translated to VTK or PyVista. For plotting, there is also support for K3D, Matplotlib and Plotly.
The data model is built around Awkward, which makes it possible to attach nested, variable-sized data to the points or the cells in a mesh, also providing interfaces to other popular libraries like Pandas or PyArrow. Implementations are fast as implementations rely on the vector math capabilities of NumPy, while other computationally sensitive calculations are JIT-compiled using Numba.
Here and there we also use NetworkX, SciPy, SymPy and scikit-learn.
Motivating examples
Mesh assembly
One of the strongest sides of the library is mesh management. This example assembles a mesh of four separate bunnies using all kinds of transformations, each with their own separate pointcloud.
from polymesh import PolyData
from polymesh.examples import download_bunny_coarse
import numpy as np
import pyvista as pv
mesh = PolyData()
mesh["bunny_1"] = download_bunny_coarse(tetra=False, read=True)
mesh["bunny_2"] = (
mesh["bunny_1"]
.spin("Space", [0, 0, np.pi/2], "XYZ", inplace=False)
.move([0.2, 0, 0])
)
mesh["bunny_3"] = (
mesh["bunny_2"]
.spin("Space", [0, 0, np.pi/2], "XYZ", inplace=False)
.move([0.2, 0, 0])
)
mesh["bunny_4"] = (
download_bunny_coarse(tetra=True, read=True)
.rotate("Space", [0, 0, 3*np.pi/2], "XYZ")
.move([0.6, 0, 0])
)
The following call centralizes the pointcloudes and revires the topologies.
mesh.to_standard_form()
Handling of jagged topologies
PolyMesh is able to handle the topologies of mixed meshes and return them as Awkward or NumPy arrays. In the previous example, one of the bunnies is a tetrahedral mesh, the others are surface triangulations.
mesh.topology()
[[0, 1, 2],
[2, 1, 3],
[4, 5, 6],
[7, 5, 4],
[4, 8, 7],
[9, 10, 11],
[12, 13, 14],
[14, 15, 16],
[15, 14, 13],
[17, 13, 12],
...,
[1182, 1163, 1715, 2285],
[1162, 1163, 1182, 2285],
[1121, 1983, 1641, 1074],
[2189, 1232, 1613, 1585],
[1642, 1121, 1667, 2286],
[2122, 1945, 2175, 2103],
[1739, 1925, 1742, 1740],
[1191, 1748, 1749, 2287],
[1191, 1202, 1748, 2287]]
--------------------------
type: 6769 * var * int32
type(mesh.topology())
polymesh.topoarray.TopologyArray
Similarly to NumPy arrays, a TopologyArray
instance has a shape property which generalizes for jagged topologies nad coincides with NumPy for regular ones.
mesh.topology().shape
(6769, array([3, 3, 3, ..., 4, 4, 4], dtype=int64))
Calling to_array
on a TopologyArray
either returns an Awkward or a NumPy array.
mesh.topology().to_array()
Visualization
PolyMesh provides a mechanism to easily configure the blocks of a mesh to be plotted using PyVista:
mesh["bunny_1"].config["plot"] = dict(color="red", opacity=0.9)
mesh["bunny_2"].config["plot"] = dict(color="green", opacity=0.9)
mesh["bunny_3"].config["plot"] = dict(color="blue", opacity=0.9)
mesh["bunny_4"].config["plot"] = dict(color="yellow", opacity=0.9)
plotter = mesh.plot(
notebook=True,
config_key=["plot"],
return_plotter=True,
theme=pv.themes.DarkTheme(),
show_edges=False,
lighting=True
)
plotter.camera.tight(padding=0.1, view="xz", negative=True)
plotter.show(jupyter_backend="static")
Values can be assigned to the cells
for cb in mesh.cellblocks():
n = len(cb.topology())
cb.celldata["scalars"] = np.random.rand(n)
mesh["bunny_1"].config["plot"]["opacity"] = 1.0
mesh["bunny_2"].config["plot"]["opacity"] = 1.0
mesh["bunny_3"].config["plot"]["opacity"] = 1.0
mesh["bunny_4"].config["plot"]["opacity"] = 1.0
plotter = mesh.plot(
notebook=True,
config_key=["plot"],
return_plotter=True,
theme=pv.themes.DarkTheme(),
show_edges=False,
lighting=True,
scalars="scalars",
show_scalar_bar = False
)
plotter.camera.tight(padding=0.1, view="xz", negative=True)
plotter.show(jupyter_backend="static")
and to the points
n = len(mesh.coords())
scalars = np.random.rand(n)
mesh.pd.db["scalars"] = scalars
plotter = mesh.plot(
notebook=True,
config_key=["plot"],
return_plotter=True,
theme=pv.themes.DarkTheme(),
show_edges=False,
lighting=True,
scalars="scalars",
show_scalar_bar = False
)
plotter.camera.tight(padding=0.1, view="xz", negative=True)
plotter.show(jupyter_backend="static")
Passing data between points and cells
Values defined on the cells can also be aggregated to the nodes, creating a smoothing mechanism:
plotter = mesh.plot(
notebook=True,
config_key=["plot"],
return_plotter=True,
theme=pv.themes.DarkTheme(),
show_edges=False,
lighting=True,
scalars=mesh.pd.pull("scalars"),
show_scalar_bar = False
)
plotter.camera.tight(padding=0.1, view="xz", negative=True)
plotter.show(jupyter_backend="static")
Import and export
The heart of the database of a mesh is the combination of nested dictionaries equipped with Awkward records. Thanks to that, the data of a mesh can be easily converted to and from various data formats.
from polymesh import PointData
mesh.pointdata.to_parquet("bunny.parquet")
mesh.pointdata = PointData.from_parquet("bunny.parquet")
Documentation
The documentation is hosted on ReadTheDocs, where you can find more examples.
Installation
PolyMesh can be installed from PyPI using pip
on Python >= 3.7:
>>> pip install polymesh
Testing
>>> python -m unittest
License
This package is licensed under the MIT license.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for polymesh-0.0.21-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 396b7b6333963a6d8bcc0ec7448ebbb51e4632def787e668c2dfd48b9d4968b0 |
|
MD5 | 74b45fc67b54f17f63bfb4e4dcd6d542 |
|
BLAKE2b-256 | 0adf5b856bcd03d4caedc2a2eb5860b1e8eac857fe1e105f652c56acb575e29c |