mmgpy 0.9.0

Python bindings for the MMG software

mmgpy

mmgpy brings the power of MMG mesh adaptation to Python. Generate, optimize, and refine 2D, 3D, and surface meshes with a clean API.

import mmgpy

mesh = mmgpy.read("input.vtk")
mesh.remesh(hmax=0.1)
mesh.save("output.vtk")

Try It

No installation needed — run directly with uvx:

# Remesh a mesh file
uvx mmgpy input.stl -o output.mesh -hmax 0.1

# Launch the interactive UI
uvx --from "mmgpy[ui]" mmgpy-ui

Installation

The recommended way to install mmgpy:

uv pip install mmgpy

This uses pre-built wheels from PyPI that bundle all native libraries (MMG, VTK) — no compiler needed.

Other install methods

# pip
pip install mmgpy

# conda-forge
conda install -c conda-forge mmgpy

# With UI support
uv pip install "mmgpy[ui]"

Using uv for project management

uv add mmgpy                 # add to project dependencies
uv tool install mmgpy        # install CLI tools globally
uv tool install "mmgpy[ui]"  # install CLI tools + UI globally

PyPI vs conda-forge

	PyPI (pip/uv)	conda-forge (Linux/macOS)
Install speed	Fast (pre-built wheels)	Slower (solver + download)
Dependencies	Bundled (self-contained)	Shared across packages
Disk usage	Larger (duplicate VTK/libs)	Smaller in conda environments
Lagrangian motion	No (ELAS library not bundled)	Yes (includes iscd-linearelasticity on Linux/macOS)
Best for	Quick setup, isolated use, CI	Scientific stacks sharing VTK/NumPy

Use PyPI (uv pip install) for the fastest setup. Use conda-forge when you already have a conda environment with VTK, PyVista, or other scientific packages — it also includes the ELAS library (via iscd-linearelasticity) for Lagrangian motion on Linux and macOS.

Features

• Multi-dimensional — 2D triangular, 3D tetrahedral, and surface meshes
• Local refinement — Control mesh density with spheres, boxes, cylinders
• Anisotropic adaptation — Metric tensors for directional refinement
• Level-set discretization — Extract isosurfaces from implicit functions
• Lagrangian motion — Remesh while tracking displacement fields
• PyVista integration — Visualize and convert meshes seamlessly
• 40+ file formats — VTK, STL, OBJ, GMSH, and more

Usage

Basic Remeshing

import mmgpy

mesh = mmgpy.read("input.mesh")
result = mesh.remesh(hmax=0.1)

print(f"Quality: {result.quality_mean_before:.2f} → {result.quality_mean_after:.2f}")
mesh.save("output.vtk")

Local Sizing

mesh = mmgpy.read("input.mesh")

# Fine mesh near a point
mesh.set_size_sphere(center=[0.5, 0.5, 0.5], radius=0.2, size=0.01)

# Fine mesh in a region
mesh.set_size_box(bounds=[[0, 0, 0], [0.3, 0.3, 0.3]], size=0.02)

mesh.remesh(hmax=0.1)

Typed Options

from mmgpy import Mmg3DOptions

opts = Mmg3DOptions(hmin=0.01, hmax=0.1, hausd=0.001)
mesh.remesh(opts)

# Or use presets
mesh.remesh(Mmg3DOptions.fine())

Visualization

mesh.plot()  # Quick plot with edges

# Or for custom plotting:
import pyvista as pv
plotter = pv.Plotter()
plotter.add_mesh(mesh.vtk, show_edges=True, color="lightblue")
plotter.show()

Command Line

MMG executables are included and available after installation:

# Auto-detect mesh type
mmg input.mesh -o output.mesh -hmax 0.1

# Or use specific commands
mmg3d input.mesh -o output.mesh -hmax 0.1
mmgs surface.stl -o refined.mesh -hausd 0.001
mmg2d domain.mesh -o refined.mesh -hmax 0.05

# Check versions
mmg --version

The _O3 suffix variants (mmg3d_O3, etc.) are also available for compatibility.

Gallery

Documentation

kmarchais.github.io/mmgpy

• Quick Start
• Tutorials
• API Reference
• Examples

Contributing

Contributions are welcome! See CONTRIBUTING.md for development setup, coding standards, and the pull request process.

License

MIT