EdelweissFE: A light-weight, platform-independent, parallel finite element framework.

See the documentation.

EdelweissFE aims at an easy to understand, yet efficient implementation of the finite element method. Some features are:

• Python for non performance-critical routines
• Cython for performance-critical routines
• Parallelization
• Modular system, which is easy to extend
• Output to Paraview, Ensight, CSV, matplotlib
• Interfaces to powerful direct and iterative linear solvers

EdelweissFE makes use of the Marmot library for finite element and constitutive model formulations.

Installation

EdelweissFE runs on Linux, macOS (version 14 or newer) and Windows. Installing it takes four steps and about 15 minutes, most of it waiting for downloads.

Just want to run simulations? With conda installed (step 1), one command installs a released EdelweissFE including Marmot, no compiler needed:

conda create -n edelweissfe -c https://repo.prefix.dev/matthiasneuner/edelweiss -c conda-forge edelweissfe

The steps below set up a development installation instead: use them to change EdelweissFE or Marmot, add Marmot modules, or compile for your own CPU. Keep the two in separate environments.

1. Install conda

EdelweissFE and everything it needs (Python, numerical libraries, compilers) are installed with conda, a package manager. If you do not have conda yet, install Miniforge, a free distribution of conda:

• Linux or macOS: open a terminal and run

  curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
  bash Miniforge3-$(uname)-$(uname -m).sh
  

  then close and reopen the terminal.
• Windows: download and run the installer from conda-forge.org/download. Windows also needs Microsoft's C++ compiler, which conda cannot install: install the free Visual Studio 2022 Build Tools and select Desktop development with C++. Afterwards, run all commands below in the Miniforge Prompt (find it in the Start menu).

2. Create the EdelweissFE environment

An environment is a separate folder that contains everything EdelweissFE needs, in exactly the versions it is tested with, without touching anything else on your computer. Create it with:

conda create -n edelweissfe -c https://repo.prefix.dev/matthiasneuner/edelweiss -c conda-forge edelweissfe-dev=2026.11.09

Always use this separate environment for EdelweissFE. If it ever breaks, delete it (conda env remove -n edelweissfe) and run the command again.

3. Download and install EdelweissFE

conda activate edelweissfe
git clone https://github.com/Edelweiss-Numerics/EdelweissFE.git
cd EdelweissFE
pip install -e .

conda activate edelweissfe switches to the environment. You need it again in every new terminal before using EdelweissFE.

4. Check that it works

Run the test suite (it takes a few minutes and should end with Tests failed: 0):

• Linux or macOS:

  PYTHON_GIL=0 run_tests_edelweissfe ./testfiles/edelweiss-only/
  

• Windows:

  set PYTHON_GIL=0
  run_tests_edelweissfe .\testfiles\edelweiss-only\
  

PYTHON_GIL=0 lets EdelweissFE compute in parallel on several processor cores.

Running a simulation

conda activate edelweissfe
edelweissfe my_simulation.inp

Optional: Marmot

Marmot adds further elements and material models. Build it into the same environment, then reinstall EdelweissFE (on Windows, see the installation documentation for the commands):

conda activate edelweissfe
git clone --recurse-submodules https://github.com/MAteRialMOdelingToolbox/Marmot/ ../Marmot
cmake -S ../Marmot -B ../Marmot/build -DCMAKE_INSTALL_PREFIX=$CONDA_PREFIX -DCMAKE_PREFIX_PATH=$CONDA_PREFIX
cmake --build ../Marmot/build -j && cmake --install ../Marmot/build
pip install -e .
PYTHON_GIL=0 run_tests_edelweissfe ./testfiles/marmot/

More details, including how the environment is pinned, how to change dependencies and troubleshooting, are in the installation documentation.