Installation

EGTtools is distributed via PyPI (pip wheels) and conda (Anaconda.org). Prebuilt packages are available for Linux (x86_64), macOS (x86_64 and arm64), and Windows (x86_64 and arm64), for Python 3.10 – 3.13.

Platform	Architectures	Python versions	OpenMP
Linux	x86_64	3.10 – 3.13	✅
macOS	x86_64, arm64 (M1–M4)	3.10 – 3.13	✅
Windows	x86_64, arm64	3.10 – 3.13	❌

Install with conda or mamba (recommended)

The conda package bundles all native dependencies (BLAS/LAPACK, OpenMP) and works out of the box on Linux and macOS. It is hosted on the maintainer’s personal Anaconda.org channel.

miniforge, mambaforge, and miniconda are all free for any use, including large organisations.

conda install -c socrats egttools

Or with mamba (faster dependency solver):

mamba install -c socrats egttools

To avoid typing -c socrats every time, add the channel permanently to your conda configuration:

conda config --add channels socrats
conda config --set channel_priority strict

After that, conda install egttools works without a channel flag.

Note

EGTtools is not in the conda-forge or defaults channels — it is maintained on the author’s personal Anaconda.org channel so the author retains full ownership of the recipe and release schedule.

Install with pip

pip install egttools

To upgrade to the latest release:

pip install -U egttools

Note

On macOS inside a conda environment, installing via pip can trigger ABI mismatches between pip-provided and conda-provided NumPy/SciPy. Using the conda package above avoids this entirely. If you must use pip, install the heavy dependencies via conda first:

conda install numpy scipy matplotlib networkx seaborn plotly
pip install egttools --no-deps

Build from source

Requirements

• Linux, macOS (Monterey or later), or Windows

• CMake ≥ 3.27

• A C++17-capable compiler (GCC ≥ 9, Clang ≥ 10, MSVC 2019+)

• Eigen ≥ 3.4 (or 5.x)

• Boost ≥ 1.82 (header-only multiprecision)

• Python ≥ 3.10

The easiest way to satisfy the C++ dependencies is via vcpkg (bundled as a git submodule) or via conda.

With vcpkg (default for local development)

git clone --recurse-submodules https://github.com/Socrats/EGTTools.git
cd EGTTools
pip install .

With conda dependencies (SKIP_VCPKG=ON)

# Create and activate an environment with all build and runtime deps
conda create -n egtenv python=3.11 numpy scipy matplotlib networkx \
    seaborn plotly eigen boost-cpp libblas liblapack pybind11 \
    scikit-build cmake ninja
conda activate egtenv

git clone --recurse-submodules https://github.com/Socrats/EGTTools.git
cd EGTTools
SKIP_VCPKG=ON pip install . --no-build-isolation

Development mode

pip install -e .  # editable install; rebuilds the extension on demand

Python distributions

conda / mamba

Use conda install -c socrats egttools as described above. For Apple Silicon (arm64), the conda package is the recommended installation method since it links against native arm64 BLAS/LAPACK and libomp.

PyPy

Recent versions of PyPy are supported by the pybind11 project and should thus also be supported by EGTtools.

Other

For any other Python distribution, pip install egttools should work. Please open a GitHub issue if you encounter problems.

Troubleshooting

If you run into problems, please create a GitHub issue or write to elias.fernandez.domingos@ulb.be.

Outdated pip

If installation hangs or errors immediately, update pip first:

pip install --upgrade pip

ImportError after install

If import egttools fails with an ImportError, the most common cause on macOS is a Python/NumPy ABI mismatch between pip and conda packages. Use the conda package or the --no-deps + conda-deps approach above.