Skip to content

Installation

Requirements

  • Python: >= 3.11, < 3.15
  • C++ Compiler (source builds only): C++14 compatible (gcc, clang, MSVC)
  • Eigen (source builds only): Linear algebra library (version 3.x)
  • Matplotlib (optional): install via pip install vbpca_py[plot]

Pre-built wheels are available for Python 3.11–3.14 on Linux, macOS, and Windows:

pip install vbpca-py

With plotting support:

pip install vbpca-py[plot]

From source

Building from source requires Eigen and a C++14-compatible compiler.

Install Eigen

sudo apt-get install libeigen3-dev

brew install eigen

conda install -c conda-forge eigen

Download from eigen.tuxfamily.org and set: 

export EIGEN_INCLUDE_DIR=/path/to/eigen3

Eigen is located automatically via EIGEN_INCLUDE_DIR, $CONDA_PREFIX/include/eigen3, /opt/homebrew/include/eigen3, /usr/include/eigen3, or /usr/local/include/eigen3.

Build and install

git clone https://github.com/yoavram-lab/VBPCApy.git
cd VBPCApy
pip install .

Optional extras

# Development tools (pytest, ruff, mypy, just)
pip install .[dev]

# Plotting utilities (matplotlib)
pip install .[plot]

# Optional data utilities (pandas)
pip install .[data]

# Analysis dependencies (matplotlib, pandas, scikit-learn)
pip install .[analysis]

# Benchmark + plotting stack (joblib, pandas, scikit-learn, seaborn)
pip install .[benchmark]

# Documentation (mkdocs-material, mkdocstrings)
pip install .[docs]

# Install everything
pip install .[dev,plot,data,benchmark,docs]

Using uv

uv is the recommended environment manager for development:

# Core developer environment
uv sync --extra dev --extra data

# Include benchmark + plotting
uv sync --extra dev --extra data --extra benchmark

# Include Octave Python bridge (optional)
uv sync --extra dev --extra data --extra benchmark --extra octave