eclosure

eclosure provides Python bindings and research utilities for e-Closure based multiple testing procedures. The package wraps a compiled C++ backend for the core closed e-BH and closed BY routines and ships example datasets used in the paper.

• Documentation: https://neilzxu.github.io/eclosure/
• Source: https://github.com/neilzxu/eclosure
• Paper: https://arxiv.org/abs/2509.02517

Installation

Install the core package from PyPI:

python -m pip install eclosure

Optional extras:

# experiment and real-data helpers
python -m pip install "eclosure[experiments]"

# plotting helpers used by the experiment code
python -m pip install "eclosure[plotting]"

Supported releases ship binary wheels for common CPython platforms. If you build from source, you will need a working C++17 toolchain.

Quick start

import eclosure

eclosure.closedeBH([1.0, 100.0], alpha=0.05)
eclosure.closedBY([0.01, 0.5], alpha=0.05)

The top-level API focuses on the stable testing procedures. The experiments workflow is available through the packaged module:

python -m eclosure.experiments --mode real --alpha 0.1 --out-dir figures

Learn more

The full package documentation includes:

• API reference for closedBY, closedeBH, and packaged data helpers
• Script documentation for python -m eclosure.experiments
• Testing guidance for unit, real-data, and timing suites
• Notes on repository-only exploratory scripts that are not part of the package

See https://neilzxu.github.io/eclosure/ for the full docs.

Citation

@article{xu2025bringing,
  title={Bringing Closure to False Discovery Rate Control: A General Principle for Multiple Testing},
  author={Xu, Ziyu and Solari, Aldo and Fischer, Lasse and de Heide, Rianne and Ramdas, Aaditya and Goeman, Jelle},
  journal={arXiv preprint arXiv:2509.02517},
  year={2025}
}