PyART — Python Analytical Relativity Toolkit

PyART (Python Analytical Relativity Toolkit) provides a unified interface to access and work with numerical relativity (NR) waveform catalogs for compact-object binary mergers. Different NR groups maintain their own data formats and access conventions, which can make comparative or downstream analyses cumbersome. PyART addresses this by offering a lightweight abstraction layer over several major public catalogs.

In addition to catalog access, PyART includes basic tools for waveform comparison—such as time-domain alignment and frequency-domain mismatch calculations—for both single-mode and multi-mode data (and few more).

Supported Catalogs

  • Simulating eXtreme Spacetimes (SXS)

  • Maya

  • CoRe (Computational Relativity)

  • RIT

  • GR-Athena

  • ICCUB NR

  • SACRA

  • RWZ

  • and others

Warning

PyART is under active development. The codebase is mostly well-behaved, but some of its dependencies are temperamental, and things may break without warning. Think of it as a promising postdoc — brilliant, but occasionally unpredictable.

PyART is currently available only from source. Clone the repository and install the base library with:

git clone https://github.com/RoxGamba/PyART.git
cd PyART
pip install .

This covers the core Waveform interface, analysis tools (matched filtering, eccentricity, scattering angles, …), and the PN/BHPT analytic expressions. Optional/soft dependencies (e.g. EOBRun_module, lalsimulation) are imported inside try/except, so a missing one only disables the specific model or catalog that needs it, without breaking the rest of the package.

On top of the base install, three extras add optional functionality:

Extra

Install with

What it gives you

catalogs

pip install ".[catalogs]"

Downloading/scraping NR waveform catalogs — SXS (sxs, romspline) and RIT (requests, beautifulsoup4). Needed to run the test suite, which exercises SXS/RIT downloads.

models

pip install ".[models]"

Optional analytical/semi-analytical waveform models — TEOBResumS (teobresums), IMRPhenomX/X_AS (phenomxpy), and SEOBNR (pyseobnr).

docs

pip install ".[docs]"

Building the Sphinx documentation locally (make -C docs html) — Sphinx, MyST, furo, sphinx-autoapi, and seaborn for the executed tutorial notebooks. Not needed to install or use the library.

Extras can be combined, e.g.:

pip install ".[catalogs,models]"

Note: PyART depends on pycbc, which currently has two incompatibilities with numpy 2.x. After installing, apply the one-time patch:

python scripts/patch_pycbc_numpy2.py

This patches pycbc’s events/threshold_cpu.py and filter/matchedfilter.py in-place. It is idempotent and safe to re-run after pycbc upgrades.

We promise to make available on PyPI in near future.

Contents: