Rapid Iterative FiTting for gravitational-wave parameter inference
RIFT estimates the properties of gravitational-wave sources by combining parallel likelihood evaluations, marginalization over extrinsic parameters, and iterative fitting of the intrinsic likelihood. It supports compact-binary analyses, multiple waveform models, GPU acceleration, and extensions for population and equation-of-state inference.
Documentation · Installation · Tutorials · Science & citations · Release notes
Warning
Planning a publication or production analysis? Read Science and citations before starting, including the production-use guidance and collaboration/authorship policy.
| Your goal | Branch | Release series |
|---|---|---|
| Stable analyses and maintenance updates | rift_O4c |
0.0.17.xx |
| New features and ongoing short-author research | rift_O4d |
0.0.18.xx development |
Use rift_O4c for the stable line. main and master are not the stable
branch recommendation. O4c is maintained separately, will not be a descendant
of master, and has no planned merge back into it. See the
branch maintenance policy for details.
Install LALSuite and the dependencies described in INSTALL.md, then install from the branch appropriate to your analysis. For the stable line:
git clone --branch rift_O4c https://github.com/oshaughn/research-projects-RIT.git
cd research-projects-RIT
python -m pip install -e .
export GW_SURROGATE=''For development, select rift_O4d when cloning. GPU execution requires a
compatible CuPy/CUDA environment; cluster jobs also require the appropriate
accounting configuration. See optional dependencies.
Start with the tutorials and pipeline documentation. The worked examples cover pipeline, hyperparameter, population, and simulation workflows.
| Task | Start here |
|---|---|
| Analyze an event | util_RIFT_pseudo_pipe.py — pipeline examples |
| Coordinate analyses with Asimov | RIFT / Asimov integration |
| Infer EOS or population hyperparameters | Hyperpipe examples |
| Manage simulation campaigns and archives | Simulation-manager design and backends |
| Explore applications and choose citations | Science and citations |
Features and examples vary between release branches; consult the documentation in your selected checkout.
The rift_O4d root pixi.toml provides local development environments:
pixi install
pixi run install-rift
pixi run import-checkThe default environment pins SWIG below 4.4.0. Compatibility environments are
available as swig-pre44 and swig-post44; run swig-version and import-check
in each when testing binding compatibility.
Read the architecture guide and API reference for the source layout and interfaces. Repository contributors should also follow AGENTS.md and .travis/PRECOMMIT.md for module design and validation requirements. For help, see troubleshooting or open an issue.
RIFT and rapid_pe share their origin in the parallel inference method of Pankow et al. (2015). The RIFT project is also hosted at rapidpe-rift/rift.
Distributed under the MIT license. Scientific references and the collaboration policy are collected in SCIENCE_AND_CITATIONS.md.