Dingo (Deep Inference for Gravitational-wave Observations) is a Python program for analyzing gravitational wave data using neural posterior estimation. It dramatically speeds up inference of astrophysical source parameters from data measured at gravitational-wave observatories. Dingo aims to enable the routine use of the most advanced theoretical models in analyzing data, to make rapid predictions for multi-messenger counterparts, and to do so in the context of sensitive detectors with high event rates.
The basic approach of Dingo is to train a neural network to represent the Bayesian posterior conditioned on data. This enables amortized inference: when new data are observed, they can be plugged in and results obtained in a small amount of time. Tasks handled by Dingo include
- building training datasets;
- training normalizing flows to estimate the posterior density;
- performing inference on real or simulated data; and
- verifying and correcting model results using importance sampling.
To install using pip, run the following within a suitable virtual environment:
pip install dingo-gwThis installs Dingo and its runtime dependencies, as specified in
pyproject.toml.
Optional functionality can be enabled via extras, for example:
pip install "dingo-gw[wandb,pyseobnr]"Dingo is also available from the conda-forge repository.
To install using conda, first activate a conda environment, then run:
conda install -c conda-forge dingo-gwIf you would like to make changes to Dingo or contribute to its development, install it from source.
First clone the repository:
git clone git@github.com:dingo-gw/dingo.git
cd dingo-gwWe recommend using uv for development installs, as it
provides fast, reproducible dependency resolution.
Create a virtual environment and install all development dependencies:
uv syncThis installs Dingo in editable mode along with development, documentation, and typing dependencies. To also install optional dependencies, use
uv sync --extra wandb --extra pyseobnrIf you prefer pip, create and activate a virtual environment:
python3 -m venv dingo-venv
source dingo-venv/bin/activateInstall Dingo in editable mode with development tools:
pip install -e ".[dev]"Optional user-facing features can be enabled via extras, for example:
pip install -e ".[dev,wandb]"For instructions on using Dingo, please refer to the documentation.
There is a CLAUDE.md
file in this repository with basic instructions for coding agents. All agents
read multiple files for instructions that have different scopes. It is
recommended to leave the provided CLAUDE.md as is and put additional
instructions into the user-specific files ~/.claude/CLAUDE.md (systemwide) and
CLAUDE.local.md (repo specific) which should then be added to the .gitignore
file. If you use other agents, please follow the instructions for
Codex or
Antigravity and again put
additional files in the .gitignore.
Dingo is based on the following series of papers:
- https://arxiv.org/abs/2002.07656: 5D toy model
- https://arxiv.org/abs/2008.03312: 15D binary black hole inference
- https://arxiv.org/abs/2106.12594: Amortized inference and group-equivariant neural posterior estimation
- https://arxiv.org/abs/2111.13139: Group-equivariant neural posterior estimation
- https://arxiv.org/abs/2210.05686: Importance sampling
- https://arxiv.org/abs/2211.08801: Noise forecasting
- https://arxiv.org/abs/2407.09602: Binary neutron star inference
Dingo was used also in https://arxiv.org/abs/2404.14286 to find evidence for eccentric binaries.
If you use Dingo in your work, we ask that you please cite at least https://arxiv.org/abs/2106.12594.
Contributors to the code are listed in AUTHORS.md. We thank Charlie Hoy, Vivien Raymond, and Rory Smith for acting as LIGO-Virgo-KAGRA (LVK) review chairs. Dingo makes use of many LVK software tools, including Bilby, bilby_pipe, and LALSimulation, as well as third party tools such as PyTorch and nflows.
For questions or comments please contact Maximilian Dax or Stephen Green.