argosim is a lightweight, modular, and open-source Python package for simulating radio interferometric observations. It is developed as part of the ARGOS project, a European initiative to build a next-generation, sustainable radio interferometer.
- 📡 Antenna array simulation – Generate custom arrays or load predefined layouts.
- 🔭 Aperture synthesis – Compute baseline distributions and uv-sampling masks.
- 🌀 Sky model generation – Create or load input sky models.
- 🧼 Dirty beam and image reconstruction – Simulate dirty observations and reconstruct images using CLEAN.
- 📊 Beam and image quality metrics – Evaluate beam shape, sidelobe level, and image fidelity.
- ⚡ JAX backend – Accelerated, differentiable computations for high-performance and ML applications.
- 🐳 Docker support – Easy reproducibility and cross-platform compatibility.
The imaging forward model — baselines, UV-track generation, Kaiser–Bessel convolutional gridding, FFT — is implemented end-to-end in JAX, so the path from antenna positions all the way to the dirty image is differentiable. Combined with an optax optimiser, this turns array layout design into a gradient-based problem: you can directly minimise a loss on the dirty beam (or any downstream observable) with respect to the antenna coordinates.
The animation below shows 20 antennas being optimised with Adam to match a Gaussian target beam, under multiple physical constraints (minimum spacing, maximum site radius). The reusable building blocks live in argosim.optim_utils; see notebooks/layout_optimisation.ipynb for the full example.
You can install the latest version of argosim via pip:
pip install argosimOr use the development version directly from GitHub:
git clone https://github.com/ARGOS-telescope/argosim.git
cd argosim
pip install -e .- Python 3.11+
- NumPy, Matplotlib, JAX, scikit-image
- PyQt6 (optional, for running the GUI)
argosim provides an optional PyQt6-based GUI for interactive simulations. It supports array configuration, uv-coverage visualization, and imaging—all without writing code.
Ideal for education, outreach, and rapid prototyping.
Running the GUI requires PyQt6 which can be installed as follows:
pip install "argosim[gui]"To launch the GUI and have access to the custom templates you need to clone the argosim repository:
git clone https://github.com/ARGOS-telescope/argosim.git
cd argosimLaunch the GUI with:
python app/argosim-gui.pyComprehensive documentation is available at:
Four interactive Jupyter notebooks are provided to help you get started:
These notebooks cover the main functionalities of argosim and provide practical examples.
A pre-built Docker image is available:
docker pull ghcr.io/argos-telescope/argosim:mainThis ensures reproducibility and simplifies setup across different environments.
We use continuous integration to maintain code quality and robustness:
- Unit tests with
pytest - Linting with
blackandisort - Documentation building with
sphinx
If you use argosim in your research, please cite it as follows:
@software{argosim,
author = {Ezequiel Centofanti and Emma Ayçoberry and Samuel Farrens and Samuel Gullin and Manal Bensahli and Jean-Luc Starck and John Antoniadis},
title = {argosim: a Python package for radio interferometric simulations},
version = {2.0.0},
year = {2026},
url = {https://github.com/ARGOS-telescope/argosim}
}fill in the version number with the current version of argosim.
This software is developed at the CosmoStat laboratory, CEA, Paris-Saclay, in collaboration with the ARGOS project.
For questions, bugs, or feature requests, feel free to open an issue.