astro-rubix 0.0.4

A toolkit for simulating and analyzing integral field spectroscopic data cubes of astronomical sources.

Welcome to RUBIX

RUBIX is a versatile Integral Field Unit (IFU) tool designed for astrophysical simulations. It transforms any particle based galaxy model (e.g. cosmological hydrodynamical simulation outputs) into realistic mock IFU cubes, enabling both forward and inverse modeling. Built on JAX, RUBIX leverages GPU acceleration and automatic differentiation, allowing users to perform gradient-based optimization for inverse modeling alongside traditional forward modeling.

Key features include:

• Mock IFU Cube Generation: Convert simulation data into realistic IFU cubes.
• GPU-Accelerated Computations: Built on JAX for high-performance GPU support.
• Gradient-Based Inverse Modeling: Utilize gradients for efficient inverse modeling techniques.
• Flexible and Extensible: Designed to easily integrate with existing pipelines and astrophysical analysis tools.

Installation

Rubix is on PyPi. You can install it via

pip install astro-rubix

If you need GPU acceleration, you can install it via

pip install astro-rubix[cuda]

Alternatively, the Python package rubix is published on GitHub and can be installed alongside its runtime dependencies (including JAX) by choosing the relevant extras. For a CPU-only environment, install with:

git clone https://github.com/AstroAI-Lab/rubix.git
cd rubix
pip install .

If you need GPU acceleration, please add the optional dependence with [cuda] (or install jax[cuda] following the JAX instructions before installing Rubix).

Development installation

If you want to contribute to the development of rubix, we recommend the following editable installation from this repository:

git clone https://github.com/AstroAI-Lab/rubix.git
cd rubix
python -m pip install --editable .[tests,dev]

Having done so, the test suite can be run using pytest:

python -m pytest

This project depends on jax. For the pytests we only test the cpu version. For installation instructions with gpu support, please refer to here or simply use the cuda option when pip installing.

Configuration overview

Rubix ships with two YAML files in rubix/config/: rubix_config.yml (constants, SSP templates, dust recipes, handler mappings, etc.) and pipeline_config.yml (pipeline graphs such as calc_ifu and calc_dusty_ifu). There is no configuration wizard — your runtime settings must supply a dictionary with the following blocks:

• pipeline.name: Identifies the pipeline from pipeline_config.yml (e.g., calc_ifu, calc_dusty_ifu, or calc_gradient).
• galaxy: Must provide dist_z and a rotation section (type or explicit alpha, beta, gamma).
• telescope: Requires name, psf (currently only the gaussian kernel with size and sigma), lsf (sigma), and noise (signal_to_noise plus noise_distribution, choose from normal or uniform).
• ssp.dust: Must declare extinction_model and Rv before calling the dusty pipeline (see rubix/spectra/dust/extinction_models.py for the supported models such as Cardelli89).
• data.args.particle_type: Should include "stars" (and "gas" if you want the gas branch) so the filters and rotation functions know which components exist.

The tutorials and notebooks assume square spaxels, so the default telescope factory currently only supports pixel_type: square. For a working example, inspect notebooks/rubix_pipeline_single_function_shard_map.ipynb, which runs the exact pipeline used in the tests.

Documentation

Sphinx Documentation of all the functions is currently available under this link.

Contribution

Contributions to rubix are welcome and greatly appreciated! Whether you're fixing bugs, improving documentation, or suggesting new features, your help is valuable to us.

Please see here for contribution guidelines.

Thank you for helping improve rubix!

Citation

Please cite both of the following papers (Cakir et al. 2024, Schaible et al. 2025) if you use Rubix in your research:

  @ARTICLE{2024arXiv241208265C,
       author = {{{\c{C}}ak{\i}r}, Ufuk and {Schaible}, Anna Lena and {Buck}, Tobias},
        title = "{Fast GPU-Powered and Auto-Differentiable Forward Modeling of IFU Data Cubes}",
      journal = {arXiv e-prints},
     keywords = {Astrophysics - Instrumentation and Methods for Astrophysics, Astrophysics - Astrophysics of Galaxies, Physics - Computational Physics, Physics - Data Analysis, Statistics and Probability},
         year = 2024,
        month = dec,
          eid = {arXiv:2412.08265},
        pages = {arXiv:2412.08265},
          doi = {10.48550/arXiv.2412.08265},
        archivePrefix = {arXiv},
       eprint = {2412.08265},
        primaryClass = {astro-ph.IM},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2024arXiv241208265C},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
  }

  @ARTICLE{2025arXiv251117110S,
       author = {{Schaible}, Anna Lena and {{\c{C}}ak{\i}r}, Ufuk and {Buck}, Tobias and {Mack}, Harald and {Obreja}, Aura and {Oguz}, Nihat and {Oliver}, William H. and {C{\u{a}}r{\u{a}}mizaru}, Horea-Alexandru},
        title = "{RUBIX: Differentiable forward modelling of galaxy spectral data cubes for gradient-based parameter estimation}",
      journal = {arXiv e-prints},
     keywords = {Astrophysics of Galaxies},
         year = 2025,
        month = nov,
          eid = {arXiv:2511.17110},
        pages = {arXiv:2511.17110},
          doi = {10.48550/arXiv.2511.17110},
        archivePrefix = {arXiv},
       eprint = {2511.17110},
      primaryClass = {astro-ph.GA},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2025arXiv251117110S},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
  }

Contributors

Ufuk Çakır 💻 🖋 🔣 📖 🎨 💡 🤔 🚇 🚧 🔌 📆 💬 🔬 👀 🔧 ⚠️ 📢 📓

anschaible 💻 🖋 🔣 📖 🎨 💡 🤔 🚇 🚧 🔌 📆 💬 🔬 👀 🔧 ⚠️ 📢 📓

Tobias Buck 💻 🖋 🔣 📖 🎨 💡 🤔 🚇 🚧 🔌 📆 💬 🔬 👀 🔧 ⚠️ 📢 📓 🧑‍🏫

Robin Janssen 💻

nihatog 💻 📖 💡 🔬 👀 ⚠️ 📓

Aura Obreja 💬 👀 🔧 📓

Harald Mack 💻 📖 🎨 🚇 💬 👀 🔧 ⚠️ 📓

Horea Caramizaru 💻 🎨 🚇 💬

Will 💬 📓

Licence

MIT License

Acknowledgments

This repository was set up using the SSC Cookiecutter for Python Packages.