Repository showing how to use Neural Networks (NN) to test beyond ΛCDM models against Cosmic Microwave Background (CMB) data, at the level of the angular power spectra. All the analysis is found within the corresponding jupyter notebooks.
The repository is structured as follows. Within the neural_networks folder, you can find the folders:
modified_gravityfeature
Each folder contain the corresponding jupyter notebooks
- Required software:
python - Dependencies:
numpy,matplotlib,tensorflow,shap - It is recommended to run the notebooks to train the Neural Networks in a cluster with GPUs
- The data to train the Neural Networks are found at these Zenodo repositories: modified gravity and primordial feature.
# Example to get it running
pip install numpy matplotlib tensorflow
git clone https://github.com/IndiraOcampo/CMB_ML_based_model_selection.git
cd neural_networksIf you are using the content provided in this repository to do your own analysis, please cite this repository and the manuscript:
@misc{CMB_ML_based_model_selection,
author = {Ocampo, I},
title = {CMB ML based model selection},
year = {2024},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/IndiraOcampo/CMB_ML_based_model_selection.git}},
}@article{Ocampo_2025,
doi = {10.1088/1475-7516/2025/02/004},
url = {https://dx.doi.org/10.1088/1475-7516/2025/02/004},
year = {2025},
month = {feb},
publisher = {IOP Publishing},
volume = {2025},
number = {02},
pages = {004},
author = {Ocampo, I. and Cañas-Herrera, G. and Nesseris, S.},
title = {Neural Networks for cosmological model selection and feature importance using Cosmic Microwave Background data},
journal = {Journal of Cosmology and Astroparticle Physics}.
}
This project is licensed under the MIT License - see the LICENSE file for details.
This reaseach acknowledges support from the ESA Archival Research Visitor Programme: a programme to increase the scientific return from ESA space science missions by supporting scientists interested in pursuing research based on publicly available data in the ESA Space Science Archives. This work is also supported by the fellowship LCF/BQ/DI22/11940033 from “la Caixa” Foundation (ID 100010434) and acknowledges the use of the Finis Terrae III Supercomputer which was financed by the Ministry of Science and Innovation, Xunta de Galicia and ERDF (European Regional Development Fund). This research uses ESA Planck Legacy archives.