spherinator 0.4.0

Rotation-Invariant Hyperspherical Variational Convolutional Autoencoder

Spherinator

Spherinator and HiPSter are tools that provide explorative access and visualization for multimodal data from extremely large astrophysical datasets, ranging from exascale cosmological simulations to multi-billion object observational galaxy surveys.

A variational autoencoder (VAE) will be trained using PyTorch Lightning to compress the structural data into a low-dimensional spherical latent space.

Installation

pip install spherinator

Documentation

Read The Docs

Acknowledgments

Funded by the European Union. This work has received funding from the European High-Performance Computing Joint Undertaking (JU) and Belgium, Czech Republic, France, Germany, Greece, Italy, Norway, and Spain under grant agreement No 101093441.

Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European High Performance Computing Joint Undertaking (JU) and Belgium, Czech Republic, France, Germany, Greece, Italy, Norway, and Spain. Neither the European Union nor the granting authority can be held responsible for them.

License

This project is licensed under the Apache-2.0 License.

Citation

If you use Spherinator in your research, we provide a citation to use:

@article{Polsterer_Spherinator_and_HiPSter_2024,
author = {Polsterer, Kai Lars and Doser, Bernd and Fehlner, Andreas and Trujillo-Gomez, Sebastian},
title = {{Spherinator and HiPSter: Representation Learning for Unbiased Knowledge Discovery from Simulations}},
url = {https://arxiv.org/abs/2406.03810},
doi = {10.48550/arXiv.2406.03810},
year = {2024}
}