Published June 12, 2025 | Version v3
Software Open

Jun-Young Lee/CosmoTopo: v.1.0.1

Authors/Creators

Description

Purpose

This software enables the generation of combinatorial complexes on cosmological datasets for the purpose of accurately predicting cosmological parameters. Topological Neural Networks (TNN), which enable higher-order message passing while preserving E(3)-invariance, are trained with full hyperparameter optimizations in multi-node, multi-GPU configurations.

 

Instructions

For overall instruction and guidance, please refer to the README.md file. For detailed information, please refer to the docstrings attached to individual files.

 

Version Update

This release adds docstrings to each of the files, which includes summaries, notes, and references for individual files. Code structure remains identical to v.1.0.0.

Files

CosmoTopo-1.0.1.zip

Files (3.7 MB)

Name Size Download all
md5:7cb6dcca6b6dd66d064bc8b8447fe062
3.7 MB Preview Download

Additional details

Related works

Is supplement to
Software: https://github.com/Byeol-Haneul/CosmoTopo/tree/v1.0.1 (URL)
Preprint: arXiv:2505.23904 (arXiv)

Dates

Updated
2025-06-12

Software

Repository URL
https://github.com/Byeol-Haneul/CosmoTopo
Programming language
Python , Shell