English

Clusternets: A deep learning approach to probe clustering dark energy

Cosmology and Nongalactic Astrophysics 2025-02-03 v2 General Relativity and Quantum Cosmology

Abstract

Machine Learning (ML) algorithms are becoming popular in cosmology for extracting valuable information from cosmological data. In this paper, we evaluate the performance of a Convolutional Neural Network (CNN) trained on matter density snapshots to distinguish clustering Dark Energy (DE) from the cosmological constant scenario and to detect the speed of sound (csc_s) associated with clustering DE. We compare the CNN results with those from a Random Forest (RF) algorithm trained on power spectra. Varying the dark energy equation of state parameter wDEw_{\rm{DE}} within the range of -0.7 to -0.99, while keeping cs2=1c_s^2 = 1, we find that the CNN approach results in a significant improvement in accuracy over the RF algorithm. The improvement in classification accuracy can be as high as 40%40\% depending on the physical scales involved. We also investigate the ML algorithms' ability to detect the impact of the speed of sound by choosing cs2c_s^2 from the set {1,102,104,107}\{1, 10^{-2}, 10^{-4}, 10^{-7}\} while maintaining a constant wDEw_{\rm DE} for three different cases: wDE{0.7,0.8,0.9}w_{\rm DE} \in \{-0.7, -0.8, -0.9\}. Our results suggest that distinguishing between various values of cs2c_s^2 and the case where cs2=1c_s^2=1 is challenging, particularly at small scales and when wDE1w_{\rm{DE}}\approx -1. However, as we consider larger scales, the accuracy of cs2c_s^2 detection improves. Notably, the CNN algorithm consistently outperforms the RF algorithm, leading to an approximate 20%20\% enhancement in cs2c_s^2 detection accuracy in some cases.

Keywords

Cite

@article{arxiv.2308.03517,
  title  = {Clusternets: A deep learning approach to probe clustering dark energy},
  author = {Amirmohammad Chegeni and Farbod Hassani and Alireza Vafaei Sadr and Nima Khosravi and Martin Kunz},
  journal= {arXiv preprint arXiv:2308.03517},
  year   = {2025}
}

Comments

12 pages, 6 figures, 6 tables; data available at https://doi.org/10.5281/zenodo.8220732; version accepted to MNRAS

R2 v1 2026-06-28T11:49:47.581Z