Deep Learning for Cosmological Parameter Inference from Dark Matter Halo Density Field

Deep Learning for Cosmological Parameter Inference from Dark Matter Halo Density Field
Zhiwei Min, Xu Xiao, Jiacheng Ding, Liang Xiao, Jie Jiang, Donglin Wu, Qiufan Lin, Yin Li, Yang Wang, Shuai Liu, Zhixin Chen, Xiangru Li, Jinqu Zhang, Le Zhang, Xiao-Dong Li
arXiv:2404.09483v1 Announce Type: new
Abstract: We propose a lightweight deep convolutional neural network (lCNN) to estimate cosmological parameters from simulated three-dimensional DM halo distributions and associated statistics. The training dataset comprises 2000 realizations of a cubic box with a side length of 1000 $h^{-1}{rm Mpc}$, and interpolated over a cubic grid of $300^3$ voxels, with each simulation produced using $512^3$ DM particles and $512^3$ neutrinos . Under the flat $Lambda$CDM model, simulations vary standard six cosmological parameters including $Omega_m$, $Omega_b$, $h$, $n_s$, $sigma_8$, $w$, along with the neutrino mass sum, $M_nu$. We find that: 1) within the framework of lCNN, extracting large-scale structure information is more efficient from the halo density field compared to relying on the statistical quantities including the power spectrum, the two-point correlation function, and the coefficients from wavelet scattering transform; 2) combining the halo density field with its Fourier transformed counterpart enhances predictions, while augmenting the training dataset with measured statistics further improves performance; 3) achieving high accuracy in inferring $Omega_m$, $h$, $n_s$, and $sigma_8$ by the neural network model, while being inefficient in predicting $Omega_b$,$M_nu$ and $w$; 4) compared to the simple random forest network trained with three statistical quantities, lCNN yields unbiased estimations with reduced statistical errors: approximately 33.3% for $Omega_m$, 20.0% for $h$, 8.3% for $n_s$, and 40.0% for $sigma_8$. Our study emphasizes this lCNN-based novel approach in extracting large-scale structure information and estimating cosmological parameters.arXiv:2404.09483v1 Announce Type: new
Abstract: We propose a lightweight deep convolutional neural network (lCNN) to estimate cosmological parameters from simulated three-dimensional DM halo distributions and associated statistics. The training dataset comprises 2000 realizations of a cubic box with a side length of 1000 $h^{-1}{rm Mpc}$, and interpolated over a cubic grid of $300^3$ voxels, with each simulation produced using $512^3$ DM particles and $512^3$ neutrinos . Under the flat $Lambda$CDM model, simulations vary standard six cosmological parameters including $Omega_m$, $Omega_b$, $h$, $n_s$, $sigma_8$, $w$, along with the neutrino mass sum, $M_nu$. We find that: 1) within the framework of lCNN, extracting large-scale structure information is more efficient from the halo density field compared to relying on the statistical quantities including the power spectrum, the two-point correlation function, and the coefficients from wavelet scattering transform; 2) combining the halo density field with its Fourier transformed counterpart enhances predictions, while augmenting the training dataset with measured statistics further improves performance; 3) achieving high accuracy in inferring $Omega_m$, $h$, $n_s$, and $sigma_8$ by the neural network model, while being inefficient in predicting $Omega_b$,$M_nu$ and $w$; 4) compared to the simple random forest network trained with three statistical quantities, lCNN yields unbiased estimations with reduced statistical errors: approximately 33.3% for $Omega_m$, 20.0% for $h$, 8.3% for $n_s$, and 40.0% for $sigma_8$. Our study emphasizes this lCNN-based novel approach in extracting large-scale structure information and estimating cosmological parameters.