Deep learning for cosmological parameter inference from a dark matter halo density field

Abstract

We propose a lightweight deep convolutional neural network (lCNN) to estimate cosmological parameters from simulated three-dimensional dark matter (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 $Λ$CDM model, simulations vary standard six cosmological parameters including $Ω_m$, $Ω_b$, $h$, $n_s$, $σ_8$, $w$, along with the neutrino mass sum, $M_ν$. 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 $Ω_m$, $h$, and $σ_8$ by the neural network model, while being inefficient in predicting $Ω_b$, { $n_s$}, $M_ν$ and $w$; 4) { compared to the simple fully connected network trained with three statistical quantities, our CNN yields statistically reduced errors, showing improvements of approximately 23\% for $Ω_m$, 11\% for $h$, 8\% for $n_s$, and 21\% for $σ_8$. Additionally, in comparison with the likelihood-based analysis on $P(k)$ data, our CNN provides much tighter constraints on parameters, especially on $Ω_m$ and $σ_8$.} Our study emphasizes this lCNN-based novel approach in extracting large-scale structure information and estimating cosmological parameters.