Power Spectrum Emulators from Neural Networks and Tree-Based Methods

Andrei Lazanu

doi:10.1016/j.dark.2025.102186

paper

Power Spectrum Emulators from Neural Networks and Tree-Based Methods

arXiv:2506.07514 · doi:10.1016/j.dark.2025.102186

Abstract

We use two subsets of 2000 and 1000 Quijote simulations to build two power spectrum emulators, allowing for fast computations of the non-linear matter power spectrum. The first emulator is built in terms of seven cosmological parameters: the matter and baryon fraction of the energy density of the Universe $Ω_{m}$ and $Ω_{b}$ , the reduced Hubble constant $h$ , the scalar spectral index $n_{s}$ , the amplitude of matter density fluctuations $σ_{8}$ , the total neutrino mass $M_{ν}$ and the dark energy equation of state parameter $w$ , on scales $k \in [0.015, 1.8] h / Mp c^{- 1}$ . The power spectra can be directly determined at redshifts 0, 0.5, 1, 2 and 3, while for intermediate redshifts these can be interpolated. The second emulator is based on five cosmological parameters, $Ω_{m}$ , $h$ , $n_{s}$ , $σ_{8}$ and the amplitude of equilateral non-Gaussianity $f_{NL}^{eq}$ , at redshifts 0, 0.503, 0.733, 0.997 for $k \in [0.015, 1.8] h / Mp c^{- 1}$ . The emulators are built on machine learning techniques. In both cases we have investigated both neural networks and tree-based methods and we have shown that the best accuracy is obtained for a neural network with two hidden layers. Both emulators achieve a root-mean-squared relative error of less then 5\% for all the redshifts considered on the scales discussed.

17 pages, 6 figures, accepted version

Power Spectrum Emulators from Neural Networks and Tree-Based Methods

Abstract

References in corpus (59)