Power spectrum emulators from neural networks and tree-based methods

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 ∈ [ 0.015 , 1.8 ] h / M p 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 N L e q , at redshifts 0, 0.503, 0.733, 0.997 for k ∈ [ 0.015 , 1.8 ] h / M p 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.