COVMOS: A new Monte Carlo approach for galaxy clustering analysis

Abstract

We validate the COVMOS method introduced in our previous study allowing for the fast simulation of catalogues of different cosmological field tracers (e.g. dark matter particles, halos, galaxies, etc.). The power spectrum and one-point probability distribution function of the underlying tracer’s density field are set as inputs of the method and arbitrarily chosen by the user. To evaluate the validity domain of COVMOS at the level of the produced two-point statistics covariance matrix, we chose to target these two input statistical quantities from realistic N-body simulation outputs. We performed this cloning procedure based on the ΛCDM as well as on a massive neutrino cosmology, for five redshifts in the range of z ∈ [0, 2]. First, we validated the output real-space two-point statistics (both in the configuration and Fourier space) and estimated over 5000 COVMOS realisations per redshift and per cosmology, with a volume of 1 [Gpc/h]3 and 108 particles each. This validation was performed against the corresponding N-body measurements, estimated from 50 simulations. We found the method to be valid up to k ∼ 0.2h Mpc−1 for the power spectrum and down to r ∼ 20 Mpc h−1 for the correlation function. Then, we extended the method by proposing a new modelling of the peculiar velocity distribution, aimed at reproducing the redshift-space distortions both in the linear and mildly non-linear regimes. After validating this prescription, we finally compared and validated the resulting redshift-space two-point statistics covariance matrices in the same range of scales. We released the Python code associated with this method on a public repository, which allows for the production of tens of thousands of realisations in record time. In particular, COVMOS is intended for any study involving large-scale galaxy-survey science that requires a large number of mock realisations.