The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: a large sample of mock galaxy catalogues

TL;DR

This paper introduces a fast, two-point statistic–aware method to generate large numbers of galaxy mocks using 2LPT-based matter fields, FoF halos with a calibrated linking length, and a calibrated Halo Occupation Distribution to populate galaxies. It calibrates and validates the method against LasDamas N-body simulations and applies it to produce 600 CMASS DR9 mocks that reproduce the survey geometry, redshift distribution, and clustering signals, including monopoles, quadrupoles, and power spectra. The resulting covariance matrices agree with analytic predictions in the linear regime and reproduce key features of the data within ~10% across relevant scales, providing a practical and accurate tool for error estimation in large-scale structure analyses such as BOSS DR9 and future surveys. The approach significantly accelerates mock catalog production (roughly two orders of magnitude faster than full N-body mocks) while preserving essential non-Gaussian corrections needed for robust covariance estimation.

Abstract

We present a fast method of producing mock galaxy catalogues that can be used to compute covariance matrices of large-scale clustering measurements and test the methods of analysis. Our method populates a 2nd-order Lagrangian Perturbation Theory (2LPT) matter field, where we calibrate masses of dark matter halos by detailed comparisons with N-body simulations. We demonstrate the clustering of halos is recovered at ~10 per cent accuracy. We populate halos with mock galaxies using a Halo Occupation Distribution (HOD) prescription, which has been calibrated to reproduce the clustering measurements on scales between 30 and 80 Mpc/h. We compare the sample covariance matrix from our mocks with analytic estimates, and discuss differences. We have used this method to make catalogues corresponding to Data Release 9 of the Baryon Oscillation Spectroscopic Survey (BOSS),producing 600 mock catalogues of the "CMASS" galaxy sample. These mocks enabled detailed tests of methods and errors that formed an integral part of companion analyses of these galaxy data.

Figures (18)

• Figure 1: The Northern Galactic cap (NGC) and Sourthern Galactic cap (SGC) footprint of the CMASS DR9 galaxy sample
• Figure 2: Normalised redshift distribution of galaxies in the NGC (solid) and SGC (dashed) CMASS DR9 sample
• Figure 3: Ratio between PTHalos and N-body halo-matter cross-power spectra as a function of linking length, $b$, for the $10^6$ most massive halos. From top to bottom linking length are: 0.27, 0.30, 0.36,0.38 (in lighter color), and 0.40. N-body halos use $b=0.2$ with the corresponding mass threshold of $3.02\cdot 10^{13} M_{\odot}/h$
• Figure 4: Ratios between PTHalos and N-body halo-matter cross-power spectra as a function of halo mass threshold, for a linking length of b=0.38 (2LPT) and b=0.2 (N-body). The halo masses are given in Table \ref{['masstable']}.
• Figure 5: TOP: Ratio of the cross-power variance of PTHalos and N-body simulations for a mass threshold of $3.02\cdot10^{13}M_{\odot}/h$. The expected ratio is shown in a solid line and a 15 per cent band range is shown in dashed lines. MIDDLE and BOTTOM: Comparison of correlation coefficients of N-body (solid blue) and PTHalos (dashed red) halo-matter cross-power spectra. Middle panel $k_1=0.06$. Bottom $k_1=0.201$