Halo abundance matching: accuracy and conditions for numerical convergence

Accurate predictions of the abundance and clustering of dark matter haloes play a key role in testing the standard cosmological model. Here, we investigate the accuracy of one of the leading methods of connecting the simulated dark matter haloes with observed galaxies -- the Halo Abundance Matching (HAM) technique. We show how to choose the optimal values of the mass and force resolution in large-volume N-body simulations so that they provide accurate estimates for correlation functions and circular velocities for haloes and their subhaloes -- crucial ingredients of the HAM method. At the 10% accuracy, results converge for 50 particles for haloes and 150 particles for progenitors subhaloes. In order to achieve this level of accuracy a number of conditions should be satisfied. The force resolution for the smallest resolved (sub)haloes should be in the range (0.1-0.3)rs, where rs is the scale radius of (sub)haloes. The number of particles for progenitors of subhaloes should be 150. We also demonstrate that the two-body scattering plays a minor role for the accuracy of N-body simulations thanks to the relatively small number of crossing-times of dark matter in haloes, and the limited force resolution of cosmological simulations.