Quantifying and Controlling Biases in Estimates of Dark Matter Halo Concentration

Abstract

ABSTRACT: We use bootstrapping to estimate the bias of concentration estimates on N-body dark matter halos as a function of particle number. We find that algorithms based on the maximum radial velocity and radial particle binning tend to overestimate the concentration by 15% − 20% for halos sampled with 200 particles and by 7%-10% for halos sampled with 500 particles. To control this bias at low particle numbers we propose a new algorithm that estimates halo concentrations based on the integrated mass profile. The method uses the full particle information without any binning, making it reliable in cases when low numerical resolution becomes a limitation for other methods. This method reduces the bias to < 3% for halos sampled with 200-500 particles. The velocity and density methods have to use halos with at least ∼ 4000 particles in order to keep the biases down to the same low level. We also show that the mass-concentration relationship could be shallower than expected once the biases of the different concentration measurements are taken into account. These results show that bootstrapping and the concentration estimates based on the integrated mass profile are valuable tools to probe the internal structure of dark matter halos in numerical simulations.