Robustness of Cosmological Simulations. I. Large‐Scale Structure

The gravitationally-driven evolution of cold dark matter dominates theformation of structure in the Universe over a wide range of length scales.While the longest scales can be treated by perturbation theory, a fullyquantitative understanding of nonlinear effects requires the application oflarge-scale particle simulation methods. Additionally, precision predictionsfor next-generation observations, such as weak gravitational lensing, can onlybe obtained from numerical simulations. In this paper, we compare results fromseveral N-body codes using test problems and a diverse set of diagnostics,focusing on a medium resolution regime appropriate for studying manyobservationally relevant aspects of structure formation. Our conclusions arethat -- despite the use of different algorithms and error-control methodologies-- overall, the codes yield consistent results. The agreement over a wide rangeof scales for the cosmological tests is test-dependent. In the best cases, itis at the 5% level or better, however, for other cases it can be significantlylarger than 10%. These include the halo mass function at low masses and themass power spectrum at small scales. While there exist explanations for most ofthe discrepancies, our results point to the need for significant improvement inN-body errors and their understanding to match the precision of near-futureobservations. The simulation results, including halo catalogs, and initialconditions used, are publicly available.