Compact groups from the Millennium Simulations - I. Their nature and the completeness of the Hickson sample

Abstract

We identify compact groups of galaxies (CGs) within mock galaxy catalogues from the Millennium Simulation at z=0 with three semi-analytic models (SAMs) of galaxy formation. CGs are identified using the same 2D criteria as those visually applied by Hickson (1982) to his CGs (HCGs), but with a brightest galaxy magnitude limit, and the blending of close projected pairs. Half of the mock CGs identified in projection contain at least 4 accordant velocities (mvCGs), versus 70% for HCGs. In comparison to mvCGs, the HCGs are only 8% complete at distances < 9000 km/s, missing the CGs with small angular sizes, a strongly dominant galaxy, and (for one SAM) the CGs that are fainter and those with lower surface brightness. We explore different ways to determine the fraction of physically dense groups. Binding energy criteria turn out to be inapplicable given the segregation between galaxies and dark matter particles. We rely instead on the combination of the 3D length of the CGs (maximum real space galaxy separation) and their elongation along the line-of-sight (ratio of maximum line-of-sight to maximum projected separations), restricting ourselves in both cases to smallest quartets within the CGs. We find that between 64% and 80% (depending on the SAM) of the mvCGs have 3D lengths shorter than 200 kpc/h, between 71% and 80% have line-of-sight elongations less than 2, while between 59% and 76% have either 3D lengths shorter than 100 kpc/h or both lengths shorter than 200 kpc/h and elongations smaller than 2. Therefore, chance alignments (CAs) of galaxies concern at most 40% of the mvCGs. These CAs are mostly produced from larger host groups, but a few have galaxies extending a few Mpc beyond the host group. The mvCGs built with the Hickson selection with (without) the close projected pair blending criterion have 10% higher (lower) fractions of physically dense systems.