2019ApJ...873..124W

We use the outputs of a semianalytical model of galaxy formation run on the Millennium Simulation to investigate the prevalence of 3D compact groups (CGs) of galaxies from z = 11 to 0. Our publicly available code identifies CGs using the 3D galaxy number density, the mass ratio of secondary+tertiary to the primary member, mass density in a surrounding shell, the relative velocities of candidate CG members, and a minimum CG membership of three. We adopt "default" values for the first three criteria, representing the observed population of Hickson CGs at z = 0. The percentage of nondwarf galaxies (M > 5 x 10⁸ h^–1 M_☉) in CGs peaks near z ∼ 2 for the default set and in the range of z ∼ 1-3 for other parameter sets. This percentage declines rapidly at higher redshifts (z >= 4), consistent with the galaxy population as a whole being dominated by low-mass galaxies excluded from this analysis. According to the most liberal criteria, <=3% of nondwarf galaxies are members of CGs at the redshift where the CG population peaks. Our default criteria result in a population of CGs at z < 0.03 with number densities and sizes consistent with Hickson CGs. Tracking identified CG galaxies and merger products to z = 0, we find that <=16% of nondwarf galaxies have been CG members at some point in their history. Intriguingly, the great majority (96%) of z = 2 CGs have merged to a single galaxy by z = 0. There is a discrepancy in the velocity dispersions of Millennium Simulation CGs compared to those in observed CGs, which remains unresolved.