We present N-body simulations of galaxy groups embedded in a common halo of matter. We study the influence of the different initial conditions upon the evolution of the group and show that denser configurations evolve faster, as expected. We then concentrate on the influence of the initial radial density profile of the common halo and of the galaxy distribution. We select two kinds of density distributions, a singular profile (modeled by a Hernquist distribution) and a profile with a flat core (modeled by a Plummer sphere). In all cases we witness the formation of a central massive object due to mergings of individual galaxies and to accretion of stripped material, but both its formation history and its properties depend heavily on the initial distribution. In Hernquist models the formation is due to a ``burst'' of mergings in the inner parts, due to the large initial concentration of galaxies in the center. The merging rate is much slower in the initial phases of the evolution of a Plummer distribution, where the contribution of accretion to the formation of the central object is much more important. The central objects formed within Plummer distributions have projected density profiles which are not in agreement with the radial profiles of observed brightest cluster members, unless the percentage of mass in the common halo is small. On the contrary the central object formed in initially cusped models has projected radial profiles in very good agreement with those of brightest cluster members, sometimes also showing luminosity excess over the $r^{1/4}$ law in the outer parts, as is observed in cD galaxies.

The projected directions of the rotation axes of interacting binary disk galaxies tend to align orthogonal to each other. Sofue (1992) has suggested that this could be due to shorter merger times for galaxies with paralel spins. We show by means of N-body simulations that this suggestion is correct.

We present N-body simulations of groups of galaxies with a number of very different initial conditions. These include spherical isotropic, nonspherical anisotropic collapses and virialised spherical systems. In all cases but one the merging instability leads to the formation of a giant central galaxy in the center of the group. The initial conditions of the exception are such that no galaxies are present in the central part of the group. Thus some central seed of material is necessary to trigger the formation of a giant central galaxy. We concentrate on the properties of these giant central galaxies. Spherical virialised systems give rise to relatively round and isotropic systems, while aspherical initial conditions give rise to triaxial objects with anisotropic velocity dispersion tensors. In the latter cases the orientation of the resulting central galaxy is well correlated with that of the initial cluster. We compare the projected properties of the objects formed with the properties of real brightest cluster member galaxies. The surface density profiles are in good agreement with the observed surface brightness profiles. In the case of extended virialised groups the projected properties of the giant central galaxy are the same as the properties of cD galaxies. These include a halo of luminous material and a nearly flat velocity dispersion profile.

In this paper we test the performance of explicit simulations of groups of galaxies, (i.e. simulations in which each galaxy is treated as a mass point and the physics of the interactions is modelled by specific analand the physics of the interactions is modelled by specific analytical prescriptions for merging conditions), by comparing them with fully self-consistent simulations starting from identical initial conditions. The quality of the explicit simulations is very unequal. For somThe quality of the explicit simulations is very unequal. For some prescriptions the results are complete disagreement with the self-consistent simulations. The inclusion of other dynamical effects like dynamical friction gives, in some cases, better agreeement.like dynamical friction gives, in some cases, better agreeement. We also propose a new merging criterion, which, combined with dynamical friction, gives much better agreement with self-consistent simulations in a variety of initial conditions, but even this criterion has a limited ranof initial conditions, but even this criterion has a limited range of applicability.