Introduction

In most cosmological scenarios, galaxies form early and rapidly in the collapse of density perturbations but continue to evolve over a Hubble time both in luminosity and morphology. The ageing of the galaxy stellar population leads to a gradual dimming of their light, a signal that is now picked up in distant spirals and elliptical galaxies (e.g. Vogt et al. 1996; Kelson et al. 1997) Dynamics primarily drives the galaxy morphological evolution through: 1) internal gravitational instabilities such as the bar instability which may influence bulge formation, 2) tidal interactions which can explain much of the ``disturbed'' and irregular morphology of galaxies at high z (e.g. Oemler et al. 1997) 3) the merging of disks to form some or most of the ellipticals (Toomre 1977). Merging obviously contributes to evolution of the galaxy luminosity function.

Galaxy clusters are the best place to investigate the connection between dynamics and morphology because they contain a diverse population of galaxies that have interacted strongly many times over their lifetime. Dressler (1984) has aptly described clusters as laboratories of galaxy formation, in particular in the way they emphasize the importance of gravitational interactions. Here, I explicitly follow that lead by setting up numerical experiments of galaxy interactions in cosmological clusters.

The simulation of galaxy dynamics in a cosmological context at sufficient resolution to resolve detailed dynamical effects is now becoming feasible. Kiloparsec scales can now be resolved in the volume surrounding a collapsing cluster. Most aspects of the dynamical evolution of galaxies do not require the complicated details of dissipative galaxy formation, so there is no need for expensive hydro calculations.

A simple technique that allows studies of galaxy dynamics in clusters works as follows. First, a cosmological N-body simulation in a large volume is run and a cluster size dark halo is identified at z=0. The simulation is re-examined at early times (z=3 to 2) and all dark halos that will end up in the cluster are replaced with N-body models of disk (and possibly elliptical) galaxies scaled according the mass and circular velocity of the halos with at least $10\times$ the resolution. The rationale is that galactic disks should form rapidly prior to cluster collapse and the bulk of their dynamical evolution will be driven by the interactions they experience when they fall into the collapsing cluster. Simulations are then continued with the resolved galaxy models to z=0.

So far I have applied this technique in 2 simulations: the first with a poor cluster ( $\sigma = 550$ km/s) containing 100 well-resolved disk galaxies inserted at z=2 (Dubinski 1998) and a Virgo-scale cluster ( $\sigma=800$ km/s) with 200 disks and 15 ellipticals inserted at z=3. The main results of these simulations are discussed below.