Numerical studies of gas dynamics in clusters of galaxies

Modern developments in hydrodynamics, parallel processing, and multi-grid solvers are combined into a new hydrodynamic code for cosmology. It features a time-variable moving grid which adjusts itself to a curl-free flow. In such a frame, the mass per grid cell is kept approximately constant, which allows us to address problems with a large dynamic range in length scales when objects collapse due to gravitational instability. The code simulates dark matter on the curvilinear grid using the particle-mesh (PM) algorithm, and gas using a total variation diminishing (TVD) scheme. Newtonian gravity is solved using a multigrid iterator. The code has been optimized for execution on a symmetric multi-processor machine. We present a series of extensive tests comparing the new code to previously developed algorithms, and discuss various aspects of the error analysis.; Using this code we study the formation and evolution of clusters of galaxies. We first address the statistical properties of clusters. We confirm the fact that the cold dark matter model (CDM) produces too many bright clusters and over-estimates their temperatures for {dollar}rm Xsb8sbsp{lcub}sim{rcub}{lcub}>{rcub}{dollar} 0.5. Models with a cosmological constant {dollar}Lambda{dollar} = 0.65 or hyperbolic models with a density parameter {dollar}Omegasb0{dollar} = 0.35 appear consistent with the current observational constraints. We show that the evolution of the temperature-luminosity relation can in principle distinguish different cosmological scenarios.; We then study the intrinsic properties of individual galaxies. The gas biasing relative to dark matter in the cores of galaxies is measured. We find the gas less dense than the dark matter be a factor of two relative to their global ratio in the innermost 200 kpc. Nevertheless these two phases trace each other very closely exterior to approximately 1 Mpc. The degree of virialization and the presence of substructure is quantified, and we find that while there is some dependence on the cosmological parameters, the scatter within each cosmology is quite large. A low density universe will produce almost as many clusters with significant substructure as a critical density universe, thus it would require a large observational sample to distinguish between models.