abstract
- We report on work in progress by the Virgo consortium, a collaboration set up to carry out large simulations of the formation of galaxies and large-scale structure exploiting the latest generation of parallel supercomputers. We show results of $256^3$ particle N-body simulations of the clustering evolution of dark matter in four cold dark matter models with different cosmological parameters. The high resolution and large volume of these simulations allows us to determine reliably the mass autocorrelation function for pair separations in the range $40\hkpc$ to $20\hmpc$. Comparison of these with the observed galaxy correlation function shows that for any of these models to be viable, the distribution of galaxies must be biased relative to the distribution of mass in a non-trivial, scale-dependent fashion. In particular, low $\Omega_0$ models require the galaxies to be more ``weakly'' clustered than the mass at small and intermediate pair separations. Simulations which include the evolution of gas show that cold gas knots form with approximately the abundance expected on theoretical grounds, although a few excessively massive objects grow near the centres of rich clusters. The locations where these cold gas knots form are, in general, biased relative to the distribution of mass in a scale-dependent way. Some of these biases have the required sign but they are, for the most part, weaker than is necessary for agreement with observations. The antibias present in our low $\Omega_0$ N-body/SPH simulation appears to be related to the merging and disruption of galaxies in rich clusters.