GALAXY FORMATION AND PHYSICAL BIAS

We have supplemented our code, which computes the evolution of the physical state of a representative piece of the universe to include, not only the dynamics of dark matter (with a standard PM code), and the hydrodynamics of the gaseous component (including detailed collisional and radiative processes), but also galaxy formation on a heuristic but plausible basis. If, within a cell the gas is Jeans' unstable, collapsing, and cooling rapidly, it is transformed to galaxy subunits, which are then followed with a collisionless code. After grouping them into galaxies, we estimate the relative distributions of galaxies and dark matter (physical "bias") and the relative velocities of galaxies and dark matter ("velocity bias"). In a large scale CDM run of 80 h-1 Mpc size with 8 x 10(6) cells and dark matter particles, we find that physical bias b = (deltaN/N)gal/(deltaM/M)tot on the 8 h-1 Mpc scale is about 1.6 and increases towards smaller scales, and that velocity bias, b(upsilon) = (DELTAupsilon(rms, gal)/DELTAupsilon(rms, DM)) is about 0.8 on the same scale. The comparable HDM simulation is highly biased with b = 2.7 on the 8 h-1 Mpc scale. Implications of these results are discussed in the light of the COBE observations which provide an accurate normalization for the initial power spectrum. CDM can be ruled out on the basis of too large a predicted small scale velocity dispersion at greater than 95% confidence level; DELTAupsion(rms, 1D) = 715 +/- 135 km s-1 compared with observations DELTAupsilon(rms, 1D) = 340 +/- 40 km s-1 on the 1 h-1 Mpc scale.