We present results of large N-body-hydrodynamic simulations of galaxy forma
tion. Our simulations follow the formation of galaxies in cubic volumes of
side 100 Mpc, in two versions of the cold dark matter (CDM) cosmogony: the
standard, Omega = 1 SCDM model and the flat, Omega = 0.3 Lambda CDM model.
Over 2000 galaxies form in each of these simulations. We examine the rate a
t which gas cools and condenses into dark matter haloes. This roughly track
s the cosmic star formation rate inferred from observations at various reds
hifts. Galaxies in the simulations form gradually over time in the hierarch
ical fashion characteristic of the CDM cosmogony. In the Lambda CDM model,
substantial galaxies first appear at z similar or equal to 5 and the popula
tion builds up rapidly until z = 1 after which the rate of galaxy formation
declines as cold gas is consumed and the cooling time of hot gas increases
. In the SCDM simulation, the evolution is qualitatively similar, but is sh
ifted towards lower redshift. In both cosmologies, the present-day K-band l
uminosity function of the simulated galaxies resembles the observations. Th
e galaxy autocorrelation functions differ significantly from those of the d
ark matter. At the present epoch there is little bias in either model betwe
en galaxies and dark matter on large scales, but a significant anti-bias on
scales of similar to1 h(-1) Mpc and a positive bias on scales of similar t
o 100 h(-1) kpc is seen. The galaxy correlation function evolves little wit
h redshift in the range z = 0-3, and depends on the luminosity of the galax
y sample. The projected pairwise velocity dispersion of the galaxies is muc
h lower than that of the dark matter on scales less than 2 h(-1) Mpc. Apply
ing a virial mass estimator to the largest galaxy clusters recovers the clu
ster virial masses in an unbiased way. Although our simulations are affecte
d by numerical limitations, they illustrate the power of this approach for
studying the formation of the galaxy population.