Abstract
How did large-scale structures form, starting from a perfectly homogeneous initial state of the Universe, with only fluctuations of relative amplitude 10-5at z = 1 000 (the last scattering surface of the cosmic ray background) ? It can be shown that, in an expanding universe, a density fluctuation can only grow by its own gravity in proportion to the characteristic size of the Universe, i.e. as 1/(1 + z). So, starting at z = 1 000, fluctuations today could only reach an amplitude of 10-2, i.e. still not non-linear. If we want to explain the existence of galaxies today, we absolutely must have exotic dark matter, which does not interact with photons. This matter can then collapse under the effect of its self-gravity well before z = 1 000, and form potential wells, or " dark galaxies ", into which baryons from z = 1 000 will fall to form stars. The power spectrum of these structures can be calculated, and corresponds well to an initial scale-independent spectrum (expected in inflation theories, at the origin of primordial fluctuations). The observed spectrum comes from a modification, a reversal (or " tilt ") due to the fact that certain scales could not develop, being smaller than the horizon, in the period dominated by radiation. Photons prevent these structures from collapsing, as they travel at speed c. When radiation is diluted by expansion, and matter dominates, these structures can develop. The development of these structures can be calculated analytically, as long as they are not highly non-linear (Press-Schechter formalism). Cosmological simulations are then necessary. These are very successful on large scales, as long as baryon physics does not play a major role. On the galaxy scale, however, agreement with observations is not good, and feedback phenomena due to star formation and active nuclei must be taken into account to account for the size of galaxies (always too large in simulations).
