Large- and small-scale constraints on power spectra in Omega = 1 universes

The cold dark matter (CDM) model of structure formation, normalized on large scales, leads to excessive pairwise velocity dispersions on small scales. In an attempt to circumvent this problem, we study three scenarios (all with $Omega=1$) which have more large-scale power and less small-scale power than the CDM model: 1) an admixture of cold and hot dark matter; 2) cold dark matter with a non-scale-invariant, power-law primordial power spectrum; and 3) cold dark matter with coupling of dark matter to a long-range vector field. Despite the {reduced} small-scale power, when such models are evolved in the nonlinear regime to large amplitude, the velocities on small scales are actually {it increased} over CDM with the same value of $sigma_8$. This `flip-over', in disagreement with the expectation from linear perturbation theory, arises from the nonlinear coupling of the extra power on large scales with shorter wavelengths. However, due to the extra large-scale power, the recent COBE DMR results indicate smaller amplitudes for these models, $sigma_8 sim 0.5 - 0.7$, than for CDM (for which $sigma_8 sim 1.2$). Therefore, when normalized to COBE on large scales, such models do lead to reduced velocities on small scales and they produce fewer halos compared with CDM. Quantitatively it seems, however, that models that produce sufficiently low small-scale velocities fail to produce an adequate distribution of halos.