Galaxy formation in the Planck cosmology - III. The high-redshift universe

We present high-redshift predictions of the star formation rate distribution function (SFRDF), UV luminosity function (UVLF), galactic stellar mass function (GSMF), and specific star formation rates (sSFRs) of galaxies from the latest version of the Munich semi-analytic model L-GALAXIES. We find a good fit to both the shape and normalization of the SFRDF at z = 4-7, apart from a slight underprediction at the low-SFR end at z = 4. Likewise, we find a good fit to the faint number counts for the observed UVLF at brighter magnitudes our predictions lie below the observations, increasingly so at higher redshifts. At all redshifts and magnitudes, the raw (unattenuated) number counts for the UVLF lie above the observations. Because of the good agreement with the SFR we interpret our underprediction as an overestimate of the amount of dust in the model for the brightest galaxies, especially at high redshift. While the shape of our GSMF matches that of the observations, we lie between (conflicting) observations at z = 4-5, and underpredict at z = 6-7. The sSFRs of our model galaxies show the observed trend of increasing normalization with redshift, but do not reproduce the observed mass dependence. Overall, we conclude that the latest version of L-GALAXIES, which is tuned to match observations at z <= 3, does a fair job of reproducing the observed properties of galaxies at z >= 4. More work needs to be done on understanding observational bias at high redshift, and upon the dust model, before strong conclusions can be drawn on how to interpret remaining discrepancies between the model and observations.