We present a comparison of 14 galaxy formation models: 12 different semi-analytical models and 2 halo-occupation distribution models for galaxy formation based upon the same cosmological simulation and merger tree information derived from it. The par
ticipating codes have proven to be very successful in their own right but they have all been calibrated independently using various observational data sets, stellar models, and merger trees. In this paper we apply them without recalibration and this leads to a wide variety of predictions for the stellar mass function, specific star formation rates, stellar-to- halo mass ratios, and the abundance of orphan galaxies. The scatter is much larger than seen in previous comparison studies primarily because the codes have been used outside of their native environment within which they are well tested and calibrated. The purpose of the `nIFTy comparison of galaxy formation models is to bring together as many different galaxy formation modellers as possible and to investigate a common approach to model calibration. This paper provides a unified description for all participating models and presents the initial, uncalibrated comparison as a baseline for our future studies where we will develop a common calibration framework and address the extent to which that reduces the scatter in the model predictions seen here.
We adapt the L-Galaxies semi-analytic model to follow the star-formation histories (SFH) of galaxies -- by which we mean a record of the formation time and metallicities of the stars that are present in each galaxy at a given time. We use these to co
nstruct stellar spectra in post-processing, which offers large efficiency savings and allows user-defined spectral bands and dust models to be applied to data stored in the Millennium data repository. We contrast model SFHs from the Millennium Simulation with observed ones from the VESPA algorithm as applied to the SDSS-7 catalogue. The overall agreement is good, with both simulated and SDSS galaxies showing a steeper SFH with increased stellar mass. The SFHs of blue and red galaxies, however, show poor agreement between data and simulations, which may indicate that the termination of star formation is too abrupt in the models. The mean star-formation rate (SFR) of model galaxies is well-defined and is accurately modelled by a double power law at all redshifts: SFR proportional to $1/(x^{-1.39}+x^{1.33})$, where $x=(t_a-t)/3.0,$Gyr, $t$ is the age of the stars and $t_a$ is the loopback time to the onset of galaxy formation; above a redshift of unity, this is well approximated by a gamma function: SFR proportional to $x^{1.5}e^{-x}$, where $x=(t_a-t)/2.0,$Gyr. Individual galaxies, however, show a wide dispersion about this mean. When split by mass, the SFR peaks earlier for high-mass galaxies than for lower-mass ones, and we interpret this downsizing as a mass-dependence in the evolution of the quenched fraction: the SFHs of star-forming galaxies show only a weak mass dependence.
We introduce a new physical recipe into the De Lucia and Blaizot version of the Munich semi-analytic model built upon the Millennium dark matter simulation: the tidal stripping of stellar material from satellite galaxies during mergers. To test the s
ignificance of the new physical process we apply a Monte Carlo Markov Chain parameter estimation technique constraining the model with the $K$-band luminosity function, $B-V$ colours and the black hole-bulge mass relation. The differences in parameter correlations, and in the allowed regions in likelihood space, reveal the impact of the new physics on the basic ingredients of the model, such as the star-formation laws, feedback recipes and the black hole growth model. With satellite disruption in place, we get a model likelihood four times higher than in the original model, indicating that the new process seems to be favoured by observations. This is achieved mainly due to a reduction in black hole growth that produces a better agreement between the properties of central black holes and host galaxies. Compared to the best-fit model without disruption, the new model removes the excess of dwarf galaxies in the original recipe with a more modest supernova heating. The new model is now consistent with the three observational data sets used to constrain it, while significantly improving the agreement with observations for the distribution of metals in stars. Moreover, the model now follows the build up of intra-cluster light.
