galics– III. Properties of Lyman-break galaxies at a redshift of 3

Mon. Not. R. Astron. Soc. 352, 571–588 (2004) doi:10.1111/j.1365-2966.2004.07947.x GALICS – III. Properties of Lyman-break galaxies at a redshift of 3 Jérémy Blaizot,1,2 Bruno Guiderdoni,1 Julien E. G. Devriendt,2,4 François R. Bouchet,1 Steve J. Hatton1 and Felix Stoehr3 1 Institut d’Astrophysique de Paris, 98 bis boulevard Arago, 75014 Paris, France 2 Oxford University, Denys Wilkinson Building, Keble Road, Oxford OX1 3RH 3 Max-Planck-Institut für Astrophysik, Karl-Schwarzschild-Strasse 1, 85741 Garching, Germany 4 CRAL, Observatoire de Lyon, 69561 St Genis Laval cedex, France Accepted 2004 April 22. Received 2004 April 21; in original form 2003 October 2 ABSTRACT Key words: astronomical data bases: miscellaneous – galaxies: evolution – galaxies: formation – galaxies: high-redshift.  E-mail:  C 2004 RAS This paper illustrates how mock observational samples of high-redshift galaxies with sophisticated selection criteria can be extracted from the predictions of GALICS, a hybrid model of hierarchical galaxy formation that couples the outputs of large cosmological simulations and semi-analytic recipes, to describe dark matter collapse and the physics of baryons. As an example of this method, we focus on the properties of Lyman-break galaxies at redshift z ∼ 3 (hereafter LBGs) in a  cold dark matter (CDM) cosmology. With the MOMAF software package described in a companion paper, we generate a mock observational sample with selection criteria as similar as possible to those implied in the actual observations of z ∼ 3 LBGs by Steidel, Pettini & Hamilton. We need to introduce an additional ‘maturity’ criterion to circumvent subtle effects due to mass resolution in the simulation. We predict a number density of 1.15 arcmin−2 at R
25.5, in good agreement with the observed number density 1.2 ± 0.18 arcmin−2 . Our model allows us to study the efficiency of the selection criterion to capture z ∼ 3 galaxies. We find that the colour contours designed from models of spectrophotometric evolution of stellar populations are able to select more ‘realistic’ galaxies issued from models of hierarchical galaxy formation. We quantify the fraction of interlopers (12 per cent) and the selection efficiency (85 per cent), and we give estimates of the cosmic variance. We then study the clustering properties of our model LBGs. They are hosted by haloes with masses ∼1.6 × 1012 M , with a linear bias parameter that decreases with increasing scale from b = 5 to 3. The amplitude and slope of the two-dimensional correlation function is in good agreement with the data. We investigate a series of physical properties: ultraviolet (UV) extinction (a typical factor 6.2 at 1600 Å), stellar masses, metallicities and star formation rates, and we find them to be in general agreement with observed values. The model also allows us to make predictions at other optical and infrared/submillimetre wavelengths, that are easily accessible though queries to a web-interfaced relational data base. Looking into the future of these LBGs, we predict that 75 per cent of them end up as massive ellipticals and lenticulars today, even though only 35 per cent of all our local ellipticals and lenticulars are predicted to have a LBG progenitor. In spite of some shortcomings that come from our simplifying assumptions and the subtle propagation of mass resolution effects, this new ‘mock observation’ method clearly represents a first step toward a more accurate comparison between hierarchical models of galaxy formation and real observational surveys. 572 J. Blaizot et al. 1 INTRODUCTION 2 S I M U L AT I N G O B S E RVAT I O N S 2.1 A brief summary of the GALICS model GALICS is a model of hierarchical galaxy formation which combines high-resolution cosmological simulations to describe the DM content of the Universe with semi-analytic prescriptions to deal with the baryonic matter. This hybrid approach is fully described in GALICS I and GALICS II, so we only briefly recall its main characteristics hereafter.  C 2004 RAS, MNRAS 352, 571–588 Models of hierarchical galaxy formation are now sophisticated enough to produce a host of predictions for the statistical properties of local and high-redshift galaxies. These models have to be tested against observational samples. However, such a comparison is not as straightforward as it might appear at first glance, because observations are affected by selection criteria and observational biases. The difficulty of the comparison is increased for samples of high-redshift galaxies that are selected only on the basis of their apparent magnitudes and colours, that is, according to properties that are far from the theoretical quantities naturally computed by models of galaxy formation. In such a case, it is not easy to track back the propagation of the selection criteria and observational biases to the ‘model’ parameter space. At this stage, the best method consists in putting model predictions into the ‘observation’ parameter space, by producing mock galaxy samples that are obtained with selection criteria and observational biases as close as possible to those that affect the real observational samples. This paper uses the predictions of the GALICS model of hierarchical galaxy formation (for Galaxies in Cosmological Simulations) that embodies the so-called ‘hybrid approach’ by coupling the description of dark matter (DM) collapse by means of cosmological N-body simulations, and the description of the physics of baryons with semi-analytic recipes, including the estimate of spectral energy distributions in the UV to submillimetre wavelength range (Hatton et al. 2003, hereafter GALICS I, and Devriendt et al., in preparation, hereafter GALICS II). We use the MOMAF package (for Mock Map Facility) described in Blaizot et al. (2003a) (hereafter MOMAF) to produce mock observing cones from which mock samples can be extracted, with selection criteria and biases that mimic those of actual observations. We hereafter illustrate such a method by addressing the constraints put on models by the so-called Lyman-break galaxies (hereafter LBGs) at redshift z ∼ 3 (Steidel et al. 1996). These galaxies are obtained using the UV drop-out technique, first employed for the selection of distant galaxies by Steidel & Hamilton (1993), that simply relies on the shift of the Lyman-break through broad-band filters. This break, being mainly due to the absorption of the UV photons by the HI of the intergalactic medium (IGM), is roughly independent from the intrinsic properties of the galaxies, thus allowing the efficient selection of a complete sample of luminous galaxies at a given redshift, with photometry in only three broad-band filters. Over recent years, this technique has led to a tremendous increase in the number of observed objects at z ∼ 3. Although the set of data is impressively large today, compared to only a few years ago, a theoretical picture has not yet clearly emerged. This is mainly (...truncated)