Numerical simulations of galaxies in cosmological volumes

Modelling galaxy formation in a cosmological context is challenging due to many physical processes and a wide range of scales involved. It is then convenient to address the problem by means of numerical techniques. Cosmological simulations are still orders of magnitude away from capturing directly the spatial scales where stars actually form. The treatment of the interstellar medium, with multiple gas phases co-existing at very different densities and temperatures, needs to be included in simulations in the form of sub-resolution effective models. In Murante et al. (2015), cosmological simulations of individual disk galaxies, carried out with GADGET-3 (Springel, 2005), are presented including the sub-resolution model MUlti-Phase Particle Integrator (MUPPI). Late-type galaxies are obtained with properties in broad agreement with observations (disk size, mass, surface density, rotation velocity, gas fraction) and also with results shown by other groups. During my PhD project, the properties of the bars in simulated galaxies are quantified and presented in Goz et al. (2015). Morphology, kinematic and radial streaming motions are in agreement with observations of local Universe (Seidel et al., 2015). In the GA galaxy, Fourier analysis of the surface density of the disk reveals that bar length and strength are remarkably independent of resolution. In the GA simulations, the onset of bar instability is found to take place when the disk is Toomre-unstable due to accumulation of mass in the disk. The complex task to calculate the emerging spectral energy distribution of a simulated galaxy, treating the reprocessing of light emitted by stars by means of dusty ISM, has been faced using the post-processing code GRASIL-3D (Dominguez-Tenreiro et al., 2014). During my PhD, GRASIL-3D has been interfaced with the outputs of the simulation and modified to better manage the quantities provided by the MUPPI algorithm. In this Thesis, results of a sample of simulated galaxies at z = 0 are shown, extracted from one large volume simulation presented by Barai et al. (2015), in which feedback and star formation are modelled by MUPPI. In broad agreement with observations of the local Universe, simulated galaxies reproduce the gas mass scaling, the main sequence, the relation between the gas-phase metallicity and stellar mass. Infrared predicted spectra, binned by stellar mass, gas metallicity, SFR and morphology, are remarkably in agreement with the available Herschel Reference Survey (Ciesla et al., 2014) templates calibrated with observations on nearby galaxies. H2 and HD are effective in cooling primordial gas below the temperature of 3c10^4 K, but usually the non-equilibrium radiative rates are only approximate in simulations, typically either assuming collisional ionization equilibrium or ignoring metal-line cooling altogether, rather than computing them explicitly. Maio et al. (2007) have extended a previous non-equilibrium calculations of Yoshida et al. (2003) in order to include in the numerical code GADGET a detailed chemical network and to follow the formation/destruction of a wide range of species, both with primordial and metal-enriched gas. During my PhD project, the chemical network, presented by Maio et al. (2007), has been extended to take into account the H2 formation onto dust grains and its UV-photodissociation. Then, the modified chemical network has been coupled in two different preliminary forms with the MUPPI algorithm within GADGET-3. The main aim, motivated by the tight observed connection between molecular hydrogen and star formation (e.g. Kennicutt et al., 2007; Bigiel et al., 2008), is to devise a way to track self-consistently non-equilibrium abundance and cooling processes of H2 and H2-based star formation in smoothed particle hydrodynamic simulations