A comparison between semi-analytical gas cooling models and cosmological hydrodynamical simulations
We compare the mass cooling rates and cumulative cooled-down masses predicted by several semi-analytical (SA) cooling models with cosmological hydrodynamical simulations performed using the AREPO code (ignoring processes such as feedback and chemical enrichment). The SA cooling models are the new GALFORM cooling model introduced in Hou et al. (2017), along with two earlier GALFORM cooling models and the L-GALAXIES and MORGANA cooling models. We find that the predictions of the new GALFORM cooling model are generally in best agreement with the simulations. For halos with $M_{\rm halo}\lesssim 3\times 10^{11}\,{\rm M}_{\odot}$, the SA models predict that the timescale for radiative cooling is shorter than or comparable to the gravitational infall timescale. Even though SA models assume that gas falls onto galaxies from a spherical gas halo, while the simulations show that the cold gas is accreted through filaments, both methods predict similar mass cooling rates, because in both cases the gas accretion occurs on similar timescales. For halos with $M_{\rm halo}\gtrsim 10^{12}\,{\rm M}_{\odot}$, gas in the simulations typically cools from a roughly spherical hot gas halo, as assumed in the SA models, but the halo gas gradually contracts during cooling, leading to compressional heating. SA models ignore this heating, and so overestimate mass cooling rates by factors of a few. At low redshifts halo major mergers or a sequence of successive smaller mergers are seen in the simulations to strongly heat the halo gas and suppress cooling, while mergers at high redshifts do not suppress cooling, because the gas filaments are difficult to heat up. The new SA cooling model best captures these effects.