A semi-analytic model comparison: testing cooling models against hydrodynamical simulations

We compare predictions of cooled masses and cooling rates from three stripped-down semi-analytic models (SAMs) of galaxy formation with the results of N-body+Smoothed Particle Hydrodynamics (SPH) simulations with gas particle mass of 3.9 × 106 h-1 M⊙, where radiative cooling of a gas of primordial composition is implemented. We also run a simulation where cooling is switched on at redshift ˜2, in order to test cooling models in a regime in which their approximations are expected to be valid. We confirm that cooling models implemented in SAMs are able to predict the amount of cooled mass at z = 0 to within ˜20 per cent. However, some relevant discrepancies are found. (i) When the contribution from poorly resolved haloes is subtracted out, SAMs tend to underpredict by ˜30 per cent the mass that cools in the infall-dominated regime. (ii) At large halo masses, SAMs tend to overpredict cooling rates, though the numerical result may be affected by the use of a standard version of SPH. (iii) As found in our previous work, cooling rates are found to be significantly affected by model details: simulations disfavour models with large cores and with quenching of cooling at major mergers. (iv) When cooling is switched on at z ˜ 2, cold gas accumulates very quickly in the simulated haloes. This accumulation is reproduced by SAMs with varying degrees of accuracy.