We study the role of feedback from supernovae and black holes in the evolution of the star formation rate function (SFRF) of z~4-7 galaxies. We use a new set of cosmological hydrodynamic simulations, ANGUS (AustraliaN GADGET-3 early Universe Simulati
ons), run with a modified and improved version of the parallel TreePM-smoothed particle hydrodynamics code GADGET-3 called P-GADGET3(XXL), that includes a self-consistent implementation of stellar evolution and metal enrichment. In our simulations both Supernova (SN) driven galactic winds and Active Galactic Nuclei (AGN) act simultaneously in a complex interplay. The SFRF is insensitive to feedback prescription at z>5, meaning that it cannot be used to discriminate between feedback models during reionisation. However, the SFRF is sensitive to the details of feedback prescription at lower redshift. By exploring different SN driven wind velocities and regimes for the AGN feedback, we find that the key factor for reproducing the observed SFRFs is a combination of strong SN winds and early AGN feedback in low mass galaxies. Conversely, we show that the choice of initial mass function and inclusion of metal cooling have less impact on the evolution of the SFRF. When variable winds are considered, we find that a non-aggressive wind scaling is needed to reproduce the SFRFs at z>4. Otherwise, the amount of objects with low SFRs is greatly suppressed and at the same time winds are not effective enough in the most massive systems.
