Below the subgrid: uncertainties in supernova input rates drive qualitative differences in simulations of galaxy evolution
Feedback from core collapse supernovae (SNe), the final stage of evolution of the most massive stars, has long been a key element in simulations of galaxy formation. In this paper, we examine how simplifying assumptions made in approximating the SN rates along with underlying physical uncertainty in those rates can lead to large variations in the overall evolution of simulated Milky Way-like galaxies. We find that the clustering of star formation is strongly impacted by the delay between star formation and SN feedback. In addition, the choice to use a realistic delay time distribution or instantaneous injection for SN can have a significant impact on the galaxy. These effects appear even when identical sub-grid models are used for coupling SN energy and momentum, and the total SN energy budget is kept constant. In addition, we show that the uncertain minimum SN progenitor mass has a significant impact on the SN energy budget and injection timescale, and can completely change the overall evolution of the galaxy. These underlying uncertainties mean that despite advances in the sub-grid modelling of SN feedback, there are still serious difficulties in constraining the effects of SN feedback. This complicates the task of comparing different simulations to each other, as well as comparing simulations to observations. We conclude by providing practical limits on the parameters of subgrid models for SN feedback, which bound the uncertainty arising from SN progenitor physics for future predictions from galaxy simulations.