Galaxy formation simulations: "sub-grid" vs. physics
Cosmological simulations of structure formation have undergone rapid development over the last decade and evolved from purely gravitational computations of large scale structure to full hydrodynamical simulations which include a plethora of complex baryonic physics. Recent successes of these simulations have been impressive, with several independent groups obtaining models of present day galaxies with a morphological mix in broad accord with observation, which has been a long standing goal for more than 20 years. Yet, due to many unknowns and the prohibitively large mass and spatial resolutions needed to solve the problem ab initio, cosmological simulations have adopted so-called "sub-grid" physics models, which are employed on small, unresolved scales. These invariably contain a number of free or poorly constrained parameters which can crucially affect the main simulation results. In this talk I will critically review how much cosmological simulations are subject to fine tuning due to these "sub-grid" models, which consequences this has for our understanding of the underlying physics and what are the future prospects of moving towards ab initio simulations.