The case for large-scale AGN feedback in galaxy formation: insights from XFABLE simulations

While cosmological simulations of galaxy formation have reached maturity and are able to reproduce many fundamental galaxy and halo properties, no consensus has yet been reached on the impact of 'baryonic feedback' on the non-linear matter power spectrum. This severely limits the precision of (and potentially biases) small-scale cosmological constraints obtained from weak lensing and galaxy surveys. Recent observational evidence indicates that 'baryonic feedback' may be more extreme than commonly assumed in current cosmological hydrodynamical simulations. In this paper, we therefore explore a range of empirical active galactic nucleus (AGN) feedback models, within the FABLE simulation suite, with different parametrizations as a function of cosmic time, host halo properties, and/or spatial location where feedback energy is thermalized. We demonstrate that an AGN radio-mode feedback acting in a larger population of black holes, with jets thermalizing at relatively large cluster-centric distances, as exemplified by our XFABLE model, is in good agreement with the latest weak lensing + kSZ constraints across all k-scales. Furthermore, XFABLE maintains good agreement with the galaxy stellar mass function, and gas fraction measurements, as well as consistency with key galaxy group and cluster properties, including scaling relations and intracluster medium radial profiles, within current observational uncertainties. Our work highlights the pressing need to model black hole accretion and feedback physics with a greater level of realism, including relativistic magnetized jets in full cosmological simulations. Finally, we discuss how a range of complementary observational probes in the near future will enable us to constrain AGN feedback models, and therefore reduce 'baryonic feedback' modelling uncertainty for the upcoming era of large cosmological surveys.