Towards a realistic population of simulated galaxy groups and clusters

We present a new suite of large-volume cosmological hydrodynamical simulations called cosmo-OWLS. They form an extension to the OverWhelmingly Large Simulations (OWLS) project, and have been designed to help improve our understanding of cluster astrophysics and non-linear structure formation, which are now the limiting systematic errors when using clusters as cosmological probes. Starting from identical initial conditions in either the Planck or WMAP7 cosmologies, we systematically vary the most important `sub-grid' physics, including feedback from supernovae and active galactic nuclei (AGN). We compare the properties of the simulated galaxy groups and clusters to a wide range of observational data, such as X-ray luminosity and temperature, gas mass fractions, entropy and density profiles, Sunyaev-Zel'dovich flux, I-band mass-to-light ratio, dominance of the brightest cluster galaxy, and central massive black hole (BH) masses, by producing synthetic observations and mimicking observational analysis techniques. These comparisons demonstrate that some AGN feedback models can produce a realistic population of galaxy groups and clusters, broadly reproducing both the median trend and, for the first time, the scatter in physical properties over approximately two decades in mass ($10^{13} \lesssim M_{500} \lesssim 10^{15}~\textrm{M}_{\odot}$) and 1.5 decades in radius ($0.05 \lesssim r/r_{500} \lesssim 1.5$). However, in other models, the AGN feedback is too violent (even though they reproduce the observed BH scaling relations), implying calibration of the models is required. The production of realistic populations of simulated groups and clusters, as well as models that bracket the observations, opens the door to the creation of synthetic surveys for assisting the astrophysical and cosmological interpretation of cluster surveys, as well as quantifying the impact of selection effects.