FLAMINGO: Calibration and Clusters

Roi Kugel and the FLAMINGO collaboration


Introducing the FLAMINGO project, a large suite of cosmological hydrodynamics simulations, with the two flagship boxes being a $(2.8~\text{Gpc})^3$ with a mass resolution of 1.09e9 Msun (5040^3 gas particles) and a $(1~\text{Gpc})^3$ with a mass resolution of 1.34e8 Msun (3600^3 gas particles). Additionally, there are simulations where we vary cosmology and AGN feedback in $(1~\text{Gpc})^3$ at 1.09 Msun resolution. In order to be able to use hydro simulations of this kind for upcoming galaxy surveys, we need to have a good grasp of the baryonic physics in the simulation. By using machine learning to emulate the stellar mass function and the gas mass fractions in clusters, we are able to use Markov chain Monte Carlo to fit our subgrid models to observations. This gives us constraints on our subgrid model parameters, and allows us to create variations that bracket the observations by $\pm2sigma$. I will finish by talking about my current project, where I am comparing different cluster mass proxies, from for example X-ray, SZ, weak lensing and optical surveys, with the true mass from the FLAMINGO simulation, in order to see how selections based on on different quantities impact the resulting selection functions.