Description
Galaxy positions and shapes as tracers of the large-scale structure of the Universe are key observables to test cosmological models in the late Universe down to the non-linear regime. Accurate cosmological constraints require a good understanding of the galaxy-halo connection, i.e. galaxy bias, which can be influenced by complex multi-scale baryonic dynamics and evolve over time. I will present a novel semi-analytic model of galaxy bias which naturally incorporates stochasticity and non-locality while ensuring the physicality and isotropy of the galaxy field modelled from the underlying matter field. I will show how this model can be calibrated from large-scale hydrodynamical simulations, such as FLAMINGO, to accurately sample galaxy populations from simulated matter fields while conditioning on various galaxy properties. I will also present how this model can be used to efficiently forward model galaxy positions consistent with hydrodynamical simulations while varying over cosmological and astrophysical parameters. Therefore, this model may enable an accurate simulation-based inference (SBI) analysis of galaxy clustering and galaxy-galaxy lensing observables from galaxy surveys such as Euclid, Rubin LSST and DESI.
