Title: Towards an emulator-based halo model in modified gravity theories
Abstract: N-body simulation-based templates, so-called emulators, will enable us to accurately predict dark matter halo properties down to highly non-linear scales. This talk presents an accurate and flexible emulated-based halo model for the real- and redshift-space galaxy two-point correlation functions. The main set of simulations — the FORGE modified gravity N-body simulation suite — covers 50 cosmologies around the Planck ΛCDM cosmology in f(R) gravity. We feed the data into neural networks to emulate the halo mass functions (HMFs), halo concentration-mass relations, halo real-space clustering and pairwise velocity moments. For example, the emulator for HMFs achieves about 1% accuracies for M > 1e12 M_solar/h. Once the halo emulators are ready, we will have the flexibility to combine them with a viable galaxy-halo connection model, such as the halo occupation distribution, to compute galaxy two-point correlation functions. In the end, we demonstrate that this halo model approach can be applied to describe the cosmological observables beyond two-point statistics and galaxy clustering.