Density reconstruction has become a powerful tool for studying the large-scale structure of the Universe and the distribution of baryons beyond galaxies. When combined with constrained simulations, it enables the interpretation of cosmological and astrophysical processes directly at the field level.
In the first part of this talk, I will present our analysis of large-scale structures at z~2 using density reconstructions from both galaxy tracers and the Lyman-α forest. Through cross-matching these maps, we identified a galaxy protocluster at z=2.3 that is not visible in the Lyman-α density map from the CLAMATO survey. Despite the absence of a corresponding absorption signal, the surrounding galaxy distribution indicates that this structure is likely the progenitor of a ~10^14 solar mass galaxy cluster. By analyzing similar systems in zoom-in simulations of galaxy clusters, we show that large-scale gas heating associated with galaxy feedback—particularly AGN jet feedback—provides a plausible explanation for the absent Lyman-α absorption in such environments.
In the second part, I will introduce a density reconstruction framework I am developing to infer the initial conditions of the Universe from galaxy surveys. The framework, dubbed Mujica and MaiGO, is designed to address stability challenges caused by complex survey geometries. Both methods rely on the same forward model of structure formation implemented with JaxPM, but follow two complementary reconstruction strategies: an optimization-based approach (Mujica) and a sampling-based approach (MaiGO). I will demonstrate their performance using mock semi-analytic galaxy catalogs based on the Millennium simulation and discuss their potential applications to galaxy surveys across different redshifts.
