Cosmological simulations from the cosmic web to supermassive black holes

This thesis presents an investigation of the evolution of supermassive black holes (SMBHs) and galaxies through cosmological hydrodynamic simulations on petascale supercomputers. With the large volume, high-resolution cosmological simulation BlueTides and ASTRID, we study the local and large scale environment of the first quasars, mass assembly and dynamics of the SMBHs over cosmic history, as well as the interplay between SMBHs and galaxy formation. We also diagnose the high redshift galaxy population predicted by cosmological simulations, and use current and future observations of high redshift galaxy abundance to test the alternative dark matter models such as the fuzzy dark matter. Finally, we also explore the possibility of using deep learning methods to model the small scale physical process for larger volume simulations, as a potential way to further extend the dynamic range that can be covered by next generation cosmological simulations.