12:00–1:00 pm ERC 301B
Yueying Ni (Carnegie Mellon University) "Cosmological simulations from the cosmic web to supermassive black holes"
The upcoming high-redshift observations will soon revolutionize the study of the early universe and allow us to better understand the growth and evolution of the first galaxies and supermassive black holes (SMBHs). Making full use of the future observations will require the next generation galaxy formation simulations to have large volumes for statistical samples of large structures as well as keep the high resolution to model the associated galaxy formation process. The involved memory load incurs huge computational difficulties and raises one of the major challenges in the field. In this talk, I will first introduce our new ASTRID simulation, one of the largest cosmological hydrodynamic simulations that concurrently model the evolution of galaxies and SMBHs down to z=2. Due to the volume limitation, current state-of-art galaxy formation simulations can hardly sample and study the extremely rare z>6 quasar populations. In the second part of my talk, I will introduce how the Constrained Realization (CR) method can be applied in small volume simulations to efficiently sample the rare massive objects conditioned on certain (user-specified) large-scale features and study the impact of large-scale structures on the early quasar growth. Finally, as a potential direction for future galaxy formation simulations, I will briefly introduce our developed super-resolution technique that uses deep learning models to extend the dynamic range covered by cosmological simulations.
