University of Pennsylvania Astronomy Seminar

Hydrodynamic simulations have a huge computational cost (~10 million CPU hours for 0.001 Gpc^3 volume), and cannot therefore be directly used in predictions for upcoming cosmological surveys which will probe 10-100 Gpc^3 volumes. Focusing on neutral hydrogen (HI), I will show that neural networks can be trained on hydro simulations to quickly generate accurate HI maps from gravity-only dark matter simulations. I will also show that the environment of a dark matter halo has a crucial effect on its HI mass, and ignoring this effect in a halo model framework will lead to underprediction of real-space HI power spectrum by ~10% on linear scales. I will present novel symbolic expressions, derived via machine learning, which encode this environmental effect and can be used to augment traditional halo based models. Towards the end, I will switch gears and talk about analytic methods for calculating covariance matrices for galaxy surveys instead of using thousands of expensive mock simulations.