AstroAI Lunch Talks - March 18, 2024 - Tri Nguyen
25 Mar 2024 - Joshua Wing
The video can be found here: https://www.youtube.com/watch?v=NCnu7AKIh-Q
Speaker: Tri Nguyen
Title: FLORAH: Emulating the merger histories of DM halos with deep generative models
Abstract: The mass assembly and merger histories of dark matter halos play a crucial role in shaping the formation and evolution of galaxies. These histories are used extensively in semi-analytic and empirical models of galaxy formation, yet current analytic methods to generate them are inaccurate and unable to capture their relationship with the halo internal structure and large-scale environment.
In this talk, I introduce FLORAH, a machine-learning framework for generating generating mass assembly histories of dark matter halos. By training FLORAH on the GUREFT and VSMDPL simulations, I demonstrate its ability to recover key properties such as stellar-halo mass relation and assembly bias of galaxies. In addition, by combining multiple networks trained on different simulations, I can construct accurate assembly histories over a wide dynamic halo mass and up to an ultra-high redshift of 20, which is currently far beyond that of a single N-body simulation. Lastly, I will present the latest results on extending FLORAH to generating full merger trees, by incorporating graph generative models.
Watch the talk below!