Learning Galaxy Properties from Merger Trees with Mangrove
Efficiently mapping between baryonic properties and dark matter is a major challenge in astrophysics.
Although semi-analytic models (SAMs) and hydrodynamical simulations have made impressive advances in reproducing galaxy observables across cosmologically significant volumes, both still require significant computation times, and are hard to succinctly analyze, representing a barrier to many applications. Graph Neural Networks (GNNs) have recently proven to be the natural choice for learning physical relations. Among the most inherently graph-like structures found in astrophysics are the dark matter merger trees that encode the evolution of dark matter halos. In this cake talk I will introduce a new, graph-based emulator framework, Mangrove, and show that it emulates the galactic stellar mass, cold gas mass and metallicity, instantaneous and time-averaged star formation rate, and black hole mass with scatters two times lower than other methods across a simulation box of side length 75 Mpc/h in 40 seconds, 4 orders of magnitude faster than a SAM and 9 orders of magnitude faster than a hydro simulation. I’ll also show how Mangrove allows for quantification of the dependence of galaxy properties on merger history, making it possible to learn about the simulations on a new level. I will also compare Mangrove results to the current state of the art in emulating the dark matter – galaxy connection and show significant improvements for all target properties.
Mangrove is publicly available at https://github.com/astrockragh/Mangrove.
