STARFORGE, or, How I Learned to Stop Cheating at Star Formation and Love the IMF

18/08/2021 - 14:00 - 15:00
Remoto, com transmissão pela internet
STARFORGE, or, How I Learned to Stop Cheating at Star Formation and Love the IMF
Mike Grudić
CIERA - Northwestern University
Star formation (SF) presents many open problems but has historically been challenging to model realistically due to the wide range of physics to account for, including gravity, hydro-and stellar dynamics, magnetic fields, dust, chemistry and cooling, radiation, and stellar feedback. I will present STARFORGE, a numerical SF simulation framework in the GIZMO code that can 1) evolve all of these physics in concert, including all important feedback mechanisms (jets, radiation, winds, and supernovae), 2) predict an IMF from these physics self-consistently, and 3) simulate the entire gas mass of the massive (1e4+Msun) GMCs responsible for most star formation. After briefly reviewing some of the numerical techniques that make this possible, I will present the results of our first simulation of a 20,000Msun GMC with the full physics package. The cloud initially collapses and fragments, heating up as it is irradiated by protostars, causing accretion onto existing stars to be increasingly favoured over fragmentation of new stars - hence the protostellar mass function becomes more top-heavy with time until the cloud is disrupted by feedback from massive stars, linking the IMF directly to feedback. The cloud is disrupted and star formation largely ceases after converting  ~7% of the GMC mass to stars, but we observe an extended late-time tail of slower star formation in expelled clumps that resist disruption. The final IMF resembles the observed IMF more closely than previous simulations without feedback, but still has a slightly shallower slope of -2 (vs. the typical -2.35), possibly indicating that further missing physics must help regulate massive star formation. Nearly all massive stars end up in binaries, while most low-mass stars are single, as observed. The final stellar system forms an unbound association as the cluster is ripped apart by the strong tides of the expanding, clumpy GMC. I will conclude by summarizing the successes of this new generation of simulation, while also pointing out some new open problems hinted at by its failures.
Mike Grudić likes to work on theoretical and computational astrophysics, with a focus on numerical methods for N-body and hydrodynamics simulations, and the application of these techniques to novel simulations of star formation (as leader of the STARFORGE Project) and galaxy formation (as a contributor to the FIRE Project). He did his PhD in Physics at Caltech with Phil Hopkins, and is now a CIERA Postdoctoral Fellow at Northwestern University.  In September 2021 he will join Carnegie Observatories as a Hubble Fellow.