Abstract
Numerical simulations of giant molecular clouds that resolve individual star formation with all relevant feedback physics have recently become feasible with the STARFORGE framework. I will first review the STARFORGE project and its numerical techniques for simulating star formation, stellar dynamics, cooling/heating physics, radiation, winds, jets, and core-collapse supernovae. I will then present a variety of results from the first ever simulation to combine all of these important physics, of a 20,000Msun giant molecular cloud comparable to stellar nurseries in the Solar neighborhood. I will focus on comparing the simulation with hallmarks of star formation: the initial mass function, star formation efficiency, star cluster kinematics, stellar multiplicity, and stellar accretion. The simulation successfully reproduces all of these phenomena - if and only if feedback is fully accounted for.
I will then use the simulation to address one basic question: how do stars get their mass? An analysis of the spatial and temporal extent of stellar accretion allows us to rule out a variety of proposed star formation theories, motivating a new picture and providing some analytic intuition for the origin of the IMF.
I will conclude by pointing out various failures of the STARFORGE model, which hint at some interesting open problems for future investigations of star formation physics.