Comparing Models for IMF Variation Across Cosmological Time in Milky Way-like Galaxies

One of the key observations regarding the stellar initial mass function (IMF) is its near-universality in the Milky Way (MW), which provides a powerful way to constrain different star formation models that predict the IMF. However, those models are almost universally "cloud-scale" or smaller -- they take as input or simulate single molecular clouds (GMCs), clumps, or cores, and predict the resulting IMF as a function of the cloud properties. Without a model for the progenitor properties of all clouds which formed the stars at different locations in the MW (including ancient stellar populations formed in high-redshift, likely gas-rich dwarf progenitor galaxies that looked little like the Galaxy today), the predictions cannot be fully explored, nor safely applied to "live" cosmological calculations of the IMF in different galaxies at different cosmological times. We therefore combine a suite of high-resolution cosmological simulations (from the Feedback In Realistic Environments project), which form MW-like galaxies with reasonable star formation properties and explicitly resolve massive GMCs, with various proposed cloud-scale IMF models. We apply the models independently to every star particle formed in the simulations to synthesize the predicted IMF in the present-day galaxy. We explore models where the IMF depends on Jeans mass, sonic or turbulent Bonner-Ebert mass, fragmentation with a polytropic equation-of-state, or where it is self-regulated by protostellar feedback. We show that all of these models, except the feedback-regulated ones, predict far more variation ($\sim 0.6-1$ dex $1\,\sigma$ scatter in the IMF turnover mass) in the simulations than is observed in the MW.