What FIREs Up Star Formation: the Emergence of the Kennicutt-Schmidt Law from Feedback

We present an analysis of the global and spatially-resolved Kennicutt-Schmidt (KS) star formation relation in the FIRE (Feedback In Realistic Environments) suite of cosmological simulations, including halos with $z = 0$ masses ranging from $10^{10}$ -- $10^{13}$ M$_{\odot}$. We show that the KS relation emerges and is robustly maintained due to the effects of feedback on local scales regulating star-forming gas, independent of the particular small-scale star formation prescriptions employed. We demonstrate that the time-averaged KS relation is relatively independent of redshift and spatial averaging scale, and that the star formation rate surface density is weakly dependent on metallicity and inversely dependent on orbital dynamical time. At constant star formation rate surface density, the `Cold \& Dense' gas surface density (gas with $T < 300$~K and $n > 10$~cm$^{-3}$, used as a proxy for the molecular gas surface density) of the simulated galaxies is $\sim$0.5~dex less than observed at $\sim$kpc scales. This discrepancy may arise from underestimates of the local column density at the particle-scale for the purposes of shielding in the simulations. Finally, we show that on scales larger than individual giant molecular clouds, the primary condition that determines whether star formation occurs is whether a patch of the galactic disk is thermally Toomre-unstable (not whether it is self-shielding): once a patch can no longer be thermally stabilized against fragmentation, it collapses, becomes self-shielding, cools, and forms stars, regardless of epoch or environment.