2009ApJ...707..954P

We use time-varying models of the coupled evolution of the H I, H₂ gas phases and stars in galaxy-sized numerical simulations to (1) test for the emergence of the Kennicutt-Schmidt (K-S) and the H₂-pressure relation, (2) explore a realistic H₂-regulated star formation recipe which brings forth a neglected and potentially significant SF-regulating factor, and (3) go beyond typical galactic environments (for which these galactic empirical relations are deduced) to explore the early evolution of very gas-rich galaxies. In this work, we model low-mass galaxies (M_baryon≤ 10⁹ M_☉), while incorporating an independent treatment of CO formation and destruction, the most important tracer molecule of H₂ in galaxies, along with that for the H₂ gas itself. We find that both the K-S and the H₂-pressure empirical relations can robustly emerge in galaxies after a dynamic equilibrium sets in between the various interstellar medium (ISM) states, the stellar component and its feedback (T ≳ 1 Gyr). The only significant dependence of these relations seems to be for the CO-derived (and thus directly observable) ones, which show a strong dependence on the ISM metallicity. The H₂-regulated star formation recipe successfully reproduces the morphological and quantitative aspects of previous numerical models while doing away with the star formation efficiency parameter. Most of the H I -> H₂ mass exchange is found taking place under highly non-equilibrium conditions necessitating a time-dependent treatment even in typical ISM environments. Our dynamic models indicate that the CO molecule can be a poor, nonlinear, H₂ gas tracer. Finally, for early evolutionary stages (T ≲ 0.4 Gyr), we find significant and systematic deviations of the true star formation from that expected from the K-S relation, which are especially pronounced and prolonged for metal-poor systems. The largest such deviations occur for the very gas-rich galaxies, where deviations of a factor ∼3-4 in global star formation rate (SFR) can take place with respect to those expected from the CO-derived K-S relation. This is particularly important since gas-rich systems at high redshifts could appear as having unusually high SFRs with respect to their CO-bright H₂ gas reservoirs. This points to a possibly serious deficiency of K-S relations as elements of the sub-grid physics of star formation in simulations of structure formation in the early universe.