We compute the star formation rate (SFR) in molecular clouds (MCs) that originate {it ab initio} in a new, higher-resolution simulation of supernova-driven turbulence. Because of the large number of well-resolved clouds with self-consistent boundary and initial conditions, we obtain a large range of cloud physical parameters with realistic statistical distributions, an unprecedented sample of star-forming regions to test SFR models and to interpret observational surveys. We confirm the dependence of the SFR per free-fall time, $SFR_{rm ff}$, on the virial parameter, $alpha_{rm vir}$, found in previous simulations, and compare a revised version of our turbulent fragmentation model with the numerical results. The dependences on Mach number, ${cal M}$, gas to magnetic pressure ratio, $beta$, and compressive to solenoidal power ratio, $chi$ at fixed $alpha_{rm vir}$ are not well constrained, because of random scatter due to time and cloud-to-cloud variations in $SFR_{rm ff}$. We find that $SFR_{rm ff}$ in MCs can take any value in the range $0 le SFR_{rm ff} lesssim 0.2$, and its probability distribution peaks at a value $SFR_{rm ff}approx 0.025$, consistent with observations. The values of $SFR_{rm ff}$ and the scatter in the $SFR_{rm ff}$--$alpha_{rm vir}$ relation are consistent with recent measurements in nearby MCs and in clouds near the Galactic center. Although not explicitly modeled by the theory, the scatter is consistent with the physical assumptions of our revised model and may also result in part from a lack of statistical equilibrium of the turbulence, due to the transient nature of MCs.
Download