2011ApJ...729..120G

We use numerical hydrodynamic simulations to investigate prestellar core formation in the dynamic environment of giant molecular clouds, focusing on planar post-shock layers produced by colliding turbulent flows. A key goal is to test how core evolution and properties depend on the velocity dispersion in the parent cloud; our simulation suite consists of 180 models with inflow Mach numbers M≡v/c_s =1.1–9. At all Mach numbers, our models show that turbulence and self-gravity collect gas within post-shock regions into filaments at the same time as overdense areas within these filaments condense into cores. This morphology, together with the subsonic velocities we find inside cores, is similar to observations. We extend previous results showing that core collapse develops in an "outside-in" manner, with density and velocity approaching the Larson-Penston asymptotic solution. The time for the first core to collapse depends on Mach number as t_coll ∝M^{-1/2}ρ_0^{-1/2}, for ρ₀ the mean pre-shock density, consistent with analytic estimates. Core building takes 10 times as long as core collapse, which lasts a few x10⁵ yr, consistent with observed prestellar core lifetimes. Core shapes change from oblate to prolate as they evolve. To define cores, we use isosurfaces of the gravitational potential. We compare to cores defined using the potential computed from projected surface density, finding good agreement for core masses and sizes; this offers a new way to identify cores in observed maps. Cores with masses varying by three orders of magnitude (∼0.05-50 M_☉) are identified in our high-M simulations, with a much smaller mass range for models having low M. We halt each simulation when the first core collapses; at that point, only the more massive cores in each model are gravitationally bound, with E_th+ E_g< 0. Stability analysis of post-shock layers predicts that the first core to collapse will have mass M ∝ v^–1/2ρ^–1/2₀ T ^7/4, and that the minimum mass for cores formed at late times will have M ∝ v^–1ρ^–1/2₀T ², with T being the temperature. From our simulations, the median mass lies between these two relations. At the time we halt the simulations, the M versus v relation is shallower for bound cores than unbound cores; with further evolution the small cores may evolve to become bound, steeping the M versus v relation.