 |
|
| EPoS Contribution |
|
Turbulent fragmentation and the stellar IMF
Troels Haugboelle STARPLAN, Copenhagen, DK | |
|
It has been suggested that the stellar initial mass function (IMF) is primarily
the result of fragmentation driven by supersonic turbulence, with gravity being
dominant only locally, where a dense region becomes unstable and collapses into
a protostar. Idealized numerical experiments of driven supersonic MHD
turbulence and no self-gravity has allowed us to confirm some of the
assumptions and results of the model, as far as the turbulent fragmentation is
concerned, but it cannot provide much insight on the subsequent evolution of the
mass distribution of dense unstable cores locally controlled by self-gravity.
We have taken the scenario a step further, by including self-gravity and following the collapse of individual cores down to a scale of 20 AU, using adaptive mesh refinement (AMR). Beyond that scale, we create accreting sink particles, so the origin and evolution of the stellar IMF can be tracked in detail by the accretion history of the sink particles. In order to sample the whole IMF, we simulate a slightly overdense region with a size of 3.5 pc and a total mass of 3,000 solar masses, requiring a maximum AMR resolution of 16,3843. By varying the root grid size, the number of AMR levels, and the internal parameters in our sink particle model we prove the numerical convergence of the slope of the Salpeter range, of the mass of the IMF peak, and of the relative abundance of brown dwarfs. These converged values agree well with our theoretical predictions and with the Chabrier-Salpeter fit to the observed IMF derived from field stars. While the role of stellar feedbacks (stellar radiation, jets, outflows, winds, HII regions) remains to be tested by this model, the current experiments include the main ingredients (MHD, turbulence, and self gravity), and demonstrate that the full IMF can be understood as the result of the turbulent fragmentation of isothermal, self-gravitating gas. Other effects are either of secondary importance, or conspire to recreate the same end result. | |
 | |
| Caption: Log column density from one of the runs is shown to the left, with black circles indicating sink particles. To the right is one of the many systematic comparisons done between the more than 30 runs. The upper plot shows the stellar IMF when using 3 or 5 AMR levels above a root grid of 5123. While both could be taken as in reasonable agreement with the Chabrier IMF, the lower plot shows otherwise through a powerful neighbour statistics: A new sink particle is only allowed to be created at least (in this case) 8 cells away from any other sinks; the exclusion radius. In the plot we show distribution of distances at creation for the two models. This illustrates, in this case, the minimum resolution and density needed for creating a realistic distribution of sink particles, which do not have a non-physical peak in the neighbour diagram near the exclusion radius. | |
| Collaborators: Paolo Padoan, ICREA & ICC-UB, Spain |
Suggested Session:
Turbulence |