We perform a series of three-dimensional, adaptive-mesh-refinement (AMR) magnetohydrodynamical (MHD) simulations of star cluster formation including gravity, turbulence, magnetic fields, stellar radiative heating and outflow feedback. We observe that the inclusion of protostellar outflows (1) reduces the star formation rate per free-fall time by a factor of ~2, (2) increases fragmentation, and (3) shifts the initial mass function (IMF) to lower masses by a factor of 2.0 +/- 0.2, without significantly affecting the overall shape of the IMF. The form of the sink particle (protostellar objects) mass distribution obtained from our simulations matches the observational IMFs reasonably well. We also show that turbulence-based theoretical models of the IMF agree well with our simulation IMF in the high-mass and low-mass regime, but do not predict any brown dwarfs, whereas our simulations produce a considerable number of sub-stellar objects. Our numerical model of star cluster formation also reproduces the observed mass dependence of multiplicity. Our multiplicity fraction estimates generally concur with the observational estimates for different spectral types. We further calculate the specific angular momentum of all the sink particles and find that the average value of 1.5x10^19 cm^2/s is consistent with observational data. The specific angular momentum of our sink particles lies in the range typical of protostellar envelopes and binaries. We conclude that the IMF is controlled by a combination of gravity, turbulence, magnetic fields, radiation and outflow feedback.
The simulation shows a comparison of the NOWIND (no jets/outflows; left) with the same simulation, but including OUTFLOWS (right). The top panels show column density, while the bottom panels show gas temperature, highlighting the effects of radiation feedback.
We thank the anonymous reviewer for their comments and valuable suggestions, which helped to improve the paper. We thank Pavel Kroupa for valuable discussions regarding the system IMF. We further thank Piyush Sharda for helpful comments on the multiplicity algorithm and Donghee Nam for assistance with the python code for reproducing the Hopkins IMF. C.F. acknowledges funding provided by the Australian Research Council (Future Fellowship FT180100495), and the Australia-Germany Joint Research Cooperation Scheme (UADAAD). We further acknowledge high-performance computing resources provided by the Leibniz Rechenzentrum and the Gauss Centre for Supercomputing (grants pr32lo, pr48pi and GCS Large-scale project 10391), the Australian National Computational Infrastructure (grant ek9) in the framework of the National Computational Merit Allocation Scheme and the ANU Merit Allocation Scheme. The simulation software FLASH was in part developed by the DOE supported Flash Center for Computational Science at the University of Chicago.