2014MNRAS.439.3039L

We have simulated star formation in pre-stellar cores, using smoothed particle hydrodynamics and initial conditions informed by observations of the cores in Ophiuchus. Because the observations are limited to two spatial dimensions plus radial velocity, we cannot infer initial conditions for the collapse of a particular core. However, with a minimum of assumptions (isotropic turbulence with a power-law spectrum, a thermal mix of compressive and solenoidal modes, a critical Bonnor-Ebert density profile) we can generate initial conditions that match, in a statistical sense, the distributions of mass, projected size and aspect ratio, thermal and non-thermal one-dimensional velocity dispersion, observed in Ophiuchus. The time between core-core collisions in Ophiuchus is sufficiently long, that we can simulate single cores evolving is isolation, and therefore we are able to resolve masses well below the opacity limit. We generate an ensemble of 100 cores, and evolve them with no radiative feedback from the stars formed, then with continuous radiative feedback and finally with episodic radiative feedback. With no feedback the simulations produce too many brown dwarfs, and with continuous feedback too few. With episodic radiative feedback, both the peak of the protostellar mass function (at ∼ 0.2M_☉) and the ratio of H-burning stars to brown dwarfs are consistent with observations. The mass of a star is not strongly related to the mass of the core in which it forms. Low-mass cores (Mcore ∼ 0.1M_☉) tend to collapse into single objects, whereas high-mass cores (Mcore ≳ 1M_☉) usually fragment into several objects with a broad mass range.