Context. Lacking a paradigm for the onset of star formation, it is important to derive basic physical properties of prestellar cores and filaments like density and temperature structures. Aims. We aim to disentangle the spatial variation in density and temperature across the prestellar core L1689B, which is embedded in a filament. We want to determine the range of possible central densities and temperatures that are consistent with the continuum radiation data. Methods. We apply a new synergetic radiative transfer method: the derived 1D density profiles are both consistent with a cut through the Herschel PACS/SPIRE and JCMT SCUBA-2 continuum maps of L1689B and with a derived local interstellar radiation field. Choosing an appropriate cut along the filament major axis, we minimize the impact of the filament emission on the modeling. Results. For the bulk of the core (5000-20000 au) an isothermal sphere model with a temperature of around 10 K provides the best fits. We show that the power law index of the density profile, as well as the constant temperature can be derived directly from the radial surface brightness profiles. For the inner region (< 5000 au), we find a range of densities and temperatures that are consistent with the surface brightness profiles and the local interstellar radiation field. Based on our core models, we find that pixel-by-pixel single temperature spectral energy distribution fits are incapable of determining dense core properties. Conclusions. We conclude that, to derive physical core properties, it is important to avoid azimuthally-averaging core and filament. Correspondingly, derived core masses are too high since they include some mass of the filament, and might introduce errors when determining core mass functions. The forward radiative transfer methods also avoids the loss of information owing to smearing of all maps to the coarsest spatial resolution. We find the central core region to be colder and denser than estimated in recent inverse radiative transfer modeling, possibly indicating the start of star formation in L1689B.

We study the stability of filaments in equilibrium between gravity and internal as well as external pressure using the grid based AMR-code RAMSES. A homogeneous, straight cylinder below a critical line mass is marginally stable. However, if the cylinder is bent, e.g. with a slight sinusoidal perturbation, an otherwise stable configuration starts to oscillate, is triggered into fragmentation and collapses. This previously unstudied behavior allows a filament to fragment at any given scale, as long as it has slight bends. We call this process 'geometrical fragmentation'. In our realization the spacing between the cores matches the wavelength of the sinusoidal perturbation, whereas up to now, filaments were thought to be only fragmenting on the characteristical scale set by the mass-to-line ratio. Using first principles, we derive the oscillation period as well as the collapse timescale analytically. To enable a direct comparison with observations, we study the line-of-sight velocity for different inclinations. We show that the overall oscillation pattern can hide the infall signature of cores.

We extend our previous work on simulations with the code ramses on accretion-driven turbulence by including self-gravity and study the effects of core formation and collapse. We show that radial accretion on to filaments drives turbulent motions which are not isotropic but radially dominated. In contrast to filaments without gravity, the velocity dispersion of self-gravitating filaments does not settle in an equilibrium. Despite showing similar amounts of driven turbulence, they continually dissipate their velocity dispersion until the onset of core formation. This difference is connected to the evolution of the radius as it determines the dissipation rate. In the non-gravitational case filament growth is not limited and its radius grows linearly with time. In contrast, there is a maximum extent in the self-gravitational case resulting in an increased dissipation rate. Furthermore, accretion-driven turbulence shows a radial profile which is anticorrelated with density. This leads to a constant turbulent pressure throughout the filament. As the additional turbulent pressure does not have a radial gradient it does not contribute to the stability of filaments and does not increase the critical line-mass. However, this radial turbulence does affect the radius of a filament, adding to the extent and setting its maximum value. Moreover, the radius evolution also affects the growth time-scale of cores which compared to the time-scale of collapse of an accreting filament limits core formation to high line-masses.

We present a solution for the observed core fragmentation of filaments in the Taurus L1517 dark cloud which previously could not be explained (Hacar & Tafalla 2011). Core fragmentation is a vital step for the formation of stars. Observations suggest a connection to the filamentary structure of the cloud gas, but it remains unclear which process is responsible. We show that the gravitational instability process of an infinite, isothermal cylinder can account for the exhibited fragmentation under the assumption that the perturbation grows on the dominant wavelength. We use numerical simulations with the code RAMSES, estimate observed column densities and line-of-sight velocities, and compare them to the observations. A critical factor for the observed fragmentation is that cores grow by redistributing mass within the filament and thus the density between the cores decreases over the fragmentation process. This often leads to wrong dominant wavelength estimates, as it is strongly dependent on the initial central density. We argue that non-linear effects also play an important role on the evolution of the fragmentation. Once the density perturbation grows above the critical line-mass, non-linearity leads to an enhancement of the central core density in comparison to the analytical prediction. Choosing the correct initial conditions with perturbation strengths of around 20%, leads to inclination corrected line-of-sight velocities and central core densities within the observational measurement error in a realistic evolution time.