Consistent SPH Simulations of Protostellar Collapse and Fragmentation

We study the consistency and convergence of smoothed particle hydrodynamics (SPH), as a function of the interpolation parameters, namely the number of particles $N$, the number of neighbors $n$, and the smoothing length $h$, using simulations of the collapse and fragmentation of protostellar rotating cores. The calculations are made using a modified version of the GADGET-2 code that employs an improved scheme for the artificial viscosity and power-law dependences of $n$ and $h$ on $N$, as was recently proposed by Zhu et al., which comply with the combined limit $N\to\infty$, $h\to 0$, and $n\to\infty$ with $n/N\to 0$ for full SPH consistency, as the domain resolution is increased. We apply this realization to the "standard isothermal test case" in the variant calculated by Burkert & Bodenheimer and the Gaussian cloud model of Boss to investigate the response of the method to adaptive smoothing lengths in the presence of large density and pressure gradients. The degree of consistency is measured by tracking how well the estimates of the consistency integral relations reproduce their continuous counterparts. In particular, $C^{0}$ and $C^{1}$ particle consistency is demonstrated, meaning that the calculations are close to second-order accuracy. As long as $n$ is increased with $N$, mass resolution also improves as the minimum resolvable mass $M_{\rm min}\sim n^{-1}$. This aspect allows proper calculation of small-scale structures in the flow associated with the formation and instability of protostellar disks around the growing fragments, which are seen to develop a spiral structure and fragment into close binary/multiple systems as supported by recent observations.