Formation of Sink Particles using a subgrid model
Star formation simulations involve physics across a wide range of scales, a fact that impedes capturing all the relevant physics in a single simulation. Molecular clouds in which stars are born are typically larger than 10s of parsecs whereas the small (relatively speaking) droplets of gas that precede an individual star’s birth are on the order of 0.1 parsecs. After its birth, a star continues to grow by accreting matter from its protoplanetary discs, which itself is typically only 1000s of AU (or ~0.005 pc). If we were to use only a single grid cell for the entire protostar and its disc, our computational domain would need to be 2000 grid cells across, which is a resolution that approaches the limits of computation. Even with such high resolving power, a single grid cell would be unable to capture the complexity of the interactions between a protostar and its disc.
Therefore we utilise “subgrid models” - that is, models which reproduce behaviour that occur at resolutions below that which our grid can resolve. One such subgrid model is the “sink particle” (Federrath et al. 2010) which identifies when a region of gas becomes gravitationally unstable and would collapse to form a star.
This is a run of the Boss-Bodenheimer test which is a standard means to test sink particle creation. We perform this test in FLASH and use the sink particle submodule implemented by Federrath et al. (2010). The test involves a sphere of dense gas surrounded by a less dense ambient medium. The sphere initially has solid body rotation, as well as a sinusoidal density perturbation in the x-y plane. This perturbation has two peaks which collapse, resulting in the formation of two sink particles.