Modeling collisional fragmentation with Fraggle, a high fidelity fragment generation model developed for the Swiftest n-body project

We have developed a new collisional fragmentation model, called Fraggle, as part of an effort to develop a new version of the Swift family of n-body integration codes, which we call Swiftest. When two massive bodies collide in a Swiftest simulation, the collision is evaluated to determine the kind of collisional outcome that is expected. Collisions can lead to either pure mergers or pure hit-and-run, which do not generate new fragments. However, if the collision would lead to a fragmentation, Fraggle generates a set of new bodies with self-consistent distributions of mass, position, velocity, and rotation. The new particles are constrained to strictly conserve angular and linear momentum of the pre-collision system, and (less strictly) reduce the system energy (the sum of potential, kinetic, and gravitational binding energies) by an amount prescribed by the collisional outcome model. The position and velocity distributions are also determined to be consistent with that expected from the three main collisional regimes that Fraggle currently models: disruption, in which the target body is left largely intact and the fragments are ejected from the impact site, supercatastrophic disruption, in which both the target and projectile bodies are disrupted, and disruptive hit-and-run, in which the projectile is fully disrupted but the fragments continue along the projectile’s original trajectory with some dispersion.

Because collisions between growing planetary bodies often have steep-sloped size-frequency distributions, the number of collisional fragments generated by a single fragmentation event can easily be many orders of magnitude larger than what is computationally feasible in an n-body simulation.Though based on the earlier Swifter code (which itself a Fortran 90 rewrite of the earlier Fortran 77 Swift), the data structures used by Swiftest have been overhauled to improve the performance of integrations in a multi-CPU shared memory parallel environment. This allows Swiftest versions of integrators, such as SyMBA, to simulate many more particles in a given amount of wall time than previous versions, allowing us to simulate planetary accretion with collisional fragmentation with higher fidelity than previous models were capable of.

Here we demonstrate the use of Fraggle in n-body simulations of late stage planet formation and the formation of martian satellites from a debris disk. We compare our results to earlier studies of similar systems, both with and without collisional fragmentation. We also demonstrate the capabilities of Swiftest, particularly its parallel performance.