In the late stages of terrestrial planet formation, pairwise collisions between planetary-sized bodies act as the fundamental agent of planet growth. These collisions can lead to either growth or disruption of the bodies involved and are largely responsible for shaping the final characteristics of planets. Despite their critical role in planet formation, an accurate treatment of collisions has yet to be realized. Therefore, we simulated a new set of 10,700 smoothed-particle hydrodynamics (SPH) simulations of pairwise collisions between rotating, differentiated bodies. This dataset is an order of magnitude larger than any previous collision dataset and is intended to serve as a training dataset for machine learning models, so that accurate surrogate models for planetary-scale collisions can be developed.

Each collision is uniquely defined by a set of 12 pre-impact parameters. Together, these parameters define the geometry of the impact and the physical and rotational characteristics of the bodies involved in the collision. The colliding bodies are referred to as the target and projectile, where the target is the more massive of the two. The pre-impact parameter space was constructed using Latin hypercube sampling (LHS). For the LHS10K set, a LHS-based adaptive surface response method (ARSM) was used to progressively enrich the sample from an initial LHS sample of 1000 collisions.

The collisions were simulated with Gasoline, a massively-parellel smoothed-particle hydrodynamics (SPH) code (Wadsley et al. 2004). The number of particles in each collision is set by the ratio of the projectile mass to target mass, whereby the projectile always contains 10,000 particles and the number of particles in the target is scaled relative to its mass. Each collision was simulated for 100 times its collision timescale, which is equivalent to the crossing time of the encounter.

The files provided here include the Gasoline runtime parameter files (collision.param), initial conditions (collision.std), final outputs (outcome.std), and SKID group finding files (*.grp). The parameter and initial condition files are sufficient to run the simulations with Gasoline and the appropriate Tillotson equation of state (EOS). The associated outputs and group finding files are those on which the analysis was carried out. Users can explore and analyze the input and output files using a number of open-source software packages, but we suggest Python's pynbody package (https://pynbody.github.io/pynbody/).

In the course of simulating the datasets, a small number of collisions failed to complete. These failures were either a result of hardware failures during integration, interruptive maintenence downtimes, or the result of a pre-impact body (i.e., target or projectile) that was rotationally unstable. Therefore, in the LHS10K set, there are 9978 collisions, 499 collisions in the LHS500 dataset, and 199 collisions in the LHS200 dataset.