Building a complete coherent model of planet formation has proven difficult. There are gaps in the observational record, difficult physical processes that we have yet to fully understand, such as planetesimal formation, and an extensive list of observationally determined constraints that the model must fulfil. For example, the diversity of extrasolar planets detected to date is staggering – from single hot-Jupiters to multiple planet systems with several tightly packed super-Earths. In addition, the characteristics of the host stars are broad from single solar-mass stars to tight binaries and low mass, low metalicity stars. Even more surprising, perhaps, is the frequency of detection and thus, the implied efficiency of the planet formation process. Any theoretical model must not just be able to explain how planets form but must also explain the frequency and diversity of planetary systems. So why is planet formation so prolific? What parameters determine the type of planetary system that will result? How important are the initial parameters of the protoplanetary disk, such as composition, versus stochastic effects, such as gravitational scattering events, that occur during the evolution of the planetary system?
Current observations of extrasolar planets provide snapshots in time of the earliest and latest stages of planet formation but do not show the evolution between the two. It is at this point that we must rely on numerical models to evolve proto-planetary disks into planets. But how can we validate the results of our numerical simulations if the middle stages of planet formation are effectively invisible? Collisions are a core component of planet formation. Planetesimals, the building blocks of planets, collide with one another as they grow and evolve into planets or planetary cores and are viscously stirred by larger protoplanets and fully-formed planets. The range of impact parameters encountered during growth from planetesimals to planets span multiple collision outcome regimes: cratering, merging, disruption, and hit-and-run events. Most of these collisions produce significant debris and dust. If we have a good understanding of the production of collisional debris we can use it as an indirect tracer of on-going planetary evolution even if the planets themselves are not directly detectable.
In this paper I will show how numerical simulations of planet formation including realistic collision modelling can be used to predict, and be constrained by, observations.