Generally, most computer impact models set our Sun, the planets, and a small representative subset of potential impactors in an initial start state, and then run iteratively, accounting at each step for the forces active on this population.

Even using a representative subset, the computing demands are enormous, and one pass through such models takes long periods of computer time. While the task could be handled with programs which are parallelized, generally there is not sufficient funding to allow the use of parallel machines, or of parallel coding. Even with the use of parallel techniques, the task remains daunting, at least equal to if not exceeding the difficulty of modeling weather.

In attempts to give the computer models a better approximation to reality, different models have incorporated different physical mechanisms, including the effect of sunlight and rotation (Yarkovsky) and outgassing.

While it has been possible to improve the modeling of forces acting upon the system, in as much as the nature of the formation processes for the potential impactors is still poorly understood, the initial start state of potential impactors is usually infered from their current state, leaving the definition of the start state arbitrary to too large a degree.

Generally, the solutions for impacts for any planet will be bound by the elliptical toroid defined by a planets orbit around the Sun. In as much as impacts can effect a planets orbit, that toroid can change, and while known, due to complexity this is never set as a start state.

For example, in the case of the Earth, the formation of the Earth-Moon has only recently been modeled adequately, and to my knowledge the start state for impact models is never set before the formation of the Moon.

In the case of Mars, there is the Hellas impact: this must have had a considerable effect on the orbit of Mars around the Sun over time. Another problem has to be wiggle and wobble, as Mars assumes a new shape following the impact. The elliptical toroid changes. I don't believe any program accounts for these.

Another item coming up for modeling is that the relationship between impacts and volcanism, which has only recently started to be considered in any detail. For these studies, the comparison of Mars with the Earth should be of great help in defining those mechanisms, particularly the case of Hellas and Olympus.

Finally, the study of global blast effects has only recently started to come under detailed study, and again in this case data from Mars should prove useful in sorting out the problem.

EP