[Cincinatus]

Early runs tracked 3,000 pieces--leaving the iron core of the moon to be represented by just a single piece. Even the slightest computational imprecision could vastly overstate the iron content, in which case the computer compensated by reducing the impact angle. The result was a bias toward heavy impactors and light proto-Earths. Because Canup and Asphaug use 30,000 particles, they get by with a much smaller impactor. Everything--mass, iron, momentum--clicks into place.

Well, isn't that convienient? They just needed to simulate 30,000 particles, rather than 3,000. This -- for a process that actually involved trillions of particles.

Computer models of the "giant-impact hypothesis" for the origin of the moon are similar to computer models of global warming -- they can be tweaked to yield whatever answer you desire.

[jbstrick]

I need to renew my bubscription to SciAm. I miss the light reading...

[jlogajan]

Particle collisions are much easier to model than the weather and global warming. And it is not true that you can "tweak" any model to prove anything you want. If that were true there'd be no dilemma in the first place.

[NorthMountain]

What is your experience in finite element modeling?