Friday, February 12, 2016

[mfwjsbox] Long term solar system simulation

Create a computational model of the solar system (or similar collection of orbiting planets) which is accurate over hundreds of millions of years.  In particular, model how orbits of planets shift gradually (or suddenly).  Assume a realm of integration that is not chaotic: things are not highly dependent on initial conditions.  Main problem is to avoid a systematic error introduced by the integration technique, which could drown out the accuracy of a simulation when compounded over long time scales.

Inspired by the Nice model and the theory that Neptune and Uranus have switched places, so such models must have already been constructed: probably Kepler with perturbations.

No comments :