Compactifying collisions in the N-body problem

Richard Montgomery

2.00 pm-4.00 pm May 13 (MONDAY!)

383N (3rd floor lounge) at STANFORD

Triple collisions act like an essential singularity for the Newtonian three-body problem: one cannot continue the dynamics through triple collision in any reasonable way. However, one can prove that triple collisions are approached asymptotically along one of five possible limiting shapes -- the central configurations, with analogous results for collisions in the N-body problem. The basic analytical tool was invented by R. McGehee in 1974. It is a real blow up of N-tuple collision, combined with a clever rescaling of velocities and time. The result is the addition of a manifold with boundary to phase space. This boundary is called the collision manifold and the flow extends smoothly to it. The flow on the collision manifold is gradient-like: it admits a Liapanov function The collision flow helps to organize the dynamics near collision. The only rest points for the full extended flow lie on the collision manifold. They are the `central configurations' alluded to above. There are exactly five central configurations for N = 3, and these were discovered by Euler and Lagrange. For N = 4 or greater it is an open problem to show that the central configurations are finite in number. My goal is to motivate and explain McGehee's coordinates, describe its underlying geometry, and present some applications of the collision manifold methods.
Annotated references