Key points are not available for this paper at this time.
Ideas from the theory of dynamical systems are applied to a collection of N identical point masses interacting via one-dimensional ``gravity, '' with a two-body force F=-G sgn (x₀-x₁). This Hamiltonian system is viewed as a geodesic flow on a curved, but conformally flat, N-dimensional space, the curvature of which exhibits certain interesting features. In the large-N limit, the Ricci tensor can be written approximately in the form R₀^b=-₀^b+B₀^b, where is a smooth, negative-definite quantity which, oftentimes, can be viewed as nearly constant, and B₀^b is a positive sum of functions of codimension 1. The curvature K (u, ) ==R₀₁₂₃u^bu^dx^ax^c associated with a small change x^a in the N-velocity u^a of the system is not negative definite. However, in the large-N limit, the smooth piece of K associated with a ``generic'' change x^a of a generic u^a is in fact negative, the fraction of u^a's and x^a's leading to a positive K decreasing at least as fast as N^-1. If the -function contributions to K could be ignored, as they can for D-dimensional ``gravity'' with D2, this would imply an average ``mixing'' very much akin to the astrophysicists' ``violent relaxation'' on a time scale given essentially as the free-fall time for the system. This conclusion fails, however, because of the singular contributions to K reflecting physical collisions which, for one-dimensional gravity, cannot be ignored. The similarities and differences between one- and three-dimensional ``gravitational'' systems are discussed, and it is argued that numerical simulations of one-dimensional systems need say little about real three-dimensional systems.
Henry E. Kandrup (Fri,) studied this question.