I would like to understand basic concepts of the general relativity in 2+1 spacetime. As far as I know, GR predicts that such a spacetime is flat everywhere except for the point masses which create angular deficit proportional to their mass. Flatland with one point mass is like surface of cone. I imagine that when one adds other point masses the Flatland can be folded to a (convex) polyhedron (then there is the constraint on total masses, since total angular deficit is 720 degrees) (see note #1). I assume that a 2d Flatlander would not (at least locally) notice crossing the edges when moving from one face of polyhedron to another.
The problem I have with this model is that when one heavy body which defines the Flatland is set to motion, its mass must change and - more surprisingly from a local point of view - also masses of the neighboring bodies to keep the total of 720 degrees. The image shows cube with a vertex moving along edge to its middle with corresponding angular deficits.
On the other hand, I know that 2+1 gravity and motion of point masses have been considered seriously by Gott (in his two strings time machine), Caroll, Guth, t' Hooft and others. Where is error in my naive model?
Edited: Given the first answer and comments I should maybe be more precise:
Is a motion which requires change of angular deficit (and hence mass) of the surrounding point masses possible, or is possible only motion when all angular deficits are kept constant? Anyway, for a Flatender living on the polyhedron surface the situation looks like there is an interaction between the point masses, despite the fact the spacetime is flate in between them. Or is such a configuration (initial condition) simply impossible?
Edited: I have overlooked the fact that a point mass cannot be just "set to motion" by a miracle - total momentum must be conserved. I will think it over and prepare a better example.
Edited: This papers by 't Hooft may contain answer:
Notes (added in later edits):
1) Gott & Alpert: General Relativity in a (2+1)-Dimensional Space-Time (Gen. Relat. Gravit. 16:243-247, 1984):
"Consider a convex polyhedron with a finite number of faces. The faces and edges have no intrinsic curvature and represent solutions to the vacuum field equations. The vertices each have an angle deficit (like the vertex of a cone) and represent point masses. For example, a universe shaped like the surface of a cube represents a vacuum with 8 point masses of $M = \pi/2$ each (three squares meet at each vertex giving each an angle deficit of $\pi/2$). The Einstein static universe of equation (6) may be approximated by a polyhedron of many faces containing many vertices each with small angle deficits. The total mass in such a closed universe is always $M_u = 4\pi$."
In my opinion, there are also some nonconvex polyhedra which work well.