I call
coords = dofmap.tabulate_all_coordinates(mesh).reshape(-1,gdim)
to get all coordinates ot the mesh. How is this vector distributed among the processes in a parallel MPI run? How can I scatter the coordinate vector to all processes?