I cannot be a local optimum: moving the axle just sligthly towards the center of the disk will produce a slightly rounder, and better, shape. It seems that the

I cannot be a local optimum: moving the axle just sligthly towards the center of the disk will produce a slightly rounder, and better, shape. It seems that the optimization is botched here.