[Bruce Hartley]

I have used the Lagrangian solver by setting a central potential, finding its gradient to give the force and then setting particle velocities and tracing the paths. This is esentially in two dimensions. I find that when I set a radial velocity I can trace a nice circle, as expected, with a bit of numerical round-off error giving a slightly different path the second time round. When I reverse the velocity (i.e. starting with the negative of the original velocity) the system traced a circle but with a very different radius. This should not be the case. Has anyone else found this asymmetry in tracing particles using the Lagrangian solver or with the function for finding the gradient of a potential field? Thanks.