[gmx-users] initial velocities

[James Rustad]

I am trying to integrate forward and backward in time from the same point.
When I print out the initial velocities at <<zero>> time (either read
in on the input file or generated with the same seed)  they are not exactly
the same for dt=-0.001 as for dt = +0.001 (as set in the mdp file).
Is there a way to strictly enforce the velocities to be the same?