We would like to monitor trajectories by segmenting long MD runs
into short sections and sampling the positions and velocities.
We do this by a series of “run pre no” commands.
The problem is that we find that the sampling seems to alter the trajectory.
We see this by comparing two MD runs. The first run is sampled every 100th
time step. The second run is sampled every 10th step.
The positions of second run ought to compare well with those from the first
run where the time is the same, but we find that this is not the case, and that
the difference is on the order of 8-9 decimal places, which is too high for what
we are trying to do.