On Fri, Oct 13, 2017, William Marquardt wrote:
>
> I've run a short simulation of azide anion in vacuo, and am having a bear
> of a time generating force vectors that match what the .mdfrc output reads.
> dt=0.001, and I am printing forces and coordinates each timestep for the
> duration of this short NVE simulation.
A key possible complication here (and I don't know the answer) is knowing
exactly which coordinates were used to compute the forces that are being
dumped out. Because Amber uses a leap-frog like code (although the actual
propagation is velocity-Verlet), there is a non-zero chance that the forces
being dumped are one step out of sync with the coordinates.
If you are sure that your bending calculations are correct (and don't need to,
say, try just a diatomic molecule as an even simpler test case), see if using
the coordinates on step n-1 might match the forces that are dumped.
Someone on the list might actually know the answer, so being a little patient
might help. Also, Dan might (as per usual) be able to point you to a cpptraj
command that calculates the forces from input coordinates, and hence removes
this potential source of problems.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Oct 13 2017 - 14:00:03 PDT
