VMD-L Mailing List

From: Seth Axen (seth.axen_at_gmail.com)
Date: Wed May 18 2016 - 14:04:29 CDT

Christopher,
   Thanks for the tips! I found that the optimization step had moved the
molecule into one of the local minima but not the global minimum the PES
revealed. By re-optimizing into the global minimum and repeating the
subsequent stages of analysis, I was able to fit the dihedral parameters.
   Thanks also for your development of ffTK! It's an extremely useful tool
that's really lowered the barrier to basic forcefield parameterization.
   Seth

On Tue, May 17, 2016 at 1:45 PM, Mayne, Christopher G <cmayne2_at_illinois.edu>
wrote:

> Seth,
>
> The problem that you are encountering is that the global minima are
> offset. Since the QME and MME profiles are not on the same absolute scale,
> we rescale the global minimum. Here, the global QME minimum is much lower
> than the MME minimum, causing the whole MME profile to be offset too low.
> There are a few things to look into.
>
> 1) Use the torsion explorer to inspect the conformation of the QME global
> minimum to check for anything strange going on with the structure.
> 2) Check the bond and angle parameters for something that is raising the
> MME energy
> 3) Split the optimization into two phases, one of which isolates the
> parameters associated with the QM scan that is generating the global
> minimum.
>
> Regards,
> Christopher Mayne
>
>
> On May 17, 2016, at 1:23 PM, Seth Axen <seth.axen_at_gmail.com> wrote:
>
> Hello,
> I am using ffTK to fit several dihedral parameters. For several of my
> molecules, through iterative sampling, I am able to fit the MM potential
> energy surface curve to the shape of the QM curve. Generally, the global
> minima also match in height. However, additional simulated annealing or
> downhill, regardless of the tolerance, is unable to fit the heights of the
> curves. I was hoping someone could shed some light on what might be the
> cause of this. Attached is an example.
> Seth
> <multiplot.png>
>
>
>