Hi Albert,
The labels on the graph are indeed correct!
You're looking at the two peaks, which correspond to the high-energy,
unimportant structures that have been weighted as lower priority in the
fit. Focus instead on the correspondence between the Amber energies (red or
blue) and the QM data (black) for the other structures and you'll see that
the fitting does greatly improve the match between QM and Amber energy.
If you look at the graph in the previous section, you'll see how the high
energy structures that are unlikely to be accessible in simulation will
excessively bias the fit, resulting in poor correspondence with QM energy
for all of the structures.
I'd recommend carefully reading section 6 for a more thorough explanation
of why making the fit worse for those two structures isn't a problem.
Best,
Robin
On Fri, Jul 29, 2016 at 12:50 PM, Albert <mailmd2011.gmail.com> wrote:
> Hello:
>
> I am reading the Paramfit tutorial:
>
> http://ambermd.org/tutorials/advanced/tutorial23/paramfit_6.html
>
> I notice that there are not so much improvement when we compare the
> initial Amber (blue line) and the fit Amber (red line). Both of them are
> far far away from the QM results (black line). Probably there is some
> problem of the figure legend, and the black should be "initial Amber"
> whereas blue and red should be "QM" and "fit Amber"?
>
> Thanks a lot
>
> Albert
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jul 29 2016 - 13:30:03 PDT
