VMD-L Mailing List

From: Daniel Fellner (dfel694_at_aucklanduni.ac.nz)
Date: Sun Sep 27 2020 - 20:15:46 CDT

Hi everyone, I'm encountering some weird results when optimising charges
for two unrelated indole compounds (other scaffolds have been relatively
straightforward compared to these):

I need to increase the water shift distances significantly more than other
compounds (up to +2.4 in this case, where most other compounds were fine
with <1.0) to reduce the objective value numbers, and I notice in the
output file that the difference between the MM and QM distances is always
positive and quite large (ranging from 0.4 to 2.3, averaging 1.6 – 1.8).
The energies seem to converge well, but the objective function for the
distance is throwing huge numbers in comparison (e.g. 64.49 for energy v.s.
1092.45 for distance *even with distance weight set at 0.5*). Setting the
water shift distances to an intermediate range (i.e. 1.4 – 1.6) makes the
objective value contributions from energy and distance more equal, but I
notice many of the interactions hit their maximum distance and the overall
objective value is no better.

Should I ignore the huge distance objective value numbers and just focus on
the energy? Or reduce the distance weight lower than 0.5? Or is there
something I'm missing? I've looked at the log files and the waters seem to
settle at reasonable distances. The few that don't I've excluded from the
QM data.

*Daniel Fellner BSc(Hons)*
PhD Candidate
School of Chemical Sciences
University of Auckland
Ph +64211605326