On Fri, Mar 04, 2016, Taylor Harville wrote:
> We assigned parameters in leap using “set default PBradii
> mbondi3’” and used igb = 8. The “blow up” begins with the
> initial step where the Antechamber-parameterized phenyl ring fragments,
> sending carbon and hydrogen tens of angstroms away within a few steps.
> The energies in the heating output appear reasonable, except for the
> bonding terms where the energies look much higher than expected. The gas
> phase simulations and the explicit solvent simulations are very stable
> on the microsecond timescale, so we are confident in everything, but the
> GB parameters. As far as the energies go, the minimization energy is
> roughly 49 kJ/mol higher than the energy at step 0 of the heating.
First step is to carefully examine the mdout file for minimzation and MD, and
figure out why the starting energy from MD is not the same as the ending
energy from minimization.
Secondly, try a more general GB model, like igb=1. The igb=8 model is
specific to proteins and nucleic acids, and would require much careful
analysis to apply it to something like carbon nanotubes. The more generic GB
model is a lot more likely to be transferable. At a minimum, this would be a
good debugging step in helping to localize the problem.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Mar 04 2016 - 10:00:03 PST
