Hi Dave,
> It turns out that the default behavior for sqm has gradients that are
> not
> accurate enough. If you set "grms_tol=0.0001, tight_p_conv=1,
> scfconv=1.d-10," in the namelist input for sqm, you then get good
> behavior
> with the TCNG minimizer, as expected, and as we see with MM gradients
> (which
> are essentially exact.)
I think I suggested this several months ago but we never acted on it. Right
now the default settings are what is generally good for MD. You can conserve
energy reasonably well with the defaults although if I want to run a good
NVE simulation I normally turn tight_p_conv on and set SCF_CONV=1.0D-08 at
the same time I change dsum_tol and the shake tolerance. It sounds like we
need this at least for minimization though.
The reason it is not this tight for MD by default is because during MD you
can sometimes get convergence problems when you get strained structures and
the convergence tolerance is set this tight. This can cause all sorts of
problems not only destroying performance. Hence why I did not have it this
tight normally. What is set for the default right now is equivalent to what
Dynamo and Divcon use during MD.
Thus we should probably increase it for SQM by default, I would suggest
trying 10D-8 and tight_p_conv as the default. 10D-10 might be too tight for
some poor starting structures. Or we write some code that tightens up the
convergence criteria as the structure optimizes which might be a good idea.
> I can/will change antechamber to set stricter parameters to get AM1-BCC
> charges...we'll see if this helps reduce the dependence on compiler.
It might do. I never checked this but it would make sense that it could.
> But this also leaves open the question of what the defaults should be
> for
> sqm as used in QM/MM molecular dynamics. I don't know how much testing
> was
> done, but it is of concern that the default for tight_p_conv is zero,
> since
> Istvan's analyis shows that this is dangerous, at least for
> minimization.
> I'm not changing things right now, since that would probably involve
> updating
> every one of 100 or so QM/MM test cases (sigh), but it is something
> that
> I'd like to get comments on.
I have a script for updating these test cases automagically if you want to
change things but of course with the caveats above.
All the best
Ross
/\
\/
|\oss Walker
| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
|
http://www.rosswalker.co.uk |
http://www.wmd-lab.org/ |
Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Dec 17 2009 - 13:00:02 PST
