On Sun, Jan 15, 2012, FyD wrote:
> Dear Dr Case,
>
> The problem of threshold in geometry optimization & charge
> reproducibility was defined in R.E.D. version I in 2002/2003 and
> reported at Anaheim in 2004 -
> http://oasys2.confex.com/acs/227nm/techprogram/P705829.HTM - but
> related to the Gaussian & GAMESS programs.
>
> In short, geometry optimization always use Opt=Tight in Gaussian (4
> criteria of convergence) and OPTTOL=1.0E-06 in GAMESS (2 criteria of
> convergence) - SCF convergence 10-8, while SCF convergence is set to
> 10-6 for bio-organic molecules in MEP computation. In these conditions
> the reproducibility of charge value achieved is ~1.10-4 e.
> See also the section "Limitations of the strategy presented" at:
> http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2918240/
This is interesting information, but has little to do with the question of
mopac vs. sqm. Maybe we missed something, but we were never able to get mopac
to achieve the sort of tolerances you are describing.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Jan 15 2012 - 09:00:03 PST
