Dear Jason,
> One of the things that make metals harder to parametrize is, as Francois
> mentioned, the fact that the spin multiplicity is often not a singlet and
> also not often obvious. When you tell Gaussian that the spin multiplicity
> is 2, Gaussian will pair up every electron except for 1. If your molecule
> is not _actually_ a doublet, then the wavefunction (and resulting electron
> density and charges) will be wrong.
I continue ;-) another difficulty is that for transition metal
complexes even when using open shell systems and DFT the wavefunction
of the obtained optimized geometry might not stable. From our tests
this appends quite frequently for multiplets. We have developed a new
algo in R.E.D. Python to solved this problem and compare energy values
for the optimized geometry and for MEP computation. This feature is
available through the MOD_GAUSSIAN_JOB = "Complex".
See the keyword in:
http://q4md-forcefieldtools.org/Help/SO4/ (MOD_GAUSSIAN_JOB="Normal"
in this case)
http://q4md-forcefieldtools.org/Help/SO4/Configuration.py
This algorithm handles a database of error messages and as the
function of the detected error messages different QM inputs are
generated insuring SCF convergence (even in complexes cases) as well
as correct optimized geometries. Obviously, this complex procedure
takes far more time than the 'Normal' one, but the idea behind it is
to run the job and let the algo. going to the right optimized
geometry. Gaussian jobs are also automatically stopped, when they
takes too much times and automatically restarted. This saves a looot o
cpu time.
regards, Francois
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Oct 18 2013 - 12:30:02 PDT
