The ABINIT Users Mailing List ( CLOSED )

[Szieberth Denes <dino@iris.inc.bme.hu>]

On Wed, 14 Sep 2005 paul.boulanger@umontreal.ca wrote:

> Hi Denes (and Hi Mikami),
>
>                  The system considered is very interesting.

yeah, that's why I'm trying to investigate it :)


> Usually, to minimize the numbers of atoms we can replace CH3 by H only, 
> which diminishes the number of atoms by 60 in your case. This permutation 
> does not change the electronic properties too much;

Thanks, I see now that it would seriously be a reduction... a bit hard to 
forget the full electron GTO based reflex that H and C holds so much less 
electrons than the others...  for the purpose of finding the calculation 
parameters that work this substitution should be OK, I'm already doing it 
:). For the "final" purpose  (which is trying to differentiate between two 
possible structures) it may be necessary to take into account the effect 
of the methyl groups. But since I still can't achieve convergence of the 
wave function even for the smaller case it's not very pressing...

> Another point, that can be important before relaxing your system, is to 
> inspect the geometries of every site. We see that the bond between the 
> Cu and P at one of your Mo site seems a little strange, considering the 
> other sites.
> Maybe manual cleaning will help the convergence of the relaxation and
 of the SCF.

Hmm. Which one is that? I usually use cut3d/xcrysden for periodic 
structures, it  makes to see the bonds passing through the unit 
cell boundaries easier. For my eyes the 3 P-Cu bonds are similar but not 
equivalent, but since two are "chain-forming" and the third is toward a 
"side-group" it is understandable. Of course finding an error in the 
initial geometry would nicely explain the SCF convergence failure, but as 
yet I don't see this error.

cheers,
Denes