[molpro-user] (TiO)2+

[Gerald Knizia]

On Thursday 24 September 2009 09:59, Omololu wrote:
> I am trying to perform a Hartree-Fock energy calculation for the system
> mentioned above with aug-cc-pvtz for O and aug-cc-pwCVTZ-DK (downloaded
> from EMSL basis set site) for Ti. I use the "wf,28" option in MOLPRO and
> "ICHARG=2" one in GAMESS.

One note: the pwCVTZ varieties of basis sets are intended for core-valence 
correlation calculations. The -DK indicates that it is to be used in 
conjunction with Douglas-Kroll Hamiltonians (which you need to turn on 
explicitly if using them). Are you sure that it is a good idea to use one of 
these on Ti and a regular, non-DK valence-only set on O in conjunction? (I 
personally have no experience with this).

> The problem is that at large separations (say, 9 
> bohrs), I get different results between GAMESS (-920.706634 Hartrees) and
> MOLPRO (-920.4296532 Hartrees). At short separations (say, 2 bohrs) GAMESS
> & MOLPRO agree. Does anyone know if this is a pathological system? Is there 
> a way to make  MOLPRO give the lower energy?

One thing you could try is
  set,chage=2
  {rhf; start,atden}
This would prevent Molpro from using the orbitals from previous calculations 
as starting guess for RHF (and thus, most likely get a HF wave function 
rather similar to them), and start from atomic densities instead.
If this doesn't work, you can try obtaining a RHF wave function for the 4+ or 
3+ cationic species or the neutral molecule first, and use this as a starting 
guess for the +2 molecule.

> Is there a way to force the system to go to Ti4+ and O2- at long distance?

There may be a way to force that by doing separate Ti4+ and O2- calculations, 
and assembling starting orbitals from them using the orbital merge 
functionality (see manual on orbital merging). This is non-trivial however, 
so you should try other options first.
-- 
Gerald Knizia