[molpro-user] problem with cc-pVnz-PP basis set

Kirk Peterson kipeters at wsu.edu
Fri Sep 2 21:38:54 BST 2005

Dear Peng,

I am sure it is not a problem of balance with the basis set, and I've 
never seen a symptom like this due to a poor choice of basis.  Whenever 
this has happened to me it is always due to a problem with the 
underlying HF orbitals (perhaps converged to wrong state?).  Of course, 
it could also be due to which electrons you are correlating in the 
QCISD calculation.  These basis sets were not designed for correlating 
the 3d electrons of bromine, however I would be very surprised to get 
this particular effect even if you were.  Are you sure the ECP (and 
corresponding new nuclear charge) is correctly picked up by the program 
when you specify the basis?

best regards,


On Sep 2, 2005, at 1:26 PM, Peng Zou wrote:

> Dear Molpro users,
>  We met this UNREASONABLE NORM. CAL. STOPPED problem during energy 
> calculation of a bromine compound using QCISD(T) with ECP basis set 
> (cc-pvnz-pp developed by Dr. Kirk Peterson and coworks). I tried to 
> search the email archives and found that it could be the basis set may 
> be so unbalanced on bromine. However, the HF calculation is fine with 
> cc-pvnz-pp basis sets. And I've tried to use smaller basis set (3-21G, 
> and 6-311G) as starting point, and read in the orbitals for later ECP 
> basis sets calculation. It failed. Part of the output is attached 
> below. The Norm. parameter seems to have a huge jump on Iter. 3. The 
> change is actually very big at Iter. 2. Any suggestions are highly 
> appreciated.
>  Best regards,
> Peng
>   DEN1      VAR(S)    VAR(P)  DIIS     TIME
>   1      1.27517057    -0.39536036  -492.68948796    -0.39536036    
> -0.02172057  0.17D-02  0.63D-01  0  0     2.23
>   2      2.55280534    -0.41495025  -492.70907784    -0.01958988    
> -0.07537923  0.42D-03  0.74D+00  0  0     4.03
>   3     18.33412068    -0.43777980  -492.73190740    -0.02282955    
> -0.87898236  0.62D-03  0.89D+01  1  1     5.83

More information about the Molpro-user mailing list