CCL:G: Discrepancies between Jaguar and Gaussian

Dear Ken,
as far as i am aware of differ both programs in the contraction schemes for certain basis sets of the transition metals.
Both manuals refer to the papers of Hay en Wadt, but the implementation is slightly different.
----- Original Message -----
Sent: Tuesday, February 06, 2007 5:07 PM
Subject: CCL:G: Discrepancies between Jaguar and Gaussian

We are testing and comparing outputs obtained from Jaguar and Gaussian03 suites.  We have determined that the geometries obtained by optimizing with either program are almost identical.  However, the energies differ significantly. 


The things we have done so far to try to narrow down the reason for the discrepancies in the energies are:

1)       Compared different methods (i.e., HF vs B3LYP) and regardless of the method the energy difference remains suggesting that the error is not dependent on DFT or ab initio methods

2)       Compared basis sets (i.e., 6-31G(d), 6-311+G(3df,3pd)) although Jaguar uses pseudospectral method on the smaller basis set it uses analytic for the larger basis set.  Suggesting that the error is not dependent on either analytic or pseudospectral methods.

3)       Compared the definition of the basis sets (i.e., comparing basis set libraries) and these are identical between Jaguar and Gaussian

4)       Compared using 6D and 5D orbitals and still the difference remains.


Any assistance in determining the source of the difference in energies would be greatly appreaciated.


Ken Hunter