Posted by Steven Vancoillie on January 17, 2015 at 20:18:04:
In Reply to: Re: CASPT2 convergence posted by Roser on January 14, 2015 at 09:35:40:
: Can I compare results of two compounds, one obtained using the shift but the other without it?
No, both calculations would need to be done with the shift.
: I will do rassi and single aniso calculations before the caspt2. And I want to compare two compounds (Dy and Er complexes).
: The caspt2 calculations of Dy(III) complex converged without shift, but for Er complex I must use shift.
: I wanted to observe if the shift gives different results for the same compound. I repeated the caspt2 for Dy complex with shift=0.4. Results differ in energies, g-tensors... Then I think I can't compare results.
Indeed, again, you need to consistently use the same shift.
: Are convergence problems due to intruder states? Why a caspt2 calculation may not converge?
That could be the problem. The "proper" way is to increase the size of your active space. You can get an indication for the importance of certain excitations in the output of caspt2. Can you give us some more details or post your output log for Dy/Er complex?
grtz
Steven