Firefly and PC GAMESS-related discussion club


 
Learn how to ask questions correctly  
 
 
We are NATO-free zone
 



Re^3: Any experience with cc-pV5Z without h-functions for 2nd row elements?

Thomas Pijper
thomaspijper@hotmail.com


Hi Slawomir,

>Your experience with pc-4 is interesting. I am not very happy with the entire pc-x family of basis sets either. They seem to converge slower to the CBS limit than cc-pVxZ or xZVPD and HF calculations with the augmented sets are nearly impossible to converge beyond aug-pc-2.

With RHF and the pc set, I've seen cases where the energy converged slower towards the CBS (compared to cc-pVxZ) while the geometry converged faster. The set seems to have its quirks. I do use it occasionally though.

aug-pc-x can indeed be though to converge, though aug-cc-pVxZ can in my experience pose problems as well. xZVPD always converges much better, but then again it is not a full augmentation (and it is not designed for general purpose use). :-)


Kind regards,
Thom


On Thu Nov 28 '13 11:33pm, Slawomir Janicki wrote
-------------------------------------------------
>Hi Thom,

>Thanks for the feedback on cc-pV(5-h)Z. It looks like I would need to test it before making any commitments.

>Your experience with pc-4 is interesting. I am not very happy with the entire pc-x family of basis sets either. They seem to converge slower to the CBS limit than cc-pVxZ or xZVPD and HF calculations with the augmented sets are nearly impossible to converge beyond aug-pc-2.

>All the best,
>Slawomir
>
>
>On Thu Nov 28 '13 10:36pm, Thomas Pijper wrote
>----------------------------------------------
>>Hi Slawomir,

>>My experience with cc-pV(5-h)Z is limited, but in the cases where I have used it I found it always yields energies and geometries closer to the complete basis set than cc-pVQZ. On the other hand, I found its use limited due to the amount of basis functions involved, even with h functions removed.

>>Experience with the full cc-pV5Z sets I do not have. I recall Frank Jensen writing that high angular momentum functions are important for recovering electron correlation in post-Hartree-Fock methods, so it might perform well with those. If someone has used these sets I too would like to learn about their experiences.

>>By the way, I've on occasion used pc-4 with h functions removed. I found it to behave erratically in its convergence towards the CBS, though this is based on a very small amount of results.
>>
>>
>>Kind regards,
>>Thom
>>
>>
>>
>>On Thu Nov 28 '13 7:53pm, Slawomir Janicki wrote
>>------------------------------------------------
>>>Does anyone have any experience using cc-pV5Z without h-functions for 2nd row elements? I was thinking about putting together a Firefly basis set library like that.

>>>Apparently GAMESS-US was ignoring the h-functions until release 2013R1 rather than stopping like Firefly. I wonder how do the results compare to the full cc-pV5Z basis set.

>>>Going to check the turkey in the oven...

>>>Cheers,
>>>Slawomir


[ Previous ] [ Next ] [ Index ]           Fri Nov 29 '13 0:05am
[ Reply ] [ Edit ] [ Delete ]           This message read 542 times