[CP2K:3475] Basis set confusion

hut... at pci.uzh.ch hut... at pci.uzh.ch
Fri Sep 2 17:07:20 UTC 2011


>Question 1: When I use the DZVP-GTH sets, it works nicely. When I try
>a better basis set (QZV2P-GTH), the SCF no longer converges. Why is
>this and how do I resolve it?

This might be due to the change in condition number for the overlap
matrix. Try a better preconditioner (e.g. FULL_ALL), use OT with
conjugate gradient and line search if DIIS has problems.

>Question 2: What are the augmented basis sets? Would I use them /
>instead/ of the normal basis set? When should I use the augmented
>basis sets?

Augmented normally refers to additional difuse functions of s,p.. type.
They are needed for example for negative ions or weakly bound complexes.

>Question 3: Why are some elements not available? Their
>pseudopotentials are in the POTENTIALS file, but there is no GTH basis
>set. For example, Titanium. It is available only as DZV in the tests/
>QS/BASIS_SET file. Is this sufficient?

Not all types of basis sets have been produced for all types of pseudos.
Basis sets should be balanced (of same quality on all atoms) and 
of course sufficiently accurate for your problem. This needs testing
and experience.



More information about the CP2K-user mailing list