Which auxiliary basis set should I use for ADMM calculations?

[Torstein Fjermestad]

Dear all, 

I am doing a PBE0 calculation. 

For the primary basis set, I use TZVP-MOLOPT-PBE0-GTH; for the pseudopotential, I use GTH-PBE0; and for the auxiliary basis set, I use FIT3. 

Using Si as an example, the exact numbers are the following:

PRIMARY BASIS SET

# Si [12s12p4d|3s3p1d]
# SOURCE: https://github.com/cp2k/cp2k/raw/ac0226eb549c7ef1ea50d0597d545f29d4c8fc87/data/BASIS_MOLOPT_UZH#L5260
Si  TZVP-MOLOPT-PBE0-GTH-q4 TZVP-MOLOPT-HYB-GTH-q4
1
2 0 2 4 3 3 1
  1.213658630823 -0.030815174945 -0.634072471864 -0.542641279521  0.136162907866  0.066271452739  0.848730515959  0.278715935685
  0.428809571621  0.005952863069  0.570670193748  0.150997228423 -0.777588638886 -0.496367537858  0.304840077053  0.788029670430
  0.165906813294 -0.876412895115  0.270751967113 -0.381705514748 -0.613274545168  0.667988337027 -0.181028440213  0.547416077765
  0.042600429814 -0.480536601823 -0.446072867026  0.732830934650  0.026644013324  0.550471564668  0.392374492719  0.040771356636

PSEUDOPOTENTIAL

# Si [10|2s2p]
# SOURCE: https://github.com/cp2k/cp2k/raw/ac0226eb549c7ef1ea50d0597d545f29d4c8fc87/data/POTENTIAL_UZH#L6300
Si  GTH-PBE0-q4 GTH-HYB-q4
2 2 0 0
      0.44599315   2    -6.14101714     0.02439420
     2
      0.43467858   2     8.97165588    -2.70628598
                                        3.49478023
      0.49725736   1     2.43351142

AUXILIARY BASIS SET

Si FIT3
 6
 1  0  0  1  1
  0.09648 1.0
 1  0  0  1  1
  0.24696 1.0
 1  0  0  1  1
  1.27756 1.0
 1  1  1  1  1
  0.08949 1.0
 1  1  1  1  1
  0.29563 1.0
 1  1  1  1  1
  2.20247 1.0

I have understood that I should use a primary basis set and a pseudopotential that have been generated considering the same number of valence electrons. In the example above, this is indicated with the -q4 label. 

However, shouldn't the same rule also apply when choosing the auxiliary basis set?

The FIT3 is from the file BASIS_ADMM of CP2k version 2023.1, but there is no q label attached to the names of the auxiliary basis sets. 

Given the atom type, can the auxiliary basis sets be used with any pseudopotential?

Thanks for your help.

Best regards,

Torstein

[Jürg Hutter]

Hi

Yes, all basis sets used in a calculation, the primary orbital basis, RI (aux) basis sets, and ADMM basis sets should be consistent with the pseudopotential used.
The -qX nomenclature was introduced to make it easier to spot inconsistencies in the choice of basis sets. However, some older libraries don't have these labels and it is up to the user to check for consistency.

You can find some newer ADMM basis sets (with the -qX label) in BASIS_ADMM_UZH.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Torstein Fjermestad <tfjer...@gmail.com>
Sent: Thursday, November 6, 2025 7:39 PM
To: cp2k
Subject: [CP2K:21955] Which auxiliary basis set should I use for ADMM calculations?

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/26c8c114-f50c-4c1f-85e2-4f811c89fc49n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/26c8c114-f50c-4c1f-85e2-4f811c89fc49n%40googlegroups.com?utm_medium=email&utm_source=footer>.