Choice of RI basis set fo RPA calculation of argon dimer

[Riccardo Pezzetta]

Dear CP2K community,

I am trying to calculate the binding energy of the argon dimer using RPA. I used the basis set TZV2P-MOLOPT-PBE-GTH-q8 from the file BASIS_MOLOPT_UZH with auxiliary basis set RI_85_7_6_5_5_0_0_0_8.0e-08 from the file BASIS_RI_MOLOPT:

BASIS_SET TZV2P-MOLOPT-PBE-GTH-q8

BASIS_SET RI_AUX RI_85_7_6_5_5_0_0_0_8.0e-08

With this basis I get -17.9 meV as binding energy. Then I tried to remove RI_AUX in front of the auxiliary basis set:
BASIS_SET TZV2P-MOLOPT-PBE-GTH-q8

BASIS_SET RI_85_7_6_5_5_0_0_0_8.0e-08

With these parameters i read in the output file:
Orbital Basis Set                  RI_85_7_6_5_5_0_0_0_8.0e-08TZV2P-MOLOPT-

and

RI Auxiliary Basis Set             Ar-RI-AUX-RI_85_7_6_5_5_0_0_0_8.0e-08TZV

It seems that the program mixes the two basis sets, but I don't know how. The result is -10.6 meV which is closer to the experimental value (which is -12.3 meV) than the result I obtained writing RI_AUX in front of the auxiliary basis set. Could anyone explain me what happens with this basis mixing and tell me if the result based in this parameters is reliable? I have attached to this message the input and output files.

Thank you in advance.

Kind regards,

Riccardo Pezzetta

[Frederick Stein]

Dear Riccardo,

You need to use the structure "BASIS_SET RI_AUX <basis set name>" to specify the RI-basis sets (see https://manual.cp2k.org/trunk/methods/post_hartree_fock/mp2.html).

Best,

Frederick

[Jürg Hutter]

Hi

the syntax for defining a basis set is

BASIS "TYPE" "NAME"

If "TYPE" is missing it defaults to ORB. If there is more than one BASIS definition for ORB
the basis sets are combined (e.g. allowing to augment a standard basis with a difuse set).

Your second input therefore was combining the two basis sets to create a huge ORB basis
and then used a an automatic procedure to generate a RI basis based on this huge ORB set.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Riccardo Pezzetta <richi...@gmail.com>
Sent: Wednesday, April 1, 2026 4:09 PM
To: cp2k
Subject: [CP2K:22149] Choice of RI basis set fo RPA calculation of argon dimer

--
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/dff85d15-2f99-485d-9556-3ff9bd58c3adn%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/dff85d15-2f99-485d-9556-3ff9bd58c3adn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Riccardo Pezzetta]

Dear professor Hutter,

thank you for your answer. Since the only RI basis for argon are the ones in the file BASIS_RI_MOLOPT, would you recommend using the  TZV2P-MOLOPT-PBE-GTH-q8 basis set and  RI_AUX RI_85_7_6_5_5_0_0_0_8.0e-08 basis set or a correlation consistent basis set like ccGRB-Q-q8 and an auxiliary basis set generated with  auto_basis? With the latter choice I get a worse result, but maybe it is because I have to include the basis set superposition error?

Kind regards,

Riccardo Pezzetta

[Frederick Stein]

Dear Riccardo,

I did run calculations on Argon and Neon crystals in the past. I found that you need additional diffuse basis functions and a BSSE correction for good accuracy. For more information on the basis sets, check https://www.mdpi.com/1420-3049/25/21/5174 . For the augmentation, you can simply use the ones from the Ahlrich basis sets.

Regards,

Frederick