[CP2K-user] [CP2K:12701] optimize RI basis

[Katarina S.]

Dear Juerg,
thank you for your answer. I played a bit with the suggested method, and 
with optimization of auxillary basis via AutoAux procedure and I obtained 
the same results as with my original combination of RI-cc-QZ/cc-TZ basis 
set. The difference in adsorption eneriges between these two approaches was 
around 1 kJ/mol. Thus, no matter what I do, my BSSE is still around 15-20 
kJ/mol. Is there any way how I can do better? Because if I cannot bring 
BSSE down, RPA is for me not usable at all... (it is for benchmark 
purposes). Also, I am somehow surprised that BSSE is at PBE level around 
2kJ/mol but for RPA it is almost 10x more. Am I doing a mistake somewhere?

Attached I put two example input files - standard BSSE input file for 
RI-cc-QZ/cc-TZ procedure and my input for AutoAux procedure.

Thank you for any suggestions 
Kind regards,
Katarina

Dňa piatok, 3. januára 2020 11:47:29 UTC+1 jgh napísal(a):
>
> Hi 
>
> the easy way out here is to use the automatic RI basis option. 
> See e.g. tests/QS/regtest-ri-mp2/RI_MP2_CH3_auto.inp 
> or 
> CP2K_INPUT / FORCE_EVAL / DFT / AUTO_BASIS 
>
> This generates slightly bigger RI basis sets (see also the paper 
> cited for more information). Tests will still be needed as not all 
> combinations of PP and Basis sets have been considered. 
>
> You cannot use all electron basis sets with pseudopotentials. 
>
> regards 
>
> Juerg 
> -------------------------------------------------------------- 
> Juerg Hutter                         Phone : ++41 44 635 4491 
> Institut für Chemie C                FAX   : ++41 44 635 6838 
> Universität Zürich                   E-mail: h... at chem.uzh.ch 
> <javascript:> 
> Winterthurerstrasse 190 
> CH-8057 Zürich, Switzerland 
> --------------------------------------------------------------- 
>
> -----c... at googlegroups.com <javascript:> wrote: ----- 
> To: "cp2k" <c... at googlegroups.com <javascript:>> 
> From: "Katarina S." 
> Sent by: c... at googlegroups.com <javascript:> 
> Date: 01/01/2020 04:35PM 
> Subject: [CP2K:12701] optimize RI basis 
>
> Dear all, 
> I am studying a binding of water in aluminosilicate systems using RPA. 
> However, available basis set (cc-QZ for H and O and cc-TZ) gives too big 
> BSSE - it is between 15-20 kJ/mol. BSSE on PBE level is small (< 5 kJ/mol) 
> so I assume it is due to auxiliary basis. Therefore, I would like to 
> re-optimize RI basis sets for Al and Si atoms. I had look on regtests: 
> cp2k-3.0/tests/QS/regtest-ri-mp2/opt_basis_O_auto_gen.inp.  and 
>  opt_basis_O_num_func.inp and they run smoothly. I looked also on the 
> tutorial given in 
> https://www.cp2k.org/_media/events:2015_cecam_tutorial:wilhelm_posthf.pdf 
> However, I am still not sure how to continue with my system. I have 
> several questions: 
> 1) Because I will run RPA, shall I turn on  the PBE functional in XC 
> section for basis set optimization? 
> 2) If I optimize basis set for a simple system such as Al atom to which 
> extend are the parameters reliable/transferable? 
> 3) Alternatively I looked on BASIS_def2_QZVP_RI_ALL. Is it possible to 
> combine all electron basis with pseudopotentials?  I run the calculations 
> in combination with GTH potentials and I see that the results are very 
> different than when I use cc-QZ basis (adsorption energy of water differs 
> about 20-30 kJ/mol), so I do not trust them much. 
>
>  Thank you for your help! 
> Kind regards, 
> Katarina 
>   
>   -- 
>  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 c... at googlegroups.com <javascript:>. 
>  To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/1e8ffd8d-716f-490b-a6be-9116b9781098%40googlegroups.com. 
>
>   
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200111/8bf8fbbc/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rpa_H2O_AutoAux.inp
Type: chemical/x-gamess-input
Size: 2276 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200111/8bf8fbbc/attachment.inp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rpa_f.1+H2O_bsse.inp
Type: chemical/x-gamess-input
Size: 3132 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200111/8bf8fbbc/attachment-0001.inp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: BASIS_RI_cc-TZ_modified
Type: application/octet-stream
Size: 20112 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200111/8bf8fbbc/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: f.1+H2O_final_cc-QZ.xyz
Type: chemical/x-xyz
Size: 2649 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200111/8bf8fbbc/attachment.xyz>