[CP2K-user] [CP2K:17054] TDDFPT using PBE0 in version 9.1 versus in 4.1

[Anna Hehn]

Dear Maria

the choice of GGA exchange functionals for the first-order correction term
of ADMM is restricted. Rerunning your input choosing EXCH_CORRECTION_FUNC
PBEX in the ADMM section however works fine for me.
Recommendations regarding the cutoff radius for exchange are the same as
before; standardly half of the smallest cell vector is a good start.

Best regards

Anna

Am Do., 26. Mai 2022 um 16:01 Uhr schrieb Maria Fumanal <mfumanal at gmail.com
>:

> Hello,
>
> If I add the truncation section to the previous input that works well
> (pbe0_jh.inp/out) as:
>
> &FORCE_EVAL
> METHOD Quickstep
> &PROPERTIES
> &TDDFPT
> KERNEL FULL
> NSTATES 5
> MAX_ITER 50
> CONVERGENCE [eV] 1.0e-7
> &END TDDFPT
> &END PROPERTIES
> &DFT
> &QS
> METHOD GPW
> &END QS
> &AUXILIARY_DENSITY_MATRIX_METHOD
> METHOD basis_projection
> ADMM_PURIFICATION_METHOD none
> &END
> &XC
> &XC_FUNCTIONAL PBE0
> &END
> &HF
> &INTERACTION_POTENTIAL
> POTENTIAL_TYPE TRUNCATED
> CUTOFF_RADIUS 5.0
> T_C_G_DATA t_c_g.dat
> &END INTERACTION_POTENTIAL
> &END HF
> &END XC
>
> it complains with the "derivatives bigger than 1 do not work correctly"
> error (see attached pbe0_jh_hf.inp/out)
>
> I understood that for periodic systems it is necessary to include this
> truncation of the HF with cutoff radius half of the smallest cell vector...
> Is this different in this version?
>
> Many thanks,
>
> Maria
>
> El jueves, 26 de mayo de 2022 a las 13:02:34 UTC+2, jgh escribió:
>
>> Hi
>>
>> you can specify a truncated potential:
>>
>> &INTERACTION_POTENTIAL
>> POTENTIAL_TYPE TRUNCATED
>> CUTOFF_RADIUS 5.0
>> T_C_G_DATA t_c_g.dat
>> &END
>>
>> The long range correction only makes a significant contribution for very
>> small
>> cutoff radius. I would not use it.
>>
>> regards
>>
>> Juerg Hutter
>>
>> ________________________________________
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Maria
>> Fumanal <mfum... at gmail.com>
>> Sent: Thursday, May 26, 2022 10:59 AM
>> To: cp2k
>> Subject: Re: [CP2K:17041] TDDFPT using PBE0 in version 9.1 versus in 4.1
>>
>> Thanks a lot for your answer, this input works well with 9.1 and gives
>> (almost) same results than with the 4.1 version
>>
>> I see however that now HF truncation is no longer specified, is it still
>> considered though?
>> Another question I have is if it would be possible to use Longe Range
>> correction and how to specify it?
>>
>> Many thanks,
>>
>> El miércoles, 25 de mayo de 2022 a las 14:15:15 UTC+2, jgh escribió:
>> Hi
>>
>> this input should work with the current Trunk version of CP2K. You can
>> find many
>> examples in tests/QS/regtests-tddfpt-*
>> Please note that we recently fixed a bug in the HFX-tddft code.
>>
>> best, Juerg
>>
>> &FORCE_EVAL
>> METHOD Quickstep
>> &PROPERTIES
>> &TDDFPT
>> KERNEL FULL
>> NSTATES 5
>> MAX_ITER 50
>> CONVERGENCE [eV] 1.0e-7
>> &END TDDFPT
>> &END PROPERTIES
>> &DFT
>> &QS
>> METHOD GPW
>> &END QS
>> &AUXILIARY_DENSITY_MATRIX_METHOD
>> METHOD basis_projection
>> ADMM_PURIFICATION_METHOD none
>> &END
>> &XC
>> &XC_FUNCTIONAL PBE0
>> &END
>> &END XC
>> ........
>>
>> ________________________________________
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Maria
>> Fumanal <mfum... at gmail.com>
>> Sent: Tuesday, May 24, 2022 5:22 PM
>> To: cp2k
>> Subject: [CP2K:17029] TDDFPT using PBE0 in version 9.1 versus in 4.1
>>
>> Dear CP2K community,
>>
>> I wanted to reproduce the TDDFPT results of a small system obtained with
>> PBE0 with CP2K 4.1 and 9.1,
>> but I have seen that the TDDFPT input section has changed a lot and I do
>> not manage to make it work to obtain the same results.
>>
>> with CP2K 4.1 the simplest TDDFPT input I used to use is (see
>> pbe0_41.inp/out):
>>
>> &TDDFPT
>> NSTATES 3
>> MAX_ITER 2000
>> CONVERGENCE [eV] 1.0e-4
>> &MGRID
>> CUTOFF 200
>> &END
>> &END TDDFPT
>>
>> This above does not work with the 9.1 version anymore.
>> The only way I have manage to make it work without failing is specifying
>> the &XC subsection this way:
>>
>> &TDDFPT
>> &XC
>> &XC_FUNCTIONAL
>> &PBE
>> SCALE_X 0.75
>> SCALE_C 1.0
>> &END PBE
>> &END XC_FUNCTIONAL
>> &END XC
>> NSTATES 3
>> MAX_ITER 2000
>> CONVERGENCE [eV] 1.0e-4
>> &MGRID
>> CUTOFF 200 # separate cutoff for TDDFPT calc
>> &END
>> &END TDDFPT
>>
>> It works but the TDDFPT results I obtained are very far from the ones I
>> get with the older version. (see pbe0_91_XC.inp/out)
>>
>> If I do not specify this &XC section it fails (see pbe0_91.inp/out, error
>> "derivatives bigger than 1 do not work correctly ")
>>
>> If I specify the same &XC settings than for the Ground State including
>> the &HF section, &vdW_POTENTIAL, etc it fails (see pbe0_91_XC_HF.inp/out,
>> error "ADMM is not implemented for a TDDFT kernel XC-functional which is
>> different from the one used for the ground-state calculation. A
>> ground-state 'admm_env' cannot be reused.)
>>
>> Does anyone know why this is happening and how to solve it?
>>
>> Thanks a lot in advance,
>>
>> Maria
>>
>> --
>> 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... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
>>
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/a07282d7-2ccf-4847-9213-2ff78a3a09bdn%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/a07282d7-2ccf-4847-9213-2ff78a3a09bdn%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>
>>
>> --
>> 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... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
>>
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/376870b1-e57d-4984-b755-73d05732634dn%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/376870b1-e57d-4984-b755-73d05732634dn%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>
>>
> --
> 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+unsubscribe at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/0cb361af-307a-4216-ab94-038d6b0de624n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/0cb361af-307a-4216-ab94-038d6b0de624n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
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+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CACM8JBnLeKFUqOhy9v1WgVvEeC0Q0P9cbiq9pM5WsP_Co9XSjA%40mail.gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220527/5d44a14f/attachment-0001.htm>