[CP2K-user] [CP2K:15365] Problem converging SCF for a system containing Fe2+
Pierre-André Cazade
pierre.a... at gmail.com
Fri May 14 12:55:30 UTC 2021
Hi Salome,
Sorry, basd copy/paste. Here is a much better email.
I agree with Matt. Try OT:
&SCF
MAX_SCF 20
SCF_GUESS ATOMIC
&OT T
MINIMIZER CG
PRECONDITIONER FULL_ALL
&END OT
&OUTER_SCF T
MAX_SCF 100
&END OUTER_SCF
&END SCF
and if you still prefer to use the default solver, use Broyden Mixing
instead of P-mixing.
&SCF
SCF_GUESS ATOMIC
&MIXING
METHOD BROYDEN_MIXING
ALPHA 0.4
BETA 0.5
NBROYDEN 8
NMIXING 2
&END MIXING
&END SCF
It should be CHARGE 2 and MULTIPLICITY 0.
Try increasing the CUTOFF to 500 or 600.
Regards,
Pierre
On Fri, May 14, 2021 at 1:54 PM Pierre-André Cazade <
pierre.a... at gmail.com> wrote:
> Hi Salome,
>
> I agree with Matt. Try OT:
>
>
> and if you still prefer to use the default solver, use Broyden Mixing
> instead of P-mixing.
>
> &SCF
> SCF_GUESS ATOMIC
> &MIXING
> METHOD BROYDEN_MIXING
> ALPHA 0.4
> BETA 0.5
> NBROYDEN 8
> NMIXING 2
> &END MIXING
> &END SCF
>
> It should be CHARGE 2 and MULTIPLICITY 0.
>
> &SCF
> MAX_SCF 20
> SCF_GUESS ATOMIC
> &OT T
> MINIMIZER CG
> PRECONDITIONER FULL_ALL
> &END OT
> &OUTER_SCF T
> MAX_SCF 100
> &END OUTER_SCF
> &END SCF
>
> Try increasing the CUTOFF to 500 or 600.
>
> Regards,
> Pierre
>
> On Fri, May 14, 2021 at 1:29 PM Matt W <mattwa... at gmail.com> wrote:
>
>> Reduce the ALPHA value in &SCF / &MIXING
>>
>>
>> https://manual.cp2k.org/cp2k-8_1-branch/CP2K_INPUT/FORCE_EVAL/DFT/SCF/MIXING.html
>>
>> the default is much to large and causes the oscillations.
>>
>> OT is often a faster / more reliable solver for non-metals.
>>
>> Matt
>>
>> On Friday, May 14, 2021 at 12:37:44 PM UTC+1 salo... at gmail.com
>> wrote:
>>
>>> Hi Pierre,
>>>
>>> Thank you so much for your answer,
>>>
>>> I have added all your suggestions:
>>> - I am using DZVP-MOLOPT-SR-GTH for all atoms.
>>> - I have added CHARGE 0 and MULTIPLICITY 2 ( I am not completely sure
>>> about this but I have 55 atoms.).
>>> - I have expanded the Cutoff as follows:
>>>
>>> &MGRID
>>>
>>> CUTOFF 400
>>>
>>> REL_CUTOFF 60
>>>
>>> NGRIDS 5
>>>
>>> &END MGRID
>>> I still have the same problem. I am worried because the Convergence of
>>> the SCF optimisation up instead of decreasing.
>>>
>>> Step Update method Time Convergence Total energy
>>> Change
>>>
>>>
>>> ------------------------------------------------------------------------------
>>>
>>> 1 P_Mix/Diag. 0.40E+00 25.1 1.97116897 -352.7972578538
>>> -3.53E+02
>>>
>>> 2 P_Mix/Diag. 0.40E+00 22.1 1.15920278 -353.8713026582
>>> -1.07E+00
>>>
>>> 3 P_Mix/Diag. 0.40E+00 22.5 0.70468277 -354.5020638040
>>> -6.31E-01
>>>
>>> 4 P_Mix/Diag. 0.40E+00 22.5 1.81519648 -354.7042785765
>>> -2.02E-01
>>>
>>> 5 P_Mix/Diag. 0.40E+00 22.6 1.40667909 -353.4830224789
>>> 1.22E+00
>>>
>>> 6 P_Mix/Diag. 0.40E+00 22.6 1.61367789 -352.6169777702
>>> 8.66E-01
>>>
>>> 7 P_Mix/Diag. 0.40E+00 22.8 9.77791081 -349.7222190083
>>> 2.89E+00
>>>
>>> 8 P_Mix/Diag. 0.40E+00 22.7 3.04606642 -338.8207963019
>>> 1.09E+01
>>>
>>> 9 P_Mix/Diag. 0.40E+00 22.9 10.99358257 -288.8580020934
>>> 5.00E+01
>>>
>>> 10 P_Mix/Diag. 0.40E+00 22.9 5.98455238 -271.1871548265
>>> 1.77E+01
>>>
>>> ...
>>> And after this remains between 4 and 5 for a long time.
>>>
>>>
>>> Thanks,
>>>
>>> Salome
>>>
>>> On Friday, 14 May 2021 at 13:03:25 UTC+2 pier... at gmail.com wrote:
>>>
>>>> Hi Salome,
>>>>
>>>>
>>>>
>>>> Two things come to my mind, first you should specify the charge and
>>>> multiplicity of your system. Then, the cutoff and rel_cutoff might be too
>>>> small.
>>>>
>>>>
>>>>
>>>> Actually, there’s a third: why mixing MOLOPT and MOLOPT-SR? Just go
>>>> with the latter, they usually work fine.
>>>>
>>>>
>>>>
>>>> Regards,
>>>>
>>>> Pierre
>>>>
>>>>
>>>>
>>>> *From: *c... at googlegroups.com <c... at googlegroups.com> on behalf of
>>>> salo... at gmail.com <salo... at gmail.com>
>>>> *Date: *Friday, 14 May 2021 at 11:53
>>>> *To: *cp2k <c... at googlegroups.com>
>>>> *Subject: *[CP2K:15362] Problem converging SCF for a system containing
>>>> Fe2+
>>>>
>>>> Dear all,
>>>>
>>>>
>>>>
>>>> I am quite new to CP2K and working with transition metals.
>>>>
>>>>
>>>>
>>>> I am trying to reproduce some results of an O2 molecule dissociating
>>>> from an heme group (Fe2+). I am struggling to converge the SCF wavefunction
>>>> even in a singlepoint calculation and I am not sure what I am doing wrong
>>>> here.
>>>>
>>>>
>>>>
>>>> As an initial guess, I am using PBE with the following combination of
>>>> basis sets and pseudopotentials:
>>>>
>>>> - Fe: BASIS_SET DZVP-MOLOPT-SR-GTH & GTH-PBE-q16.
>>>>
>>>> - All the other atoms: DZVP-MOLOPT-GTH with GTH-PBE-q* .
>>>>
>>>>
>>>>
>>>> I have attached the input file and the initial geometry.
>>>>
>>>>
>>>>
>>>> Can you help me out?
>>>>
>>>>
>>>>
>>>> Thank you very much in advance,
>>>>
>>>>
>>>>
>>>> Salome
>>>>
>>>> --
>>>> 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 cp... at googlegroups.com.
>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/cp2k/f894e8c6-c46a-417b-83a6-6343cf0aff64n%40googlegroups.com
>>>> <https://groups.google.com/d/msgid/cp2k/f894e8c6-c46a-417b-83a6-6343cf0aff64n%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 cp... at googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/b43fc558-0c90-4f56-b932-3e9f71f880f1n%40googlegroups.com
>> <https://groups.google.com/d/msgid/cp2k/b43fc558-0c90-4f56-b932-3e9f71f880f1n%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210514/c3a1a793/attachment.htm>
More information about the CP2K-user
mailing list