[CP2K-user] Energy changes abnormally in OT method during the GEO_OPT

[Travis]

Hi,

I was approaching convergence for the low spin case in 400 steps changing 
the ALPHA parameter above to 0.02 with energy differences between SCF 
iterations down to 1e-5 Ha. For the high spin case, you'll need to improve 
the quality of your initial guess, else it will take even more cycles to 
converge. For that you can add the MAGNETIZATION keyword in &KIND for the 
relevant atoms. The combination of the ATOMIC guess with the high 
multiplicity causes your system's magnetic moment to fluctuate pretty 
wildly initially, contributing to the very slow convergence behavior you're 
seeing.

-T


On Wednesday, November 13, 2019 at 3:23:51 AM UTC-5, Xiaokun wrote:
>
> Hi Travis,
>
> I have tried your recommended DIAG method for my system. But unfortunately 
> it failed. Even though I changed  EPS_SCF 1e-05 and MAX_SCF 400, it's hard 
> to converge. The energy change seems to be very large (1E0~1E-1) and the 
> 'Convergence' also is a large number during the SCF WAVEFUNCTION 
> OPTIMIZATION progress.  How can I solve this problem?
>
> Yours,
>
> Xiao-kun
>
>
> 在 2019年11月10日星期日 UTC+8上午4:35:22，Travis写道：
>>
>> Hi,
>>
>> For high spin materials like this, I recommend switching away from the OT 
>> method. Sometimes the combination of high spin and use of GGAs makes your 
>> system half-metallic.
>>
>>   &SCF
>>    EPS_SCF 1e-06
>>    MAX_SCF 200
>>    SCF_GUESS RESTART
>>    ADDED_MOS  200
>>    &OUTER_SCF  F
>>    &END OUTER_SCF
>>    &SMEAR  T
>>     METHOD  FERMI_DIRAC
>>     ELECTRONIC_TEMPERATURE     3.0000000000000000E+02
>>    &END SMEAR
>>    &MIXING  T
>>     METHOD  BROYDEN_MIXING
>>     ALPHA     4.0000000000000002E-01
>>     BETA     1.5000000000000000E+00
>>     NMIXING  5
>>     NBUFFER  8
>>    &END MIXING
>>   &END SCF
>>
>> -T
>>
>>
>> On Friday, November 8, 2019 at 11:40:34 AM UTC-5, Xiaokun wrote:
>>>
>>> Hi，everyone！
>>>
>>> I am new one to CP2K and I am doing a Geo-optimization in a MOF. In the 
>>> input file, I use the OT method, constrain the most atomes obtained from 
>>> the experiment and relax the NO3- in the MOF cell. I want to find the 
>>> resonable positions of the NO3- , cause the experiment cannot sure where 
>>> NO3- it is.  And there seems to some errors in my output file. The energy 
>>> changes abnomally in a scf. Did I set some keywords wrong？Can anyone give 
>>> me some suggestions about my claculation? Attached are my input and output 
>>> files. Thank you.
>>>
>>> Best wishes.
>>>
>>> Xiao-kun
>>>  
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191113/64233c58/attachment.htm>