Multiplicity jumps back and forth between 1 and 3

[Lance]

Dear All,

I am a beginner of quantum chemical calculation and CP2K. I am performing an AIMD simulation using CP2K 7.1 on one polysulfide (Li2S6) in the solvent environment (DOL & DME) to see how S6 ion dissociates. I use a high temperature of 2000 K to accelerate the reaction for testing purpose. The initial structure was optimized using PBE. Considering that the S6 ion may dissociate into two S3 radicals, during which the multiplicity (or say state?) could change. Actually, the multiplicity could keep changing during the AIMD simulation, especially when any S-S bond is broken. To realize what I want, the initial multiplicity of the system is set to 1 and I turn on the UKS and RELAX_MULTIPLICITY. However, after several hundreds of steps, the SCF cannot converge (I suppose the S-S bond is about to break) and the multiplicity jumps back and forth between 1 and 3. I have attached the input files, including the restart file at which the SCF does not converge (wfn file is too large to be uploaded). And the relevant sample outputs are shown below:

SCF WAVEFUNCTION OPTIMIZATION

 

  Step     Update method      Time    Convergence         Total energy    Change

  ------------------------------------------------------------------------------

     1 Broy./Diag. 0.40E+00    2.2     0.01297368     -1774.0321771624 -1.77E+03

 

 *** WARNING in qs_mo_occupation.F:326 :: Multiplicity changed from 1 to 3 ***

 

     2 Broy./Diag. 0.40E+00    3.0     0.39035194     -1774.1912880872 -1.59E-01

 

 *** WARNING in qs_mo_occupation.F:326 :: Multiplicity changed from 3 to 1 ***

 

     3 Broy./Diag. 0.40E+00    3.1     0.39486999     -1774.6563932332 -4.65E-01

     4 Broy./Diag. 0.40E+00    3.0     0.00184080     -1774.0623889601  5.94E-01

     5 Broy./Diag. 0.40E+00    3.1     0.00117617     -1774.0445921719  1.78E-02

     6 Broy./Diag. 0.40E+00    3.1     0.00071690     -1774.0406131978  3.98E-03

     7 Broy./Diag. 0.40E+00    3.1     0.00060508     -1774.0442535992 -3.64E-03

     8 Broy./Diag. 0.40E+00    3.1     0.00028427     -1774.0448595937 -6.06E-04

     9 Broy./Diag. 0.40E+00    3.1     0.00023203     -1774.0459817967 -1.12E-03

    10 Broy./Diag. 0.40E+00    3.1     0.00014194     -1774.0468706246 -8.89E-04

    11 Broy./Diag. 0.40E+00    3.1     0.00009891     -1774.0478670248 -9.96E-04

    12 Broy./Diag. 0.40E+00    3.1     0.00171644     -1774.0487633110 -8.96E-04

 

 *** WARNING in qs_mo_occupation.F:326 :: Multiplicity changed from 1 to 3 ***

 

    13 Broy./Diag. 0.40E+00    3.1     0.39094929     -1774.0604047458 -1.16E-02

 

 *** WARNING in qs_mo_occupation.F:326 :: Multiplicity changed from 3 to 1 ***

 

    14 Broy./Diag. 0.40E+00    3.1     0.39607259     -1774.5995020947 -5.39E-01

    15 Broy./Diag. 0.40E+00    3.1     0.00384091     -1774.0151299392  5.84E-01

I searched in the cp2k google group for relevant questions and am not sure if the WFN_MIX keyword can help. There is little detailed information about how to use this keyword. Or are there any other solutions? Additionally, I am a beginner and have limited knowledge of quantum chemistry. Any suggestions or comments on how should I solve this problem will be greatly appreciated!

Kind Regards, 

Lance

[Anton S. Lytvynenko]

Dear Lance,

try to play with the value of relax_multiplicity -- say, from 0.00001 to 0.1.

Yours,

Anton

17.09.21 04:10, Lance пише:

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/c94ef51b-3434-4880-b846-f8ff9146065dn%40googlegroups.com.

[Lance]

Dear Anton,

Thank you very much for your suggestion! I just tried RELAX_MULTIPLICITY 0.1 (by restarting the simulation) but the results seem more unreasonable, which showed "Multiplicity changed from 1 to 7", "Multiplicity changed from 7 to 17", "Multiplicity changed from 17 to 9", "Multiplicity changed from 9 to 7", etc. and the SCF still didn't converge. It seems that changing the value of RELAX_MULTIPLICITY could not help. Anyway, thank you for your comments!

Kind Regards,

Lance

[atiqa razzaq]

Dear all,

Hope all are doing well.

I am the new beginner of cp2k and I don't know how to run cp2k. I want to calculate the diffusion rate in solid electrolytes for solid state batteries..

Any one who is willing to help me can reply that what are the initial steps that I should take.i have installed cp2k and also downloaded the basis sets..

Can any body help me how to start ab initio md for electrolytes???

My electrolyte composition is li7GeS5Br. 

Thank you

@t!q@

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/171cda59-46e9-1bff-2c7a-c97c3b038026%40gmail.com.

[Anton S. Lytvynenko]

Dear Lance,

this behavior is reasonable:

"RELAX_MULTIPLICITY {Real}
Tolerance in Hartrees. Do not enforce the occupation of alpha and beta MOs due to the initially defined multiplicity, but rather follow the Aufbau principle. A value greater than zero activates this option. Larger tolerance values increase the probability for a spin flip. This option is only valid for unrestricted (i.e. spin polarised) Kohn-Sham (UKS) calculations."

Larger tolerance values increase the probability for a spin flip -- so, if you feel that spin value is changing unreasonably often in your calculation and this compromises the convergence, you should gradually decrease it, hampering the flipping. However, if you decrease it too much, the spin will actually be "frozen" in its initial value. So, bearing in mind your previous tryouts, if I were in your shoes, I would consider 1E-4 as the next value for this parameter, and further decrease it if the calculation continues to be unstable.

Hope that helps.

Yours,

Anton

17.09.21 14:55, Lance пише:

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/635d89a6-174a-4d03-bb32-34dcc8443b4bn%40googlegroups.com.

[Lance]

Dear Anton,

Thank you very much for your further comments. I actually tried many different RELAX_MULTIPLICITY values. As expected, if the value is too big, the spin flip occured; if the value is too small, the SCF would never converge. Finally, I narrow down the range of the RELAX_MULTIPLICITY value, which is between 0.000050 and 0.000055. However, I don't think it's possible to find the optimal value that can make things work. I am wondering if there are any other settings I can adjust.

Kind Regards,

Lance

[Anton S. Lytvynenko]

Dear Lance,

did I understand correctly that you got these issues not on the first MD step, but rather on later ones?

Did you try any "typical" methods to facilitate convergence?

I would start from alpha 0.1 in your mixing section.

Yours,

Anton

21.09.21 14:05, Lance пише:

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/21c270b3-3a3a-4efe-9a1a-dd52eebec3bcn%40googlegroups.com.