[CP2K-user] how to run brokn spin symmetry calculation with OT

[Tamas K. Stenczel]

Thank you Matt, your answer was helpful and it worked!
The BS section allowed me to set up an excited state of each atom for the 
atomic guess, which is then rescaled to the number of electrons in each 
spin channel. However, I found that the resultant wavefunction can converge 
to some local minima. For example, while testing an O2 binding energy 
curve, there were some points that initially converged to ~1eV higher than 
the neighboring ones, however restarted form the wavefunctions of the 
distances adjacent to them they gave me energies consistent with the ones 
around them. 

Do you have any idea how to make my calculations, and perhaps the initial 
guess more robust for molecular systems?

Thanks,
Tamas

On Sunday, 29 December 2019 16:58:15 UTC, Matt W wrote:
>
> Yes, you need a broken symmetry guess. 
>
> You can use the BS section in the KIND section (many threads on how to do 
> this it is non intuitive) or you can use the 
>
>
> https://manual.cp2k.org/cp2k-6_1-branch/CP2K_INPUT/FORCE_EVAL/DFT/PRINT/WFN_MIX.html
>
> section to set up your calculation as you are attempting here. Again, it 
> is not all that easy to use...
>
> Matt
>
> On Sunday, December 29, 2019 at 2:38:45 PM UTC, Tamas K. Stenczel wrote:
>>
>> Dear cp2k developers and users,
>>
>> I am working modelling the combustion of small hydrocarbons and have 
>> issue with spin. I am using CP2K 5.1 right now.
>>
>> I would like to calculate my system in UKS singlet state, with broken 
>> symmetry of the alpha-beta channels.  I have understood that this is the 
>> expected behavior of a large system with many oxygen molecules and the 
>> result should be better than the singlet with RKS or with spin-symmetric 
>> UKS (these two give me the same results actually).
>>
>> Is this possible with the OT methods, how to do it? I have been using 
>> OT%DIIS for its performance with SCF convergence. Diag and mixing methods 
>> were not as reliable as this so far, but can perform relaxation of 
>> multiplicity, which has given me asymetric spins but I cannot use the 
>> wavefunctions of that to restart OT from, due to the error:
>>
>>> *******************************************************************************
>>>  *   ___                                                                       *
>>>  *  /   \                                                                      *
>>>  * [ABORT]                                                                     *
>>>  *  \___/    Number of occupied MOs on restart unit larger than allocated MOs. *
>>>  *    |                                                                        *
>>>  *  O/|                                                                        *
>>>  * /| |                                                                        *
>>>  * / \                                                          qs_mo_io.F:818 *
>>>  *******************************************************************************
>>>
>>>
>>
>> some of the input. I have attached an input file of an O2 molecule for 
>> example.
>> &DFT
>>     POTENTIAL_FILE_NAME GTH_POTENTIALS
>>     UKS TRUE
>>     MULTIPLICITY 1
>>     BASIS_SET_FILE_NAME COMBINED_BASIS_TKS32
>>     &SCF
>>       MAX_SCF 25
>>       EPS_SCF 0.0001
>>       SCF_GUESS RESTART
>>       &OT
>>         MINIMIZER DIIS
>>       &END OT
>>       &OUTER_SCF
>>         EPS_SCF 0.0001
>>         MAX_SCF 4
>>       &END OUTER_SCF
>>     &END SCF
>>
>>
>> Thank you very much,
>> Tamas K Stenczel
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200102/070c58e5/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: opt.inp
Type: chemical/x-gamess-input
Size: 2063 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200102/070c58e5/attachment.inp>