[CP2K-user] Change arrangement of electrons in POTENTIALS file

[Daniele Ongari]

Dear all, 
for some high-throughput calculation on Metal Organic Frameworks I need to 
specify the MAGNETIZATION in order to impose the highest ferromagnetic 
state at a given oxidation state. It is an assumption to treat all the 
metals in a coherent way. 

The problem I'm facing is that, when the electrons of the potential are 
arranges such as:
- all the orbitals are filled 
or
- more than one orbital is partially filled

CP2K fails with the error 
*******************************************************************************
 *   ___                                                                    
   *
 *  /   \                                                                  
    *
 * [ABORT]                                                                  
   *
 *  \___/        Magnetization value cannot be imposed for this atom type  
    *
 *    |                                                                    
    *
 *  O/|                                                                    
    *
 * /| |                                                                    
    *
 * / \                                                    
qs_kind_types.F:2790 *
 *******************************************************************************

because, I think, MAGNETIZATION can not decide which electrons to unpair. 
I could use the section &BS but it gets too complicate, so the question is: 
am I allowed to change the original ordering in the POTENTIAL file?

To explain better, let's take the elements of the 10th column that lead to 
failure, Pt and Pd:
(row 3) Ni is arranged "4 6 8", and it is fine as magnetization will unpair 
electrons form the partially occupied d orbitals
(row 4) Pd is arranged as "3 6 9" and fails because both s and d orbitals 
are partially occupied. Am I allowed to modify it as "4 6 8"?
(row 5) Pt is arranged as "2 6 10" and fails because all orbitals are fully 
occupied. Am I allowed to modify it as "4 6 8"?

Technically this trick works and the CP2K calculation starts an even 
converges, but do you see any theoretical problem in doing this?

Thanks for your attention.

Daniele 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210104/f5e47b01/attachment.htm>