[QE-users] Partial charges and smearing with SIC scheme

[JULIEN, CLAUDE, PIERRE BARBAUD]

Dear QE users,

I have been trying to use the self-interaction correction (sic) scheme implemented in QE as described in the references:
https://link.aps.org/doi/10.1103/PhysRevLett.129.126401
https://journals.aps.org/prb/abstract/10.1103/PhysRevB.106.125119

However, I ran into some trouble when trying to simulate partial charges (to verify Janak’s condition). To do so, I simply modified the built-in examples on the MgO system by changing the charge and magnetization of the system to fractional values (0.5). Additionally, since we have fractional occupancies, a form of energy broadening will be required by QE to perform a simulation without facing numerical issues. 
However, I could not find a broadening scheme that seemed to work well with the sic options. The ‘occupations=smearing’ option results in the following error, irrespective of the type of smearing chosen (gaussian, mp, mv…):
“Error in routine sic_init (1):
  gaussian smearing not allowed”

This type of broadening seems incompatible with the way sic was implemented. I then tried the tetrahedron method. This time, the program successfully went through the init routine but fails at the first iteration with error:
“Error in routine electrons (1):
charge is wrong: smearing is needed”

Have I made a mistake, or is there a specific way to go about treating partial charges with sic?


On a related note, I am also quite confused by the use of the keyword ‘sic_energy’, since it seems that it always assumes a full hole/electron (i.e. the preliminary calculation seems to always add/substract a full electron from the polaronic system, which leads to obvious problems when trying to simulate partial charges). I have therefore set it to .false. for my tests. I don’t think I saw this preliminary energy correction mentioned in the papers, is there any material that goes into more detail for how this works?

Please find attached the input and output files to reproduce the issue.


thanks in advance,
Julien


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: MgO_3_0.5.30463068.out
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20240206/65a7ad33/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: MgO_3_0.5.in
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20240206/65a7ad33/attachment-0001.ksh>