[Wien] The problem for atomic forces in the supercells of lanthanides compounds

[Artem Tarasov]

Dear wien2k users,

Recently, I have noticed a strange problem related to the forces 
inside the supercell for compounds containing 4f-atoms with 4f-states 
placed on the core level ("open core approximation" for Tb, Dy, Ho, Ce 
etc.). If we relax the slab of a cell with a period of 1x1 and a 
symmetry group of P4mm, and then make a cell with a doubled lateral 
period of 2x2 from the relaxed structure, then after converging the 
density for the 2x2 cell the forces on all atoms in the supercell grow 
significantly compared to the case of the 1x1 cell. Up to a jump from 
0.05 to 68 mRy/bohr on individual atoms. I thought that the problem 
may be related to the shift of the k-grid, because when I move to the 
supercell, I change the k-grid from 12x12 to 6x6, and I had supposed 
that such a k-grid may not be enough for an accurate calculation. 
However, even if I don't use the k-grid shift in both cases, this 
problem persists. My calculation is a slab of CeB6 with symmetry P4mm 
with Ce4f states given in case. inc as

16 0.70  0  NUMBER OF ORBITALS (EXCLUDING SPIN), SHIFT, IPRINT
…
4, 3,0.43               ( N,KAPPA,OCCUP)
4,-4,0.57               ( N,KAPPA,OCCUP)

Thank you in advance for your help.

Best regards,
Artem Tarasov
Department of Solid State Electronics
Saint Petersburg University