[Wien] Difference in DOS (LDA+U) between ver.09.1 and ver.14.2

[s-taro]

Dear Wien users,

I calculated DOS of a perovskite-type oxide, LaCoO3(R-3C). For simple spin-polarized calculation, I got substantially the same DOS output with WIEN2k ver.09.1 and ver.14.2 using the same input parameters.

But when I used LDA+U(SIC) method to improve the description of Co 3d electrons after the calculation without LDA+U converged, the DOS with WIEN2k ver. 14.2 was different from the one with WIEN2k ver. 09.1.

The difference is most clearly observed in the PDOS for Co-dz2 down-spin.  For example, the peak of the PDOS from WIEN2k ver.14.2 and 09.1 locate at the Fermi level and 1 eV, respectively, with the U of 2.7 eV.  Moreover, the PDOS from ver.14.2 stays at the Fermi level regardless of the U value (ranging between 2.7 eV and 7.5 eV), while the one from ver.09.1 well responds to the U value, shifting to higher energy with the U.

With ver.14.2, a similar PDOS sticking to the Fermi level is observed also for Co(dx2y2-dxy) and (dxz-dyz) in up-spin.

My question:
Why were the DOS's different between WIEN2k ver. 09.1 and WIEN2k ver. 14.2 although I used the same input parameters ?

Some of the input information is attached below. I would appreciate if you could give me some advices or comments.

Sincerely,

Taro


----------------------------------------------------------------------------------
The input files and parameters are following:

case.struct:
  Space group: R-3c

Lattice parameters in bohr
  a=b=10.2849
  c=24.741924
  α=β=90
  γ=120

Inequivalent Atoms: 3
  Atom 1: La RMT=2.37
  Pos 1: x=y=z=0.25
  Pos 2: x=y=z=0.75

Atom 2:Co RMT=1.87
  Pos 1: x=y=z=0
  Pos 2: x=y=z=0.5

Atom 3: O RMT=1.65
  Pos 1: x=0.20030001 y=0.2997     z=0.75
  Pos 2: x=0.75       y=0.20030001 z=0.2997
  Pos 3: x=0.2997     y=0.75       z=0.20030001
  Pos 4: x=0.79969999 y=0.7003     z=0.25
  Pos 5: x=0.25       y=0.79969999 z=0.7003
  Pos 6: x=0.7003     y=0.25       z=0.79969999

----------------------------------------------------------------------------------
Initialize Calc.:
  x nn: distance factor = 2
  inst_gen: u(default)
  case.in1: nband = 106
             emax = 2.5(WIEN2k ver. 09.1)
             de = 2.0(WIEN2k ver. 14.2)(because of EF = 0.5)
  case.in1_st: nband = 106
                emax = 2.5(WIEN2k ver. 09.1)
                de = 2.0(WIEN2k ver. 14.2)(because of EF = 0.5)
  case.in2: EMIN = -9.0
  case.in2_st: EMIN = -9.0
  case.in2_ls: EMIN = -9.0
  case.inm: BROYD/PRATT scheme = MSEC1
             mixing factor = 0.05
  k points: 10000

command: runsp_lapw -ec 0.00001 -cc 0.00001 -i 200 -p

----------------------------------------------------------------------------------
For LDA+U method:
case.indm:
  -9.                      Emin cutoff energy
   1                       number of atoms for which density matrix is calculated
   2  1  2      index of 1st atom, number of L's, L1
   0 0           r-index, (l,s)index

case.inorb:
    1  1  0                     nmod, natorb, ipr
  PRATT  1.0                    BROYD/PRATT, mixing
    2 1 2                          iatom nlorb, lorb
    1                              nsic 0..AFM, 1..SIC, 2..HFM
     0.1985 0.00        U J (Ry)   Note: we recommend to use U_eff = U-J and J=0

command: runsp_lapw -ec 0.00001 -cc 0.00001 -i 200 -p -orb