[Pw_forum] problem: mix non-relativistic pp with fully-relativistic pp when turning on spin-orbit
Jiawei Zhou
zhoujw20 at gmail.com
Thu Jul 3 20:35:50 CEST 2014
Dear all,
I want to calculate the band structure of MnTe above the Neel
temperature (which is around room temperature and is a paramagnetic
phase). Originally I wanted to include spin-orbit coupling because Te is
a heavy element. So I used pslibrary to generate rel-pz-n-nc type
fully-relativistic pseudopotential. But when I calculated the band using
'bands' option it gives me errors:
Error in routine cdiaghg (262):
S matrix not positive definite.
I searched the forum and found suggestions to change pseudopotential.
But the thing is I don't want to exclude spin-orbit coupling for element
Te. So I tried mixing non relativistic Mn pp with fully relativistic Te
pp and still turned on spin-orbit calculation. The calculation goes on
well. But I don't know how the spin-orbit calculations are done in the
pw. I wonder in this case how the material is modeled? (only Te includes
spin-orbit coupling because its pseudopotential is in fully-relativistic
form?) Thanks a lot.
script is shown below:
&control
calculation = 'scf'
wf_collect=.true.
restart_mode='from_scratch'
prefix='MnTe-scf'
tstress = .true.
tprnfor = .true.
pseudo_dir = '$PSEUDO_DIR'
outdir = '$OUTDIR'
/
&system
ibrav=0, celldm(1)=7.8
nat= 4, ntyp= 2,
ecutwfc =60 , occupations='smearing', degauss=0.005, smearing='mv',
lspinorb=.true.,
noncolin=.true.,
starting_magnetization(1)=0.0
starting_magnetization(2)=0.0
/
&electrons
diagonalization='david'
mixing_mode = 'plain'
mixing_beta = 0.7
conv_thr = 1.0d-12
/
ATOMIC_SPECIES
Mn 54.938 Mn.pz-sp-hgh.UPF
Te 127.60 Te.rel-pz-n-nc.0.3.1.UPF
ATOMIC_POSITIONS (alat)
Mn 0.000000000 -0.000000000 0.000000000
Te -0.000000000 0.568151165 0.294296057
Mn -0.000000000 -0.000000000 0.588590750
Te 0.492033342 0.284075582 0.882885442
K_POINTS { automatic }
20 20 12 1 1 1
CELL_PARAMETERS
0.984066685 -0.000000000 0.000000000
-0.492033342 0.852226749 0.000000000
0.000000000 -0.000000000 1.177181498
Best,
Jiawei
More information about the users
mailing list