Page 1 of 1

Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Mon Jul 03, 2023 12:27 pm
by nirmal_ganguli2
Facing convergence problem with bandstructure calculation with spin orbit coupling(Nonlinear calculation). The scf calculation with SOC converges very well in the previous step with the same input files with K-mesh 9X9X5 but problem arises in band calculation only. here is the INCAR inputs for the bandstructure calculation:
SYSTEM

ICHARG = 11
ENCUT = 500
EDIFF = 1.0E-5
NELM = 500
LREAL = .FALSE.
LMAXMIX = 6
# ====================
ISYM = -1
IVDW = 1
ISPIN = 2
# ====================
ISMEAR = 0
SIGMA = 0.01
PREC = Accurate
LWAVE = .FALSE.
LORBIT = 11
# ====================
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 2 2 -1
LDAUU = 4.0 2.0 0.0
LDAUJ = 1.0 0.5 0.0
LDAUPRINT = 1
# ====================
LNONCOLLINEAR = .TRUE.
LSORBIT =.TRUE.
SAXIS = 0 1 0
NBANDS = 150
LORBMOM = .TRUE.
MAGMOM = 0 4 0 0 4 0 0 -3 0 0 -3 0 0.0*36

with the following output:
entering main loop
N E dE d eps ncg rms rms(c)
DAV: 1 0.927567145911E+03 0.92757E+03 -0.37831E+04176310 0.125E+03
DAV: 2 0.105001249495E+03 -0.82257E+03 -0.76748E+03199404 0.416E+02
DAV: 3 -0.396459225369E+02 -0.14465E+03 -0.13732E+03198588 0.157E+02
DAV: 4 -0.581741617322E+02 -0.18528E+02 -0.17676E+02221136 0.731E+01
DAV: 5 -0.600711932147E+02 -0.18970E+01 -0.19245E+01182652 0.177E+01
DAV: 6 -0.760512943188E+02 -0.15980E+02 -0.18325E+02211932 0.124E+01
DAV: 7 -0.113609902468E+03 -0.37559E+02 -0.82801E+02162150 0.152E+01
DAV: 8 -0.985874259088E+03 -0.87226E+03 -0.94860E+03204810 0.131E+01
DAV: 9 -0.464005815582E+03 0.52187E+03 -0.55314E+03171576 0.586E+01
DAV: 10 -0.138792286343E+04 -0.92392E+03 -0.13733E+04200988 0.277E+01
DAV: 11 -0.112129457720E+04 0.26663E+03 -0.11979E+04191688 0.685E+01
DAV: 12 -0.199637552035E+04 -0.87508E+03 -0.20276E+04193806 0.587E+01
DAV: 13 -0.235218736224E+04 -0.35581E+03 -0.24235E+04194154 0.768E+01
DAV: 14 -0.249744021273E+04 -0.14525E+03 -0.25547E+04192828 0.762E+01
DAV: 15 -0.207311871566E+04 0.42432E+03 -0.21251E+04194664 0.790E+01
DAV: 16 -0.292853737671E+04 -0.85542E+03 -0.29822E+04193014 0.809E+01
DAV: 17 -0.614044639303E+04 -0.32119E+04 -0.62060E+04193980 0.803E+01
DAV: 18 -0.842480225483E+04 -0.22844E+04 -0.84925E+04194478 0.833E+01
DAV: 19 -0.395249637874E+04 0.44723E+04 -0.40210E+04195462 0.854E+01
DAV: 20 -0.429544420185E+04 -0.34295E+03 -0.43595E+04195786 0.848E+01
DAV: 21 -0.319132549645E+04 0.11041E+04 -0.32527E+04195066 0.850E+01
DAV: 22 -0.297026336971E+04 0.22106E+03 -0.30274E+04196314 0.822E+01
DAV: 23 -0.267431255028E+04 0.29595E+03 -0.27399E+04194820 0.814E+01
DAV: 24 -0.217039354464E+04 0.50392E+03 -0.22383E+04194886 0.815E+01
DAV: 25 -0.257862521668E+04 -0.40823E+03 -0.26330E+04193968 0.769E+01
DAV: 26 -0.240666325057E+04 0.17196E+03 -0.24613E+04194646 0.757E+01
DAV: 27 -0.157446378783E+04 0.83220E+03 -0.16212E+04194058 0.754E+01
DAV: 28 -0.192430766802E+04 -0.34984E+03 -0.19798E+04194388 0.741E+01
DAV: 29 -0.177250873020E+04 0.15180E+03 -0.18136E+04192618 0.755E+01
DAV: 30 -0.130590179545E+04 0.46661E+03 -0.13330E+04193098 0.725E+01
DAV: 31 -0.177756969053E+04 -0.47167E+03 -0.18217E+04193062 0.755E+01
DAV: 32 -0.185061677664E+04 -0.73047E+02 -0.18866E+04192252 0.729E+01

WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
...

Kindly suggest the possible solutions for the above problem.

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Mon Jul 03, 2023 1:18 pm
by jonathan_lahnsteiner2
Dear nirmal_ganguli2,


With the information you are giving I can not tell for sure the origin of your issues.
Can you please supply all relevant input files as POSCAR, INCAR, KPOINTS and POTCAR file. It is mentioned
in the wiki guidelines https://www.vasp.at/forum/viewtopic.php?f=4&t=17928 how to append the files to your post.

Without input files I can only refer you to the following posts reporting similar issues:
https://www.vasp.at/forum/viewtopic.php?t=1192
https://www.vasp.at/forum/viewtopic.php?t=5894
https://www.vasp.at/forum/viewtopic.php?t=3206

With kind regards
Jonathan

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Mon Jul 10, 2023 11:55 am
by nirmal_ganguli2
Dear jonathan_lahnsteiner2,

Please find the required files for scf calculations with SOC and the KPOINTS with relevant high symmetry path for the band calculation are attached herewith.

Thank you.

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Tue Jul 11, 2023 6:31 pm
by jonathan_lahnsteiner2
Dear nirmal_ganguli2,

I am not 100% sure if I understand your question properly. So what you are intending to
do is to compute a band structure? This is what I guess from your sentence "problem arises in band calculation only".
To do a band structure calculation your first have to compute a converged CHGCAR file. For more info on the CHGCAR file
check here:
https://www.vasp.at/wiki/index.php/CHGCAR
After obtaining the CHGACR file you have to set the tags:

Code: Select all

ICHARG = 11 #charge read file
LORBIT = 11
When checking your INCAR file the tag ICHARG was not set. For more information you could also check this two pages of
our wiki:
https://www.vasp.at/wiki/index.php/Si_bandstructure
https://www.vasp.at/wiki/index.php/Band ... unctionals

As mentioned already I am not sure if I understand the question properly.
Nevertheless, I hope this helps solving your issue. If not please contact us again.

All the best Jonathan

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Wed Jul 12, 2023 11:37 am
by nirmal_ganguli2
Dear jonathan_lahnsteiner2,

The scf calculation with SOC converged very smoothly and converged CHGCAR file is also provided in the bandstructure calculation. If you please check the zip file where band file consist of INCAR, POSCAR, POTCAR and KPOINTS for the band calculation. The ICHARG=11 tag is set in the INCAR file for band calculation.
As I mentioned in the question, the convergence problem is arising in the band calculation only with following error in the output:

WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
WARNING: Sub-Space-Matrix is not hermitian in DAV 9

Thank you.

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Thu Jul 13, 2023 12:33 pm
by jonathan_lahnsteiner2
Dear nirmal_ganguli2,

I checked your calculation again. You can avoid the

Code: Select all

WARNING: Sub-Space-Matrix is not hermitian in DAV 9
-2.658760842063281E-005
warning by switching to a different electronic minimization algorithm. Currently you are using the
blocked Davidson iteration scheme, which is the default. You can switch to the RMM-DIIS minimization scheme
by putting

Code: Select all

IALGO=48
in your INCAR file in the band folder you sent. You don't have to repeat the calculation in the SOC folder.
For further information on the ALGO/IALGO tag check the wiki
wiki/index.php/ALGO
wiki/index.php/IALGO

I hope this helps to resolve your issue.

All the best Jonathan

Re: Convergence problem in bandstructure calculation including spin orbit coupling using VASP.6.2.2

Posted: Sun Jul 16, 2023 10:36 am
by nirmal_ganguli2
Dear Jonathan,

I have repeated the band calculation with IALGO=48, then the calculation is getting converged. In the obtained bandstructure some noise is coming after 6 eV energy range but within -6 to +6 eV range the bandstructure is correct.
Thank you so much.