Dear Avinash,

Normally for a SOC calculation, You can use LSORBIT and LNONCOLLINEAR (if your system is having non-colinear moment.)

try this link 

http://cms.mpi.univie.ac.at/vasp/vasp/Non_collinear_calculations_spin_orbit_coupling.html

I tried it  but it is not working for me do you have incar then please give me

Thanks a lot Sir 

Can i get your mail for further query? #Artem R. Khabibullin

Hi Avanish,

For SOC calculations LSORBIT tag works fine, but this tag will not work with the normal vasp setting in the makefile. Here you need to compile the vasp for full fft mashes not for reduced mashes in some direction or Gamma-point.

So compile your vasp without -DNGXhalf and -DNGZhalf flags and then run your calculations with LSORBIT tag.

thank you

bahadur

Hi Bahadur,

How do I compile vasp without _DNGXhalf and -DNGZhalf flags before? While running  it says ''non collinear calculations require that VASP is compiled without the flag -DNGXhalf and -DNGZhalf . 

Tika

Thank you Darshana. 

Can anyone help me to understand the "spin orbit coupling matrix" written at the end of OUTCAR?

My second question is, How to obtain spin and orbital magnetic moment in VASP calculation? 

Spin-orbit coupling matrix is basically the matrix element for each atom's p, d, f projected orbitals. 

To get the spin and orbital moment, you should write LORBIT = 10 or 11 and LORBMOM = .TRUE. tags in your INCAR file. 

to know details about LORBIT tag please go though VASP manual. 

 Hi Tikaram,

I am facing the same problem. Can you please elaborate that how can we compile _DNGXhalf and -DNGZhalf in VASP?

Hi Payal 

For older version of vasp (5.3.5 or less) open your Makefile and look for -DNGXhalf, -DNGZhalf and remove them, only these particular tags you should remove and compile it again. Once compiled you can use it. 

For newer version (which you can download from VASP, licensed one) vasp 5.4.1 or 5.4.4 you just compile as "make all" or "make ncl" (non-collinear) you will get an executable in "bin" folder (created inside vasp folder) named as "vasp_ncl" that would be enough for doing SOC calculations. 

The last point to say "make all" will give you there executables, vasp_std for normal vasp calculations, vasp_ncl for soc calculations and vasp_gam for gamma point calculations. It's better to do make all, it may take some time to compile all 

Best

Avanish

Thank you Sir.

Can i get your email id for any other query? 

 Hi Artem,

I have tried to perform the spin orbital coupling calculation in cobalt hcp( 2atoms/cell),  using the file INCAR2, but nothing happens, I don't have any datas in the OUTCAR and OSZICAR, what's the problem please?

Thanks,

Ulrich

Dear  Artem R. Khabibullin & Avanish Mishra!

I have few questions, please answer.

The code in these INCARs take you to the mag. ground state (ISPIN=1). what if we skip going on ground state and avoid taking extra relaxation steps(NSW=24)? 

second, instead of taking LREAL=F, if we consider LREAL=Auto. then ?

3rd, IBRON=-1 will work..?

Dear Faiza Uzma

It would be good if you can DM me your email ID I can share the whole steps and you try to follow that.

For doing SOC I do not think you need NSW (make NSW=0). You must have taken already relaxed and optimized structure. 

LSORBIT   = .TRUE.

SAXIS     = 0 0 1

MAGMOM    = 666*0

GGA_COMPAT = .FALSE.

LMAXMIX    =  4

ADDGRID    = .TRUE.

please go through this link:-

http://cms.mpi.univie.ac.at/vasp/vasp/LSORBIT_tag.html

Dear All,

What is the right way to do spin orbit coupling (SOC) calculation ? We should relax system with SOC or we should relax without SOC and then single point calculation (nscf) for spin orbit coupling effect.

Thank You All

Shilendra

First, relax the system without SOC. Then perform SCF (no relaxation) with the required tag to include SOC.

Dear Avanish and Priyanka,

For doing spin orbit coupling (SOC) calculation we should relax system with spin polarization (ISPIN=2) and LORBIT=11 or without spin we should relax the system.

What about LORBMOM tag for orbital and spin moments. we should give this tag.

Thank you for your kind help !

Shilendra

Yes, you have to relax with ISPIN = 2 for magnetic systems. Relaxation does not depends on LORBIT. LORBIT tag is to get density of states. If you want to plot DOS, then yes you need LORBIT = 10 or 11. For details, see VASP manual.

LORBMOM = .TRUE. is needed to get orbital moments when SOC is added.

I got different result for relaxation of ionic positions for with and without including SOC tags.

is SOC exclusion in relax run is recommended for memory saving purpose? or technically it is incorrect?

No need to do relaxation with SOC. Do relaxation without SOC and then use SOC tags in SCF and other later calculations wherever required.

Should the symmetry operations get changed on the inclusion of SOC?

When, i am performing band structure calculations with spin-orbit coupling (SOC) then, its symmetry operations are getting reduced than the one without SOC.

Suppose in calculations without SOC, we have 12 symmetry operations and with SOC, it gets reduced to 2.

So, can somebody please put some light whether symmetry operations should change with the inclusion of SOC or not?

How can I find whether a system is ferro mag or anti ferro mag. My system consists of 16 atoms.There are three different type of elements in my systm?

Dear Asif,

If you want to confirm your system that it is ferro or anti ferro magnetic you need to check the long range and short range magnetic ordering using, AFM and FM calculations. You need to run the calculations with AFM and FM (MAGMOM) tags and compare the stable states from energy comparison.

For the confirmation of Ferro and Anti-ferro magnetic system.

Step 1. You will add LORBIT = 11 in the INCAR-file and relax the system.

Step 2. After the completion of step1 .You will check the OUTCAR-file and find the magnetization of each orbital (see the example below). If tot is positive (+) and have some value also than, it will represent Ferro magnetic or If tot is negative (-) and have some value also than, it will represent Anti-ferro magnetic.

Magnetization (x)

# of ion s p d tot

----------------------------------------

1 0.000 0.000 0.000 0.000

2 0.000 0.000 0.000 0.000

3 0.000 0.000 0.000 0.000

------------------------------------------------

tot 0.000 0.000 0.000 0.000

e

Payal Wadhwa Its normal to see reduced symmetry operations and safe side it to switch off symmetry since number of k-points (which depend on sym) do change for different magnetic orientation which might verse final outcome.

Dear All,

The process of SOC calculation is described below.

1. First do the normal optimization

2. Then make the normal DOS calculation.

3. From there make the SOC calculation (SCF only)

open the tag a) LSORBIT = .T. ,b) GGA_COMPAT = .F. and c)MAGMOM in x, y, z format.

What is the meaning of E_soc in OUTCAR of VASP?

Can anyone help me in getting rid of segmentation fault while calculating SOC in quantum Espresso?

Hello! I tried SOC (scf calculations) calculations after relaxing the structure, I'm only able to get DOS for up spin (even though ISPIN = 2) but if I do the same thing removing SOC, I get DOS for both the spins. Can anyone help me?

Thanks, in advance.@

Sromona, when you include spin orbit coupling effects, spin is no longer a single quantum number, but is instead coupled with the angular momentum (due to relativistic effects) and so there is no spin up or spin down, just total spin. This might help: http://xbeams.chem.yale.edu/~batista/vvv/node26.html. Note that SOC effects are more pronounced for structures/molecules containing heavy atoms (~Z>50).

How can I get DOS for both the spins for a SOC system? I have seen people doing soc calculation for a spin polarized system. What should I do to do that?