[Wien] Query about SOC in WIEN2k

[Peter Blaha]

>>>Increase the k-mesh (for the final low-symmetry struct file) to a very
> good one AND TEST the convergence. of your results.
> I can generate the P1 structure for the system, but how to get the
> low-symmetry structure which is valid for all direction (001, 100, 010,
> etc).
> Is there any specific way of doing?

There is no unique way, but how to do this explicitly depends on your 
specific case. The general strategy is always to run  init_so (with 
spin-pol. and accept/examine the struct files until symmetso does not 
further reduce the symmetry.
In some cases you can simply make all equivalent atoms non-equivalent 
(and "label" them as Fe1,...), in some other cases you have to make 
a,b,c inequivalent during these initializations (but change them back to 
the original values for the scf cycles), ....

You will need to read AND UNDERSTAND the output file of symmetso.

>
>>> Since the E differences are very small, set the DEs  in case.in1 to a
> smaller value like:
> 0   -4.09      .0001 STOP 1
>                ^^^^^^^  DE
> Here do I need to set small increment for all the atomic orbitals which
> has DE .ne. 0, or for only the specific orbitals which has STOP conditions?

For all, which have DE ne 0

>>> During the process of initso_lapw
>
> Here I have confusion in generating the case.inso file. I need more
> clarification whether I should copy from SRC_templetes and modify
> accordingly or we should generate case.inso with
> interactive command initso_lapw?

For the setup and determination of the common symmetry, for sure you 
should use initso_lapw.

Once you have setup everything correctly and have run the 
scalarrelativistic calculations for the low symmetry case and the first 
SO-direction, you can later on simply edit case.inso and change the 
direction.

Can you explain case.inso file
> generation, Like: when to select RLO for atoms?

When you need them. It depends mostly if you have semicore p-states of a 
heavy atom. But anyway, test it with/without RLOs.

Do I need to keep
> kpoints same for all directions (001, 010 etc)?

What do you guess ???  As I explained, you should use the same struct 
file, so what about k-mesh ???

Should
> I generate case.inso file same way for the "force method" also, because
> as far as I understood the force theorem is base on non-scf calculation,
> so when we choose the new symmetry the
> Eigenvalues for different directions changes a lot resulting bad
> anisotropy.

As I said before, once you have the proper struct file and 
initialization, there is no difference between force-method and E-tot. 
Follow my previous explanation (use the save/restore commands).


-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--------------------------------------------------------------------------