[Wien] [SPAM?] same k vector, different eigenvectors

[Peter Blaha]

Two possible explanations, which depends on the diagonalization:

a) You have degenerate eigenvalues. In that case any linear combination 
of the 2 (or 3) vectors are again a proper solution.

b) The eigenvectors are determined only up to a phase exp^(i phi), which 
depends on the diagonalization. It will very often change.

Conclusion: The complex eigenvector is not a quantity which you can 
easily compare.
Any property you calculate from that vectors (density, partial charges, 
...), however, should be the same.


Am 20.04.2023 um 11:07 schrieb 曹迎迎 via Wien:
> Dear all,
> 
> I am runing a one-step calculation to get the eigenvector of 
> ferromagnetic BCC-Fe at k=(1,1,1) from a self-consistent charge density. 
> The command used is ' runsp_lapw -so -p -s lapw1 -e lapwso'.
> 
> 
> First i get the self-consistent charge density by using 'runsp_lapw -ec 
> 0.0001 -cc 0.0001 -p -i 100 -so'. Then i copy the scf results to two 
> directories and change the klist respectively. In one directory, i use 
> only two k-points (0,0,0) and (1,1,1) in klist as:
> 
> 
>           1         0         0         0         1  1.0 -7.0  1.5      
>     1 k, div: (  1  1  1)
>           2        10        10        10        10  1.0
> END,
> 
> 
> and in another directory i use101 k-points including (1,1,1) as:
> 
> 
> ...
> 
>          99        10         9         9        10  8.0
>         100        10         8        10        10  4.0
>         101        10        10        10        10  1.0
> END.
> 
> 
> As a consequence, I get two results about eigenvectors at k=(1,1,1) in 
> these two directory. I find that the two results are not identical. In 
> the 2-kpts case, the lowest eigenvalue and eigenvector in vectorsoup 
> file is:
> 
> 
>             1  -5.83325860040616
>   -0.149115553408975       5.906541789778055E-002 -0.149115710346805
>    5.906548006164127E-002 -0.149115710346904       5.906548006168073E-002
>   -0.149115710346889       5.906548006167468E-002 -0.149115710346984
>    5.906548006171229E-002 -0.149115553409014       5.906541789779599E-002
> ...,
> 
> 
> while in the 101-kpts case, it is:
> 
> 
>             1  -5.83325860040616
>   -2.301711645066824E-004  0.160387402501481      -2.301714067522111E-004
>    0.160387571302459      -2.301714067523196E-004  0.160387571302571
>   -2.301714067523697E-004  0.160387571302556      -2.301714067524974E-004
>    0.160387571302664      -2.301711645069168E-004  0.160387402501536
> ....
> 
> 
> In my understanding, the hamiltonian is only k-dependent but not k-mesh 
> denpendent. So i should get the same hamiltonian in the two directories 
> at same k-point, so as the eigenvector. Did i do something wrong? Orwhat 
> misunderstanding do i have? If not, why are not the eigenvectors exactly 
> the same?
> 
> 
> I am wondering if anyone has encountered a similar issue or has any 
> insights on why this may be happening. Any suggestions or advise would 
> be greatly appreciated.
> 
> Thank you in advance.
> 
> Best regards, Yingying Cao
> 
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html

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