[Wien] Ghost Band error

[Lyudmila Dobysheva]

21.05.2022 09:46, shamik chakrabarti wrote:
>                  I have obtained converged structure (Volume vs. E) of 
> LiCoSbO6 by using spin polarization only for Co. However, while doing 
> a,b,c, Gamma variation for monoclinic lattice I have again encountered 
> Ghost Band error. In this case I have found that keeping the spin 
> polarization for every atoms keep the simulation going without any 
> error. Please suggest am I doing something wrong?

The idea to take magnetic Co and nonmagnetic other atoms allows to start 
calculation from another initial point, and thus to avoid the Ghost Band 
error if it is occasional. But the final solution in most cases is 
unique, and the program will find it. So you do not use "spin 
polarization only for Co" or keep "the spin polarization for every 
atoms": when you make SP calculation all atoms are spin-polarized. You 
can see this in the SCF file (:MMI001-:MMI005 - they do not depend on 
which atoms were magnetic in the inst file).
In some systems there exist few magnetic solutions, and you should 
consider this possibility. Just playing with the file inst will give 
these different states.

So, if your system has certainly only one magnetic solution, the result 
does not depend on the inst file.

Best wishes
Lyudmila Dobysheva
------------------
http://ftiudm.ru/content/view/25/103/lang,english/
Institute of Physics and Technology,
Udmurt Federal Research Center, Ural Br. of Rus.Ac.Sci.
426000 Izhevsk Kirov str. 132
Russia
---
Tel. +7 (34I2)43-24-59 (office), +7 (9I2)OI9-795O (home)
Skype: lyuka18 (office), lyuka17 (home)
E-mail: lyuka17 at mail.ru (office), lyuka17 at gmail.com (home)