[Wien] General question about ghost bands

[Peter Blaha]

In most cases, ghostbands can be avoided easily (and actually should not 
appear at all). Mainly 3 parameters cause problems:

What are your RMT values, what is RKMAX and where's EFERMI ?

Often the RMTs are too "different", RKMAX is too large, or EF is very 
low/high because of empty space/ heavy elements. Then some corrections 
to the defaults are necessary.

Eventually you can send the struct file to my private email.


jadhikari at clarku.edu schrieb:
> Dear WIEN2k users,
> 
> I fix the ghost band problem on one of the atoms (QTL-B values vanishes)
> and it switches to other atom. And when the ghost band on the second atom
> is gone, it starts on another atom. This continues for many atoms and
> sometimes repeating on the already solved atoms. I have been removing
> ghost bands by changing energy as suggested by scf2 or scf1 or output
> files and also by varying RMT and RKmax and using different k-points in
> IBZ. Few of the times I changed Gmax and XC-potential and the Emax-Emin
> values. The smearing method is TEMP with 0.002 (after trying other
> options)and mixing of 0.10 or 0.05.
> 
> My general question is-
> 
> " Is it a usual and common outcome to get GHOST BANDS on atoms from one to
> another repeatedly and even on the ones already solved?"
> 
> The case is with the acidic form of Potassium Calcium Niobate, where I get
> the ghost band for Hydrogen atom with QTL-B of 6.2 (the minimum that could
> be obtained till today). After altering the energy level and changing RMT
> it is gone but there starts a ghost band on Calcium atom. It dissappears
> for a while and starts on Oxygen atom. Eventually it keeps rotating on
> different atoms. This is with 3 k-points in IBZ as the calculation with 1
> k-point in IBZ converged without any ghost band. 6 k-points in IBZ did not
> move forward from lapw2 step as the previous calculation had a small QTL-B
> value.
> 
> Any suggestions would be gratefully appreciated.
> Thanking you.
> 
> Sincerely,
> 
> Subin Adhikari
> Clark University
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien