[Wannier] several problems

Jonathan Yates jry20 at cam.ac.uk
Mon Mar 3 20:18:07 CET 2008

[n.b. In a follow-up to this email you refer to a reply from Nicola -  
I haven't seen that reply]

On 3 Mar 2008, at 06:25, nafise rezaei wrote:
> 1)How can I choose centre of projections?
> 3)In first example, in example/example1 directory ,centre of  
> projection is on the As atom.
> in this state ,bandstructure have oscillation.but when i get centre  
> of projection on the Ga atom,
> band is good and resemeble pwscf bandstructure.

In example1 (valence bands of GaAs) the trial projections are 4 sp3  
functions centred on the As atom. We could have chosen 4 's' like  
projections located near the bond centres (see example5 - diamond for  
a demonstration of how to do this). sp3 functions centred on the Ga  
should also be fine.
  The projections just provide an initial guess for the unitary  
transformation between the original bloch states - all of the above  
projections should provide a reasonable guess - and I would expect  
all of them to lead to the same set of MLWF. If you find this to not  
be the case, you should provide us with more details about your  
  [you might also try setting
    in some cases this can help get you into the right minimum]

> 4)I have another problem , when I generate *.xsf and plot wannier  
> function
> for si(in tutorial_files/w90-ex1) and also for GaAs(in example/ 
> example1) ,my wannier function isn't
> simalar to your Gallary. what is problem?

You'll need to give more detailed information before I could answer  
that (exactly how is it different!)

  If you run example1 the MLWF should look similar to the image on  
the webpage - but not quite the same. You need to understand the  
keyword wannier_plot_supercell. Also read the hint on page 4 of the  


Dr Jonathan Yates         |    Theory of Condensed Matter Group
Corpus Christi College    |    Cavendish Laboratory
Cambridge, CB2 1RH, UK    |    Cambridge, CB3 OHE, UK
email jry20 at cam.ac.uk     |    Tel +44 (0)1223 337461

More information about the Wannier mailing list