[Wannier] several questions about wannier calculation and settings

[Jonathan Yates]

> another stupid question about projection setting:    the setting Nicola 
> gave for PbTiO3 in last mail  is according its band's structure , and 
> the projection functions are s/p type-like orbitals.  i wanna know  
>  whether there are some guidelines for setting up initial projection 
> functions ?

  The projections functions are just a starting guess. In an ideal world 
the wannier90 minimiser would be so good that it didn't need a starting 
guess. Sometimes this is true, but often the guess helps.
  The projections should reflect the character of states you are working 
with.

  So lets take Anatase.

Your file has num_wann=24.
  I'm assuming this means you are including the Ti 3s/3p semi-core 
states. While this is good for the ab-initio calculation you probably 
don't want WF for those states (just because they aren't so interesting)
  After running the nscf calculation with 24 bands you could exclude the 
semi-core states from the pw2wannier90 and all subsequent calculations 
by putting
   exclude_bands = 1 2 3 4
  in the wannier90 file. (with num_wann=16).

So now you want 16 WF to describe the valence bands. Chemistry tells you 
that these are combinations of O 2s/2p states.
  So you could try
O:sp3
or
O:sp2;pz
or
O:s,p
  maybe all of these work equally well, maybe one is better.

There are some existing papers on MLWF for TiO2 polymorphs

Phys. Rev. B 69, 121101(R) (2004)
Phys. Rev. B 74, 125113 (2006)

  Hope this helps

    Yours
     Jonathan

ps. I made plenty of assumptions about your calculations, if I got 
anything wrong you'll have to correct me and give us more details.

-- 
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