[Wannier] charge of wannier function

[Jonathan Yates]

On 3 Jul 2012, at 10:03, Hu Shuanglin wrote:
> 
> It maybe says that the wave function from UNK files are already not
> normalized. Is it
> correct?
> 
> 
> So is there anything wrong with my understanding and method? Can I use the
> renormalized wannier function to construct total charge density?

You are correct that the WF for the valence states should give the same charge density as the bloch states. But there can be a couple of tricky points when using ultrasofts or PAW.

To plot the WF Wannier90 asks for the periodic parts of the bloch states (unk). For a normconserving pseudopotential code these are normalised and so the wannier functions that are plotted will give a charge of 1. For USPs or PAW some codes write only the soft wavefunction into the unk file (QE does this - I guess VASP does too). As this is not normalised the WF that you plot will not give unit charge.

 Renormalising a USP or PAW wavefunction in the way that you describe is not physical. To get a correctly normalised UNK file from a USP or PAW calculation would be a bit difficult (not impossible, just tedious coding). However, note that it makes no difference to any of the WF properties computed with Wannier90 - it only concerns making a realspace representation for plotting.

 Jonathan

 



**Job Advert**
Postdoc position in First Principles Simulation at Oxford University
https://www.recruit.ox.ac.uk/pls/hrisliverecruit/erq_jobspec_version_4.display_form
**************

-- 
Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK
tel: +44 (0)1865 612797                http://users.ox.ac.uk/~oums0549/