[Pw_forum] Details of partial density of states calculation

[Lorenzo Paulatto]

On 05/14/2013 10:26 PM, Holzwarth, Natalie wrote:
> Dear Stefano and Lorenzo,
>        Thanks for your emails.     There are actually 2 threads to my 
> email.   The first is to understand what is coded now and the second 
> is to add something in addition to the code if it is desired.    On 
> the first point, you say the projection is onto the reference atomic 
> functions, but I could not figure out which functions from the PAW 
> dataset were being taken for this projection. From "reading" the code 
>  It looks like it the  beta functions which are in fact the 
> projectors.     I am not sure if this is true and/or if these 
> functions are renormalized in some way.   Or perhaps it is a different 
> function that is taken for this.

The core of the projwfc code is the subroutine projwave, at line 230 of 
file projwfc.f90. Ignore all the IF(gamma_only) parts and all the 
symmetrization, and it becomes relatively clear. The beta functions are 
only used to orthogonalize the atomic pseudo wavefunctions of distinct 
atoms centers: these are ultrasoft-like wafunctions, hence they are 
orthogonal in the generalized sense, with the S matrix

Applying this S matrix should be equivalent to transform the "project on 
atomic wavefunction" operator from the all-electron form to the paw 
(pseudo-electron + all-electron^1 - pseudo^1) form. Here the one-center 
terms should cancel each other, however I never thought about this 
before now; it is possible I'm missing something.

>        About calculating the partial charges on the basis of the 
> charge in the augmentation charge, I will be glad to help if that is 
> convenient.   In addition to terms in the augmentation sphere there 
> would be a contribution from the plane wave expansion (or pseudo wave) 
> portion of the wavefunction.   What we did in the past is to just 
> calculate the charge within the sphere by integrating the Bessel 
> function expansion of the pseudo density for that state.  If you want 
> to keep the l-components it would be necessary to modify that pseudo 
> term according to their l-components.   Of course there are other 
> (perhaps more simple) ways to do this.    In principle, this is not 
> hard, but I would need to understand to code better....

I'm not sure I understand this part, I hope Stefano can be more helpful

>                                   Thanks, Natalie
>

you're welcome


-- 
Dr. Lorenzo Paulatto
IdR @ IMPMC -- CNRS & Université Paris 6
phone:+33 (0)1 44275 084 / skype: paulatz
www:  http://www-int.impmc.upmc.fr/~paulatto/
mail: 23-24/4é16 Boîte courrier 115, 4 place Jussieu 75252 Paris Cédex 5