[Pw_forum] wave function coefficients

[Gabriele Sclauzero]

Dear Priya,

Johari, Priya wrote:
> The calculations were for graphene, where I used vacuum of 12A, k-mesh = 45x45x1,
> ecutwfc = 40.0 and  ecutrho = 400.0.

You missed to give the most important information, namely if you are using ultrasoft or 
norm-conserving pseudopotentials... I guess USPPs, since you are using ecutrho > 4*ecutwfc.

Hence you need the S matrix, which is not stored at the end of the run, as you learned 
from PG. Its pieces (D coefficients + screening, beta projectors) are computed in various 
initialization routines of the PW code or of the PP programs (see for instance 
read_file.f90 in PW/, it should be quite instructive). The main initializations for S are 
performed in init_us_1, calbec in the becmod and newd (maybe there is something else I 
left out...).

GS

> 
> For some post processing purpose I was needed plane wave coeeficients and thats why I
> converted evc.dat, gkvectors.dat files into ASCII format first but found the problem
> mentioned below (in previous mail).
> 
> 
> PJ, Department of Engineering, Brown University, Providence, RI. USA.
> 
> 
> -----Original Message----- From: pw_forum-bounces at pwscf.org on behalf of Gabriele
> Sclauzero Sent: Mon 2/15/2010 9:09 AM To: PWSCF Forum Subject: Re: [Pw_forum] wave
> function coefficients proble Hello,
> 
> Johari, Priya wrote:
>> Hi,
>> 
>> It seems that the wavefunction coefficients are not normalized.  (i.e. sum of the
>> magnitudes squared of the coeffs does not equal 1).
> 
> This seem quite unlikely to me.
> 
>> For example, for first four states in the directory k00001/evc.dat I am getting
>> magnitudes, for example:
>> 
>> state   mag^2 1          1.08 2          0.96 3          0.92 4          0.92
>> 
>> What's strange is that the state vectors are still orthogonal to each other (i.e.
>> state 1 is perpendicular to states 2,3, 4 etc.)  It's like the coefficients are off
>> by a scale factor.   It seems too large to be a rounding error.
> 
> Can you tell us which procedure you followed to compute these magnitudes, for which 
> system, ...? Remember that with USPP the famous S overlap matrix should be taken into
> account!
> 
> And please supply your affiliation, thanks.
> 
> 
> GS
> 

-- 


o ------------------------------------------------ o
| Gabriele Sclauzero, PhD Student                  |
| c/o:   SISSA & CNR-INFM Democritos,              |
|        via Beirut 2-4, 34014 Trieste (Italy)     |
| email: sclauzer at sissa.it                         |
| phone: +39 040 3787 511                          |
| skype: gurlonotturno                             |
o ------------------------------------------------ o