[Q-e-developers] PW lists, PW coefficients

[Paolo Giannozzi]

Dear Warren

do you need to extract data during the run, or from files written by
PWscf at the end of the calculation? In the latter case, my advice is
to use one of the many postprocessing codes that read data, modify
them according to your needs, without messing up with the actual
format of files. You may also save a file in a straightforward format
using the "--pw2casino" option, producing output for CASINO code. It
is described in some detail in the user guide.

Variables you may need are:
G(i,n) = cartesian component i of G-vector n, in 2pi/a units
rho%of_g(n) = charge density for G-vector n (au, defined so that
rho%of_g(G=0)*Omega=Nelectrons, where Omega = unit cell volume)
evc(j,m) = component j of wavefunction m. Component j corresponds to
k(:) + G(:,igk_k(j,k)) where k is the index of k-point xk(:,k), igk_k
are indices computed at the beginning.

The reason why the charge density is written in real space is its
well-defined ordering, once the three crystal axis are defined.
Ordering of G-components is not uniquely defined due to shell
degeneracy. We have a solution for that, but until now nobody has
found the time and motivation to do that, or more exactly: to do that
before other things. I can explain much more on why variables and
files are organized the way they are, but it will take 5 hours instead
of 5 minutes.

Paolo

On Fri, Apr 28, 2017 at 11:06 PM, Warren Pickett <wepickett at ucdavis.edu> wrote:
> I (my group) needs to get from QE things that must be running around the
> code in several places, to allow further analysis:
>
> * the list of plane waves used in an scf calculation.
>
> * the coefficients of rho, and the electronic potential (without the
> pseudopotential) V_H + V_xc
>
> I suppose if we are told which arrays in which subroutines, we can write
> them out. I am surprised they are not saved in that form.
>
> I am told QE retains the latter only on a real space mesh. I'm from the old
> school where we stored Fourier coefficients. Group members made some sort of
> effort to get help from users, without success.  For someone who knows the
> code, there is a 5 minute explanation. I looked through relevant subroutines
> but without enough sustained time to figure out much.
>
> Thanks.
>
> Warren E. Pickett
> Distinguished Professor of Physics
> Department of Physics
> University of California Davis
> Davis CA 95616, USA
>   Cell: 530-220-2138
>
> _______________________________________________
> Q-e-developers mailing list
> Q-e-developers at qe-forge.org
> http://qe-forge.org/mailman/listinfo/q-e-developers
>



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222