[Pw_forum] wavefunction coefficients

[giannozz at democritos.it]

Quoting Karoly Nemeth <nemeth at anl.gov>:

> I have tried out to generate the list of wavefunction coefficients by
> running pw2casino.x, on the same number of processors (Nproc), the pw.x
> calculation was running with. It generated Nproc pieces of files of the
> Cu111.pwfn.data* kind, each containing a list of wavefunction coefficients
> for a certain fraction of the G vectors. Each of these files indicates the
> number of G vectors contained, under "Number of G-vectors". Surprisingly,
> these numbers are not the same as what are listed on the top of the output
> of the pw.x calculation (under "columns G" and "(wavefct grid)"), for each
> processor. What is the reason for this difference?

I think the reason is that pw.x uses k=G vectors for each k-point, 
while the casino code for which this small utility [*] was written 
requires the same set of G vectors for all k-points, or something like 
this. Not sure that it works in parallel, though (it should but it will 
produce distributed output files). If you want to be on the safe side, 
follow Luke's advice: use wf_collect and run pw2casino
on a single processor

[*]I believed that pw2casino was a "inutility" and I am pleased to 
learn that it is actually useful to somebody.

Paolo Giannozzi
---
Democritos and Udine University

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.