[Pw_forum] Why does dos.x look for evc.dat files?

[Andrea Ferretti]

Dear J.D.,

in principles you are right, the wfcs would not be needed.

Anyway, since dos.f90 is started according to all the postprocessing 
codes, it does some operations on wfcs.
In particular, at the end of the routine

~espresso/PW/read_file.f90

you would find
CALL pw_readfile( 'wave', ierr )

which is probably where the error is issued.

The possibility to restart without reading the wfcs could be added
(maybe a simple flag in pw_restart that comments out the above lines 
would work, but it may be also much more involved)...

deleting lines from data_file.xml  makes it probably 
unreadable and is not a good idea in general.

For your specific case, in order not to re-run the calculation you can try
(no guarantee) the following quick and dirty procedure:

* put the correct data_file.xml back in place
   (if you do not have this file I would start considering re-running the
   calculation or computing the DOS by yourself (I probably have a stupid
   f90 code for that, anyway) )

* comment out the line
   CALL pw_readfile( 'wave', ierr )  in   PW/read_file.f90

   recompile and try to run it...

* immediatly after remove the comment and re-compile !!!

andrea

BTW: if it takes more then 5 minutes to try all the above,
      I would go for a direct calculation of the DOS from the
      eigenvalues written in the output file. (let me know, in case)


> 
> I’m trying to run dos.x (using v4.2.1) but it keeps coming up with an 
> iotk error that it can’t find any evc.dat files in the K* directories of 
> prefix.save. Since the cell is pretty big I deleted the evc.dat files 
> some time ago (after I calculated the 
> projected LDOS) so my admin wouldn’t complain about using so much 
> hard-drive space. I just wanted to do a quick calculation of the total 
> DOS calculation with a smaller degauss.
> 
>  
> 
> 1.       Why would a calculation of the *total* DOS, not projected LDOS, want the eigenvectors?? The total DOS should only depend on the eigenvalues, k-point weights and a gaussian
> smearing factor.
> 
> 2.       Is there a way to get around this without having to recalculate the eigenvectors? I thought I could just delete any references to the evc.dat files in data-file.xml, but
> that just came up with another iotk read error.
> 
>  
> 
> I know I can just grab the eigenvalues and weights and do this myself (just add up a series of properly weighted gaussians, one at each eigenvalue) … but that seems to defeat the
> purpose of having a standardized code. J
> 
>  
> 
> Cheers,
> 
> J. D.
> 
>  
> 
> ************************************
> 
> J. D. Burton, Ph.D.
> 
> jdburton1 at gmail.com
> 
> University of Nebraska Lincoln
> 
> Physics and Astronomy
> 
> Office Ph. (402) 472 2499
> 
> Mobile Ph. (402) 419 9918
> 
> 310A Jorgensen Hall
> 
> CV: http://tinyurl.com/2avltsc
> 
> ************************************
> 
> "The job of a scientist is to generate wrong ideas as fast as possible."
> 
> -- Murray Gell-Mann
> 
>  
> 
> 
>

-- 
Andrea Ferretti
Oxford University, Department of Materials
Parks Road, Oxford OX1 3PH, UK
Tel: +44 (0)1865 612792;  Skype: andrea_ferretti
URL: http://quasiamore.mit.edu

Please, if possible, don't send me MS Word or PowerPoint attachments
Why? See:  http://www.gnu.org/philosophy/no-word-attachments.html