[Pw_forum] filelph question

[Miguel Martinez]

Dear collegues,

I have (finally) writen my script to read phonon frequencies and el-ph 
coefficients and put them in a lambda.x readable form. Apart from 
deleting the double delta lines it seemed to be fine until...

$ ~/espresso/pwtools/lambda.x < lista
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
      from lambda : error #         2
      inconsistent DOS(Ef) read
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
      stopping ...

Well, pwtools/lambda.f90 checks many things, and this yells because the 
DOS of the first file was different to the DOS in the second file. As 
the differences were really small (on the very last number, could be due 
to numerical accuracy) on only a line, I changed the values on the 2nd 
and the 3rd filelph. No big deal.

However, the 4th file (q=0.5 0.5 0.5) still fails. This time, the 
differences are larger (4th decimal digit in DOS, a few tenths of a meV 
in the fermy energy). I certainly don't want to touch this.

Is this caused by numerical accuracy? Or is this, as I fear, a lack of 
convergence of my calculations? The thing is, runs were performed with 
the following cutoffs:

   ecutwfc =40.0, ecutrho=250,

I would gladly hear any comments. If you ask about the k-point grid, it 
is a 20 20 20 M-P grid (fcc cell), which is not as large as I would 
like, but the compiler bug I suffered kept me from using a larger one 
(24 24 24 or beyond).

Thank you very much in advance,

Miguel

-- 
----------------------------------------
Miguel Martínez Canales
Dto. Física de la Materia Condensada
UPV/EHU
Facultad de Ciencia y Tecnología
Apdo. 644
48080 Bilbao (Spain)
Fax:  +34 94 601 3500
Tlf:  +34 94 601 5437
----------------------------------------

2005 International Year of Physics