[Pw_forum] phonon errors

vtmtrinh at caltech.edu vtmtrinh at caltech.edu
Wed Oct 7 09:07:18 CEST 2009


Dear PWSCF Users,

I have 2 questions:

1. I did a test phonon calculation for Si bulk, 2atom-cell, following
example_06.  I first did an scf calculation with k-point mesh 10x10x10 Si.
 Next, I wanted to do a phonon calculation along [001] direction (i.e q=0,
q=0.1, 0.2,...).  If I specified 'ldisp=.true.', it was supposed to do
nscf calculation and then phonon calculation in a series.  However, it
stopped with errors.  When I checked the output file, after it had
finished calculation at Gamma point and nscf calculations, when it entered
the phonon calculations for q !=0, it stopped with the error:
     task #         8
     from openfilq : error #         1
     file si.wfc not found
The run is supposed to run continously as I specified ldisp=.true.  Now
suddenly, it stopped with that error, so I could not understand why.  When
I tried with different q-point mesh, e.g. 4x4x4.  I ahd the same problem.

Here is my input for the phonon calculation:

phonons of Si bulk
 &inputph
  tr2_ph=1.0d-14,
  prefix='si',
  ldisp=.true.,
  nq1=1, nq2=1, nq3=10   !(or nq1=4, nq2=4, nq3=4 for second test)
  amass(1)=28.086,
  outdir='./',
  fildyn='si.dyn',
 /

2. If I run each single q-point,i.e. did scf, then Gamma calculation, then
nscf and phonon calculation for each single q-point, it was fine. I
obtained 6 dyn files for 6 q-points from 0 to 0.5. However, I had problem
when I used q2r.x.  In the input file, q2r.in, if I specified the q-point
mesh: 1 1 10, and below it I provided a list of 6 dyn files that I
calculated above, I got an error:

  task #         0
     from init : error #         1
     q not allowed

I am not sure what is the correct input for the q-point mesh in q2r.in
that I should use in this case.

Below is my q2r.in:

 &input
   fildyn='si.dyn', zasr='crystal', flfrc='si.fc'
 /
1 1 10
6
si.dyn1
si.dyn2
si.dyn3
si.dyn4
si.dyn5
si.dyn5

I am using QE version 4.0.3.

Thank you very much,

MyTrinh
JPL/CakTech






More information about the Pw_forum mailing list