[Pw_forum] About electron-phonon calculation on one point with version 3.1.1

Yingli Niu niuyingli at gmail.com
Wed Nov 8 03:03:40 CET 2006

Dear all
I have read example07/run_example  in espresso-3.1.1.
It is convenient to calculate the q-points for a full dispersion with
ldisp=.true. .
But before el-ph calculation, it is better to make a convergence test
on one or several q-points.
Then I have to use the script example07/run_example in espresso-3.0.
Because there is no example showing how to calculate el-ph on one
q-point in espresso-3.1.1.

I try to use example07/run_example  in espresso-3.1.1 to calculate
el-ph on one q-point:
1.  make a scf calculation on dense k-mesh for el-ph calculation
(case.scf.fit.in   with   la2F = .true.)
2.  make a scf calculation for phonon calculation(case.scf.in)
3.  make el-ph calculation:

cat > li.elph.in_ecut << EOF
Electron-phonon coefficients for Li
  reduce_io = .true.
  tr2_ph       = 1.0d-14
  prefix        = 'li'
  fildyn        = 'li.dyn'
  fildvscf     = 'li.dv'
  amass(1) = 6.941
  outdir       = '$TMP_DIR/'
  elph         = .true.
  trans       = .true.
  lnscf        = .true.
  ldisp        = .false.
0.00 -0.25 0.25

I thought that it would calculate phonons on q-point(0.00 -0.25
0.25)because of "trans=.true." and "lnscf=.true.".
But I failed.

I make a calculation as in espresso-3.0
1.  make a scf calculation on dense k-mesh for el-ph calculation
(case.scf.in  without  la2F = .true.)
2.  make a nscf calculation for phonon calculation (case.nscf1.in,
3.  make a phonon calculation ( with  elph=.flase. ,  trans=.true. ,
4.  make a nscf calculation for elph calculation (case.nscf1.in,
5.  make a el-ph calculation (with   elph=.true. ,  trans=.true. ,
I got the results!

My question is:
1. Must we calculate one q-point el-ph using run_example in espresso-3.0?
    That is to say the method which I test (scf.fit, scf, el-ph) is wrong!

2. If we got error result for el-ph calculation  on some q-point with
ldisp=.true., when should we
    set restart_mode = 'restart'? In case.scf.fit.in or case.scf.in
in espresso-3.1.1?

3. What's wrong with my script ?
    It should have calculated phonon and el-ph because it set
trans=.true. ,elph=.true. to calculate
    on 0.00 -0.25 0.25?

More information about the users mailing list