[QE-users] root not converged in phonon calculation

[Marco Infantino]

Hello !

I am doing a phonon calculation with one of my structure, but I always 
have the following error

"     Electric Fields Calculation
      kpoint   1 ibnd 425 solve_e: root not converged  1.291E-02
      kpoint   1 ibnd 425 solve_e: root not converged  1.961E-02
      kpoint   1 ibnd 425 solve_e: root not converged  9.242E-03
      kpoint   2 ibnd 425 solve_e: root not converged  2.195E-02
      kpoint   2 ibnd 425 solve_e: root not converged  1.554E-02
      kpoint   2 ibnd 425 solve_e: root not converged  1.200E-02

       iter # 171 total cpu time :  1405.0 secs   av.it.: 192.3
       thresh= 1.000E-02 alpha_mix =  0.500 |ddv_scf|^2 = 5.291E+57
      kpoint   1 ibnd 425 solve_e: root not converged  1.993E-02
      kpoint   1 ibnd 425 solve_e: root not converged  7.971E-03
      kpoint   1 ibnd 425 solve_e: root not converged  1.208E-02
      kpoint   2 ibnd 425 solve_e: root not converged  1.705E-02
      kpoint   2 ibnd 425 solve_e: root not converged  8.628E-03
      kpoint   2 ibnd 425 solve_e: root not converged  1.024E-02 "


It seems from previous similar error in the forum, that I have to play a 
bit with the value of alpha_mix (I have tried with bot alpha_mix=1.00 
and alpha_mix=0.500) and I have to reduce the value tr2_ph (I am using 
1e-7 at the moment).

So far anything did change and this error always appears. I see also 
that the |ddv_scf|^2 (is ti one of the element of the hessian ?) it 
totally crazy. Honestly my structure seems to be very reasonable and 
easely converges with an SCF calculation.

Is there some suggestion to overcome this problem ?

Thanks,

Marco
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20240620/c04fc162/attachment.html>
-------------- next part --------------
phonons (vibrational modes) of structure
 &inputph
  recover = .true.
  outdir='./',
  prefix='scf_image_4_ts_001_no_charge',
  amass(1)=1.00784,
  amass(2)=15.999,
  fildyn='dyn.dynG',
  tr2_ph = 1e-7
  alpha_mix = 0.500
  trans= .true.
  epsil=.true.,
 /
0.0 0.0 0.0