[Pw_forum] help for convergence in "Computing Second order response"

[hyello25 Qgrui]

hi everyone,
     When I do raman calculation of Li2O, I found the "second order
response" process is not convergent. The log is below:

      iter #   1   av.it.:   8.1      thresh= 0.100E-01 alpha_mix =
0.700|ddv_scf|^2 =
0.208E-03
      iter #   2   av.it.:  13.1     thresh= 0.144E-02 alpha_mix =
0.700|ddv_scf|^2 =
0.905E-01
      iter #   3   av.it.:  11.0     thresh= 0.100E-01 alpha_mix =
0.700|ddv_scf|^2 =
0.228E-03
      iter #   4   av.it.:   8.8      thresh= 0.151E-02 alpha_mix =
0.700|ddv_scf|^2 =
0.178E-03
      iter #   5   av.it.:  10.9      thresh= 0.133E-02 alpha_mix =
0.700|ddv_scf|^2 =
0.118E-08
      iter #   6   av.it.:  12.6      thresh= 0.343E-05 alpha_mix =
0.700|ddv_scf|^2 =
0.155E-07
      iter #   7   av.it.:   9.6      thresh= 0.124E-04 alpha_mix =
0.700|ddv_scf|^2 =
0.279E-08
      iter #   8   av.it.:  11.9      thresh= 0.529E-05 alpha_mix =
0.700|ddv_scf|^2 =
0.191E-08
      iter #   9   av.it.:  11.9      thresh= 0.437E-05 alpha_mix =
0.700|ddv_scf|^2 =
0.402E-07
      iter #  10   av.it.:  10.9      thresh= 0.201E-04 alpha_mix =
0.700|ddv_scf|^2 =
0.437E-08
.............................(omitted)
      iter #  42   av.it.:  10.7      thresh= 0.855E-07 alpha_mix =
0.700|ddv_scf|^2 =
0.201E-11
      iter #  43   av.it.:  11.9      thresh= 0.142E-06 alpha_mix =
0.700|ddv_scf|^2 =
0.181E-12
      iter #  44   av.it.:  13.6      thresh= 0.425E-07 alpha_mix =
0.700|ddv_scf|^2 =
0.546E-10
      iter #  45   av.it.:  10.2      thresh= 0.739E-06 alpha_mix =
0.700|ddv_scf|^2 =
0.190E-11
      iter #  46   av.it.:  11.0      thresh= 0.138E-06 alpha_mix =
0.700|ddv_scf|^2 =
0.111E-11

and the ph input file as:
&inputph
   tr2_ph=1.0d-14,
   prefix='li2O',
   amass(1)= 6.941,
   amass(2)=15.999,
   fildyn='li2O.dynG',
   epsil=.true.,
   trans=.true.,
   lraman=.true.,
   outdir='/home/qgr/_work_/li2O/raman/temp',
/
0 0 0
    And in the scf input file, conv_thr =  1.0d-12 with my own Li and O PPs
generated using fhi98PP code.
    How can I solve the problem?
    What is more wried, this situation never happened days before when using
the exactly same input file and PPs. That is , I cannot repeat the result
again with all the same input file though I check all the input file
carefully. The scf result are the same, but the ph.x gave different
convergence process. Besides, I got different raman frequencies.

Thanks in advance.

Bests, Qian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20071026/67fcb15d/attachment.html>