[QE-users] "good" and "bad" q-points (ph.x)

[Christoph Wolf]

Dear all,

I have observed that the convergence of phonon calculations sometimes
strongly depends on the choice of q-point without being able to understand
which points should better be avoided.

Take for example graphene. I sample it with 33x33x1 k points for the scf
step. This choice of grid contains, among others

     .....
        k(  106) = (   0.3030303   0.5248639   0.0000000), wk =   0.0110193
        k(  107) = (   0.3030303   0.5598548   0.0000000), wk =   0.0220386
   **   k(  108) = (   0.3333333   0.5773503   0.0000000), wk =   0.0036731
**

The latest k point (in cartesian units) is the K-point where the diract
point lies. Subsequently I find that a phonon calculation that contains
q=k(108) such as a 15x15x1 grid

     Dynamical matrices for (15,15, 1)  uniform grid of q-points
     (  27 q-points):
       N         xq(1)         xq(2)         xq(3)
       1   0.000000000   0.000000000   0.000000000
       2   0.000000000   0.076980036   0.000000000
....
      25   0.266666667   0.461880215   0.000000000
      26   0.266666667   0.538860251   0.000000000
  *** 27   0.333333333   0.577350269   0.000000000 ***

will fail as follows:

     Representation #  3 modes #   3  4
     Self-consistent Calculation
      iter #   1 total cpu time : 64587.1 secs   av.it.:   7.1
      thresh= 1.000E-02 alpha_mix =  0.700 |ddv_scf|^2 =  2.917E+03
      ......
      iter #  14 total cpu time : 65207.4 secs   av.it.:  45.2
      thresh= 1.000E-02 alpha_mix =  0.700 |ddv_scf|^2 =  5.244E+40

(playing with alpha_mix(niter) does not seem to help much here).

Now the question I have been asking myself is if there is a "good" and a
"bad" choice of SCF k points and phonon q-points? If someone has any
experience related to the proper choice of points I would be very glad to
hear!

Thanks in advance and with best regards,
Chris

-- 
Postdoctoral Researcher
Center for Quantum Nanoscience, Institute for Basic Science
Ewha Womans University, Seoul, South Korea
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181001/c7925c50/attachment.html>