[QE-users] Convergence issue in HP code with nbnd

[Iurii TIMROV]

Dear K C Bhamu,


> The best overall accuracy which was reached :
>       diff =    0.0000114935  iter = 273


Just adjust a bit the threshold from

conv_thr_chi = 1.0d-5

to

conv_thr_chi = 2.0d-5


> Is it okay, if I simply use alpha_mix(1)=0.2 (or some smalled value) ?


Yes. The value of 0.2 will be used for all iterations.


HTH


Iurii


--
Dr. Iurii TIMROV
Senior Research Scientist
Theory and Simulation of Materials (THEOS)
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Dr. K. C. Bhamu <kcbhamu85 at gmail.com>
Sent: Monday, November 29, 2021 7:26:08 AM
To: Quantum Espresso users Forum
Subject: [QE-users] Convergence issue in HP code with nbnd

Dear Iurii,

As per our previous conversation,  I am running my Fe3O4 case with nbnd and without nbnd. It seems that calculations without nbnd is running file (second self-consistent calculations has finished based on U+V from 1st SCF) but with nbnd, I see HP calculations are not converging for 2nd q-point of 5th atom.

I am perturbing three atoms in my Fe3O4 system as below:
     List of  3 atoms which will be perturbed (one at a time):

        1      Fe1    55.8450   tau( 1) = (  0.00000  0.00000 -0.00000  )
        5      Fe2    55.8450   tau( 5) = (  0.75736  0.43726  0.30920  )
        7      O      15.9994   tau( 7) = (  0.49572  0.28620  0.20238  )

I am getting below message:

atom #  5   q point #   2   iter # 300
"NOT MENTIONING SOME LINES HERE TO KEEP THE EMAIL SHORT"

      Average number of iter. to solve lin. system:   10.6
      Total CPU time : 60285.4 s
      Convergence has not been reached after 300 iterations!
      The best overall accuracy which was reached :
      diff =    0.0000114935  iter = 273


Here is my HP input file:

 &inputhp
   prefix = 'pwscf',
niter_max=300
   outdir = './temp/',
   nq1 = 2, nq2 = 2, nq3 = 2,
   conv_thr_chi = 1.0d-5,
find_atpert = 1
docc_thr=1.d-3
!disable_type_analysis=.true.,
/

I understand that I may need to reduce the mixing factor of alpha_mix(i). But could you please elaborate some more, how how can I take/understand this i-th iteration?
 Because in the description, it is mentioned that alpa_mix is applied in some for the i-th iteration.

Is it okay, if I simply use alpha_mix(1)=0.2 (or some smalled value) ? Or do I need to take care of something else?

Thanks and Regards

K C Bhamu
Postdoctoral Fellow
University of Ulsan
ROK
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20211129/64a2cd63/attachment.html>