Hello GUYS,
I Need your help realted to Quantum Espresso. Iam running Spin orbit caluculations for CaTe. things got well for the case when we not taking SOC but while considering SOC Iam not getting the results. Means DOS and BAND giving a single line. Then I checked my Scf.out . There is a message with each iterations.
Davidson diagonalization with overlap
c_bands: 1 eigenvalues not converged
You can check input and output files below.
Any help will be appreciated.
Thanks,
Abhinav Nag
Thanks Sanjiv and Bojidarka. I will repeat the calculations by lowering the mixing_beta value.
Abhinav Nag
Hi Abhinav, did lowering the mixing_beta work for your system? I am currently facing the same issue. However, I am using Conjugate Gradient (CG) style diagonalization (and not Davidson).
Could you please tell whether you got rid of the warning/s in the output? Thanks in advance!
-Shalmalee.
I am facing the same problem for nscf calculation. Lowering the mixing_beta is not working also.
Anish Das Hi. I did get rid of that problem. I don't remember exactly how, right now. But I played with the parameters, changed the diagonalization and/or conversion threshold as well. I changed some of the parameters from &SYSTEM too, just to check if the warning goes away. I don't know whether you have solved it or not, but no worries, the warning doesn't affect the calculation. Hope this helps.
Sorry for the delay in response, but I missed the notification. Please tag me next time if you have any more doubts, that way I can get notifications and can respond in time. Thanks.
Shalmalee Kapse Hi, yes i also got rid of the problem. Four ways to slove it 1) Increase the cut off (ecutwfc, hence ecutrho), 2) decrease conv_thr, 3) Do both simulteneously, 4) Decrease k meshing. Any one of the above four will solve the problem.
Yes you are right, generally warning does not affect the result, however, too many warnings eventually crash the programme. This happened for me.
Set the number of nbnd appropriately , ususally you would most likely be using a small nbnd value. check the number of khon sham states and use appropriately
Anoop A Nair my unit cell calculation='bands' didn't show any errors of eigenvalues not converging but supercell with defect is showing these errors. I didn't even specify nbnd in unit cell calculation neither did I in the supercell. How do I specify it to get rid of this problem? Thanks
- Badges
- Science topic
- Similar topics
- Correction