QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: abc123def on October 24, 2016, 15:28
-
Is there any way to include noncollinear spin-orbit coupling while performing a geometrical optimization?
Thank you.
-
Nothing special is needed for this. Your problem (in the other post) is related to TB09, not spin-orbit.
-
I tried with SOGGA, PBE this time. Still having the same problem. I have attached the new code and log.
Please help.
-
I see you are using an older version of ATK - the script actually works fine in 2016. However, it will still give wrong results. The pseudpotential you have chosen (FHI) does not contain spin-orbit coupling. In 2015 your only option is to use the OMX pp/basis sets, in which case you should remember to use a very high mesh cut-off (maybe 300 H), because the OMX potentials are very hard.
Do you really think the spin-orbit effect is important for the geometry, though? I would optimize the geometry using GGA/LDA and then compute the electronic structure with spin-orbit turned on.
-
The main reason I am trying to incorporate spin orbit coupling is that I mas working with a metastable state of Ta. In other codes like quantum espresso and Siesta, the optimized value of lattice constant is coming to be around 4.2 - 4.3 Angstrom using GGA, when the experimentally observed value is 4.4 Angstrom. I am trying to see if soon orbit coupling plays any role in determining lattice constant.
I have one more question: can I update Atk to 2016 version without damaging the license configurations and parallelization environment easily or do I have to reinstall everything? Because, I found out that there are some changes in OMX paeudopotentials.
-
Ok, well, then it makes sense. Actually I would then try rather than SG15 pseudpootentials in 2016, which are softer and give very nice results, usually. You can always have many versions of ATK installed, they don't interfere and just installing the software doesn't mess up any settings.
-
Thanks a lot for yournhelp