Hello,
Many thanks for your answers, Francois.
There two types of potentials, ONCV_PBE-1.2 and HSCV_PBE-1.0, for qbox and I'd like to know differences between them.
Would you let me know to what applications they are more suitable and/or needed essenitially, respectively?
It seems that ONCV needs more computational cost than HSCV.
For spin-orbit coupling calculation, would ONCV be better than HSCV?
Can I use HSCV not only for example cases but also for liquid water AIMD with a rather high accuracy for hydrogen-bond and/or IR spectrum analysis?
Best regards,
Young
potential files of qbox
Forum rules
You must be a registered user to post in this forum. Registered users may also post new topics if they consider that their subject does not correspond to any topic already present on the forum.
You must be a registered user to post in this forum. Registered users may also post new topics if they consider that their subject does not correspond to any topic already present on the forum.
-
fgygi
- Site Admin
- Posts: 164
- Joined: Tue Jun 17, 2008 7:03 pm
Re: potential files of qbox
Hello Young,
Both the HSCV and ONCV potentials should be adequate for water simulations. The PBE400 dataset http://www.quantum-simulation.org/refer ... /index.htm and the SCAN330 dataset http://www.quantum-simulation.org/refer ... /index.htm were generated using the HSCV potentials. The ONCV potentials have additional non-local projectors, which should make them more accurate. This incurs an additional computational cost. However, the HSCV potentials require a higher plane wave cutoff (85 Ry) than the ONCV (which is apparently good with 65 Ry). I do not have accurate comparisons of timings of these potentials in the same conditions.
Regarding spin-orbit interactions, note that Qbox does not currently compute such interactions. If you mean "spin-polarized", this is supported by setting the variable
If you are interested in MD simulations for IR spectra, HSCV may be ok, but you should use an 85 Ry cutoff. I would still recommend using ONCV in spite of a possible addiional cost.
Note also that if you want to compute the stress tensor, the plane wave cutoff should be increased. Please refer to the PBE400 and SCAN330 papers for details (see links above).
Francois
Both the HSCV and ONCV potentials should be adequate for water simulations. The PBE400 dataset http://www.quantum-simulation.org/refer ... /index.htm and the SCAN330 dataset http://www.quantum-simulation.org/refer ... /index.htm were generated using the HSCV potentials. The ONCV potentials have additional non-local projectors, which should make them more accurate. This incurs an additional computational cost. However, the HSCV potentials require a higher plane wave cutoff (85 Ry) than the ONCV (which is apparently good with 65 Ry). I do not have accurate comparisons of timings of these potentials in the same conditions.
Regarding spin-orbit interactions, note that Qbox does not currently compute such interactions. If you mean "spin-polarized", this is supported by setting the variable
nspin to 2.If you are interested in MD simulations for IR spectra, HSCV may be ok, but you should use an 85 Ry cutoff. I would still recommend using ONCV in spite of a possible addiional cost.
Note also that if you want to compute the stress tensor, the plane wave cutoff should be increased. Please refer to the PBE400 and SCAN330 papers for details (see links above).
Francois