Hi,
I encounter warning: " PSMAXN for non-local potential too small". I saw from previous discussions that ENCUT is too large compared to ENCUT and has to be decreased or harder potentials need to be used. I have two questions which I could not find answers in the forum
1- Despite this warning, VASP run flawlessly. Can one trust the results?
2- I am having elements Ta Ni Se and there are no harder potentials. I would like to have high accuracy on energies and test ENCUT. What would you recommend in such a case? If I use high ENCUT more than 600eV I got the warning.
Regards,
Burak
PSMAXN is too small
Moderators: Global Moderator, Moderator
-
michael_wolloch
- Global Moderator
- Posts: 110
- Joined: Tue Oct 17, 2023 10:17 am
Re: PSMAXN is too small
Dear Burak,
Whether you can trust your results depends on many more things than whether a warning is printed by the code or not. However, the warning should inform you that you are not operating in the parameter space for which your setup (structure, potential, etc.) was designed. So caution is necessary.
You do not state which potentials you use, but I assume the PBE potentials for Ta, Ni, and Se (ENMAX between 212 and 270), without any other suffixes. There are harder potentials available. In the potpaw_PBE.64 potential library, there are e.g. Ta_sv_GW (ENMAX=286), Ni_sv_GW(ENMAX=390), and Se_sv_GW(ENMAX=469), which have higher ENMAX and lower core radii.
If you provide a minimum reproducible example and state what properties you are interested in, I could help you more with questions about accuracy. Note that energy differences (e.g. between different structures or at different volumes) will converge much faster concerning the plane-wave basis compared to absolute energies. If you are not already doing it, always use energy differences in convergence studies.
All the best, Michael
Whether you can trust your results depends on many more things than whether a warning is printed by the code or not. However, the warning should inform you that you are not operating in the parameter space for which your setup (structure, potential, etc.) was designed. So caution is necessary.
You do not state which potentials you use, but I assume the PBE potentials for Ta, Ni, and Se (ENMAX between 212 and 270), without any other suffixes. There are harder potentials available. In the potpaw_PBE.64 potential library, there are e.g. Ta_sv_GW (ENMAX=286), Ni_sv_GW(ENMAX=390), and Se_sv_GW(ENMAX=469), which have higher ENMAX and lower core radii.
If you provide a minimum reproducible example and state what properties you are interested in, I could help you more with questions about accuracy. Note that energy differences (e.g. between different structures or at different volumes) will converge much faster concerning the plane-wave basis compared to absolute energies. If you are not already doing it, always use energy differences in convergence studies.
All the best, Michael