Why did GGA+U give different energies with the same input?

[zongtan_fang1]

I am runnning the gas solid reation on a surface of NMC cathode material and found out that the GGA+U give different energies with the same input starting from the same Input. The different energies I am thinking are due to different magnetic moments of some transition metals.
What causes that problem? Wrong U values?
How do I know which one is right? Also how can I know if I get the global minimum?

Thanks a lot.

Zongtang

[alex]

Hi!

That's a lot of questions. I'm trying to get to them one by one.
VASP is using a random wavefunction as initial guess. This might lead to different spin states.
You choose the one with the lowest energy. That's just basic quantum mechanics.
Well, is it the lowest possible? Excellent question. Use your chemical knowledge and check right and left from the solution you have.

Hth, alex

[zongtan_fang1]

Hi Alex,

Thanks a lot for your reply.
What I am doing is scanning a range of possible spin states by manipulating different MAGMOM values.
Hopefully will get some reasonable results.

Zongtang

[rolando_saniz1]

The use of PBE+U on magnetic systems can be problematic because the localization enforced by U may lead the system to be stuck in a metastable state. If the energy difference between metastable states is small, different runs may end up in different states. Look at the paper Allen & Watson, PCCP 16 (2014) 21016 for a suggested workaround.

Best,
Rolando