Energies in the OUTCAR file. What is what?

[Tyx]

Hello everyone,

Could someone please clarify once and for all, what each energy corresponds to in the output? Specifically...:

-If energy(sigma->0) is the internal energy U, does that mean that the 'total free energy' above it corresponds to the Helhmoltz free energy F=U-TS?

-What if you were able to set T=0 for your calculations, does that mean that the 2 values above would be equal?

-What about the 'energy without entropy' term, isn't that the same as F, according to F's definition? (No it isn't.)

- And if energy (sigma->0) is the internal energy U as was said by the head admin in another thread, when enthalpy H=U+PV is output in OUTCAR, why is it that the 'total free energy' value is used instead, to calculate it?

Some people I have asked always take the (sigma->0) value as their groundstate energy, others take the 'total free energy' value. And noone is 100% sure about their choice. The VASP manual is not very clear on it either, I'm afraid.
And in a recent thread, someone said that he takes the
'energy without entropy' when he does calculations in metallic systems (?)

Could you please try to give a definite answer to my torrent of questions??

Thanks a lot!

P.S. My calculations are mostly on metallic systems, so I use ISMEAR=1 and SIGMA=0.2. (For the MP method the sigma->0 energy was pointed as the value to use by the admin in another thread. Why?)

<span class='smallblacktext'>[ Edited ]</span>

[forsdan]

I'm not this is an accurate explaination, but your goal is to calculate the total energy at 0 K. This means that the Fermi function / occupation probability is given by a a step function. With a discete set of k-points the evaluation of integrals with the fermi function will have slow convergence. In order to improve the convergence one introduce a smooth smearing funtion, i.e. you set a small temperature. The calculations are then carried out and then you interpolated the result back to zero smearing width. The value you therefore should use for the total energy is the energy (sigma->0). You can read more about it here

http://cms.mpi.univie.ac.at/vasp/vasp/node191.html
http://cms.mpi.univie.ac.at/vasp-worksh ... points.pdf

Best regards

Dan Fors

[Tyx]

Thanks Dan,

Unfortunately that doesn't really answer my question about what each value corresponds to, in thermodynamic terms..

You suggest to use the sigma->0 energy, which as far as i can tell corresponds to the true Kohn-Sham ground state energy, in other words to U. Does that mean then that the 'total free energy' corresponds to F? or maybe G?

However, as I said above, when enthalpy is calculated (usually when you set a fixed pressure it is output automatically in OUTCAR), the 'total free energy' value is used instead, as U. And not only that, all the forces are calculated as a derivative of that value, not the sigma->0 one.

And all this makes people use that energy for everything else, as well. *shrug*

Also, I still am not clear on what does the 'energy without entropy' correspond to, and why it was said it's the one to use for metallic systems.

Any more input would be appreciated

[Tyx]

....

I found a lot of answers in G. Kresse, J. Furthmuller/ Computational Materials Science 6 (1996) 15-50 (Section 2)

However I'm still not sure if energy(sigma->0) is the definite right choice for everything. The article says their difference is negligible and confirms what I said that it actually uses the 'total free energy TOTEN' instead of the 'physical' energy(sigma->0) to calculate forces and other properties.

Can anyone say whether it's very important to choose sigma->0 instead of TOTEN, or not? My feeling is that if you want this energy for comparison of structures, or to use in calculating something like an equation of state, and you are consistent in choosing one kind, it will not make a difference to your final result. Am I terribly wrong in thinking that?

[ihask]

Tyx, you have the exactly same questions I have.
I will read the article you recommended. Thanks!!