Which potential energy from VASP is passed into I-PI ?

[袁亮（Leon Yuan）]

Hello,

I am running i-PI with VASP. Using the “pot_component_raw” tag, I obtained my system's potential energy in the i-PI output file. However, I found that this value is slightly different from all the energy terms printed to the OUTCAR of VASP, which are (1) “free energy    TOTEN”, (2) “energy without entropy”, and (3) “energy(sigma->0)”. May I please ask which potential energy from VASP is passed into I-PI?

Below are more specific numbers, for example:

(1) In the OUTCAR of VASP,

free  energy   TOTEN  =      -494.79749887 eV

energy  without entropy=     -493.56282018  energy(sigma->0) =     -494.18015953

(2) In i-PI output,

the “pot_component_raw{electronvolt}” gives -4.94175282e+02 

Thank you for your help.

Liang

[marian...@gmail.com]

Dear Leon,

Great question. I am guessing you are looking at a metallic system, given the large difference between the free energy and sigma->0. The free energy is what should be passed to i-PI and I am guessing there are some differences in conversion factors going on. 
I am not an expert in VASP but it seems that the VASP interface has a few issues. Could you please make sure you are using the latest VASP patches? As you can see they were updated 3 weeks ago: https://github.com/i-pi/i-pi/pull/250

If you can confirm you are using the latest implementation of the VASP<->i-PI communication and you still see this issue, please write it as an issue/bug in the github repo and provide example inputs and outputs to reproduce. We will fix it. Thanks!

[marian...@gmail.com]

Sorry - autocorrect -- I meant to write "Dear Liang..." ;-)