# VASP: Meaning of Certain Output Tags

This is just a quick question about the output tags in VASP. My output file is of the form:

           N       E                     dE             d eps       ncg     rms          rms(c)
DAV:   1    -0.127734350174E+04   -0.64220E-03   -0.73566E-02 10044   0.270E-01    0.621E-02
RMM:   2    -0.127734356359E+04   -0.61849E-04   -0.92296E-04  9392   0.279E-02    0.399E-02
RMM:   3    -0.127734355849E+04    0.51033E-05   -0.60427E-05  8579   0.665E-03    0.161E-02
RMM:   4    -0.127734355886E+04   -0.37419E-06   -0.11561E-05  7758   0.281E-03    0.279E-03
RMM:   5    -0.127734355917E+04   -0.31447E-06   -0.73221E-07  5335   0.891E-04
32 F= -.13021567E+04 E0= -.13021567E+04  d E =-.129375E-02
BRION: g(F)=  0.112E-02 g(S)=  0.000E+00 retain N=  3 mean eig= 1.81
eig:   4.418  0.476  0.522


I understand that F is the total free energy and dE is the change in energy between steps, but what about the next lines (BRION, eig)? I assume this has something to do with the method prescribed by IBRION (in this case I have set IBRION = 1).

I cannot seem to find information relating to the meanings of g(F), g(S), etc. I have noticed that when I set EDIFFG = 1E-3 eV the calculation only converges when both |dE| and |g(F)| are less than this value.

Is there any information available relating to the meanings of these terms?

Thanks!

Edit: Here is the INCAR file:

INCAR:
ENCUT = 400.000000
EDIFF = 1.00e-06
EDIFFG = 1.00e-03
ALGO = fast
GGA = PE
PREC = Normal
ISMEAR = 0
ISTART = 1
NELM = 120
IVDW = 12
NCORE = 4
LREAL = AUTO
IBRION = 1
NSW = 200

• @frobeniusThis is a very good question. Can you please share your INCAR with us so we can help?
– Tom
Commented Feb 3 at 15:55
• @Tom Hello, thank you for your response; I have edited my original post to include the INCAR file. Commented Feb 3 at 16:13

The answer by E.I. Joe is quite informative, I just would like to extend on that a bit. Starting from the first row:

N     : The number of electronic self-consistency iterations performed at the current ionic step.
E     : The total energy at the current electronic self-consistency iteration.
dE    : The change in total energy (E) between consecutive electronic self-consistency iterations.
d eps : The change in the charge density between consecutive electronic self-consistency iterations. it quantifies the convergence of the electronic structure or charge density.
ncg   : The number of conjugate-gradient (CG) iterations performed at the current electronic self-consistency iteration. CG iterations are used to solve the Kohn-Sham equations.
rms   : The root mean square (RMS) residual of the Kohn-Sham equations at the current electronic self-consistency iteration. This term quantifies the deviation of the electronic structure from self-consistency.
rms(c): The root mean square residual of the charge density at the current electronic self-consistency iteration.


Next lines stands for the method or algorithms used to solve the eigenvalue problem in the electronic structure calculations, in your case you have used AlGO=fast, thus there is a comnination of DAV which stands for Davidson diagonalization, and RMM which stands for residual minimization method.

Coming next is g(F) and g(S), as mentioned in the Joe's answer, those are related to the convergence criteria for the ion relaxation process which is defined by you via the EDIFFG term in INCAR file. specifically, g(F) stands for the convergence criterion for the forces acting on the ions, while g(S) represents the convergence criterion for the ionic stress tensor. These criteria are used to determine when the ionic relaxation has converged.

retain N: simply is an indicator for the number of electronic steps that were used to compute the electronic structure at the current ionic step. It shows how many steps were retained for the calculation.

mean eig: is the mean eigenvalue of the electronic Hamiltonian. It shows the average value of the eigenvalues of the occupied states (eig) at the current ionic step. Where eig are the eigenvalues of the occupied electronic states

• This was exactly what I was looking for, thanks. Commented Feb 20 at 0:24