NELECT and non-self consistent steps

Message

salva · #1 Post by **salva** » Tue Mar 20, 2007 11:05 pm

Hi, I am getting what I believe to be a "benign warning." I am performing a calculation on a ionized isolated system and I obtain a "very serious problem" right before I go for the first self-consistent upgrading of the Kohn-Sham hamiltonian:

DAV: 5 -0.710277187630E+03 -0.17486E+03 -0.17091E+03 2010 0.732E+01BRMIX: very serious problems
the old and the new charge density differ
old charge density: 500.00000 new 501.00000
0.626E+01
DAV: 6 -0.667205447730E+03 0.43072E+02 -0.14472E+02 1970 0.349E+01 0.454E+01
DAV: 7 -0.635470661548E+03 0.31735E+02 -0.22935E+02 2060 0.582E+01 0.338E+01

I guess it is there when part the algorithm "realizes" I asked for an extra electron in he system. I also guess this message is benign in my case, but would like to ask for a confirmation that this is just fine and I don't need to worry about. I know there may be convergence problems but my point is, adding electrons to the system I will always get this message (unless NELMDL=0) and I should (should I?) not worry about it.

Your comments and assistance will be greatly appreciated,
Salvador Barraza-Lopez.

#2 Post by **admin** » Thu Mar 22, 2007 1:06 pm

yes, the message is written at the first step of charge density mixing,
and at that point vasp recognizes that you calculate a system that is not neutral.

salva · #3 Post by **salva** » Thu Mar 22, 2007 1:25 pm

Thank you!
Salvador.

Vlad · #4 Post by **Vlad** » Fri Apr 06, 2007 5:41 pm

This may not be a benign problem. The number of electrons may be incorrect. Count the number of electrons using the numbers from POTCAR file, and adjust for the charge. This is the correct number for your calculation. Then look at the number of electrons obtained by integrating the density at the end of each SCF step, especially before and after the NELMDL steps (easiest to look for a string "magnetization" in the OUTCAR file, even if your calculation is not spin-polarized. The number of electrons will be to the left.) Make sure you get the correct number at the end of the calculation.

Vladimir

#5 Post by **admin** » Tue Apr 10, 2007 1:58 pm

if the number of electrons is changed (NELECT) such that the system is not neutral, this warning can be ignored.
In all other cases, change the the k-point density and/or the BZ integration method.