Incorrect core charge?

[yeehaw]

Hi Graeme,

On http://theory.cm.utexas.edu/vtsttools/bader/vasp.php you explained that one needs "a fine fft grid to accurately reproduce the correct total core charge." Just out of curiosity I processed AECCAR0 and AECCAR2 for a job separately using bader, i.e., I did

bader AECCAR0
bader AECCAR2

In the ACF.dat produced by the latter the total number of electrons is equal to the total number of valence electrons. But in the ACF.dat produced by the former I got a number roughly 2x the total number of core electrons there really is. My question is, is my core charge correct? How do I properly understand these results?

Thanks.

[graeme]

The total charge in the AECCAR0 will be the core charge in the limit of a fine grid. But since you have such large values of charge density near the core, the integrated value can very significantly depending upon how this maximum is aligned with the grid. This is why you will have such large fluctuations in the total core charge density by integrating AECCAR0.

There is no such issue for the smooth density in the valence electrons. There, the integrated charge should be correct -- as you found.

[yeehaw]

Thanks, Graeme. You are right. I doubled NGXF, NGYF, and NGZF and the total number of core electrons closely approached the correct number. Happy New Year!