Differences between CHG and AECCAR files

Message

marinok · #1 Post by **marinok** » Mon Sep 22, 2008 5:41 pm

What is the difference between the CHG and AECCARs files? I know the CHG contains only the valence density and that AECCAR0 contains the core charge density from PAW potentials, but what about AECCAR1 and AECCAR2?

#2 Post by **admin** » Tue Sep 23, 2008 9:58 am

please read the README file incuded in the vasp.4.6 directory for the data written on the 3 AECCAR files:
AECCAR0: core charge density
AECCAR1: atomic AE (all electron) charge density (overlapping atomic charge density)
AECCAR2: AE charge denisty

Danny · #3 Post by **Danny** » Tue Mar 06, 2012 3:13 pm

[quote="admin"]please read the README file incuded in the vasp.4.6 directory for the data written on the 3 AECCAR files:
AECCAR0: core charge density
AECCAR1: atomic AE (all electron) charge density (overlapping atomic charge density)
AECCAR2: AE charge denisty[/quote]

Is this information still correct? I tried integrating the charge in those files (using the same method as I use for the CHGCAR file) and I find in case of a CeO2 unit cell a total charge of > 1360e in AECCAR0, and ~25e for AECCAR1 and 2.
Integration of CHGCAR gives 24e (12 from Ce + 2x 6 from O) so this ssems to work correctly, then why do I get such strange numbers? I would expect ~74e but not >1360 or 25.

What is actually being stored in those files?

kind regards
Danny

graeme · #4 Post by **graeme** » Tue Mar 06, 2012 3:57 pm

The AEC* files also do not seem to have the same normalization as the CHGCAR file. They only integrate to the correct total charge in the limit of a fine FFT grid. Particularly for the AECCAR0, which has the charge density peaks in the atomic cores, you have to use an extremely fine FFT grid to start approaching the correct total core charge.

Danny · #5 Post by **Danny** » Wed Mar 07, 2012 3:32 pm

After doing some additional calculations with extremely fine grids, I get results which are only of 0.5e (which is 1% error!!). So apparently this is only doable for very small systems (not let's say a molecule or atom in free space :s ) due to memory limitations. Good to know though.

Danny