Page 1 of 1

Problem compiling with DOS Projection vasp 5.2

Posted: Tue Sep 04, 2012 5:53 pm
by mrkent
I followed the directions specified here (http://theory.cm.utexas.edu/vtsttools/dos/) to compile this on Kraken (http://www.nics.tennessee.edu/computing ... es/kraken/). It compiled successfully, but when I try to run the example Cu_3Pt inputs, but I get this error when running vasp:

internal ERROR RSPHER:running out of buffer 0 0
18 1 0

1. I'm wondering if you guys have experienced this problem?
2. Is this even required to accurately use bader?

The reason I ask is, I had run bader on the CHGCAR from one of my calculations (without using -ref CHGCAR_sum), but ACF.dat returned 0 charge on 2 of the 5 hydrogen atoms in my system, which seemed suspicious.

35 4.7932 4.5111 17.2078 1.0432 0.3466 15.3204
36 -2.3233 4.7297 17.8280 0.9866 0.2360 14.5578
37 -3.4100 6.1288 17.6966 0.9538 0.2735 14.3639
38 -3.3276 5.8379 15.2675 0.0000 0.0000 0.0000
39 -2.2818 4.6061 15.4115 0.0000 0.0000 0.0000

Thanks.

Re: Problem compiling with DOS Projection vasp 5.2

Posted: Tue Sep 04, 2012 8:54 pm
by graeme
You do not have to modify vasp to use the bader code. The link that you mentioned is so that you can do a local projection of the DOS without writing a CHGCAR for each electronic state. To find the charges around atoms, you can just write the CHGCAR and analyze it with our Bader code.

The problem that you refer to is almost certainly due to missing core charge. See: http://theory.cm.utexas.edu/vtsttools/bader/vasp.php

Re: Problem compiling with DOS Projection vasp 5.2

Posted: Wed Sep 05, 2012 7:06 pm
by mrkent
Thanks for the quick response. I ran vasp with
LAECHG = .TRUE.
LADDGRID=.TRUE.
added to the INCAR, but I neglected the NGX, NGY, NGZ and NGXF, NGYF, NGZF-tags.
Bader ran successfully and gave perhaps reasonable results.

What are appropriate range of NG values?
I got a total electron number of 174. As long as this adds up correctly, the analysis is accurate?

Re: Problem compiling with DOS Projection vasp 5.2

Posted: Wed Sep 05, 2012 7:26 pm
by graeme
The total number of electrons will always be correct. You must include the core charges and look for convergence of the Bader charges with respect to the grid density.