bader - cube files

[egillsk]

Hi

I am using the DFT code dacapo and needs to do a bader analysis. I have created a cube file that seems to be a little bit wrong though, but I am not sure what that is. Here is the beginning of the file:

__________________________________________
ASE CUBE FILE
OUTER LOOP: X, MIDDLE LOOP: Y, INNER LOOP: Z
37 0.000000 0.000000 0.000000
56 0.287768 0.000000 0.000000
32 0.000000 0.290751 0.000000
120 0.000000 0.000000 0.299207
78 0.000000 2.394552 1.112488 -4.426455
78 0.000000 7.766228 1.112488 -4.426455
78 0.000000 13.137905 1.112488 -4.426455
78 0.000000 -0.291287 5.764497 -4.426455
78 0.000000 5.080390 5.764497 -4.426455
78 0.000000 10.452067 5.764497 -4.426455
78 0.000000 -0.291287 2.663158 -8.812407
78 0.000000 5.080390 2.663158 -8.812407
78 0.000000 10.452067 2.663158 -8.812407
78 0.000000 -2.977125 7.315167 -8.812407
78 0.000000 2.394552 7.315167 -8.812407
78 0.000000 7.766228 7.315167 -8.812407
1 0.000000 2.484362 4.550102 4.812978
8 0.000000 -0.639360 -0.507518 6.370046
1 0.000000 -0.628098 -0.371972 4.373707
1 0.000000 -2.614689 -0.546165 7.045489
8 0.000000 8.475466 4.325710 6.728781
1 0.000000 6.684127 4.384091 7.233466
1 0.000000 8.390250 4.297987 4.830336
8 0.000000 2.249751 4.716513 6.699019
1 0.000000 1.402619 3.124592 7.158088
1 0.000000 0.350609 7.169208 6.741237
8 0.000000 11.072309 -0.458173 7.984686
1 0.000000 10.045053 -1.963466 7.485919
1 0.000000 10.118210 1.087497 7.502348
1 0.000000 -3.088315 7.309223 1.771711
1 0.000000 10.492937 2.683965 1.767513
1 0.000000 2.482154 7.411052 1.817344
1 0.000000 7.759139 7.412407 1.778183
1 0.000000 -0.335993 2.790077 1.820318
1 0.000000 5.101106 2.685168 1.831855
78 0.000000 -0.305903 -0.430700 0.339320
78 0.000000 5.079182 -0.424834 0.052433
78 0.000000 10.441772 -0.430193 0.049143
78 0.000000 -2.976114 4.216673 0.042763
78 0.000000 2.403892 4.221560 0.215319
78 0.000000 7.771946 4.225212 0.224912
1.120684e+03 8.414287e+02 1.004524e+03 1.726496e+03 2.221752e+03
1.746311e+03 1.073006e+03 6.461677e+02 4.067080e+02 2.767300e+02
2.110100e+02 1.887976e+02 2.051091e+02 2.628623e+02 3.646985e+02
4.720659e+02 5.864030e+02 8.054477e+02 1.174628e+03 1.794801e+03
2.681170e+03 3.380116e+03 3.286508e+03 2.416282e+03 1.462200e+03
8.012483e+02 3.963032e+02 1.929722e+02 9.493900e+01 4.862765e+01
2.346863e+01 1.224508e+01 6.400654e+00 3.039787e+00 1.900422e+00
1.502254e+00 2.174876e-01 5.304427e-01 6.969707e-01 6.338792e-01
.
.
.
______________________________________________________


and when I run the bader program I get this:

_______________________________________________________
CHOOSE 1,2,3 or 4 : 4

ECHO INPUT
CHARGE DENSITY FILE: out-run5.0.cube
OUTPUT: NOT WRITING OUT ANY BADER VOLUMES

OPEN ... out-run5.0.cube
GAUSSIAN-STYLE INPUT FILE
forrtl: severe (64): input conversion error, unit 100, file /export/data/home/egillsk/bader_prufur/out-run5.0/out-run5.0.cube
Image PC Routine Line Source
bader 080879F5 Unknown Unknown Unknown
bader 08058D61 Unknown Unknown Unknown
bader 08059358 Unknown Unknown Unknown
bader 08071813 Unknown Unknown Unknown
bader 080707E3 Unknown Unknown Unknown
bader 0804BFF8 Unknown Unknown Unknown

Stack trace terminated abnormally.
___________________________________________________


I did your example of H2O_HF_aug-cc-pVDZ I found on the web and that works fine. The beginning of the cube file looks like this:

___________________________________________________
H2O Density Calculation.
SCF Total Density
3 -7.558904 -7.558904 -7.558904
257 .059054 .000000 .000000
257 .000000 .059054 .000000
257 .000000 .000000 .059054
1 1.000000 .000000 1.430429 .984140
8 8.000000 .000000 .000000 -.123017
1 1.000000 .000000 -1.430429 .984140
8.36042E-12 8.85444E-12 9.37371E-12 9.91931E-12 1.04923E-11 1.10938E-11
1.17251E-11 1.23872E-11 1.30815E-11 1.38091E-11 1.45715E-11 1.53699E-11
1.62057E-11 1.70804E-11 1.79954E-11 1.89523E-11 1.99526E-11 2.09979E-11
2.20899E-11 2.32302E-11 2.44206E-11 2.56628E-11 2.69589E-11 2.83105E-11
2.97197E-11 3.11885E-11 3.27189E-11 3.43130E-11 3.59730E-11 3.77011E-11
________________________________________________________


Can you tell me what is wrong in my cube file?

I can see that the charge is many orders of magnitude higher then it should be and the columns are 5 but not 6. I also noticed that I get 0.0000 in the same row as the number of atoms are specified, where in the example you get -7.558904 three times. And then I noticed that I get 0.0000 in the column between where the atoms are specified and the vector of the atoms instead of getting e.g. 1.0000, 8.00000 and 78.0000.

best regards,

- egillsk

[graeme]

There is a different convention between vasp files and cube files, and this probably applies to dacapo files as well. In one file type, the values in the charge density grid are actually densities, whereas in the other, the values are the number of electrons in the cell. You might try multiplying your values by the volume of a cell in the grid.

[egillsk]

Hi Graeme

I have multiplied the charge density with the volume of the cell divided with the number of grid points and changed it into atomic units and the bader program works with these cube files I am able to produce from dacapo NetCDFiles. The number of electrons is a little bit higher than 2, however, I am getting asymmetry in the H2 molecule that is strange. 92 per cent of the density belongs to one of the atoms and 8 per cent to the other. Have you any idea what is wrong? Here is the output from the bader analysis of the H2 molecule:

# X Y Z VORONOI BADER %
-----------------------------------------------------------------------------
1 8.0377 7.5591 7.5590 0.1877 0.1903 8.4060
2 9.4611 7.5587 7.5589 2.0764 2.0738 91.5940

NUMBER OF BADER MAXIMA FOUND: 15895
SIGNIFICANT MAXIMA FOUND: 1620
NUMBER OF ELECTRONS: 2.26415

Best regards,

Egill

[graeme]

I don't know what's going on, but I bet a cube file visualizer might help. One guess is that the cube format is [-x/2 .. x/2] where as vasp is [0..x]. If Dacapo also has boundaries of [0..x] the charge values may need to be shifted by half a box so they line up with the atoms. An easy way to check this is to look at the locations of the bader maxima, in BCF.dat, with the largest charge. Are those coordinates consistent with the locations of the nuclei?

Another convention which is different between cube and vasp charge density files is the location of each charge density value within the cell. I think that the cube files have values for the corner (origin) of the cell whereas vasp is in the middle. This doesn't usually matter for atomic charges but it does affect dipoles, because the electronic charge is displaced dx/2,dy/2,dz/2 from the nuclei (where each cell size is dx,dy,dz).