[gmx-users] Units of density in the g_spatial output

[Andrew DeYoung]

Hi, 

Does anyone know what the units of density are in the output (Gaussian 98
cube format) of the spatial distribution function in g_spatial?  Is it a
number density or a mass density?  If it is a number density, then I would
guess that the units might be atoms/nm^3, since nm is the unit of length in
Gromacs.  If it is a mass density, then I would guess that the units might
be (atomic mass unit)/nm^3.

When I look in gmx_spatial.c, I see a line with the declaration "static
const double bohr=0.529177249;  /* conversion factor to compensate for VMD
plugin conversion... */" [line 55].  It seems that this value is in the
denominator in subsequent calculations [for example, lines 267-270].  I am
not sure what this means.  The comment in the code seems to imply that VMD
does it own processing of the cube file.  But when I asked on the VMD
mailing list, I got this response:

"The units are whatever the units of the data you loaded were. The values in
the isovalue slider correspond directly to the data values that the
associated VMD molfile plugin read from the input file, unless some kind of
unit conversion was done on the fly, but in the case of volumetric data,
none of the existing plugins do any unit conversions...  The reason the
units aren't specified is that many of the existing volumetric file formats
don't include any provision to label the units  in the file, and/or the
programs that generate the volumetric files don't  bother to write the units
even if the file format supports it. The units of the isovalue slider in VMD
are therefore directly related to the per-voxel scalar values stored in the
volumetric data file, e.g. a density value (not the cartesian coordinates,
not lengths, etc)."

Indeed, when I open a cube file that is output by g_spatial, I do not see
any units specified in the file (probably Gaussian98 does not support it).

Do you have any suggestions?

Thank you!

Andrew DeYoung
Carnegie Mellon University