Dear Rainer:
I am comparing charge densities given by siesta. Due to the closeness
of the densities I am comparing, I would like the grid points at which
the densities are given to line up. To this end I need to be able to
tune the grid used for output to the .RHO file. I noticed that
increasing MeshCutoff (the equivalent plane wave cutoff for the grid)
increases the grid fineness in all 3 dimensions. Is there a way in
which I can tune the grid fineness in each dimension seperately? -
ideally specifying explicitly the number of grid points in each
dimension.
No, there isn't.
One thing you could do is to perform the
calculation in a fine enough mesh, and then
interpolate to the grid of your choice.
In 1D, an interpolation like this
is carried put in the Utility package Macroave
(see Util/Macroave/Src).
For an interpolation in 2D, see the Utility
package Contour (included in Util/Contour)
And another related question: what are the units of the numbers in the
.RHO output? I have tried to figure this out, but have come to the
conclusion that the numbers aren't normalized to anything, because
adding up the total density by adding all points in .RHO results in
different totals for the same system with different nr of grid points.
The units of the numbers in the .RHO output are
electrons/bohr**3.
If you integrate the charge density over the whole
unit cell you will obtain the number of electrons
in your system. This integral is performed, for example,
in Macroave (remember that
you have to multiply the value of RHO at a given point
by the volume/point in the mesh; trace the variable QREN
in Util/Macroave/Src/macroave.f)
Hope this helps,
Javier
