Hi again,
This is just to clarify my previous comments.
i) the problem comes from the numerical evaluation
of the radial part of the gradient.
ii) Just by the way we calculate things the problem
that we have localized now only comes from the
L=0 part of the overlap matrix (JLM=1 in the
On Thu, 26 Jan 2006, Daniel Sanchez Portal wrote:
| Dear all,
|
| I have been checking a little bit the problem commented by
| Andrei. From my point of view it seems a problem related to the use of a
| interpolation from a (not very dense) grid.
| We always have numerical values of S(R) and
Dear all,
I have been checking a little bit the problem commented by
Andrei. From my point of view it seems a problem related
to the use of a interpolation from a (not very dense) grid.
We always have numerical values of S(R) and numerical derivatives.
The derivatives given by the splines
Jose, Emilio:
with respect to the matel issue discussed yesterday,
I suggest now a more elegant substitution of my yesterday's
fast and dirty fix. It explicitly uses the symmetry,
that is that the spherical harmonics change with R - -R as
(-1)**l whereas gradients as (-1)**(l-1). The changes in
4 matches
Mail list logo
