Hi

The ADF openshell TDDFT requires that the orbitals have integer
occupation, ie are fully occupied (remember that in unrestricted
calculations one electron fills most orbitals). If in your calculation,
you have a partially occupied degenerate orbital (e.g and e orbital with
one electron or a t orbital with one or two electrons) ADF spreads the
electrons evenly among the components of the degenerate orbital, eg
if you
have one electron in an e orbital you will get 1/2 an electron in the
e_x
and 1/2 an electron in an e_y. This is done to preserve the
degeneracy of
the e orbital I guess. It means that the ground state of the system is
degenerate which is not so easy to completely correctly treat with a
single determinent method. With your small clusters of Ag atoms it's
quite
likely they will have high enough symmetry and appropriate
occupations to
give the 1/2 e_x and 1/2 e_y situation. The present implementation of
TDDFT in ADF cannot handle fractional occupations like this. I believe
that some formulations of openshell TDDFT (e.g that by Guan, Salahub and
Casida) allow for fractional occupation numbers. I implemented their
fractional occupations equations into ADF in an attempt to treat
situations just like yours. It gave unsatisfactory results in that
symmetry was broken where it should not be and other problems. I don't
imagine fractional occupations with TDDFT will be available in ADF
anytime
soon for this reason.

An alternative approach where you can get the excitation energies you
want
with ADF while avoiding the problem with fractional occupations can be
found in the Journal of Chemical Physics, volume 123, 144105 (2005). You
probably need oscillator strengths if you want to simulate the spectra.
I'm working on that problem right now. If you like, I can let you know
if/when the appropriate publication is available.

Michael

On Sat, 29 Oct 2005, may abdelghani wrote:

> Hello, ADFers
> when I want draw the UV-visible spectrum of small
> silver clusters, I not that the spectrum of Ag3, Ag5
> and Ag7, not appear(SCM----spectra----Excitation), and
> I saw, in the log file, this message error:
> error in scfrro: fractional occupation numbers have
> not yet been implemented !,
> I know ( from ADFUserGuide), that the ADF' last
> version( ADF205.01) can calculate the " Excitation
> energies for open-shell systems, including spin-flip
> excitation..." , what is the problem?
>
Received on 2005-10-31 12:15:42