[CPMD-list] (no subject)

[Irmgard Frank]

Hi,

a few words about ROKS:

As already stated by Salomon, there is no reason to use it
with triplet states, except that one might want to stay consistent
when also calculating open-shell singlets.
For triplet states the ROKS results are almost identical to LSD.

In the restricted open-shell formalism the symmetry is
prescribed by the parameters given after the line LSE PARAMETERS.
2 -1 (default) yields the correct linear combination for singlet symmetry;
0  1 for triplet symmetry.
Any other linear combination will result in a mixture of both
symmetries. Presently only the case of two unpaired electrons
is implemented and no Vanderbilt pseudopotentials. We might
want to extend the implementation to other symmetries some time,
but combination with Vanderbilt pp's is not planned. Presently we
are using Troullier-Martins pseudopotentials.

Apart from the description of open-shell ground states, the main
application of ROKS is that to excited states, specifically
motion in excited states. (For spectra, TDDFT is a very good
alternative.) However, it should be emphasized that dynamics in
excited states is not fully a black box business. ROKS should
be applied with caution, if ground and excited states have the same
symmetry. In this case, unphysical orbital rotations are possible
with the algorithm presently implemented in the official version.
The unphysical rotation can be checked by plotting the orbitals. The
highest occupied MO of the ground state should have the same nodal
structure as one of the two singly occupied MO's of the ROKS excited
state. We have ways to deal with the problem that are not yet
published or documented; for the moment: please talk to us if you
want to do excited state calculations with ROKS.

Best regards, Irmgard