Restrictions: results not trustworthy for higher excited states

Users should be aware of another technical point: Excited states are usually calculated from a Davidson diagonalization procedure, i.e., only a small number of eigenvalues and eigenvectors describing the lowest excitations are obtained. During finite displacements, some of the higher calculated states might leave the calculated energy window, while others enter it. Hence, the character of some of the higher calculated states can change. In such a case, the numerical differentiation based on a (simple) diabatic pictures will fail for the higher states, since no mapping between the excited states for reference (equilibrium) and displaced structure can be carried out.

The solution is rather simple: Users should always ask for more excited states than they are actually interested in, and discard the data for higher states, in particular for those which could not successfully be mapped for displaced structures (look for messages in the output like 'State No. X cannot be expressed in terms of reference states').

For advanced users it should be mentioned that it is possible to set an energy window within the range of states calculated, and only the states within this energy window will be taken into account in the evaluation. See the subkey ELTHRESH and EUTHRESH of the block key VIBRON.

Furthermore, it is possible to pick out certain states from this energy window, and only perform the mapping (and diabatization, if requested) and differentiation for them. See the subkey SELSTATE of the block key VIBRON.