Problem with Calculation of Oscillator Strengths (ricc2)

[evgenii]

Dear all,

I have a problem with calculation of oscillator strengths with ricc2.
My system is a molecular dimer of cs symmetry, for which I want to calculate a CC2 spectrum:
10 states of a' and 10 states of a" irreps with corresponding oscillator strengths.

The CC2 excitation energies were calculated without problems,
but the calculation of excited state properties ended with the following message in ricc2.out:

**********************************************************************************
Warning: one or more of the second eigenvectors do not give correct excitation energies:
 Largest difference between left and right eigenvalues is  0.94E-04 a.u. for state no. 10
 
 ========================
  internal module stack:
 ------------------------
    ricc2
    cc_exci
 ========================

 PROBLEM WITH LE/RE EIGENVALUES!
 ricc2 ended abnormally
 ricc2 ended abnormally
program stopped.
*************************************************************************

Do you have any suggestions why this problem took place and how this one can be overcome?

Thank you in advance!

Evgenii

[christof.haettig]

Since for CC2 the left eigenvectors are obtained by solving a different eigenvalue problem then for the right eigenvectors, it can in rare cases happen that they don't match together, e.g. because for high lying states different roots were found.
One can check for this by restarting the calculation and requesting just right eigenvectors, but than one needs at the end and with a tighter convergence threshold (conv=<number> in $excitations), and then doing the same for the left eigenvectors (add 'leftopt' in $excitations').  If the enough right and left eigenvectors converged with matching eigenvalues one can restart the calculation for oscillator strength with the original settings.

The alternative is to let the program ignore the problem. This is done if you add 'no_strict' in the $ricc2 input section.