Imaginary RPA root

I am trying to perform a standard TDDFT calculation of several lowest excited singlet states of a largish molecule with D2d symmetry.

Such calculations go smoothly using Turbomole and the Program that Must Not Be Named :wink:
Q-Chem 6.0, however, having done the TDA initial guess comes down with the following error message:

Q-Chem error occurred in module setman/rpa_dia.C, line 81:
Imaginary RPA root detected (w^2 = -1.862e-04):

  • if w is small, tighter thresholds may help
  • if w is large, SCF may not be at a stable solution

The scf convergence threshold is 10^-9. which is rather tight. What else can I do. I am a newbie in Q-Chem, so please, help.

Best regards

Imaginary roots in RPA generally indicate triplet instabilities in the reference state. Have you run a stability analysis? I also tend to turn off symmetry (SYM_IGNORE=TRUE) at the first sign of trouble.

Well, excluding triplet states from the roots did help, but I am still at a loss as to what causes this problem. The two packages that I used previously both handled the same molecule without any problems.

Did you run a stability analysis on your reference state?

The lowest triplet state is over an electronovolt higher in energy. This ground state here is a well-defined ground state (a close shell hydrocarbon molecule). Besides, the same geoometry leads to no problems of this kind when other software is used.

Nonetheless, I will try to rund the wave-function stability test once I figure out how to do it in qchem :wink:

I have the same question as here [Symmetry change at frequency calculations], namely, are you certain that you are obtaining the same SCF solution with both codes?