Posted by Coen de Graaf on October 03, 2008 at 09:38:59:
In Reply to: Small wheight of reference WF in CASPT2 posted by Evgeniy on October 02, 2008 at 17:23:18:
The weight of the reference wave function in itself is not enough information to decide whether the caspt2 result is reliable or is affected by the appearance of intruder states. The magnitude of the weight depends on the number of electrons that you correlate and also on the size of the one-electron basis set to you use. For small molecules with small basis sets, you should have very large weights, while for enormous systems with more than hundred correlated electrons, this weight will be much smaller. In order to be sure that your calculation is ok, you have to compare the weight of the reference function in different states (eg. compare the ground state with an excited state). These weights should be very similar. Furthermore, you should also take a careful look at the list of large contributions to the first order wave function printed in the output after the "Total energy". Looking at this list, it should not be so difficult to see whether you have intruder states (normally they have large coefficients and small energy contributions).
All the best,
Coen