Once again about the weight of reference WF in CASPT2

Subject:

if B is 1s22s22p1, what is Li?

Passfield:

Comments:
: Dear Coen and All, : Thanks very much for your reponse to my previous message (/molcas/wwwboard/messages/991.html). : In this post I would like to ask you something more particular on the reliability of my CASPT2 calculation. Namely I wonder whether I should apply the shift option at all. Without shift I already get very close (although not very high) weights of the ground and first excited states which are 0.56 and 0.57 respectively. They become 0.59 and 0.60 when I apply the shift 0.1. With the shift 0.3 I get both weights ca. 0.63. The excitation energy (S0-S1) changes (increases) only by 0.05 eV on the shift 0.3 (practically no change on the shift 0.1). The largest contribution to the energy is 0.11 eV (without any shift). It becomes 0.08 eV on the shift 0.08 eV. : My system has 84 correlated electrons, 37 inactive orbitals, 7 active orbitals and 197 secondary orbitals. The weights depend very little on the active space in the CASSCF procedure. : So, I would like to ask your opinion whether my CASPT2 calculation is already ok without the shift (if not what value of the shift is better to use). : Sorry for this very particular question but I have little experience in CASPT2 calculations (which I would like to improve :)). : Many thanks in advance! : Best regards, : Evgeniy