Firefly and PC GAMESS-related discussion club
Learn how to ask questions correctly
Re^3: Discontinuous energy curve from XMCQDPT2 results
>Yes, the LIIC path was first constructed between cis-HBDI and CI, and then constructed between CI and trans-HBDI, thus the provided curve consists of two parts. However, the MCSCF curve is smooth and the CI point has the smallest delta energy between S0 and S1, whereas this is not the case for XMC-QDPT2 curve, where the delta energy of CI point is larger than that of the neighbour points. This is unusually and indicates that the optimized CI point by MCSCF is not the "real" CI in XMC-QDPT2. So, what is the source for this difference between MCSCF and XMC-QDPT2? Is it because lack of dynamic electron correlation effects in MCSCF calculations?
I recommend you to carefully check the orbitals in the active space before and after CI point. Although CASSCF curves look smooth, the may be discontinuities in the orbitals (e.g., one set of active orbitals before CI point and a different set of orbitals after CI). At worst, you'll have to merge both active spaces (which substantially blows up the problem) and tune state-averaging respectively.
Fri May 10 '13 5:28pm
This message read 808 times