Posted by angelo on January 22, 2019 at 11:26:19:
Hello,
I understand that in the lastest version of Molcas and in OpenMolcas, minimum energy conical intesections (MECIs) are optimized using a ProjectorConstrainOptimization approach which needs the computation of non-adiabatic couplings (nacs), as described in J. Chem. Theory Comput. 2016, 12, 3636−3653. In the case in which a numerical MECI optimization at the CASPT2 level is run, the nacs cannot be evaluated and in fact molcas writes “Numerical nonadiabatic coupling not implemented”. How is then optimized a MECI when performing a CASPT2 numerical MECI optimization?
Thanks