DQV Diabatic Hamiltonian Output when using Multistate Effective Hamiltonian in RASSI
Hello,
I had a question about how effective Hamiltonian inputs (Heff keyword) to RASSI from various multistate CASPT2 methods propagate and interact with the DQV diabatization code.
Looking briefly through the code, if my understanding is correct, the properties (dipole moments, quadrupole moments, etc) used as input for the DQV diabatization scheme are expressed in the basis of the eigenstates of the effective Hamiltonian. However, after determining the appropriate rotation matrix from the mixed state properties, it appears that the "Diabatic Hamiltonian" output comes from applying the rotation matrix to the effective Hamiltonian expressed in the original basis of the input CASSCF/RASSCF states. My initial thought would be that the correct diabatic Hamiltonian would come from applying the rotation to the diagonalized effective Hamiltonian (same basis as the properties). As I'm not deeply familiar with the RASSI code (or the OpenMolcas codebase in general), I'm not sure if this is a bug, if I'm missing a change of basis somewhere in the code, or have some other misunderstanding.
Could someone more experienced with the RASSI/DQV diabatization code take a look?
Thank you for all your hard work, Greg