Posted by Roland Lindh on December 21, 2005 at 12:46:22:
In Reply to: but posted by tiantian on December 21, 2005 at 12:02:55:
The conical intersection has two constraints, H_11-H_22=0 and H_12=0, where H is the Hamiltonian matrix. In this context the degeneracy can be broken in two different directions. These two direction span what we can view as a cone. Hence the term conical intersection. If we exclude spin-orbit interaction H_12 is trivially zero if two states are of different spin. The same holds if the states are of different symmetries. So we could view these cases as special cases where we only have one direction in which we can break the degeneracy. To make a distinction between a conical intersection and "simple" intersection is only of a point when it comes to the computational process, since we can explore the latter without the H_12=0 constraint. In the case of states of different spin the spin-orbit interaction will make the intersection a conical intersection. For states of different symmetry the vector which break the symmetry combined with the direction in which the energy degeneracy is removed defines the cone. So in reality all intersections are conical.
In Molcas we have not implemented the constraint H_12=0 yet. This work is in progress and will be finished soon.