“…Concerning the second step, b, the nuclear velocities are adjusted either via scaling by a scalar factor, the S method of Ref. 24, by adding a vector parallel to the real part of the non-adiabatic coupling vector as proposed originally by Tully [23], method C, or adding a vector parallel to the gradient of the difference of adiabatic electronic energies corresponding to the states the system jumps from and to [33], respectively, method G. The two methods for calculating the collapse probabilities, step a, and the three methods for the after-collapse adjustment of nuclear velocities, step b, give rise to six different schemes for the inclusion of quantum decoherence processes at the MFQ level (see Ref. 24 for details).…”