ISSN:
1089-7690
Source:
AIP Digital Archive
Topics:
Physics
,
Chemistry and Pharmacology
Notes:
We show how the dynamically nonlocal formulation of classical nuclear motion in the presence of quantal electronic transitions presented many years ago by P. Pechukas [Phys. Rev. 181, 166 (1969); 181, 174 (1969)] can be localized in time using time dependent perturbation theory to give an impulsive force which acts when trajectories hop between electronic surfaces. The action of this impulsive force is completely equivalent to adjusting the nuclear velocities in the direction of the nonadiabatic coupling vector so as to conserve energy, a procedure which is widely used in surface hopping trajectory methods [J. C. Tully, J. Chem. Phys. 93, 1061 (1990)]. This is the first time the precise connection between these two formulations of the nonadiabatic dynamics problem has been considered. We also demonstrate that the stationary phase approximation to the reduced propagator at the heart of Pechukas' theory is not unitary due to its neglect of nonstationary paths. As such mixed quantum-classical evolution schemes based on this approximation are not norm conserving and in general must fail to give the correct branching between different competing electronic states. Tully's phase coherent, fewest switches branching algorithm is guaranteed to conserve the norm. The branching between different alternatives predicted by this approach, however, may be inaccurate, due to use of the approximate local dynamics. We explore the relative merits of these different approximations using Tully's 1D two state example scattering problems for which numerically exact results are easily obtained. © 1995 American Institute of Physics.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1063/1.469428
Permalink
