Ab Initio Multiple Spawning (AIMS) is a method that allows for accurate yet efficient excited-state molecular dynamics. It portrays nuclear wavepackets as linear combinations of coupled Gaussian functions that follow classical trajectories. (See our recent review for a detailed discussion on this method.)
AIMS emerges from the (formally-exact) Full Multiple Spawning framework by invoking two central approximations related to the coupling between Gaussian functions. While AIMS has been successfully employed to simulate the photochemistry of numerous molecules, a detailed analysis of the strengths and weaknesses of its underlying approximations was missing in the literature.
Together with Dr. Benoît Mignolet (University of Liège), we investigated in great details the approximations in AIMS using a model molecule, LiH, photoexcited by an external electric field to generate interfering nuclear wavepackets. Interestingly, AIMS could capture the behavior of properties like electronic state populations and the more challenging time-dependent molecular dipole moment, even if this low-dimensional system constitutes one of the worst-case scenarios for the AIMS approximations. We furthermore proposed and validated strategies, compatible with the simulation of larger molecular systems, to overcome some potential shortcomings of these approximations.
A walk through the approximations of ab initio multiple spawning, by Benoît Mignolet and Basile F. E. Curchod, J. Chem. Phys., 148, 134110 (2018).