On Choosing the Form of the Objective Functional for Optimal Control of Molecules

Optimal control techniques can be used to direct molecular dynamics to meet specified physical goals. However, the effectiveness of finding an optimal control field depends on the nature of the control landscape, defined as the objective as a functional of the control. Extensive analysis has considered the prospect of such landscapes being free of suboptimal traps for particular cases of different objective functions in both classical and quantum systems. While many typical objective functions have been shown to yield trap-free landscapes upon satisfaction of certain assumptions, this work more broadly considers the freedom in the choice of objective functionals. The latter freedom can be exploited to possibly accelerate the search for an optimal control, but we also show that the choice of functional needs to be made carefully to avoid inducing artificial landscape traps.