Evolutionary Algorithms in the Optimization of Dynamic Molecular Alignment

This paper introduces the optimization of dynamic molecular alignment by shaped femtosecond laser pulses, and analyzes the application of various Evolutionary Algorithms to this challenging real-life high-dimensional physics problem. With an expensive simulator evaluation of 35 seconds, standard evolutionary approaches based on low-dimensional parameterizations of the electric field are applied to the task, compared among each other, and shown to be clearly inferior with respect to other methods. This numerical phase provides new insights into the problem, and is meant to be followed by a lab phase.