Niching in derandomized evolution strategies and its applications in quantum control

Abstract

Evolutionary Algorithms (EAs), computational problem-solvers, encode complex problems into an artificial biological environment, define its genetic operators and simulate its propagation in time. Motivated by Darwinian Evolution, it is suggested that such simulations would yield an optimal solution for the given problem. The goal of this doctoral work is to extend specific variants of EAs, namely Derandomized Evolution Strategies, to subpopulations of trial solutions which evolve in parallel to various solutions of the problem. This idea stems from the evolutionary concept of organic speciation. Such techniques are called niching methods, and they are successfully developed, at several levels, throughout the first part of the thesis. Controlling the motion of atoms has been a dream since the early days of Quantum Mechanics; The foundation of the Quantum Control field in the 1980s has brought this dream to fruition. This field has experienced an amazing increase of interest during the past 10 years, in parallel to the technological developments of ultrafast laser pulse shaping capabilities. The second part of this work is devoted to the optimization of state-of-the-art Quantum Control applications, both theoretically (simulations) and experimentally (laboratory). The application of the newly developed niching techniques successfully attains multiple laser pulse conceptual designs.