Optimizing Quantum Control Pulses with Gaussian Process Priors: The Spectral Way

Abstract

This study presents the Gaussian Process Prior Optimization for Pulse Shaping (GPPOPS) methodology, a novel approach to pulse shaping engineering. Its main objective is to efficiently identify laser pulse shapes that can achieve a desired task encoded in a cost function while being experimentally implementable. The AlH + molecule was utilized as a test case to find pulse shapes that maximized vibronic transitions. The results demonstrate that optimal pulses can be readily implemented using current laser technology and that their control capabilities can withstand noise. The study emphasizes the benefits of constructing a surrogate approach to the control landscape during the optimization process. This approach is expected to be versatile, efficient and readily implementable in the laboratory. Its demonstrated robustness to noise sets it apart from other numerical pulse shaping engineering methods. By reducing the required experimental labor, this method has the potential to facilitate breakthroughs in quantum engineering.