Bayesian Optimization for Wavefront Sensing and Error Correction

Abstract

Algorithms for wavefront sensing and error correction from intensity attract great concern in many fields. Here we propose Bayesian optimization to retrieve phase and demonstrate its performance in simulation and experiment. For small aberration, this method demonstrates a convergence process with high accuracy of phase sensing, which is also verified experimentally. For large aberration, Bayesian optimization is shown to be insensitive to the initial phase while maintaining high accuracy. The approach’s merits of high accuracy and robustness make it promising in being applied in optical systems with static aberration such as AMO experiments, optical testing shops, and electron or optical microscopes.