Review of development for ptychography algorithm

Abstract

With the development of synchrotron radiation technology and the improvement of light source coherence, ptychography has developed rapidly. Ptychography algorithm solves the problems of slow convergence and easily falls into the local optimal solution and stagnation of the traditional coherent diffraction imaging algorithm. It has the advantages of large imaging field of view, robustness of algorithm, high tolerance to error and wide range of applications, and is becoming a hot research direction in the field of coherent diffraction imaging. Ptychography reconstructs the complex amplitude distribution and illumination light of the sample by iterative algorithms, which can theoretically reach the resolution of the diffraction limit. It has excellent applications in the fields of wavefront detection, phase imaging and optical metrology. This paper first introduces the background of the proposed ptychography algorithm and briefly describes the problem of coherent diffraction imaging algorithm and its development, and then summarizes the development of ptychography algorithm in detail, mainly including the mainstream algorithm of ptychography and its kernel. This paper then describes in detail the improvement of algorithms corresponding to the improvement of the efficiency of ptychography experiments, correction of position errors and the effect of illumination light multi-modal, and elaborates the algorithm flow. After analyzing the possible intersection of diffraction imaging and neural networks in the field of artificial intelligence, this paper introduces new algorithms with combining ptychography with artificial intelligence. New algorithms with combining ptychography with neural networks will have new potential applications in generality, accuracy and robustness. Finally, a specific parallelization implementation of the ptychography algorithm and common software packages are presented. The logic for writing the parallelization of the algorithm implementation of each package and the corresponding advantages and disadvantages of the packages are described in detail. The characteristics and performance of each package are then listed for reference. This paper helps to establish a global perspective of the algorithm itself, artificial intelligence and computational methods in the field of ptychography, and presents an important reference for systematically developing the ptychography method.