Kenneth Shui, Monta Vista High School 

Abstract: 

Phase retrieval from a single diffraction intensity measurement for the case of complex-valued objects finds multiple applications such as coherent diffraction imaging (CDI) and wavefront sensing. Most conventional iterative 2D phase retrieval algorithms only handle real-valued data in real space. The real space data in X-ray free-electron lasers (XFELs) is complex-valued, which makes phase retrieval much more challenging. In this presentation, a direct comparison is made between various iterative 2D phase retrieval algorithms with XFEL diffraction data. Modifications are made in the real space support for the Hybrid Input Output (HIO), HIO-ER (Error Reduction), and Oversampling Smoothness (OSS) algorithms, such that they can handle both real and complex-valued data in real space. All three methods are tested on diffraction data from the Linac Coherent Light Source, at the X-ray Pump Probe beamline. Different real space support regions’ sizes and shapes are studied to check their impacts on retrieval performance based on two types of error figures of merit (FOM). The effect of different types of noise on each method’s accuracy are
also studied and summarized. It is concluded that extra real-space magnitude constraint, in addition to a relatively well-defined support region, can drastically improve the complex-valued 2D phase retrieval performance. The optimal method would be useful to analyze and quantify aberration errors in XFEL wavefront sensing.

Poster Session Link: https://gather.town/invite?token=0pEoq7VP

If you have any questions for the presenter, please contact them via email: kshui2020@gmail.com