(1) How can mathematically correct inverse problems be formulated and
effectively solved to extract information from different experimental
Recent work includes new methods for fluctuation scattering and single particle imaging for the LCLS, new methods for ptychographic reconstruction, and fast methods for SAXS, WAXS, and GISAXS.
(2) Once this information is collected, how can it be effectively analyzed?
Recent work includes imaging algorithms to auto-detect fibers and breaks in materials,
deep learning for X-ray diffraction and recognition, and new mixed-scale dense deep convolution
neural networks for image classification.
(3) What is the best way to use computing resources (embedded in detectors vs.
local hardware vs. remote supercomputers) to quickly analyze results and
guide new experiments?
Recent work includes merging new algorithms, GPU accelerators and customized workflows for
real-time streaming ptychography, and Kriging optimization to automatically steer
autonomous X-ray scattering experiments.
(4) How can algorithms, data, tools, and answers be shared across the community?
Recent work includes developing Xi-CAM, a combination GUI, python plugin environment and
remote workflow manager for synchrotron data, now in use at multiple facilities.