Dynamic coherent diffractive imaging using unsupervised identification of spatiotemporal constraints.

Dynamic coherent diffractive imaging (CDI) reveals the fine details of structural, chemical, and biological processes occurring at the nanoscale but imposes strict constraints on the object distribution and illumination. Ptychographic CDI relaxes these constraints by exploiting redundant information in data obtained from overlapping regions of an object, but its time resolution is inherently limited. We have extended ptychographic redundancy into the spatiotemporal domain in dynamic CDI, automatically identifying redundant information in time-series coherent diffraction data obtained from dynamic systems. Simulated synchrotron experiments show that high spatiotemporal resolution is achieved without a priori knowledge of the object or its dynamics.