- Charles A. Bouman, Purdue University
- Edwin Fohtung, New Mexico State University/Los Alamos National Laboratory
- Jan Ilavsky, Argonne National Laboratory
- Peter R. Jemian, Argonne National Laboratory
- Shiu Fai Frankie Li, DITTO Technologies, Inc.
- Stephen R. Niezgoda, Ohio State University
- Hemant Sharma, Argonne National Laboratory
- Kevin Yager, Brookhaven National Laboratory
- Sven C. Vogel, Los Alamos National Laboratory
Non-destructive characterization of new classes of challenging materials has been made possible with the advent of third-generation synchrotron light sources, such as the Advanced Photon Source (APS) at Argonne National Laboratory, and x-ray free electron lasers (XFEL), such as the Linac Coherent Light Source (LCLS) at Stanford. The high energy and brilliance of the x-rays produced by these light sources can probe high Z, 3D polycrystalline samples at the mesoscale (~0.1-10 μm). While providing previously inaccessible data, the increasing average brightness of these light sources also creates new challenges, such as the generation of extremely large data sets (GBs to TBs). These unconventionally large data sets are overwhelming the experimentalists. Furthermore, the fact that data is generated from multi-modal techniques such as high-energy x-ray diffraction microscopy (HEDM), computed micro-tomography (μ-CT), and coherent x-ray diffraction imaging (CXDI), makes data registration and analysis even more challenging, as multiple length and time scale information from different techniques must be combined to gain physical insight into materials behavior. Current data analysis tools are still in their infancy, resulting in a significant amount of time and resources spent on manual processing with brute force and inefficient, home-brew scripts. Moreover, this big data problem will only worsen with a growing user base. Increasing data collection rates with 4th generation light sources (APS upgrade, LCLS-II repetition rate of 2 MHz compared to 120 Hz at LCLS, EuXFEL 3000 pulses/shot at 10 Hz, etc.) are creating an urgent need for the development of efficient, fast, and user-friendly (as little as possible user intervention during reconstruction) software, which has the potential to change how experiments and data mining are performed in the field of 3D materials science.
The main topics of this workshop will be:
• Identify and explore the range of datasets that are currently being produced from experiments at various light sources to solve materials science problems.
• Discuss the latest developments in advanced data analysis tools.
• Encourage collaborations between experimentalists, modelers, and data scientists across national labs, academia, and industry.
The overall workshop goal is to improve information extraction capability from high-dimensional data and design a real-time feedback framework for driving experiments.