| Abstract Scope |
Understanding the materials structure-property relationship at the mesoscale is crucial for developing microstructure-aware physics-based models. Apart from limited available beam times, the slow speeds of data acquisition and analysis are the biggest constraints on performing impactful in situ mesoscale materials research experiments at light sources. In this talk, we will demonstrate a few case studies that uses novel data analysis framework combining advanced measurement techniques, machine learning (ML), micro-mechanics simulations, and adaptive model-independent optimization methods for fast 2D and 3D microstructure reconstruction from noisy measured diffraction patterns and corresponding inference of the 3D microstructure and micro-mechanical field evolution. We will also discuss our overarching goals to (1) facilitate real-time feedback for guiding beamline experiments, (2) enable a computationally efficient way of performing detailed large-scale simulations of engineering components, and (3) accelerate the fast extraction of insight from data, which is of substantial interest to the light source user community. |