Brookhaven AIMS Series: Data-driven discovery of dynamics from time-resolved coherent scattering

US/Eastern
Description

The Brookhaven AI/ML  Seminar (AIMS) series is about showcasing research at Brookhaven National Laboratory (BNL) and elsewhere that uses AI and Machine Learning to enhance scientific discovery and that uses domain science questions to motivate new AI developments. 

 

    • 1
      Data-driven discovery of dynamics from time-resolved coherent scattering

      Abstract: Coherent X-ray scattering (CXS) techniques, including X-ray photon correlation spectroscopy (XPCS), play a critical role in the investigation of mesoscale phenomena evolving at time scales spanning several orders of magnitude. However, obtaining accurate theoretical descriptions of complex dynamics is often limited by one or more factors – the ability to visualize dynamics in real space, computational cost of high-fidelity simulations, and accuracy of approximate models. Here, we aim to bridge the gap between theory and experiments by extracting mechanistic models of dynamics directly from CXS data. To do so, we develop a data-driven framework which employs neural differential equations to parameterize unknown real-space dynamics and a computational scattering forward model to relate real-space predictions to reciprocal-space observations. This framework is used to recover dynamics of several computational model systems, including domain synchronization, particle clustering, and source fluctuation, without solving the phase reconstruction problem for the entire time series of diffraction patterns. In addition, this approach is shown to extrapolate well beyond the maximum time seen during training, which can critically inform experimental design and planning. Finally, we present a proof-of-concept experiment which uses the framework to recover the probe trajectory from a ptychographic scan. Our framework represents a general and versatile platform to discover dynamics from data with potential applications across characterization modalities.

      Bio: Nina Andrejevic is a Maria Goeppert Mayer Fellow at Argonne National Laboratory. Her research focuses on developing physics-aware machine learning models for intelligent analysis of materials characterization data. She received her B.S. in Engineering Physics from Cornell University and her Ph.D. in Materials Science and Engineering from Massachusetts Institute of Technology. Alongside her research, she is also enthusiastic about science communication through teaching and scientific data visualization.

      Speaker: Nina Andrejevic (Argonne National Laboratory)