Tue28Sep20214:00 pmLewis Hall 101
Colloquium: Machine Learning Enabled Multi-Scale Modeling of Materials
Department of Mechanical Engineering
University of Mississippi
Machine Learning Enabled Multi-Scale Modeling of Materials
Traditional computational investigation of processing-chemistry-structure-property linkage in materials science involves the usage of specialized computational tools at discrete length scales ranging from electronic to atomic to mesoscale. Alternatively, over the past two decades, a multi-length scale approach combining simulation tools at different length scales has been adopted where electronic/atomic information from lower length scale is passed to higher length scale. However, such traditional computational approaches can provide only limited insights into a highly complex set of interactions spanning over multiple length and time scales each of which are linked to the property and performance of the materials, thus requiring an out-of-the box approach. In this presentation, I will focus on the application of machine learning tools to guide simulations at multiple length scales to augment the capabilities of traditional computational tools. Further, it will be shown that machine learning enabled computational approach provides a fast and efficient pathway to navigate the vast processing, microstructure and chemical search space for a targeted property, a departure from the traditional time consuming and expensive Edisonian trial-and-error approach based on synthesis-testing experimental cycles.
Join Zoom Meeting
Meeting ID: 919 282 27187