Wed, Mar 06, 2019 @ 03:30 PM - 04:30 PM
Aerospace and Mechanical Engineering
Conferences, Lectures, & Seminars
Speaker: Krishna Garikipati, University of Michigan
Talk Title: Mechano-Chemical Phase Transformations: Computational Framework, Machine Learning Studies and Graph Theoretic Analysis
Abstract: Phase transformations in a wide range of materials-”for energy, electronics, structural and other applications-”are driven by mechanics in interaction with chemistry. We have developed a general theoretical and computational framework for large scale simulations of these mechano-chemical phenomena. I will begin by presenting our recent work in this sphere, while highlighting some of its more insightful results. In addition to being a platform for investigating mechanically driven phenomena in materials physics, this work is a foundation to explore the potential of recent advances in data-driven modeling. Of interest to us are machine learning advances that may enhance our approaches to solve computational materials physics problems. I will outline the first of several recent studies that we have launched in this spirit. Such combinations of classical high-performance scientific computing and modern data-driven modeling now allow us to access large numbers of states of physical systems. They also motivate the study of mathematical structures for representation, exploration and analysis of systems by using these collections of states. With this perspective, I will offer a view of graph theory that places it in nearly perfect correspondence with properties of stationary and dynamical systems. This has opened up new insights to our earlier, large-scale computational investigations of mechano-chemically phase transforming materials systems. This treatment has potential for eventual decision-making for physical systems that builds on high-fidelity computations.
Krishna Garikipati is a computational scientist whose work draws upon nonlinear physics, applied mathematics and numerical methods. A very recent interest of his is the development of methods for data-driven computational science. He has worked for quite a few years in mathematical biology, biophysics and materials physics. Some specific problems he has been thinking about recently are: (1) mathematical models of patterning and morphogenesis in developmental biology, (2) mathematical and physical modeling of tumor growth, and (3) mechano-chemically driven phenomena in materials, such as phase transformations and stress-influenced mass transport.
Host: AME Department
More Info: https://ame.usc.edu/seminars/
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Tessa Yao