Artificial Intelligence, Theory, and Simulation
Artificial intelligence (AI) and the subfield of deep learning within it have become standard tools in many fields. In chemical engineering, AI shows robustness in a broad range of tasks, for example, rational design of molecules and materials, chemical properties prediction and complex system analysis.
AI enables us to learn empirical predictions and new physics from existing data. For example, neural networks can replace expensive quantum mechanical calculations, leading to new highly accurate simulations of molecular-level systems like proteins. Recent trends in areas like deep learning are creating new directions for chemical engineering research, leading to new advances in catalysis, protein structure, and molecular design.
Our department has expertise in molecules and materials machine learning. We leverage both traditional and novel machine learning methods (e.g. Bayesian Methods, Self-supervised Learning, Graph Neural Networks, etc.) to perform cutting-edge research. Our work utilizes the state-of-the-art computational resources provided through the University’s Center for Integrated Research Computing. Our excellence in applied machine learning is emphasized in our undergraduate education as well, with a core class covering computational methods, computational statistics, and a deep learning class in our upper-level electives.
Theory and Simulation
Robust theory and simulation is a core part of interdisciplinary research, especially in chemical engineering, as we develop complex new materials, study increasingly complex biochemical systems, and model sophisticated electrochemical systems. Theory and simulation provide the tools to develop detailed molecular-level understanding, offer predictions for complex systems, and enable rationale design of molecules and materials.
Recent trends in areas like data-science and machine learning are creating new directions for chemical engineering theory and simulation, leading to new advances and new funding directions. For example, the material genome initiative at NIST and NSF and the computational data-science enabled cross-cutting program at NSF are funding opportunities specifically focused on theory and simulation research that use data-science.
Our department has expertise in molecular dynamics, network theory, ab ignition quantum dynamic, electrochemical finite element modeling and empirical data-driven models.
Active Faculty / Research Areas
White: Modeling Peptide Self-Assembly; Data-Driven Molecular Simulation; Molecular Modeling Methods Development; Materials Design; Deep Learning; Artificial Intelligence in Chemical Engineering