University of Connecticut

Events Calendar

Mathematics Colloquium
Statistically Learned Kinetic Monte Carlo Models of Chemically Reactive Systems from Molecular Dynamics Data
Qian Yang (Computer Science and Engineering, UCONN)

Thursday, March 28, 2019
4:00pm – 5:00pm

Storrs Campus
MONT 214

Complex chemical processes such as the decomposition of energetic materials are typically modeled at the atomistic level using large-scale molecular dynamics simulations (MD), generating a wealth of data. It is natural to wonder whether these rich and expensive datasets can be used to do more than simply describe the particular system being studied. In our work, we have developed a statistical framework for extracting information about the fundamental underlying reaction pathways observed from MD data, using it to build kinetic Monte Carlo models (KMC) of the corresponding chemical reaction network. Our KMC models can not only extrapolate the behavior of the chemical system by as much as an order of magnitude in time, but can also be used to predict the dynamics of entirely different chemical trajectories. Additionally, we have built a new and efficient data-driven algorithm for interpretable model reduction of nonlinear dynamical systems. This allows us to reduce the complex chemical reaction networks from our learned KMC models, consisting of thousands of reactions, in a matter of minutes.

In recent work, we seek to extend the extrapolation capabilities of our framework to arbitrarily long timescales by using the fast KMC models as a numerical integrator with a prediction-correction scheme. Our "leapfrog" approach has the potential to truly enable computational simulation to reach experimental timescales. One can easily imagine a future in which MD simulations used for research are routinely archived and analyzed in order to add to and modify an existing repository of learned chemical reactions and reaction rates. This repository would form a "chemical genome" that can then be used to quickly simulate all kinds of new chemical systems. This data-driven approach has the potential to break through long-standing barriers in the accuracy versus system size and time scale tradeoff that is at the core of computational materials science.

Short Bio: Qian Yang joined the Computer Science and Engineering Department at the University of Connecticut in August 2018 as an Assistant Professor. She received her PhD in Computational and Mathematical Engineering from Stanford University in January 2018, where she was a member of the Materials Computation and Theory Group advised by Dr. Evan Reed. She holds a B.A. in applied mathematics/computer science from Harvard College. Her research focuses on the development of machine learning and computational methods for the physical sciences, and her work has been published in journals such as Chemical Science, Energy & Environmental Science, among others, as well as in a book chapter.

Contact:

Kyu-Hwan Lee

Mathematics Colloquium (primary), UConn Master Calendar

Control Panel