Colloquium – Priya Vashishta

Priya Vashishta, University of Southern California, Los Angeles

Deep learning and quantum materials Dynamics

Machine learning has become powerful tool in computational sciences. Among diverse applications, molecular dynamics (MD) simulation based on neural network (NN) has been attracting great attentions. With the highly accurate energy landscape encoded by ab-initio molecular dynamics training dataset, our goal is to develop an efficient and robust neural network quantum molecular dynamics (NNQMD) framework to perform multimillion-to-billion atom and long-time nano seconds to micro second simulations that provide unprecedented access to physical and chemical processes and properties. I will discuss applications of Deep Learning and Reinforcement Learning for a variety of materials and also to ultra slow processes that are very difficult to simulate.

Research reported here is done in collaboration with Rajiv Kalia, Aiichiro Nakano, Ken-ichi Nomura and post-docs and graduate students in the Collaboratory for Advanced Computing and Simulations at USC.

Priya Vashishta is a professor of materials science, computer science, and physics and astronomy at USC, and director of the Collaboratory for Advanced Computing and Simulations. He has edited or co-edited 11 books and is the author or co-author of more than 390 papers on topics that include high-performance computing, computer-aided design of high-temperature materials, multimillion-atom simulations of materials and devices, the bio-nano interface, and the immersive and interactive visualization of billion-atom systems. Vashishta and his research team explore the effects of extreme conditions on everyday materials. This research has led to newer and better-designed products, ranging from components for computer chips to materials less susceptible to corrosion and oxidation.

March 6, 2025 @ 4:10pm Central in Stevenson 4327

Host: S. Pantelides