AI-Driven Large Language Models to Improve Energy Storage and Conversion Applications
Venkat Viswanathan utilizes Argonne’s Polaris supercomputer for 2025 DOE INCITE program
Two U-M faculty members have each been awarded a one-year grant from the Department of Energy’s (DOE) through the 2025 Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program. Congratulations to Venkat Viswanathan representing Aerospace Engineering and Sharon Glotzer representing Chemical Engineering on their respective achievements.
The DOE’s INCITE program seeks computationally intensive, large-scale research projects with the potential to significantly advance key areas in science and engineering. So far, Viswanathan’s team has built the largest molecular foundation model to date and pre-trained their model on 2 billion molecules.
In the 2025 cycle, Aerospace Engineering Professor Viswanathan’s research towards the award looks to develop AI-driven large language models by utilizing the Argonne National Laboratory’s Polaris, a 34-petaflop supercomputer. The goal is to train a foundation model capable of predicting thermodynamics, mechanical properties, thermal properties, enabling the efficient design of potential battery electrode materials. This work will help accelerate the discovery of new materials for energy storage and conversion applications.
Last year, Viswanathan received a 2024 INCITE award providing 200,000 node hours on Polaris at Argonne National Laboratory to build a foundational model for molecules. This previous work focused on small organic molecules, mainly composed of carbon, hydrogen, oxygen and nitrogen, with relevance to energy storage and conversion applications to better predict battery electrolytes.